There is muchto be gained from a study of the many various plans on which the share contract is based. {Alfred Marshall}

To My Father
In Grateful Memory


## Preface

Four years ago I accumulated material on property laws and agricultural data from Asia. My hope was to integrate the information with economic theory for my doctorate at the University of California, Los Angeles. I would have failed in this endeavor except that my advisers persuaded me to concentrate on only two sets of property right constraints defined by law in agriculture and on the associated leasing arrangements.

Even that limited task was difficult. The rental share restriction imposed by the Taiwan government led to output increases in tenant farms, an observation which was at first sight inconsistent with economic theory. Prolonged investigations were made of the reliability of the data and of an explanation other than the rental share restriction for the output increases. When this failed, I decided to use standard economic principles to formulate hypotheses relating to share tenancy (or sharecropping). Having completed the first draft of my thesis, however, I found myself interested in land tenure arrangements in general as much as in specific agrarian reforms. Further investigations were made and revisions followed. The outcome is this book.

For generations economists and land tenure writers have sought to rank the relative efficiency of resource use under different leasing arrangements. For example, share tenancy has long been considered inefficient, as have leases of relatively short duration. Indeed, tenancy in general has been regarded as less efficient than owner cultivation. Writers in development economics have frequently asserted that standard economic theory cannot be applied to conditions in underdeveloped countries. Exceptions have been alleged, "irrational" behavior has been discerned, and special theories such as "disguised unemployment" and the "dual economy" have been advanced. These analyses, however, have ignored the relevant property right constraints. My disagreement with their conclusions is only natural.

Every transaction involves a contract. The transactions conducted in the market place entail outright or partial transfers of property rights among individual contracting parties. The contractual arrangements through which these transfers are negotiated are several and varied. Instead of ranking the relative efficiency of these arrangements, I show that such ranking without reference to the property right constraints is futile. Furthermore, I seek to explain (1) why the patterns of agricultural output differ under two sets of property right constraints, and (2) why different contracts are chosen under the same property right constraints.

As a student at UCLA, I repeatedly audited the price theory lectures of Armen A. Alchian and Jack Hirshleifer. Their influence on my thinking has been profound. They and Robert Rooney guided me in my dissertation. Others who gave me valuable encouragement include Evsey Domar, Mason Gaffney, Judith Mann, John Pippenger and Michael Thomson. Eldon Dvorak, then a colleague at California State College at Long Beach, came to my aid almost daily. For the final elaborate revision, D. Gale Johnson and Theodore W. Schultz have allowed me to consume their human capital lavishly. Harry G. Johnson read and commented on the entire manuscript. Chapter 4 was written last, and received the additional benefit of advice from Ronald H. Coase, Harold Demsetz, John McManus, and George J. Stigler. My research assistant, Patricia Kuttner, bore most of the burden of proofreading and compiling the index.

Several organizations in Taiwan kindly provided me with valuable data, including the Sino-American Joint Commission on Rural Reconstruction, the Taiwan Provincial Land Bureau, the Department of Agriculture and Forestry, and the Bank of Taiwan.

The Economics Department and the Library of California State College at Long Beach made their research facilities available to me, as did the oriental libraries at Berkeley, Stanford, UCLA, and Chicago.

I am also indebted to three sources for financial aid: to the Lilly Endowment Incorporated grant supporting the study of property rights and behavior at UCLA, to the University of Chicago, for their Post-Doctoral Fellowship in Political Economy for the academic year 1967-68，and to the Ford Foundation grant for International Studies including Agricultural Economics at the University of Chicago.

Steven Ng-Sheong Cheung

Chicago,1968

## 1. Introduction

### A. The Scope

This study consists of two parts. The first part derives a theory of share tenancy with which to explore the nature of resource allocation under one of the main forms of land tenure in agriculture. Share tenancy is a land lease under which the rent paid by the tenant is a contracted percentage of the output yield per period of time. As a rule, the landowner provides land and the tenant provides labor; other inputs may be provided by either party. Share tenancy is thus share contracting, defined here as two or more individual parties combining privately owned resources for the production of certain mutually agreed outputs, the actual outputs to be shared according to certain mutually accepted percentages as returns to the contracting parties for their productive resources forsaken. The theory, to be derived from standard economic principles, may be generalized to all forms of land tenure under similar ownership of resources.

The prevailing impression is that share tenancy results in inefficient allocation of resources.[1] It will be shown here, both theoretically and empirically, that the inefficiency argument is illusory. The implied resource allocation under private property rights is the same whether the landowner cultivates the land himself, hires farm hands to do the tilling, leases his holdings on a fixed rent basis, or shares the actual yield with his tenant. In other words, different contractual arrangements do not imply different efficiencies of resource use as long as these arrangements are themselves aspects of private property rights. Implications of alternative theories will be tested against observations obtained primarily from Asian agriculture.

It remains to inquire further into the choice of contracts and various stipulations of land leases; and I shall attempt to show that the observed leasing arrangements are consistent with the theory of choice. The allocation of resources will differ, however, if property rights are attenuated or denied as private, or if the government overrules the market process of allocation. In the same theoretical context I shall argue, in Appendix A, that several hypotheses relating to disguised unemployment and the dual economy are erroneous. Their error is attributable to neglect of the flexibility of land use and failure to consider the pertinent property laws.

The second part of this study applies the theory of share tenancy to a situation in which the rental percentage is restricted to a legal maximum by the government. Two general hypotheses will be derived. One relates to compensating payments and tenure rearrangements, which are offsetting contractual rearrangements that may render the maximum rental percentage control immaterial. Another hypothesis, relating to resource reallocation, becomes significant only if compensating payments and tenure rearrangements are effectively prohibited by law. By the later hypothesis, the theory of share tenancy implies increased farming intensity in tenant farms if the rental share of the annual yield is legally reduced. Various implications for resource reallocation will be tested against observations. In particular, evidence will be presented to confirm that under effective rental share reduction the marginal product of land in tenant farms will be higher, and the marginal products of tenant inputs will be lower, than those of similar resources employed elsewhere.

Although similar rental share restrictions have been enacted in several Asian countries, Taiwan has been chosen for the second part of this study. Two advantages may be noted. First, during the first phase of the Taiwan land reform program, the rental share reduction was carried out before other reform measures were introduced.[2] Thus, there was a period of three years in which we can investigate the reallocation of resources as affected by the rental restriction independent of other factors. And second, in Taiwan both the provisions of the rental share reduction and their enforcement are restrictive enough to reveal some of the major implications derived from the theory of share tenancy.

[1]. Classical economists aside, contemporary theses on share tenancy which imply inefficient resource allocation include: Rainer Schickele, "Effect of Tenure Systems on Agricultural Efficiency", Journal of Farm Economics (February,1941); Earl Heady, "Economics of Farm Leasing Systems", Journal of Farm Economics (August,1947); Earl Heady and Earl Kehrberg, "Relationship of Crop-Share and Cash Leasing Systems to Farming Efficiency", Research Bulletin (Iowa State College Agricultural Experiment Station, May,1952); Charles Issawi, "Farm Output under Fixed Rents and Share Tenancy", Land Economics (February,1957); N. Georgescu-Roegen, "Economic Theory and Agrarian Economics", Oxford Economic Papers (February,1960); and Amartya K. Sen, "Peasants and Dualism with or without Surplus Labor", Journal of Political Economy (October,1960).

There are exceptions, however. For example, D. Gale Johnson, in his "Resource Allocation under Share Contracts", Journal of Political Economy (April,1950), is reluctant to accept the inefficiency argument in spite of his theoretical model which supports it. John Lossing Buck and James O. Bray argue against the inefficiency thesis on the grounds that observations of land use under share contracts fail to confirm the traditional belief. See J.L. Buck, Chinese Farm Economy (Chicago: University of Chicago Press,1930); and J.O. Bray, "Farm Tenancy and Productivity in Agriculture: The Case of the United States", Food Research Institute Studies (1963). We shall discuss this in chapter 3.

[2]. In the Philippines, for example, the rental share restriction was carried out simultaneously with a program of compulsory redistribution of land. See Frate Bull, "Philippine Land Reform,1950-1958", International Cooperation Administration, USOMI Philippines (Manila,1958).

### B. The Nature and Origin of Taiwan Land Reform

The definition of a land reform has been a controversial concept in the literature of land tenure. But the term "land reform" as used here has a simple meaning. It entails changes in laws pertaining to land ownership which affect wealth distribution or resource allocation.

In Taiwan, for example, the much publicized farm land reform did not spring up at one time: it was the product of a series of elaborate adjustments and enforcements of the laws already in existence. After the Japanese retrocession of Taiwan to China in 1945, the Chinese code of law pertaining to land rights came into force. This set of laws, the crucial articles of which will be discussed in the next section, has undergone rapid modification since April 1949. In the following four years, a total of over 350 articles were appended to define the Taiwan land reform.[1] Despite its complexity, we may generally divide the reform into three phases.

The first phase involved a rent limitation program, under which the rental percentage was reduced from an estimated mean of 56.8 percent to a uniform legal maximum of 37.5 percent of the annual crop yield. During this phase, two sets of provisions were established. The first set includes eighteen articles, promulgated and enforced in Taiwan from April 14, 1949.[2] The second set includes thirty-one articles, promulgated on June 7,1951, and intended to go into effect in February 1952.[3] While the provisions of both will be examined in later chapters, it is the economic implications of the first set upon which I shall concentrate, for I could find no evidence that the second set was actually enforced.

The second phase of the land reform pertains to the sale of public lands to establish owner-farmers.[4] This program, promulgated in June 1951 and carried out from July 1951 to June 1953, covered a portion of farm lands which were under state ownership and had been leased to tenants. The magnitude of farming resources affected by this program was very small.[5] The third phase is the most complex and is generally regarded as the main part of the land reform in Taiwan. It involved the promulgation and enforcement of the "Land-to-the-Tiller Act", beginning in January 1953. Under this program, landholdings of a private landowner in excess of three hectares were compulsorily purchased by the government. The compensation offered was two and one-half times the gross annual yield of the land being purchased, paid by the issuance of land bonds and industrial stocks of government enterprises. The purchased lands were resold to tenants in stipulated parcels at the same numerical prices, to be paid for in ten-year installments at no interest charge. After this compulsory redistribution of land, transfers of land rights among individuals were prohibited.[6] Neither the second nor the third phase of the Taiwan land reform will be covered in this study.

Despite the fact that the reform provisions are several and varied, two common policy intentions can be detected. First is the policy attacking tenancy either by intervening in the terms of the contract to "protect" the tenants or by abolishing tenancy altogether. A second intention is to force a redistribution of wealth in favor of the tenants. The lawmakers and reform officials in Taiwan had little difficulty justifying their reform measures. One claimed that the existing property system was simply out of date:

Land reform in Taiwan was carried out at just the right time. The time was opportune because by then the landlords had outlived their usefulness and landownership had become an obstacle to further development of agriculture as well as industry…… With the limited farm land and the irrational tenure system, rural underemployment presented a serious problem.[7]

A second justification offered is that farming under tenancy, and share tenancy in particular, is economically inefficient on three counts[8]. (1) Land-lease contracts were usually short-term or nonperpetual. This deprived tenants of security and therefore was inefficient. (2) The tenant's incentive to farm was discouraged by a share contract, since part of every unit of output produced would go to the landowner as rent. (3) And, since the rentals collected by landowners were exorbitantly high, the tenants lacked the ability to invest.

A third justification offered by reformers is that tenants were exploited by landowners. In addition to the high rent, a share contract, they claimed, was a device of exploitation. However, the evidence they cite to support the "exploitation" argument is perhaps contradictory to their claim of inefficient tenancy: for example, landowners frequently stipulated that their tenants provide seeds and fertilizers, and required that they work hard.[9]

One finds it difficult to accept these justifications as the origin of the land reform. There is a great distance between ideology and practice.[10] The making and enforcing of new rules are costly events. After all, the same land "problems" had been discerned in China long before 1949.[11] Why, then, did the reform in Taiwan take place only after the war? The striking similarities of reform regulations in Taiwan, Japan, Korea, the Philippines, and else-where, and the fact that all started between 1946 and 1950, may suggest that these reforms were prompted by the "influence" of the United States.[12] But, this foreign "influence" could at best explain the rapidity of land reforms adopted in Asia after the war. It is doubtful that the United States originated the specific reform policies. In the historical development of land tenure, we find that reform measures similar to the "Land-to-the-Tiller Act" had been introduced several times (for short durations) in China before 1800, and in several European countries during the nineteenth century. Why they took place poses a puzzle which I will not seek to solve in this study.

[1]. See Cheng Chen, Land Reform in Taiwan (Taipei: China Publishing Co.,1961), pp.181-292.

[2]. See Hui-sun Tang, Land Reform in Free China (Taipei: Sino-American Joint Commission on Rural Reconstruction,1954), pp.221-23. It is entitled "Regulations Governing the Lease of Private Farm Lands in Taiwan Province". In 1947, a similar rental share reduction had been attempted in a few provinces in mainland China.

[3]. See Chen, Land Reform in Taiwan, pp.191-97; and Tang, Land Reform in Free China, pp.224-28. It is entitled "Farm Rent Reduction Act". The rules of enforcement were established on February 2,1952. See Tang, Land Reform in Free China pp.229-31.

[4]. Chen, Land Reform in Taiwan, pp.199-212.

[5]. See Sino-American Joint Commission on Rural Reconstruction, "JCRR Annual Reports on Land Reform in the Republic of China", composite volume, mimeographed (1965), chap.3.

[6]. The most comprehensive record of the provisions governing the Land-to-the-Tiller program is seen in Chen, Land Reform in Taiwan, pp.202-92. For its complexity in legislative development and enforcement, see JCRR, "Annual Reports", chaps.4 and 5.

[7]. T.H. Lee, "Impact of Land Reform on Agricultural Development and Rural Employment in Taiwan", (Taipei: Sino-American Joint Commission on Rural Reconstruction mimeographed paper 63-RED-M-176), pp.1-2.

[8]. These points are found in: Yen-tien Chang, "Land Reform in Taiwan", mimeographed report no.1 (Taichung: Department of Agricultural Economics,1954); Cheng Chen, An Approach to China's Land Reform (Taipei: Cheng Chung Book Company,1951), chap.1; idem Land Reform in Free China (Taipei: Free China Review,1953), chap.1; idem Records of Taiwan Land Reform (Taipei: Chung Hwa Book Company,1961), chaps.1 and 2; Hsio Cheng, The Theory and Practice of Land Reform in China (Taipei: Chinese Research Institute of Land Economics,1953); JCRR, "Annual Reports", chap.1; Sidney Klein, The Pattern of Land Tenure Reform in East Asia after World War II (New York: Bookman Associates,1958), chaps.1 and 3; Lin-Fong Pun, New Ideas of China Land Reform (Taipei: Chung Wah Cultural Publishers Association,1957); Tang, Land Reform in Free China, chaps.1 and 2; H.S. Tang and S.C. Hsieh, "Land Reform and Agricultural Development in Taiwan", in Land Tenure, Industrialization and Social Stability: Experience and Prospects in Asia, ed.W. Froehlich (Milwaukee: Marquette University Press,1961), pp.114-42; and Young-Chi Tsui, "Land-Use Improvement: A Key to the Economic Development of Taiwan", Journal of Farm Economics (May,1962). No formal analysis is offered by any of them. It is interesting that tenure writers everywhere seem to share the same views.

[9]. See the preceding footnote.

[10]. And in spite of all claims, evidence has yet to be found that farmers demanded a change in the existing system of property rights. Peasant uprisings in modern Chinese history, to my knowledge, have occurred only in periods and locations in which (1) property rights were hardly enforced, (2) the taxation was allegedly high, or (3) inflation, by raising the relative price of land, effected a redistribution of income in favor of landowners.

[11]. See, for example, Tang-Yuen Chen, Systems of Land Tenure in China (Shanghai: Commercial Press,1932), chap.18; Ro Nagamu, A Study of Land System in China (Chinese translation, Shanghai: Kuo Kwong,1934); Han-Sheng Chen, Landlord and Peasant in China (Shanghai: Kelly and Walsh,1936); Chung Sen Department of Education, Papers on Land Rent Problems in China (Shanghai: Commercial Press,1937); Lin-Ho Kim, A Study of the Economics of Chinese Farming Villages (Shanghai: Chung Wah Book Store,1937); Richard H. Tawney, Land and Labour in China (London: G. Allen and Unwin,1937); Institute of Pacific Relations, Agrarian China, Selected Source Materials from Chinese Authors (Shanghai: Kelly and Walsh,1938); Quan-Chung Yu, Land Systems and Regulations in China (Szechwan: National Szechwan University,1944), vols.1 and 2; and the writings of Yet-Sen Sun.

[12]. For general discussions of these reforms, see Klein, Pattern of Land Tenure Reform; Froehlich, Land Tenure; Kenneth H. Parsons, "Land Reform in the Postwar Era", Land Economics (August,1957); Ervin J. Long, "The Economic Basis of Land Reform in Underdeveloped Countries", Land Economics (May,1961); Elias H. Tuma, "The Agrarian-Based Development Policy in Land Reform", Land Economics (August,1963); and Doreen Warriner, Land Reform and Economic Development (Cairo: National Bank of Egypt,1955). None of these discussions, however, provide adequate information on property law changes as bases of these land reforms. For the situation in Japan, which is perhaps the most complicated, see R.P. Dore, Land Reform in Japan (London: Oxford University Press,1959); Takekayu Ogura, ed., Agricultural Development in Modern Japan (Tokyo: Japan F AO Association,1963); Ministry of Agriculture and Forestry, Agricultural Land Reform Legislation (Tokyo,1949); Supreme Commander Allied Powers, Natural Resources Section Report no.79, Farm Tenancy in Japan (Tokyo,1947); and Natural Resources Section Report no.127, Japanese Land Reform Program (Tokyo,1950). For Korea, see C.C. Mitchell, "Land Reform in South Korea", Pacific Affairs, vol.22, no.2 (June,1949). For the Philippines, see Bull, "Philippine Land Reform". For Vietnam, see J.P. Gittinger. "Vietnamese Land Transfer Program", Land Economics (May,1957). For Iran, see B.H. Kristjanson, "The Agrarian-Based Development of Iran", Land Economics (February,1960); and V. Webster Johnson, "Agriculture in the Economic Development of Iran", Land Economics (November,1960). For Iraq, see Warren Adams, "Reflections of Recent Land Reform Experience in Iraq", Land Economics (May,1963); and M.H. Hashimi Rasool and Alfred L. Edwards, "Land Reform in Iraq: Economic and Social Implications", Land Economics (February,1961). Nowhere in the literature on Taiwan land reform, however, could I find an explicit statement that the United States had sponsored the reform. Elsewhere United States sponsorship is clear. See, for example, J.P. Gittinger, "United States Policy toward Agrarian Reform in Underdeveloped Nations", Land Economics (August,1961).

### C. The Prereform Land Law

Before proceeding to the formal theory of share tenancy, let us investigate here the nature of farm land ownership in Taiwan before 1949—the system upon which the rental share reduction was imposed in the first phase of the reform.[1] We shall concentrate on the laws governing private farm lands: a total of 672,000 hectares,56 percent of which were under tenancy.[2]

The term "private ownership" as applied to Chinese law implies exclusivity in using certain rights.[3] The primary ownership right attached to private farm lands was the right to farm, which meant "the utilization of land through the application of labor and capital" (Article 80).[4] Thus, the right to farm is the right to produce agricultural products as a source of income.[5] The spatial limit or dimension of the right was specified by the Civil Code, Arts.832 and 851, covering superficies. The time limit or the length of holding, according to the same articles, was in perpetuity.

Turning to the provisions governing the transfers of rights, which are important to the understanding of resource allocation under tenancy, we encounter a somewhat complex set of legal provisions which can be simplified as follows:

Outright Transfers

According to the prereform land law, land ownership was not freely alienable; that is, the owner could not transfer the ownership of his land to anyone whom he might choose. But in practice ownership transfer was almost unrestricted. It was stipulated that "the ownership of private farm land shall be transferred only to such transferees as can cultivate the land themselves after the transfer is effected (Art.30)".[6] But since a transferee who can cultivate may not necessarily do so himself, this stipulation in effect had little or no legal restriction on outright transfers. Another notable restriction was that, under tenancy, the tenant had preferential right to purchase. The tenant, however, was required to pay "the same terms as are offered to any other person".[7] This condition virtually nullified the tenant's preferential right to purchase, since to realize this right he must meet the highest offer made by anyone else. Indeed, the landowner (lessor) might create a fictitiously high price if he chose to sell discriminately. To conclude, restrictions on outright transfers were negligible.

Mortgage and Perpetual Lease

Land rights other than ownership as provided by the Chinese civil code include mortgage, yungtien, and dien-all essentially rights of transfer and not farming rights. The last two terms require explanation.

By yungtien is meant the right permanently to use another person's land for the purpose of cultivation or pasturing by paying a rent therefor [Civil Code, Art.842]; the yungtien-holder may transfer his right to another person [Civil Code, Art.843], but he may not sublease the land to another person [Civil Code, Art.845].

In short, yungtien is a conditional perpetual lease, and "if the rent which the yungtien-holder has failed to pay amounts to the equivalent of two years'total rent, the landowner may…… revoke the yungtien" (Civil Code, Art.846).

By dien is meant the right to take possession of another person's immovable property and to use it to enjoy incomes therefrom by paying a price for it [Civil Code, Art.911]. The dien-holder may transfer the right to dien to another person [Civil Code, Art.917], or lease the property to another person under dien [Civil Code, Art.915]. If the dien-obligor fails to redeem it at the original price within two years after the expiration of the contractual period of dien, the dien-holder shall acquire the ownership of the property under dien [Civil Code, Art.923].

Thus, dien is an arrangement for transferring rights resembling mortgaging and leasing.

Ordinary Lease

In yungtien and dien, we have transfers of ownership rights which may be regarded as special cases of land lease. Turning to the general case, by the lease of farm land is meant "the use of another person's agricultural land for the purpose of cultivation by oneself by paying a rent therefor" (Art.106). Under the pre-reform land law, the following articles are worth noting.

Art.108. The lessee shall not, even with the consent of the lessor, sublease the whole or part of the leased farm land to another person.

Art.109. Any farm lease contract for a definite period shall, unless the lessor takes back the land for his own cultivation on the expiration of the contractual period, be deemed to have been renewed for an indefinite period, if the lessee continues to cultivate the said land. Art.114. Any farm lease contract for an indefinite period may be terminated…… where the lessor takes back the farm land for his own cultivation.

Art.117. Where the farm land taken back by the lessor for his own cultivation is again offered for lease, the original lessee shall have preferential right to accept the lease.

The above articles set down three main points. First, subleasing was prohibited (Art.108). But this did not prevent other disguised forms of subleasing such as joint tenancy. Second, the lessor could terminate the lease contract for his own cultivation (Art.109 and 114). This in effect imposed no restriction on the length of lease other than the contracted length, since the term "owner cultivation" is vague enough to allow the hiring of farm hands to do the tilling. It was to this provision that "insecure" tenure was frequently attributed, providing a reason for the law to be condemned as out of date. And third, though preferential right to lease was allowed to the original lessee, there existed no restriction on the lessor's right to revise the original terms upon renewal of the contract. A provision for rent control is seen in Art.110 of the prereform land law, wherein "farm rent shall not exceed 8 percent of the value of the land". According to the lawmaker, however, "due to circumstances this had never been enforced".[8] This has given rise to the claim that tenants were totally unprotected and exploited by landowners.

Transferability and Delineation of Investment Rights

Investments in land and other assets attached to land owned by the lessor were protected by the Chinese civil code.

Civil Code, Art.432. The lessee shall keep and preserve the thing [property] leased with the solicitude of a good administrator. If the thing leased has productivity, he shall also preserve its productivity.

If the lessee acts contrary to the provisions of the preceding paragraph and thereby causes loss or damage to the thing leased, he shall be liable to pay compensation therefor. But he shall not be held responsible for any damage or change caused to the thing leased through its use, or through the enjoyment of the income accruing therefrom, in such ways as are agreed upon or as are in consonance with the inherent nature of the thing itself.

Civil Code, Art.462. Where agricultural implements, livestock, or other accessories are leased together with the farm land, an inventory of the same showing their respective values shall be made…… If any accessory mentioned in the inventory be destroyed or lost owing to circumstances for which the lessee must be held responsible, he shall be liable to make it good.

Civil Code, Art.445. For the settlement of claims arising out of the lease contract, the lessor of an immovable property shall have the right of retention over the movables which belong to the lessee and are fixed to the said immovable property. But this provision shall not apply to movables that are not subject to attachment.

This last Article is thus conditional to Art.118 of Law.

The lessor shall not exercise the right of retention provided for in Art.445 of the Civil Code over such farm implements, livestock, fertilizers, and farm products of the lessee as are necessary for cultivation.

Investments in land by the lessee were protected by two Articles in Law, which read:

Art.119. By special improvements on farm land shall be meant improvements resulting from the increased application of labor and capital which, besides preserving the original qualities and utility of the land, increases its productivity or facilitates its cultivation.

The lessee may freely make the special improvement referred to in the preceding paragraph, but shall notify the lessor of the amount of the outlay incurred thereon.

Art.120. When any farm land is returned to the lessor on the termination of the lease contract…… the lessee may claim from the lessor repayment of the outlay incurred on the special improvements referred to in paragraph two of the preceding Article, but such repayment shall be limited to the cost of that part of the special improvements which has not lost its utility.

In the above articles the ownership of productive entities other than land was clearly delineated. It should be noted that, according to Art.120 above, no capital gain was explicitly allowed to the lessee for the "special improvements" made by the lessee. He might be compensated, of course, by adjusting the rental terms in the lease contract.

We may now briefly summarize the attributes of private farm land ownership in Taiwan before the reform. Although mineral rights which might be attached to land were denied as private, the right to farm was defined as exclusive. Investments might be made by the lessor or the lessee, and in either case the ownership of resources was clearly delineated. Transferability of land tenure rights was provided in the form of outright transfers, yungtien, dien, mortgage, and land lease. Under land lease, compensating payments for farming investments made by either party were not restricted. We may, therefore, designate the farming resources, affected by the first phase of Taiwan land reform, as resources under a system of private property rights.[9]

[1]. The land law effective in Taiwan in the postwar period but before the land reform was the same as that promulgated by the Chinese National Government in June 30,1930, enforced March 1,1936, and amended April 29,1946. However, the Chinese land law governing private farm lands had been basically the same long before 1930. The main revisions in 1930 and 1946 involved the addition of discriminatory provisions against aliens and precautionary articles for future national interest. Thus, there was no substantial transformation in the system of property under the National Government. All articles cited hereafter are in Chen, Land Reform in Taiwan, pp.133-90. The Chinese version of these articles is in China Land Bureau, Rules and Regulations of Land Administration (China: National Government,1947).

[2]. JCRR, "Annual Reports", p.8.

[3]. We may regard a right as a limited authority to make decisions about resources; that decision authority usually has a positive present value.

[4]. To avoid ambiguities, we may simply regard "labor" and "capital" as physical input entities.

[5]. Certain other uses or rights in land were being denied. For example, "minerals attached to any land shall not become private property, even if private ownership of the said land has been duly acquired" (Art.15). See also Art.12 and 13.

[6]. Transfers to aliens were restricted (Arts.17-24).

[7]. See Arts.33 and 107.

[8]. Chen, Records of Taiwan Land Reform, pp.17-18.

[9]. See Ronald H. Coase, "The Problem of Social Cost", The Journal of Law and Economics (October,1960); and Armen A. Alchian, "Some Economics of Property Rights", mimeographed (Santa Monica: The Rand Corporation,1963).

## 2. The Theory of Share Tenancy

In this chapter a theory of resource allocation under share contracts is derived. The analysis is based on the premise of wealth maximization subject to the constraints of private property rights in a free market, that is, one without rental restriction. Under exclusive and transferable resource rights, each contracting party is free to accept or reject the negotiated terms of a share contract. Unless otherwise indicated, a zero cost of contracting is assumed.[1]

### A. The Solution Defined

In figure 1, the abscissa of the vertical supply curve,-5, indicates the total area of land belonging to a landlord. Let h denote the land area held by a tenant, and q be the product. The marginal product of land, , diminishes as h increases, holding the farming inputs of one tenant (or one tenant family) constant. Suppose the rent charged by the landowner is 60 percent of the annual yield, that is, r=0.6. A marginal contract rent curve, r, is at 60 percent of . The vertical distance between and r is the marginal tenant income, (1-r), defined as the change in tenant income with respect to a change in land area used by the tenant. The shaded area between and r represents the total farming income received by the tenant, and the area below r represents the total rent collected by the landowner. If the tenant's income is as high as or higher than his alternative earning, he will stay in farming and will use all the land available to him on this farm as long as the marginal productivity of land is greater than zero, with all farming inputs other than land held constant. To maximize wealth, the landowner will raise the rental percentage-thus the r curve-until the tenant's income from farming equals his alternative earning.

However, the rental percentage is not the only variable which the landowner can adjust to maximize his wealth. The landlord will not allow one tenant to cultivate all the land he owns if parceling his land to several tenants will result in a higher total rent. This is illustrated in figure 2. In this figure, vertical lines T1, T2, T3…… are dividing lines of plots of land to the first, second, and third tenant respectively. As the number of tenants cultivating the available land increases, the marginal product of land shifts upward relative to the situation where there is only one tenant.

Assume for the moment that the rental percentage is the same for all tenants. The curves , …… are the marginal productivities of land for each tenant, and r, r,…… are the respective marginal contract rent curves for each tenant.[2] The income for each tenant is represented by the area between the respective and r curves for the tenant.

To maximize his wealth, the landowner will maximize the difference between the integral of the marginal productivities of land and the integral of the tenants'incomes, which is represented by the shaded area in figure 2. That is, the landowner will make the integral of marginal contract rents a maximum. This implies that the income of each tenant will not be higher than his alternative earning.

As the area of land assigned to each tenant becomes smaller, however, the rental percentage the landlord charges must be lower for the tenant to obtain his alternative earning, which calls for a decrease in marginal contract rent, r, in order to prevent the tenant from giving up the lease. This decrease in r will lead to a lower rent received from each tenant, and if the land size per tenant continues to decrease, the rental percentage will eventually become so low that the total rent from the land will decline. The solution is thus well defined: Maximization of the landowner's wealth, given his total landholdings and the cost of tenant inputs, requires the simultaneous determination of the land size per tenant and the rental percentage. In other words, with private property rights over land and tenant input resources, the terms in a share contract mutually agreed upon by the landowner and the tenant will include the rental percentage and the ratio of nonland to land input that are consistent with equilibrium.

[1]. Insofar as there is more than one individual who wants the same property resource, competition is implied, and the number of competitors includes not only those who are actually using the resource but also potential owners (users). The assumption of zero contracting cost substitutes for the sometimes dubious assumption of "pure" competition. Included in the general term "contracting cost" are the costs of negotiating and the costs of enforcing the stipulations of the contract. I shall discuss these and other problems of transaction cost in chapter 4.

[2]. Depending on the production function, these pairs of curves may not be identical. We shall discuss this later.

### B. The Solution in Mathematics

For simplicity of presentation, assume there are two homogeneous factors of production, h and t, where h is the amount of land per tenant farm and t is the amount of tenant labor per farm. Further assume that the production functions of each tenant farm are identical. With these assumptions, it follows that the land size, h, and the rental percentage, r, contracted for each tenant farm will be the same in equilibrium.

Let each tenant farm's production function be q=q (h, t) The amount of land per farm, h, is equal to the total landholding of the landowner, H, divided by the number of farms, m; that is, The landowner's total rent, R, is then equal to the number of farms times the rent per farm; that is, R=m·r·q (h, t) Under competition, Wt= (1-r) q (h, t) where W is the market wage rate of the tenant labor, t.

The problem of the landowner is then to maximize R, through the choice of m, r, and t, subject to the constraint of competition; [1] that is, max.R=m·r·q (h, t) {m, r, t} subject to Wt= (1-r) q{h, t) Forming the Lagrangian expression, the problem is thus the maximization of The necessary conditions are.

From equation (2) above, we have λ=m And noting that , equation (1) becomes that is, This indicates that rent per acre of land equals the marginal product of land in equilibrium, a condition identical to that of a fixed-rent contract.

From equation (3), we have or the marginal product of tenant labor equals the wage rate, a condition identical to that of a wage contract. Finally, solving equations (1) and (4) for r, That is, in equilibrium, the rental percentage must simultaneously satisfy the last two terms. In other words, in equilibrium, the elasticity of output with respect to land, , equals the total yield net of tenant cost (rent) as a portion of the total product,

[1]. Note that t and m need not be treated separately. Given t, an adjustment of m yields the same result as adjusting t while holding m constant. They are separated here for the purpose of conveniently deriving all the conditions in equilibrium.

### C. The Geometric Solution and Further Exposition

The results derived in the preceding section can be demonstrated geometrically. In figure 3 we have the same dimensions as in figure 2. But in figure 3 we concentrate on only one of the tenants, which means that the total land space owned by the landlord may not be exhausted. The curve q/h represents the average product of land with one tenant family employed; that is, the average product with respect to the land size while holding all other farming inputs (of one tenant family) constant. The curve f/h, or the fixed total tenant farming cost divided by land area, reveals the cost of farming inputs (other than land) which yield the expected q/h. Assume for the moment that all nonland farming inputs are borne by the tenant; the f/h curve is the total cost other than land divided by the respective land space. It includes the cost of labor, seeds, fertilizers, and farming equipment associated with the period of time of the relevant production run.[1]

That is, f/h= (pt·t+pz·z+……)/h; where f is the constant total cost other than land, and pt, pz,···are the factor prices of tenant labor, t; fertilizers, z; …… Since we hold the farming inputs constant, the f/h curve is a rectangular hyperbola. The vertical difference between q/h and f/h defines (q-f)/h, the rent per unit of land, taking into account the alternative cost of the tenant.[2]

The total amount of tenant inputs which define f/h are contractually stipulated, which is essential because the tenant would commit less if only the rental percentage were prescribed. Given any rental percentage, only a portion of every unit of output produced will go to the tenant. If farming decisions were made entirely by the tenant, it would be to his interest that the cost of incremental tenant input be less than the associated marginal product. This would result in a condition inconsistent with equilibrium. A fuller discussion of this point will be presented in the next chapter.

Under mutual agreements, economic theory implies that the total amount of tenant inputs contracted will be the amount which yields the highest (q-f)/h, or which yields the highest rent per unit of land. Since in figure 3 the marginal farming cost always equals zero, the highest value of (q-f)/h can be derived as follows. For every upward shift of f/h as a result of increasing the stipulated amount of tenant input, there is associated an upward shift of q/h. The former represents the marginal nonland cost, which increases at a constant rate (i.e., constant marginal factor cost under pure competition); the latter represents the marginal productivity of additional tenant (i.e., nonland) input, which increases at a decreasing rate (i.e., decreasing marginal product of tenant input).[3] The highest (q-f)/h-associated with a specific f/h curve-is obtained when the marginal upward shifts of f/h and q/h are equal, or when the marginal product of tenant input equals the marginal tenant cost. The associated nonland farming cost, which defines f/h, covers the level of tenant inputs consistent with productive equilibrium. To maximize wealth, the crops chosen to be planted or rotated, and the method of producing them, are, by definition, those which yield the highest present value of land to the landowner, that is, which maximize rental annuity.[4] The relevant value of (q-f)/h, or average rent, for decision making is thus the highest one derived from alternative pairs of q/h and f/h. To state it more precisely, the highest value of (q-f)/h defines the cost of land per acre as a factor of production.[5]

The marginal product of land, —or in figure 2—cuts q/h and (q-f)/h at their highest points. The equilibrium land size assigned to this tenant, T1, is where (q-f)/h is at a maximum. Maximizing rent per acre of land will maximize the rental annuity from the landlord's total landholdings. With the equilibrium tenant land size determined (T1), the equilibrium rental percentage equals (q-f)/h divided by q/h (at T1). That is, the rental percentage, r, equals ar/ap as labeled in figure 3. Given this equilibrium percentage, say 70 percent, we plot the marginal contract rent curve, r, as this percentage of at every point.

Since the tenant contracts to pay a percentage of the total product, as indicated by r, the cost of land is no longer a constraint with respect to the amount of land the tenant will use. To maximize his income, the tenant prefers to employ land to the point where is zero (while holding his farming inputs constant as contractually stipulated). The landowner, on the other hand, will limit the tenant's holding to T1, and parcel the remainder of his total holding to other tenants under similar contractual arrangements.[6] The landowner cannot successfully restrict land size to less than T1, since with rental percentage r the tenant's alternative earning will be higher elsewhere and he will give up the tenancy.

Several additional remarks can be made. First, not all tenant families are equally productive. Some will be able to produce more because of their endowments of certain "specific" factors of production, for example, varying amounts of knowledge. In competitive equilibrium, the more productive tenants will be the intramarginal tenants, and the f/h curve thus defined will include the imputed rent for the tenant. The landowners cannot success-fully discriminate among tenants of varying efficiencies even if there are no costs associated with discrimination, because landlords employing marginal tenants (least productive tenants) will bid the more productive tenants away from a discriminating landlord.

Second, even if land is homogeneous, the land size per tenant farm may not be the same. Defining homogeneous land as land physically identical and having the same rental value per unit-that is, the (q-f)/h curves in different tenant farms yield equal heights at their peaks-the land sizes for different tenants will differ if the production functions of these tenants differ.[7] This also implies that the rental percentage for different tenant farms may not be the same. That is, in maximizing the rental annuity of his landholdings, the owner will assign different land sizes and charge different rental percentages to different tenants if their production functions require different intensities of tenant inputs. In equilibrium, given homogeneous lands, the marginal productivity of land for each farm must be equal at every margin, for it is at the highest value of (q-f)/h that land sizes are divided.

Third, the farming cost (other than the cost of land) may be shared by the tenant and the landowner jointly. In this case, the f/h curve represents the combined cost. Given q/h and f/h as in figure 3, the f/h minus the landowner's input cost will be lower, thus defining a higher $(q-df)^∗ /h$ curve. This higher $(q-f)^∗ /h$ curve (not drawn) measures not only the cost of land (rent), but also the landowner's nonland farming cost.[8] The rental percentage charged by the landowner will accordingly be higher, with an upward shift of r by the same percentage at every point. The implication of this is important: it does not matter whether the landowner required the tenant to invest more in land and charges a lower rental percentage or whether the landowner invests in land himself and charges the tenant a higher rental percentage; the investment will be made if it leads to a higher rental annuity.[9] It follows that for any contract the tenant does not have to possess the required amount of input. If what he has is insufficient, the tenant may raise the farming inputs by cooperating with the landowner, by subleasing, by hiring farm hands, by borrowing, or by joint tenancy with another family. Moreover, the different tenant input requirements in farms of different land grades and production functions will match tenants with different input endowments accordingly.

Let us now review the solution conditions. The highest value of rent per unit of land, (q-f)/h, is derived when the marginal tenant cost equals the associated marginal product of tenant input. As illustrated in figure 3, given the highest value of (q-f)/h, the land size per tenant (T1) is determined, and the corresponding rental percentage, r, will be ar/ap. At the dividing line T1, the rent per acre of land, that is, ar, (q-f)/h or rq/h, equals the marginal product of land, . At T1 also, we have the unique condition for the rental percentage This is identical with our mathematical solution, where we have since in our mathematics we assume only one tenant input, f=Wt.

[1]. The important distinction here is between land and inputs other than land. The substitution between labor and farming equipment, for example, is not yet important in this theory and is ignored.

[2]. For simplicity of presentation, all values are measured in real terms. Note, however, that the annual yield, q, may include a variety of crops; given their relative market prices, the value of each may be expressed in terms of one crop, say rice. And payments for tenant inputs and land will likewise be in rice.

[3]. To avoid semantic problems, we assume here the increase of one tenant input only. If several inputs are increased, as is usually the case, the relevant "shifts" to consider are when q/h increases at a decreasing rate. In either case, the theoretical outcome is the same.

[4]. Given the appropriate discount rate, , and the future income stream from the land, Y1, Y2……, Yn, Yi not necessarily equal to Yj, there exists a , called the rental annuity, such that If the private land right is perpetual, we have , where V0 is the present value of land.

[5]. The notion that rent is a cost of production began, perhaps, with J.S. Mill. And I maintain here that it is the highest rent per unit of land which really counts, for it is from this that the land value is derived.

[6]. It remains to show how the last plot of land belonging to the same landlord is allocated, since it may be too small for one tenant family. One answer is that a tenant may be renting land from two or more landlords at the same time, employing land margins from both. Likewise, a tenant may own a small area of land himself, renting an extra margin from a landlord. This is evident in the existence of part-owner farmers. Therefore, if the land size of the last plot assigned to a tenant family is smaller than the equilibrium size, the tenant will employ part of his resources elsewhere, possibly outside the farming sector.

[7]. The term "land homogeneity" employed here, which presupposes identical market location, renders the unambiguous condition: Given the same maximum values of (q-f)/h and the same ratios of tenant inputs to land in different farms, the rental percentages will be the same. Different production functions associated with different crops will require different intensities of tenant inputs. On the other hand, given the production function, the rental percentage and the land size per tenant will depend on (1) the fertility of land and (2) the cost of tenant inputs.

[8]. Note that the peak of this higher $(q-f)^∗ /h$ curve will be to the left of T1. This is not relevant to land size division, for it will lead to a lower rent per acre. The dividing line is still T1, where the (q-f)/h, or the net return to land, is at a maximum. But the rental percentage in this case will be $(q-f)^∗ /h$ at T1 divided by q/h.

[9]. The fact that most leases under share tenancy are not perpetual has been attacked as insecure tenure which leads to misallocation of resources. But it is no more insecure than the position of hired farm hands who are subject to dismissal in case of poor performance. Indeed, a temporary lease is an effective device to insure efficient allocation. The argument that nonperpetual leases tend to discourage investment is incorrect. We find the period of lease varies according to the type of contract and the asset holdings of the contracting parties, and the frequency of tenant dismissal has been small. For a fuller discussion, see chapter 4, section C.

### D. Comparison with Fixed Rent and the Market Criteria of Share Contracting

A number of questions, both theoretical and empirical, related to share tenancy deserve further inquiry, and I shall try to press them in the remaining chapters of this book. In this section, however, answers to two questions are sought.

Let us begin with the question: What are the differences between fixed-rent and share contracts? Under private ownership of resources, the basic difference between fixed rent and share tenancy lies in how the chosen labor-land ratio (or the ratio of nonland inputs to land) is expressed. With fixed rent, given the rent per acre, the tenant states how much land he will employ, and he alone decides-subject to the constraint of competition as with share tenancy-the amount of nonland inputs to be committed for every production run. Under share tenancy, however, the landowner and the tenant mutually decide the intensity of nonland to land inputs. In either case, the wealth maximizing value determines the land size per farm as well as other inputs employed. Since the set of constraints for decision making is the same for the two types of contracts, the same resource use is implied.

To illustrate, return momentarily to figure 3. In a fixed-rent contract, the landowner will be able to charge a rent equal to the maximum of (q-f)/h per acre, or ar, and the area rented by the tenant will again be T1, where the rent per acre equals the marginal product of land. Given q/h and f/h as shown, if an area larger than T1 is rented, the rental price per acre will be higher than the marginal product of land. Conversely, if an area smaller than T1 is rented, the tenant's total income from farming will be less than his alternative earning. Thus, under fixed rent, the land size per farm and the rental annuity received by the landowner will be exactly the same as under share tenancy.[1] Furthermore, if the same crops are to be planted on the same grade of land, having the same production function, the tenant under fixed rent will have to commit the same f/h to survive competition. As in sharecropping, if one tenant does not possess the necessary amount of inputs, he may acquire them through various means or rent a farm which requires less farming inputs.

Let us also press the question: Under share tenancy, what criteria will the contracting parties use in arriving at the equilibrium land size and rental percentage for each tenant? Clearly, the most efficient type of crops to be planted on the given land must be chosen, and the tenant inputs required must be considered-all ex ante decisions with some degree of uncertainty. Surely, erroneous decisions may occur owing to wrong expectations by either party to the contract. This will result in the tenant land size or the rental percentage differing from wealth maximization. It is equally clear, however, that some simple rules exist in the market which the contracting parties will follow. Given land of a particular grade, there exist, at prevailing market prices of land and other factors of production, certain combinations of tenant input requirements, rental percentages, and tenant land sizes.

Indeed, given freely alienable (marketable) land rights under private ownership, a landowner does not have to know the details of farming himself. Competition for resource ownerships will induce efficient contracts. If a less valuable crop is planted by the tenant, if the rental percentage is too low, the land size per tenant too large, or the tenant inputs required too little, the rental annuity as a return to land for the landowner will be at a rate lower than the rate of interest. In this situation, the landowner will either make the proper adjustments, lease the land to a different tenant, choose a different contractual arrangement, or sell his landownership outright. On the other hand, if the contractual agreements are such that the tenant is receiving a share lower than his alternative earning, other landowners will bid his services away, or the tenant may turn to wage labor.

As with competition among landowners, competition among tenants will insure that the contracted amount of tenant inputs is committed. Indeed, by checking the output alone the landowner will know whether the contracted terms are respected by the tenant and will decide whether the share contract should be continued. In practice, however, the transactions are frequently handled by hired agents who specialize in approximating the true yield. And in spite of numerous claims by reformers and novelists that tenants could make no decision and therefore were exploited, it is refreshing to point out that, in China at least, share contracting was rather polite.

As soon as the grain is threshed the tenant invites the landlord to a feast, after which the grain is divided and the tenant delivers the landlord's share to his home. After it is properly delivered the landlord gives the tenant a feast. If the landlord wishes to dismiss the tenant he does so at this time. If the tenant wishes to leave, he takes the sails from the windmill just before the landlord comes to the tenant's feast as a sign to the landlord of his intentions.[2]

Our analysis thus far has primarily been confined to a condition of zero cost of contracting. This assumption is dropped in chapter 4, in which we discuss problems of transaction cost and the characteristics of lease stipulations. But let us first trace the development of the theory of share tenancy, and perform tests of implications for alternative theories.

[1]. For still another exposition on this, see chapter 3, section C.

[2]. John Lossing Buck. Chinese Farm Economy (Chicago: University of Chicago Press,1930), pp.148-49.

## 3. Traditional Views of Sharecropping and Tests of Alternative Hypotheses

Perhaps the theoretical results reached in the preceding chapter would have been obtained by earlier writers had it not been for some illusory views fostered by tradition. In particular, there is the analogy frequently drawn between share rent and an excise tax. Since under sharecropping a portion of every output unit produced is taken as rent, it yields the impression of being similar to an ad valorem excise tax-where part of every unit produced is "taxed" by the landowner (government). The distribution of output is not the same, it is believed, as with fixed rent or owner cultivation-where the tiller obtains the entire incremental product. Share tenancy, therefore, is said to result in less intensive (and less efficient) farming because the tenant's incentive to work or invest in land is reduced.[1]

It is not difficult to show that the application of the analysis of a tax to share tenancy (hereafter called the "tax-equivalent" approach) is erroneous. The hypotheses stemming therefrom fail on several counts. In the tax-equivalent approach, the writers generally fail to realize that the percentage shares and area rented under share tenancy are not mysteriously "fixed" but are competitively determined in the market. Furthermore, these writers fail to specify the nature of the system of land ownership upon which their hypotheses are constructed. Let me clarify.

First, though some classical economists discussed the division of land, they did not analyze it in a general equilibrium framework (see section A of this chapter). Since Marshall, the possibility of the landlord's allocating his total holdings to several tenants has been ignored (see section B of this chapter). While this is valid in analyzing an excise tax, under share tenancy the cost of land and the distribution of land are thereby neglected. Second, the percentage share has usually been taken as given. Under share tenancy, however, the rental percentage is a discretionary variable. Third, with a tax, the government is not contracting to maximize wealth. In other words, the tax-equivalent analysis fails to offer any explicit treatment of the terms in a share contract which the participating parties must mutually agree to abide by when the contract is formed.

We cannot analyze the way a person uses resources without first specifying the nature of his property rights. It is true that once the land size and the rental percentage are under contract to a tenant he prefers to work or invest less in land than if he cultivates his own land. But under private ownership of land, the landlord's incentive to maximize his wealth is not reduced. We may well repeat a conclusion reached in the last chapter: It does not matter whether the landowner stipulates that the tenant is to invest more in land and charges a lower rental percentage or whether the landowner invests in land himself and charges the tenant a higher rental percentage; the investment will be made if it leads to a higher rental annuity.

Yet, it would be misleading to say that all earlier analysts of sharecropping were deceived by the tax-equivalent approach and invariably arrived at the conclusion of inefficient resource use. Some did so, some expressed doubt, and some seem to have abandoned the tax approach altogether. Indeed, a survey of the literature on the subject reveals that at times even their errors are most interesting and their insights are sometimes most profound.[2]

### A. The Classical View

Noting that sharecroppers "have been so long in disuse in England that at present I know no English name for them", Adam Smith wrote of the metayers (sharecroppers) in France, which he believed were successors of "the slave cultivators of ancient times".[3] Of the productive nature of the metayage system, Smith wrote.

It could never, however, be the interest [of the metayers] to lay out, in the further improvement of the land, any part of the little stock which they might have saved from their own share of the produce, because the lord, who laid out nothing, was to get one-half of whatever it produced. The tithe, which is but a tenth of the produce, is found to be a very great hindrance to improvement. A tax, therefore, which amounted to one-half, must have been an effectual bar to it.[4]

Although the analogy to a tax which Smith drew might have led succeeding writers astray, the context in which he placed the discussion of the metayers is also significant. Smith did not focus on the metayage system itself; rather, in one full chapter he attempted to trace the development of land tenure arrangements with economic interpretations.[5]

According to Smith's view, the "slave" cultivators preceding the metayers were even less productive, because "a person who can acquire no property, can have no other interest but to eat as much, and to labor as little as possible".[6] Thus for more productive land use the metayers succeeded the "slaves". Since in his view the metayage system was also defective, Smith claimed that "by very slow degrees," the metayers were succeeded by "farmers……, who cultivated the land with their own stock, paying a rent certain to the landlord".[7] Although Smith favored fixed-rent contracts (farmers) over sharecropping, he was nonetheless concerned with the "insecurity" of the farmers because of expiration of the lease: "The possession even of such farmers, however, was long extremely precarious, and still is so in many parts of Europe".[8] He advocated "the law which secures the longest leases against successors of every kind", but such a law was, to his knowledge, "peculiar to Great Britain".[9] In Smith's view, therefore, the British leasing arrangement - a freehold with a fixed rent and a lease for life - was more highly developed than those in other parts of Europe.[10]

Although the meaning of economic efficiency was not clarified until much later, Smith's idea of analyzing the development of land tenure systems on grounds of more gainful resource use is certainly an important one; however, the approach he used is not deep enough to yield fruitful results. Once property laws define a specific set of constraints on competition, there may exist several forms of contractual arrangements which imply the same resource use. (Why different contracts are chosen will be discussed in the next chapter.) When these property laws are altered, the contractual arrangements may change. It follows that the appropriate approach in analyzing land tenure development is to trace the alterations in property laws; and not, as Smith did, to interpret (or advocate) the change in laws by tracing what might appear to be defective leasing arrangements.

Thus, Smith is in error. While his claim that "slave" cultivation is grossly wasteful may or may not be true, his view that, historically, for economic reasons sharecroppers have been gradually replaced by fixed-rent farmers, is wrong. One need only point out that share tenancy has not been replaced by fixed rent, and that in the United States similar share contracts predominate among leases of retail stores, beauty salons, gasoline stations, amusement park rentals, and even the much regulated oil and fishery industries. Indeed, the rarity of sharecropping in England as observed by Smith and later by Mill and Marshall might very well be the result of the freehold, under which a lease for life was enforced by law. Under a perpetual lease, the cost of enforcing a share contract may be so high as to make it undesirable, since tenancy dismissal is one effective device to insure against poor performance by sharecroppers.

It is, of course, difficult to evaluate Smith's influence over later writers on share tenancy. The tax-equivalent argument aside, what appears to have permeated the minds of subsequent English writers is the conviction that the British (fixed rent) system was more advanced and efficient than rental arrangements elsewhere. This conviction was reinforced by the famous Travels of Arthur Young.

Young was the secretary to the Honorable Board of Agriculture and Fellow of the Royal Society. Esteemed as an agricultural expert in England, he condemned the metayers almost every time they were mentioned in his Travels in France during the Years 1787,1788, and 1789.[11] Of the metayage system, Young wrote:

There is not one word to be said in favor of the practice, and a thousand arguments that might be used against it…… In this most miserable of all the modes of letting land, the defrauded landlord receives a contemptible rent; the farmer is in the lowest state of poverty; the land is miserably cultivated; and the nation suffers as severely as the parties themselves…… Wherever this system prevails, it may be taken for granted that a useless and miserable population is found.[12]

One hundred years later in 1892, however, a very different edition of Young's Travels appeared. The editor, Miss Betham-Edwards, author and officer of public information of France, took liberty to delete most of Young's condemnations of the metayers.[13] And to the only remaining statement that I could find-in which Young claimed that the metayage system "perpetuates poverty and excludes instruction" — Betham-Edwards added a footnote: "Complex as such an arrangement may appear at first sight, metayage must be counted as a factor of great importance in the agricultural prosperity of France".[14]

Betham-Edwards is not the only editor who challenged Young's judgment. Constantia Maxwell, who edited the Travels in 1929, made numerous corrections on Young's views in the lengthy "Editor's Notes".[15] Maxwell pointed out, with the support of many sources, that at Young's time in France there were government regulations on vine growing, heavy taxes, the aftermath of the wars of Louis XIV, and various political disturbances on the eve of the French Revolution. Certainly, Young was not ignorant of all this, and — even if we accept his view that French agriculture was "miserable" — it is difficult to understand why he blamed the metayers as the sole source of trouble.[16]

Young's condemnation of metayage notwithstanding, we find in his work one piece of evidence which seemingly is consistent with inefficient land use under sharecropping; namely the low rent of land in France as compared with England.[17] According to the tax approach, nonland inputs committed to land under share rent are less than under fixed rent, and thus given the same area of land the rental payment to the landlord will be lower. According to standard economic theory, ceteris paribus, rent will be lower if (a) the land is less fertile (which Young discussed ambiguously), or (b) the cost of tenant inputs (or the wage rate) is higher (which Young would deny since the metayers were "miserably poor"). But other things were, in fact, not equal. In addition to the political instability and regulations on farming at that time, which might well have discouraged investment in land and thus led to lower rents, a more significant factor, perhaps, is the reportedly heavy taxes imposed on the metayers.[18] Given the metayers' alternative earnings, however trivial, a higher tax imposed on the occupation will require that the landowners charge a lower rental percentage in order to keep the metayers at work. And this implies a lower rent per acre of land.

Whereas Young might have allowed his emotion to run away with his judgment, some of his observations could have hinted to later writers that fixed and share rents yield the same intensity of nonland inputs should the constraints of competition be equal. In particular, Young observed that the share percentages varied from place to place, and that the division of farm size was related to population pressure.[19] Yet to my knowledge the only subsequent economists who elaborated further on the division of farms under share tenancy are Richard Jones and John Stuart Mill.

Writing in 1831, Jones duplicated not only Smith's thesis of the development of leasing arrangements, but also Smith's conclusions.[20] Jones, however, elaborated on the adjustment of labor input through land size divisions. With more information at hand, and acknowledging Young's observations, Jones wrote:

While the metayer tenant pays nominally the same [rental percentage], his own share of the produce may be diminished in two modes: by his being subjected to a greater quantity of the public burthens: or by the size of his metairie being reduced. By this second mode of reduction, I am not aware that the French metayer suffered much……[21]

And he continued further when he came to the metayers in Italy:

Metayers are always found ready to accept a subdivision [of land]…… Their multiplication, as we have seen in the case of France, usually goes on till they are stopped by the smallness of their maintenance, or, as more often happens, by the policy of the proprietors refusing to subdivide lands, already supplied with labor beyond the point they deem most advantageous to themselves.[22]

Following this track one expects that Jones would go on to say that, at least in some cases, nonland inputs relative to land inputs were equally intensive (or output yields equally high) under metayage as under fixed-rent farming or owner cultivation. But he did not, and instead he concluded:

If the relation between the metayer and the proprietor has some advantages when compared with…… the serf……, it has some very serious inconveniences peculiar to itself. The divided interest which exists in the produce of cultivation, mars almost every attempt at improvement.[23]

It is difficult to say whether Jones's conclusion contradicts his earlier statements. By improvement or what they called "stock" in the land, classical economists seem to mean "investment" in land, but exactly what they did mean is not clear. According to our convention, investment is the balancing of consumption over time; that is, present sacrifice for future benefit. A man is investing when he tills the soil today for corn tomorrow, pulls a weed, or removes a rock. The various time lengths of the investment returns are treated in a general framework. And it is conceptually the same whether the work is done by a man or a horse, or through the use of more fertilizers, better irrigation, or other assets. It is only within the framework of a timeless input-output model that we do not speak of investment. Under our convention, therefore, to say both that the intensity of labor input (which can be used to improve land) can be freely adjusted and also that "the divided interest mars almost every attempt at improvement" is contradictory indeed.

But to Jones and his contemporaries, and even to Mill and others after him, the concept of "improvement" or "investment" was ambiguous on two counts. First, they failed to distinguish farming inputs at one moment in time from investment over time. Thus it is not always clear whether they laid the blame on the product sharing or on the nonperpetual lease. Second, instead of viewing labor and nonlabor inputs as different physical entities performing different functions in production, they viewed them as different conceptually. To them, "labor" is "short" and non-labor is "long", and "improvements" were made only by "capital" and not by "labor".

Even accepting their convention in vague terms, however, Jones might have seen that since "labor" could be adjusted so could "capital", or that "labor" could be traded for "capital". But Jones did not. Indeed, one cannot help but speculate that his abrupt conclusion was drawn not from logical reasoning, but from the preconception that the British system was superior. And it is a matter of conjecture whether Jones would have altered his conclusion had he considered the accounts on the Italian metayers written by Simonde de Sismondi some fifteen years earlier. Sismondi was himself a metayer landlord, and, of course, he wrote favorably of the system:

The system of cultivation by metayers…… contributes, more than anything else, to diffuse happiness among the lower classes, to raise land to a high state of culture, and accumulate a great quantity of wealth upon it…… Under this system, the peasant has an interest in the property, as if it were his own…… The accumulation of an immense capital upon the soil, the invention of many judicious rotations, and industrious processes,…… the collection of a numerous population, upon a space very limited and naturally barren, shows plainly enough that this mode of cultivation is as profitable to the land itself as to the peasant.[24]

This exuberant endorsement of metayage is quite contradictory to Young's condemnations. But it was not until John Stuart Mill tackled the issue that arguments were taken from both sides.[25]

With an impressive coverage of the literature, Mill noted that "the metayer system has met with no mercy from English authorities".[26] He claimed "that the unmeasured vituperation lavished upon the system by English writers, is grounded on an extremely narrow view of the subject".[27] Mill's own analysis is essentially a modification of Jones's, and, more explicitly, he also treated labor input and improvement of land as two conceptually different things.

Mill quoted and accepted Smith's view that share rent is analogous to a tax, and therefore felt that the tenant would not be interested in making "improvements".[28] Thus, "the improvements must be made with the capital of the landlord," but "custom" is "a serious hindrance to improvement".[29] In regard to labor input, Mill's argument goes from "not enough" to a possibility of "too much", which may appear inconsistent at first sight:

The metayer has less motive to exertion than the peasant proprietor, since only half the fruits of his industry, instead of the whole, are his own…… I am supposing that this half produced is sufficient to yield him a comfortable support. Whether it is so, depends on the degree of subdivision of the land; which depends on the operation of the population principle…… There is a landlord, who may exert a controlling power, by refusing his consent to a subdivision. I do not, however, attach great importance to this check, because the farm may be loaded with superfluous hands without being subdivided; and because, so long as the increase of hands increases the gross produce, which is almost always the case, the landlord, who receives half the produce, is an immediate gainer, the inconvenience falling only on the laborers.[30]

Two things should be noted. First, in this quotation Mill presupposed that no prevailing wage rate or alternative earning exists for the metayer tenant, and it is not even clear whether the popular "subsistence" income is implied as a limit to the subdivision of land. Indeed, without explicitly stating the law of diminishing returns, it would be difficult to arrive at any kind of equilibrium.[31] Second, in regard to "improvements", Mill did not allow them to be made through adjusting the rental percentage-even though he was well aware of differing rental percentages.[32]

The lack of any definite conclusion by Mill stems from the fact that he did not claim to perform any economic analysis of the subject. He began by stating that the metayage system in Europe was regulated by custom and not by competition,[33] so that "when the partition of the produce is a matter of fixed usage, not of varying convention, political economy has no laws of distribution to investigate".[34] From whom did Mill get this "custom" idea? From Sismondi:

This connexion [of input commitments] is often the subject of a contract, to define certain services and certain occasional payments to which the metayer binds himself: nevertheless the differences in the obligations of one such contract and another are inconsiderable; usage governs alike all these engagements, and supplies the stipulations which have not been expressed: and the landlord who attempted to depart from usage, who exacted more than his neighbor, who took for the basis of the agreement anything but the equal division of the crops, would render himself so odious, he would be so sure of not obtaining a metayer who was an honest man, that the contract of all the metayers may be considered as identical, at least in each province, and never gives rise to any competition among peasants in search of employment, or any offer to cultivate the soil on cheaper terms than one another.[35]

The contractual stipulations as described are implied by the theory of share tenancy, and as we shall see in greater detail, they are similar to those in China. But the statement that the metayage system "never gives rise to competition" is wrong. Indeed, the restraints on the contracting parties as visualized by Sismondi are restraints imposed by competition itself.

We may well ask: Why were the terms of a share contract viewed as "customary" and not determined by competition? The answer, I believe, is that factor prices are not explicitly stated in a share contract.[36] Under fixed rent or a wage contract, not only is the rental price of land or the wage rate explicitly stated, but also one party to the contract can buy freely any quantity of a resource by paying a high enough price. Under a share contract, where the pricing mechanism operates by adjusting the rental percentage and the ratio of nonland inputs to land, it not only yields an impression that market prices do not exist, but the mutually stipulated input intensity also yields an impression of "fixity". Sismondi and Mill were unable to see that, ceteris paribus, a reduction in farm size represents either a fall in the wage rate (or in the cost of nonland inputs) or a rise in the rental price of land; or that a fall in the rental percentage represents either a rise in the wage rate or a fall in the rental price of land. Furthermore, a change in relative factor prices in the market can be flexibly adjusted in a share contract through several dimensions, so flexibly that it may yield an impression of inflexibility. For example, a 50 percent increase in the wage rate would appear significant in a wage contract; but, under a share contract, the same increase may be accounted for by lowering the rental percentage a trifle, reducing labor input a trifle, and expanding land size a trifle.

Thus, Mill was not able to settle the issue on theoretical grounds. But one must admit that, in light of the diversified arguments confronting him, the following judgment he made is of considerable wisdom:

If the [metayer] system in Tuscany [Italy] works as well in practice as it is represented to do, with every appearance of minute knowledge, by so competent an authority as Sismondi; if the mode of living of the people, and the size of farms, have for ages maintained and still maintain themselves such as they are said to be by him, it were to be regretted that a state of rural well-being so much beyond what is realized in most European countries, should be put to hazard by an attempt to introduce, under the guise of agricultural improvement, a system of money-rents and capitalist farmers.[37]

[1]. It is interesting that oriental writers generally share the same view. See sources cited in chapter 1.

[2]. When the first draft of this study was written, I was unaware that share tenancy had been frequently analyzed. A subsequent survey of the literature convinced me that tracing the development of economic thinking on the subject would be worthwhile. The following section on the classical view is an expansion of a summary by D. Gale Johnson. See his "Resource Allocation under Share Contracts", Journal of Political Economy (April,1950), pp.112-14.

[3]. Adam Smith, Wealth of Nations (1776; New York: Modern Library edition,1937), p.366.

[4]. Ibid., p.367

[5]. Ibid., Book 3, chap.2.

[6]. Ibid., p.365.

[7]. Ibid., p.368. As D. Gale Johnson points out: "Not only did Smith object to share renting, but he proposed that taxes be used to induce landlords to use other leasing arrangement". See Johnson, "Resource Allocation", p.112; and Smith, Wealth of Nations, pp.779-88, esp.p.783.

[8]. Smith, Wealth of Nations, p.368.

[9]. Ibid., p.369.

[10]. Ibid., pp.368-72.

[11]. Young's Travels was first published in 1792, and it has since gone through a number of editions. I have at hand a Dublin reprint dated 1793, vol.2; an abridged version edited by Constantia Maxwell (Cambridge: At the University Press,1929); and Miss Betham-Edwards, ed., Arthur Young's Travels in France during the Years 1787,1788,1789 (London: George Bell and Sons,1892).

[12]. Young, Travels, Dublin edition, pp.241-42. Young offered little analysis to support his claims, and one doubts his impartiality when he stated: "The metayers were so miserably poor, it was impossible for them to cultivate well. I started some observations on the modes which ought to be pursued; but all conversation of that sort is time lost in France" (Maxwell edition, pp.202-3).

[13]. Young, Travels, Betham-Edwards edition.

[14]. Ibid., p.18.

[15]. Young, Travels, Maxwell edition, pp.361-404.

[16]. When Betham-Edwards wrote in her introduction to Young's Travels (p.vi), that "nothing has done more [than metayage] to improve the condition of the peasant and of husbandry within the last fifty years", she was not speaking of French agriculture at Young's time. Thus, Maxwell's editing method appears more appropriate. Maxwell's opinion on metayage is worth noting:

It was not perhaps so much a cause of poverty however as a result, and that it was a system that worked rather better in practice than in theory is shown by the fact that it survived the Revolution, and is still a recognized form of land-holding…… Even before the Revolution there were many French landlords who lived on excellent terms with their metayers, visiting them in their holdings and discussing agricultural matters with them (Maxwell edition, p.xxx).

[17]. See Young, Travels, Dublin edition, p.239.

[18]. Young's account of taxes is best seen in a chapter on the Revolution (Travels, Maxwell edition, pp. 327-60). On p.xxvi, Maxwell noted: "According to recent estimates 36 per cent of the peasant's income disappeared in direct taxes to the State; 14 per cent went on tithes payable to the Church; while 11 or 12 per cent was consumed by seigneurial dues at Young's time."

[19]. See ibid., pp.296-97.

[20]. Richard Jones, An Essay on the Distribution of Wealth and on the Sources of Taxation, part 1—Rent (London: John Murrary, 1831). To my knowledge, no Part 2 was ever issued. That Jones shared Smith's view of land lease development is evident throughout the entire volume, and esp.on pp.73-75.

[21]. Ibid., p.91.

[22]. Ibid., pp.98-99.

[23]. Ibid., p.102.

[24].J.C.L. Simonde de Sismondi, Political Economy 1815 (New York: Augustus M. Kelley,1966), pp.41-42.

[25]. See John Stuart Mill, Principles of Political Economy (4th ed.London: John W. Parker and Son,1857), Book 2, chap.8, "Of Metayers".

[26]. Ibid., p.367.J.R. McCulloch was another noted "English authority" who wrote:

The practice of letting lands by proportional rents…… is very general on the continent; and wherever it has been adopted, it has put a stop to all improvements, and has reduced the cultivators to the most abject poverty [Principles of Political Economy (Edinburgh,1843), p.471].

[27]. Mill, Principles of Political Economy, p.380.

[28]. Ibid., pp.366-67.

[29]. Ibid., p.367.

[30]. Ibid., pp.365-66.

[31]. Note that the last statement in this quotation is quite different from Jones's view.

[32]. See Mill, Principles of Political Economy, p.363 and the second footnote on p.364.

[33]. Ibid., p.363.

[34]. Ibid., p.364.

[35]. Mill quotes Sismondi in ibid., pp.363-64.

[36]. This is drawn from my impression of their discussion of distribution. See, for example, Jones, Distribution of Wealth; Sismondi, Political Economy, chap.3; McCulloch, Principles of Political Economy, pt.3; and Mill, Principles of Political Economy, bk.2.

[37]. Mill, Principles of Political Economy, pp.380-81.

### B. The Neoclassical View

Several analytical deficiencies stood in the way of classical writers in arriving at a general solution for resource use under share tenancy. Other than their conceptual ambiguities mentioned earlier, classical writers failed to treat land rent as part of production cost.[1] Furthermore, the marginal analysis required to reach an equilibrium was vague. These shortcomings did not handicap Alfred Marshall when he analyzed sharecropping. But whereas, before Marshall, Sismondi and Mill had not placed much weight on the tax-equivalent argument, Marshall renewed the thesis, presumably because the analogy to a tax under share rent fits rather neatly into his marginal analysis.[2] Even with such an approach, Marshall almost obtained the correct solution in a footnote.

By tracing footnotes in two chapters of Marshall's Principles,[40] one derives a diagram used by several subsequent economists. With fuller exposition, this is presented in figure 4. For simplicity, let us assume that the only tenant input is labor. In figure 4, tenant labor, t, is measured along the horizontal axis, and represents the marginal product of tenant labor with a given plot of land. The marginal tenant cost, , is horizontal in a competitive market, with W as the prevailing wage rate. If the landowner is to hire farm hands to do the tilling, equilibrium is at B, and the quantity of farm labor hired will be t2. At this equilibrium, we have the marginal equality: = . The same result obtains for owner cultivation, whether the owner works up to t 2 and works elsewhere, or whether he works to less than t2and hires additional laborers at W. The total rent as a return to land received by the landowner is represented by the area MDB, an amount equal to that of a fixed-rent contract.

Under the tax approach of analyzing share tenancy, however, the marginal tenant receipt net of rent, (1-r), will shift downward at every point. That is, if the landowner takes 40 percent of the annual yield (r=0.4) and the tenant takes 60 percent, (1-r) will be 60 percent of at every point. With the tenant's decision made at the margin, it is said, equilibrium is at A, where the marginal tenant cost equals the marginal tenant receipt; that is, = (1-r). The associated quantity of tenant labor will be t1. Under this condition, the total product is represented by the area ODJt1, with the landowner getting a rent equal to area EDJA and the tenant's share equaling OEAt1. As shown, area MEA represents the amount received by the ten-ant over and above his alternative earning (area OMAt1). With equilibrium at A, the marginal product of tenant labor is higher than the marginal tenant cost. Share tenancy is therefore inefficient (with area JAB representing the economic waste).

It was with this analysis in mind that Marshall commented:

When the cultivator has to give to his landlord half of the returns to each dose of capital and labor that he applies to the land, it will not be to his interest to apply any doses the total return to which is less than twice enough to reward him. If, then, he is free to cultivate as he chooses, he will cultivate far less intensively than on the English plan [fixed rent]; he will apply only so much capital and labor as will give him returns more than twice enough to repay himself: so that his landlord will get a smaller share even of those returns than he would have on the plan of a fixed payment.[4]

What is important here is that Marshall saw that according to this analysis the share tenant will be getting a residual return and the landlord will be getting a smaller rental income than under fixed rent. What is strange is that Marshall did not question why the landlord does not choose instead a fixed rent contract or selling his landownership to the tenant outright.[5] And what is natural is that Marshall was reluctant to let the tenant's residual earning remain unexplained:

If the tenant has no fixity of tenure, the landlord can deliberately and freely arrange the amount of capital and labor supplied by the tenant and the amount of capital supplied by himself to suit the exigencies of each special case.[6]

Marshall's view was that the metayer tenant "has practical fixity of tenure"[7], and he referred to an article by Henry Higgs.[8] Higgs noted that the rental shares do differ, and that "rigid as metayer may at first sight seem to be, it is susceptible of considerable elasticity"[9]. He nonetheless shared the "custom" idea with Sismondi and Mill. Higgs based his judgment on an empirical survey he had conducted in France, which unfortunately involved a sample of only a single farm.[10] It is the notion of "fixity of tenure", perhaps, that led Marshall to mention the possibility of adjustments only in a footnote:

If the landlord controls the amount [of capital] freely and in his own interest, and can bargain with his tenant as to the amount of labor he applies, it can be proved geometrically that he will so adjust it as to force the tenant to cultivate the land just as intensively as he would under the English tenure [fixed rent]; and his share will then be the same as under it.[11]

Marshall provided no geometric proof, and it is an interesting conjecture whether he would have altered this footnote had he done so. This conjecture is interesting because the results he conceived are correct only in certain special cases, but as a matter of generality they are incorrect. They are incorrect because Marshall did not allow the rental percentage to vary.[12] Let me explain. Following Marshall, suppose the horizontal axis of figure 4 measures not only labor but nonland inputs in general (and assume away the associated substitution problem). Suppose also the initial amount of nonland inputs is at tx. Given a certain (not any) rental percentage, and given that the landlord has been providing a sufficiently large amount of nonland "capital" input, then he may adjust the nonland inputs provided by either contracting party so that (a) the rental income under share rent is the same as under fixed rent and (b) the tenant's residual earning is exhausted-without varying the given rental percentage. As such, Marshall is correct. But suppose that the tenant is to provide all the nonland inputs, or that the landlord has been providing a part of the nonland inputs but the amount is too small; then, except by accident, the rental percentage must be varied (in addition to adjusting nonland inputs over land) to obtain the results Marshall visualized. To view the problem from another angle: given the ratio of nonland inputs to land that is consistent with wealth maximization, and given that the relative shares of nonland inputs contributed by the contracting parties are stipulated, there exists one and only one rental percentage which is consistent with equilibrium.[13] This difficult paragraph will become clearer when we come to the next section.

It should be noted that Marshall, like Smith, Jones, and Mill before him, attempted to "rank" various land tenure arrangements according to some notion of economic efficiency.[14] They did not tackle the issue by identifying a specific set of property right constraints subject to which several forms of land tenure arrangements may imply the same resource use. In their discussions of share tenancy, the freely alienable rights implicit in their analyses suggest that the constraint of private property rights was assumed.[15] But whereas Smith and Jones viewed a share lease, though wasteful, as transitional. Mill and Marshall laid the blame on "custom". Smith's prediction that fixed rents would replace share rents had failed to come true; and, as noted earlier, the terms of a share contract might yield the impression of being customarily fixed.

Among contemporary writers who performed similar "rankings" and also relied on the notion of "custom" are Rainer Schickele [16] and Earl O. Heady.[17] Schickele and Heady furthered the tax approach in their analyses of a share lease, but in a way somewhat different from Marshall. Using a diagram identical to figure 4, they concluded that equilibrium is at points, where the marginal tenant cost equals the marginal tenant receipt; that is, = (1-r). The amount of tenant input committed will be t1. According to Schickele, if tenant input is increased from t1to t2, then area ABC represents a "present" from the tenant to the landowner [18]—since the added tenant receipt (area t1ACt2) is less than the added tenant cost (area t1ABt2) by this amount. Schickele did not, however, proceed with the paradox: if tenant input is at t1, then area MEA (tenant's residual earning) represents a "present" from the landowner to the tenant. It follows from this line of reasoning that if tenant input is somewhere between t1and t2, both parties will be offering "presents" to each other!

From the above Schickele (and Heady after him) claimed that part of the nonland inputs must be shared by the landowner to attain efficient resource use. A general rule of "efficient" input subsidization by the landowner was sought. Using a 50 percent rental share (r=0.5), and using the horizontal axis of figure 4 to measure nonland inputs in general, Schickele concluded that "if the landlord would share equally in all the input costs going into crop production the intensity would be carried to t2, that is, to the same degree which corresponds to maximum efficiency"[19]. Its economic content aside, what could have been merely a geometric "accident" was compounded into an illusion by Heady. Heady converted and (1-r) into marginal (output) cost curves, but with the curves arbitrarily drawn.[20] Thus, on Schickele's proposal. Heady generalized: "The cost of variable factors (where one factor such as land is fixed) must be divided between the landlord and the tenant in the proportions that hold for the division of the product [to attain efficiency]"[21]. Let me clarify.

Return to figure 4, where we use r=0.4. Suppose the rental percentage, r, is 0.5 (that used by Schickele). Schickele seems to have thought that tx would necessarily divide the distance Ot2 into two equal parts. Thus, if the landowner contributes nonland inputs from t1 to t2, he would then be sharing exactly half the nonland input cost - or sharing it "in the proportions that hold for the division of the product". But this is wrong even if the marginal product curve is linear. The marginal tenant receipt curve, recall, is 1—r of at every point, measured vertically. And only at one point will the same percentage hold for a horizontal measure. What is interesting here is not the geometric illusion itself, but a paradoxical result one obtains without it: Given any rental percentage, and a wage rate just high enough to reduce tenant input to zero (according to the tax-equivalent analysis), the value of land is still positive (since the total product is still greater than the total wage). Would it be sensible to say, then, that in certain cases the value of land to the tenant is positive but he is not willing to offer any share payment for its use?

It is not clear what is meant by "custom" in its usage by Schickele and Heady in their discussions of land tenure. Whereas Mill had visualized "custom" only as something noncompetitive, Schickele and Heady referred to both "custom" and "competition" in their analyses of share tenancy. One interpretation is that by "custom" they meant a situation where the postulate of wealth or utility maximization does not apply. Yet without any such behavioral postulate, the meaning of competition cannot be defined. To complicate the matter, at times the postulate of maximizing behavior is implicit for the tenant but not for the landowner. This is explicitly stated by Charles Issawii in his analysis of share tenancy. Concluding also that equilibrium is at A (see figure 4), Issawii admitted that, in his analysis:

It has been implicitly assumed throughout…… that landlords do not respond readily to such economic motivations as the possibility of increasing their income by investment; if they did, the distinction between fixed rents and share-cropping would, naturally, lose most of its significance. In the past this assumption has, to a large extent, held true for most underdeveloped countries and, to a slightly lesser extent, it still holds true.[22]

This kind of analysis is quite popular in "underdevelopment" literature. And one wonders how an analyst following Marshall would fare should he discard the idea of custom. We find this in a study by D. Gale Johnson.[23] Johnson formalized Marshall's analysis in greater detail, and his equations led him to conclude:

Under a crop-share lease, if the landlord's share of the crops is half, the tenant will apply his resources in production of crops until the marginal cost of crop output is equal to half the value of the marginal output. The same tenant, however, will conduct his livestock operations, where important costs are borne by the landlord and the receipts are not shared with him, in the usual manner. The landlord will not invest in land assets unless the value of the marginal product is twice the marginal cost.[24]

Johnson noted, however, that his analysis is based on "circumstances in which…… the tenant and the landlord…… each views his interest separately"[25], which is similar to Marshall's supposition that the tenant "is free to cultivate as he chooses". This supposition, of course, renders the meaning of a contract nebulous. What is interesting here is that even if we accept this supposition, Johnson's conclusion is founded on a set of constraints which are difficult to specify. This will be shown in the next section.

But Johnson was reluctant to accept the implication of inefficient resource use under a share contract, and he devoted one section to investigating other possible adjustments.[26] He found that, "though admittedly inadequate, the available evidence indicates that the crop-share contract yields at least as much, if not more, rent per acre than does the cash lease on comparable farms"[27]. In an attempt to reconcile this apparent conflict between theory and fact, Johnson argued that with a short-term lease the tenant is not really free to cultivate in anyway he sees fit.[28] Thus, the actual intensity of tenant input "will depend upon what he thinks 'he can get by with'".[29]

It is difficult to understand why Johnson did not discard his theoretical analysis and start anew - by considering that the contracting parties are free only to accept or not to accept a contract, and that they "can get by with" only as much as the restraints of competition allow. These choices are exactly the same as fixed rent and wage contracts, which are implied by the constraints of private property rights, and which Johnson had in mind. A theory constructed on these rights will reveal that the terms of a share contract are expressed through the market-determined rental percentage and ratio of nonland inputs to land. Yet on this point Johnson was in doubt:

The process by which the landlord and tenant enter into a lease is not well understood. The price system does not function in the normal sense, for land is not necessarily rented to the tenant offering the highest rent payment. However, there is only a difference of modest degree in the role of price rationing in the share-rental market and in the cash-rental market.[30]

In the course of his analysis, Johnson had obtained - and abandoned - a condition which is quite sufficient to reveal the market terms of a share contract. Referring to the theoretical formulation, he wrote:

Is it likely that a tenant operating under these conditions will allocate his resources in exactly the same way that he would if paying a cash rent independent of the actual output? The answer is apparently in the negative. There is only one average rent per acre for which the resource allocations will be the same under a variable-share proportion and a fixed rent per acre. And there is no reason to believe that this particular rent would emerge under competitive conditions.[31]

Why not? As shown in the last chapter, the identity of the average rent under fixed and share contracts exists, in equilibrium, when rent per acre is at a maximum. This maximum rent per acre is unique because it is obtained when the marginal nonland cost equals the marginal product of nonland inputs. The associated land size per tenant, values of nonland inputs and rental percentage, will be the terms stipulated in the share contract. Thus, it is somewhat puzzling that Johnson also wrote:

With a short-term lease renters are obviously aware that landlords have the alternative of renting their land for a cash rent independent of current output. Consequently, the tenant must plan to produce an average output per acre that will provide a rental payment, if yields are average, equal to the possible cash rent……[32]

The apparent contradiction between this and the earlier quotation can perhaps be reconciled as follows: In rejecting the theoretical identity of the average rent under fixed and share contracts, Johnson was following a model in which this identity cannot be obtained. In accepting the identity, empirically his observation led him to search for another interpretation. As we saw when we discussed Marshall, the theoretical identity can be derived only if the rental percentage is treated as a variable. My own impression of Johnson's insightful work is that he maintains that although share rent is less productive than fixed rent according to a model with considerable limitations they may nonetheless be identical in practice. For this reason perhaps, he called for "an empirical verification…… of [the effects of] share leases upon resource allocation"[33].

[1]. At page proof, George J. Stigler informed me that Mill did recognize rent as a cost of production, whereas Marshall was reluctant.

[2]. See Marshall, Principles of Economics (8th ed.,1920; London; Macmillan&Co.,1956), pp.534-37.

[3]. Ibid., bk.4, chap.3; and bk.6, chap.10, p.536.

[4]. Ibid., pp.535-36.

[5]. Marshall was aware, like other writers before him, that fixed rents and owner cultivators existed together with metayers in Europe. Various estimates on the frequency of metayer farms differ greatly.

[6]. Marshall, Principles of Economics, p.536.

[7]. Ibid.

[8]. Higgs, "'Metayage' in Western France", The Economic Journal (March,1894), pp.1-13.

[9]. Ibid., p.9 and n.1.

[10]. See ibid., pp.9-13.

[11]. Marshall, Principles of Economics, p.536, n.2.

[12]. Though Marshall was aware that the rental percentage was not the same everywhere (p.535, n.1), he noted that for customary reasons variation of it "could seldom be done without an appearance of violence" (p.533). No evidence of "violence" was offered.

The theoretical difficulty Marshall encountered as a result of not varying the rental percentage was made evident when he wrote (p.536, n.2):

If [the landlord] cannot modify the amount [of capital], but still control the amount of the tenant's labour, then with certain shapes of the produce curve, the cultivation will be more intensive than it would be on the English plan [fixed rent]; but the landlord's share will be somewhat less. This paradoxical result has some scientific interest, but little practical importance.

[13]. See chapter 2, and section C of this chapter. Note, however, that if several crops are grown, multiple rental percentages may exist in one share contract due to different factor intensities required for different crops. See chapter 4, section B.

[14]. Marshall, Principles of Economics, bk.6, chap.10.

[15]. Transferability of rights among individual owners implies exclusivity in use, at least to some degree. Transfers of resource rights in the marketplace are not confined to outright transfers, but also include various leasing arrangements. A legal restriction on one form or another of these transfers may impose a higher cost of transaction, but still may not constitute a set of constraints different enough to affect resource use significantly. That freely alienable rights existed for the French metayage is evident in Young, Travels, Maxwell edition, "Editor's Notes".

When N. Georgescu-Roegen analyzed share tenancy, he referred to it as a form of land tenure under "feudalism". But we have yet to identify the property right constraint defining "feudalism". Georgescu-Roegen also employed the tax-equivalent approach and reached the conclusion that sharecropping was inefficient. See his "Economic Theory and Agrarian Economics", Oxford Economic Papers (February,1960), pp.23-26.

[16]. Rainer Schickele, "Effects of Tenure Systems on Agricultural Efficiency", Journal of Farm Economics (February,1941), pp.185-207.

[17]. Earl O. Heady, "Economics of Farm Leasing Systems", Journal of Farm Economics (August,1947), pp.659-78.

[18]. Schickele, "Effects of Tenure Systems", p.193.

[19]. Ibid., p.195.

[20]. Heady, "Farm Leasing Systems", p.672.

[21]. Ibid., p.673.

[22]. Charles Issawi, "Farm Output under Fixed Rents and Share Tenancy", Land Economics (February,1957), p.76.

[23].D. Gale Johnson, "Resource Allocation under Share Contracts", Journal of Political Economy (April,1950), pp.111-23.

[24]. Ibid., p.111. Similar conclusions are also reached in Amartya K. Sen, "Peasants and Dualism with or without Surplus Labor", Journal of Political Economy (October,1966), pp.445-46.

[25]. Johnson, "Resource Allocation", p.111.

[26]. Ibid., v., pp.118-21.

[27]. Ibid., p.118. In the associated footnote Johnson wrote:

I have estimated net rents on crop-share-rented farms in Iowa from 1925 through 1946. From 1925 through 1934 net rents on share-rented farms averaged perhaps a dollar per acre less than on cash-rented farms. From 1935 through 1939 the net rents were roughly the same. From 1940 through 1946 net rents were at least four dollars an acre more on share-rented than on cash-rented farms.

[28]. Ibid., pp.119-20.

[29]. Ibid., p.120

[30]. Ibid., p.121.

[31]. Ibid., p.118.

[32]. Ibid., p.120.

[33]. Ibid., p.123.

### C. A Correction of Error

To clarify the basic difference between the theory of share tenancy derived in the last chapter and the tax-equivalent analysis, a simple geometric presentation will suffice. This is illustrated in figure 5, which is an expanded version of figure 4. We are already familiar with the marginal product of tenant labor, , and the marginal tenant receipt under share rent, (1-r). According to the tax-equivalent approach, equilibrium is at A, with the amount of tenant input equal to t1.

In figure 5, assume a given land area which we hold fixed for . The result of the more general solution can be derived as follows. Given r=0.4 which defines (1-r), if t2 of tenant labor is contractually stipulated, the landowner's share of the total product in the form of rent equals area EDBC, and the tenant's share equals OECt2. As drawn, the tenant's share is still higher than his alternative earning (area OMBt2), with area MEA greater than area ABC, which means that the total rent (area EDBC) will be smaller than that of a wage or a fixed-rent contract (area MDB). Given the marginal tenant receipt, (1-r), the corresponding average tenant receipt will be (q/t) (1-r), which is 40 percent of the average product of tenant labor, q/t, at every point. In this case, the landowner can successfully stipulate that the tenant work up to t3 where the average tenant receipt, (q/t) (1-r), equals the wage rate, implying that the tenant's income from farming is equal to his alternative earning.

With tenant input t3, however, the total rent received by the landowner will be equal to area MDB minus area BNP, an amount of rent smaller than that of owner cultivation, a wage or a fixed-rent contract. To maximize his wealth with a share contract subject to the constraint of tenant cost, the landowner will thus raise the rental percentage to $r$. The choice of $r$ lowers the marginal tenant receipt to the dotted line GF, with area MGK equal to area KBL The corresponding average tenant receipt will be $(q/t) (1-r* )$, which necessarily cuts the marginal product of labor, , at B. At this new rental percentage, the tenant labor stipulated is t2, the landowner's total rent equals area GDBI (which equals area MDB), and the tenant's share equals area OGIt2, which is no greater than his alternative earning (area OMBt2). The equilibrium condition can be identified: at B we have where r* is the equilibrium rental percentage.[1] Indeed, one of the main sources of confusion in the tax approach is the marginal tenant receipt curve, (1-r). Important as it may seem, it is only illusory for decision making under unrestrained private property rights.

Indeed, a closer analysis reveals that position A is not an equilibrium, even if we follow the tax-equivalent approach to analysis of a share contract and let the tenant make the decision at the margin. Suppose there are many landowners and each landowner does not specify the tenant input per acre, so that the tenant can farm any way he desires. Suppose further that there is no law prohibiting the tenant from working for more than one landowner. There is no reason for the tenant to choose to work up to ti (in figure 5) for one landowner. Imagine that the marginal tenant receipt curves in each of these farms are negatively sloping all the way. To maximize his income from farming, the tenant will choose to disperse his input resource over many farms until his input endowment is exhausted, in such a way that his marginal income (or marginal tenant receipt) from different farms is equal.[2] In this case, the tenant will be receiving an aggregate income higher than his alternative earning, and other tenants will enter to compete.

Permit our imagination to reach still further. Suppose each successive tenant entering into farming works on several farms also, and likewise equates his marginal income from different farms. As the number of tenants increases, the marginal income from different farms for each tenant will fall, implying rises in the landowners' rental shares. And tenants will continue entering as long as the aggregate income for each tenant is higher than the alternative earning.[3] Given (1-r) as drawn, and assuming that the landowners contractually accept any amount of labor, competition among tenants will push labor input on each farm (one such farm is represented by figure 5) to t3. The resulted "overcrowded" tilling implies that rental shares are not at maximums. Competition again prevails. Given homogeneous factors of production, wealth maximization implies that the landowners will choose among tenants who offer rental percentages as high as $r$, while competition among landowners implies that it will not be any higher. Given r which defines $( 1-r* )$, the amount of tenant labor competitively offered and contractually accepted for each farm will be t2. The market equilibrium thus reached occurs when the marginal product of tenant labor in each farm equals the marginal tenant cost (point B in figure 5).

The resource allocation implied by the above equilibrium, while satisfying the Pareto condition, takes an interesting form which has yet to be observed: There are many tenants working on one farm with each tenant committing a trivial amount of labor; each tenant works on many farms, with his aggregate income from these farms covering his alternative earning. In substance, this is identical with the form under which each tenant commits all his labor inputs to one farm until the marginal product of tenant labor equals the marginal tenant cost. However, we never observe tenants'dispersing their inputs among many farms as described because transaction costs, and in particular the cost of contracting and the cost of moving from farm to farm, are not zero.

[1]. In a less comprehensive form, this is an alternative expression of the results obtained in chapter 2.

[2]. Suppose the marginal tenant receipt curves rise before falling, and suppose the production functions in all farms are identical. Given r, the tenant (who is now entirely free to choose) will disperse his total inputs until the average tenant receipt, (q/t) (1-r), for each farm is at a maximum, or the marginal product of land equals zero. This is an alternative view of a condition implied by the law of variable proportions under linear homogeneity.

Indeed, under private ownership of resources, it is difficult to define a set of constraints which makes position A in figure 5 an equilibrium. Under state ownership of land, however, an interesting set of conditions for position A can be specified: (a) that landownership belongs to the state and the government arbitrarily assigns land to tenants with rents collected on a share basis; (b) that the tenant is free to work elsewhere at a market wage rate and the amount of labor he uses on the state land is not stipulated; and (c) that the tenant's lease right over the state land is not transferable in any form. If strictly enforced, it appears that equilibrium will be at A, and the residual earnings for the government tenants are similar to "welfare" payments. Yet this is hardly the set of constraints which writers on sharecropping have in mind.

[3]. If the industry supply of tenant labor is rising, the entry of tenants will be associated with a gradual rise in the wage rate. This complication is ignored here, and the wage rate used in figure 5 is the one finally determined in the market.

### D. Tests of Implications

Just as D. Gale Johnson called for empirical confirmation of resource use under share tenancy in 1950, Chinese writers made a similar inquiry into tenant farming in general some twenty years earlier. Data on tenant farming was then assiduously compiled. In the late 1920s and early 1930s in China, attacks on farming under tenancy were common, and the desirability of private landownership was a subject of frequent debate. Lacking standardized economic theory to support their arguments, several Chinese organizations and independent writers resorted to empirical investigations. The debate on the tenancy issue was soon terminated by the Sino-Japanese War. And, with the exception of two noted works in the English language, both by John Lossing Buck, the greater part of these findings has since remained unknown.[1]

The aforementioned findings, inadequate as they may seem, constitute the most comprehensive body of evidence relating to agricultural land use under unrestrained private property rights that I could find. The Chinese experience, together with findings from elsewhere in Asia, will be applied in this section and the next chapter. Note, however, that in every instance we use only data collected from periods and locations where the existing system of property rights conforms to the constraint on the basis of which the theory of share tenancy is derived. Therefore, the postwar farm land reforms rule out the use of Asian agricultural data of the past twenty years in this part of the study.

Applying the implications of alternative theories of share tenancy to observations, we can perform several simple tests.

According to the standard theory of share tenancy derived in chapter 2, given the production function, the rental percentage depends upon the fertility of land and the alternative earning of the tenant. Specifically, we should observe a higher rental percentage if (a) the land is more fertile or (b) the cost of tenant inputs is lower.[2] Evidence confirming this hypothesis is strong:

1. According to an investigation which covers 641 sample farms in eleven localities in China (1921-25), J.L. Buck observed:

As rent the tenant gives the landlord one-half of the grain and straw from wheat and rice land, two-fifths of the grain and straw from rice land only, and three-tenths of the grain and straw from poor land.[3]

Likewise, it was observed in Kweichow Province (1929-30) that:

The rental shares depend on the fertility of land. On the average and roughly speaking, for upper grade land the rental share is 60 per cent; for medium grade 50 per cent; and for lower grade 40 per cent.[4]

Casual observations aside, numerical data showing the same patterns were collected by the Legislative Yuan (China,1930) and the Department of Internal Affairs (China,1932).[5] The findings of the latter have been computed and placed in Appendix B, because they encompass twenty-two provinces in China with seven grades of land.

1. In Taiwan, a maximum rental of 37.5 percent of the annual yield was enforced by the government in 1949. This maximum percentage was uniform for all tenant contracts, regardless of whether the land involved was paddy field or dry field. The data reveal that 99.4 percent of the paddy fields under tenancy were affected by this share restriction; that is, the initial rental share was higher than 37.5 percent of the yield. However, only 50.9 percent of the dry fields under tenancy were affected by the same restriction.[6] This implies that higher rental percentages were generally associated with the more fertile paddy fields under a free market.

2. Also, Buck found that the rental percentage was higher when the landowner provided part of the farming inputs (i.e., the tenant cost was lower):

The percentage of total receipts for the landlord varies from 24.6 per cent,…… where small rents are demanded to 66.6 per cent,…… where the cropper system prevails and where the landlord furnishes everything but labor and routine management.[7]

Similarly, according to another survey conducted by Ching-Moh Chen in four provinces (China,1934), an average rental percent-age of 55.98 was found in tenant farms where landowners provided seeds, fertilizers, and bullocks, as compared to an average of 46.37 percent when the tenants provided these nonland in-puts.[8]

Furthermore, given the production function, the land space rented to each tenant depends upon the fertility of land and the tenant's alternative earning. Specifically, the farm size per tenant family will be smaller if (a) the land is more fertile or (b) the ten-ant's alternative earning is lower.[9] Again, supporting evidence is strong:

1. In Korea, the average farm size over a ten-year period (1929 through 1938) was 0.58 cho for paddy-field farms and 0.97 cho for dryland farms. In the same period and location, the prices of paddy fields were more than two and one-half times as high as those of dry fields, confirming that paddy fields were generally more fertile.[10]

2. According to another investigation conducted by Buck, which covers 16,786 sample farms in 168 localities in China (1929-33), we find (a) the average farm size in the more fertile rice region is 3.09 acres, as compared to an average size in the less fertile wheat region of 5.63 acres; and (b) of the seven types of crop fields listed, the smallest average size occurs in double-cropping rice fields (2.37 acres), which are generally most fertile.[11]

Turning to the tax-equivalent approach to analysis of share tenancy, and referring again to figure 5, we may well ask: If the asserted equilibrium at A were valid, what would we observe? The following is implied:

1. We would observe lower ratios of labor and other inputs to land in tenant farms than in farms under owner cultivation or in farms cultivated by hired farm hands. It also implies that hectare yields in tenant farms would be lower than in owner farms. But as Buck observed (China,1921—25):

Contrary to the prevailing opinion that tenants do not farm as well as owners, a classification according to yields by different types of tenure shows no significant variation in yields for most localities, and for the few in which a difference does occur, it is in favor of the tenant or part owner as often as for the owner…… In some places, even, it is evident that the tenants farm better than the owners.[12]

Buck's data show the following crop indexes per acre: Owner farms,100 and 101; part-owner farms,99 and 101; and tenant farms,103 and 104.[13] Thirty years later, making no reference to Buck, James O. Bray echoed the same observation:

Underdeveloped countries intent on technical progress in agriculture should recognize the fact that much of the most impressive gain in agricultural productivity occurred in areas of the United States where the dominant form of land tenure is a share rental arrangement between landlord and tenant.[14]

With respect to the intensities of farming, we find Japanese data (1932-38) showing that the average landholding per owner farmer is 2.22 tan, as compared to 2.10 tan per part-owner farmer and 1.93 tan per tenant farmer,[15] also denying the implication of the tax approach. The higher labor-land ratio in tenant farms can be explained by the somewhat higher proportion of paddy fields under tenancy.

1. We would observe, since the rental earning is lower with equilibrium at A (under the tax approach), that the market value of land under tenant cultivation is less than the value of land under owner cultivation. Again, evidence clearly denies this implication. We find that the values of land differ according to fertility grades and locations, and that "the value of land for the three types of tenure (owner, part-owner, and tenant) in most cases varies only a few dollars".[16] Also, with equilibrium at A, we would observe that the actual rent received by the landowner is lower under share tenancy than under a fixed rent (rent per acre). As is shown in Appendix B, however, share rent is generally slightly higher than crop rent.[17] This slight difference, as I shall indicate later, might be explained as a payment to the landowner for his "risk" bearing under a share contract.
2. We would observe, as implied by the tax approach, that a higher rental percentage — with the (1-r) curve shifting downward in figure 5 — would be associated with less farming inputs per unit of land in tenant farms. We find evidence to the contrary: Tenant farms with higher rental percentages usually also display higher labor-land ratios.[18] This is due either to the land's being more fertile or to the tenants' alternative earnings' being lower, a condition implied by our theory of share tenancy.
3. Finally, if the tax-equivalent approach were correct, we would observe that most, if not all, share tenants would rent land from several landlords. But this is rarely the case. Take Taiwan, for example. In 1949 there existed only 1.24 leases per tenant family.[19] This extra .24 lease can easily be explained by the use of marginal plots as discussed in chapter 2. On the other hand, landowners issuing one or two hundred leases were regarded as common.[20]

[1]. Under the auspices of the University of Nanking, Buck directed a forty-man team in the compilation of farming data in China during 1929-36. The original data, which appear to have passed unnoticed, are available in J.L. Buck, ed., Land Utilization in China-Statistics-A Study of 16,786 Farms in 168 Localities and 38,256 Farm Families in Twenty-two Provinces in China,1929-1933 (Nanking: University of Nanking, 1937). During the preparation of this impressive volume, Buck wrote the noted Chinese Farm Economy (1930), and Land Utilization in China (1937), both of which have been distributed by The University of Chicago Press. However, Buck's earlier works are also important: An Economic and Social Survey of 102 Farms near Wuhu, Anhwei, China (Nanking, 1923); An Economic and Social Survey of 150 Farms, Yehshan County, Hopei, China (Nanking, 1926); and Farm Ownership and Tenancy in China (Shanghai: National Christian Council, 1927).

Surveys by independent writers aside, other organizations which have conducted surveys include the Department of Internal Affairs, the Real Estate Bureau, the National Government Statistics Department, the Executive Yuan, and the Legislative Yuan. I have found these independent findings generally consistent with one another.

[2]. This can be conveniently demonstrated in figure 3, chapter 2. Condition (a) is obtained by raising q/h, and condition (b) by lowering f/h. In either case, a higher value of (q-f) h is defined, resulting in a higher rental percentage. Note that these implications apply to fixed and share rents alike.

[3]. Buck, Chinese Farm Economy, p.148. No further information is provided.

[4]. Chinese National Government, Shengching Route Economic Report (1931), p.102.

[5]. See Legislative Yuan, Statistical Monthly (1930),2,5; and Department of Internal Affairs, Public Reports of Internal Affairs (1932),2,1; 5,1 and 2.

[6]. See table 4 in chapter 8 of this study. In Japan, the rental share restriction enacted in 1946 was a maximum of 25 percent of the yield for paddy, and 15 percent for upland fields, reflecting a recognition of the higher rental share for paddy fields under free market conditions. See Ministry of Agriculture and Forestry, Agricultural Land Reform Legislation (Tokyo,1949).

[7]. Buck, Chinese Farm Economy, p.149. The rental shares provided are averages of sample farms in each locality (eleven localities in China,1921-25), and the 66.6 percent cited is purely incidental. See table 2 in ibid., p.148.

[8]. These averages are computed from Chen, Land Rents of Various Provinces in China (Shanghai: Commercial Press,1936), pp.102-3.

[9]. These conditions can likewise be demonstrated by moving the q/h and flh curves in figure 3, chapter 2 accordingly.

[10]. Andrew J. Grajdanzev, Modern Korea (New York: Institute of Pacific Relations, 1944). The average farm sizes are computed from data in table 2, p.291; and the land prices are seen in table 5, p.292.

[11]. See Buck, Land Utilization in China, table 23, p.197. With minor differences in magnitude, identical patterns are seen in tables 7 and 8, pp.272-73. Note also that in ibid., p.197, owner farms (average 4.22 acres) are larger than tenant farms (average 3.56 acres). This is because tenancy occurs more frequently on paddy fields than on dry land everywhere. See Sidney Klein, The Pattern of Land Reform in East Asia after World War II (New York: Bookman Associates, 1958), p.229 ff. See also Buck, Statistics, chap.2, tables 23 and 26; and chap.7.

[12]. Buck, Chinese Farm Economy, pp.156-57. Apparently failing to see that the rental percentage is a variable to insure efficient farming, Buck proceeded to discuss a program of "fair" rent.

[13]. Ibid. Buck's finding is from a sample of 2,542 farms in fifteen localities, seven provinces, China. The slightly higher yields per acre in tenant farms are perhaps due to a higher proportion of paddy fields under tenancy, which Buck did not discern. For similar evidence, see Buck, Farm Ownership and Tenancy in China.

[14]. Bray, "Farm Tenancy and Productivity in Agriculture: The Case of the United States", Food Research Institute Studies (1963), p.25. Although Bray provided no formal theory of share tenancy, the traditional argument did not pass his intuition:

The resource-efficiency argument is somewhat academic…… Both tenant and landlord have an incentive to try to increase the marginal productivity of their own resources…… neither wholly succeeds…… For example, a landlord's suggestion that a third cultivation of the corn would pay (at no cost to him) may be met with the tenant's suggestion that the living room really would be improved by new wall-paper (at no cost to him) [ibid., p.27].

The lack of significant difference in resource use among different leases is also noticed by Walter G. Miller, Walter E. Chryst, and Howard W. Ottoson, "Relative Efficiencies of Farm Tenure Classes in Intrafirm Resource Allocation", Research Bulletin 461 (Iowa Agricultural and Home Economics Experiment Station, November,1958), pp.321-37. Evidence from the United States, however, is not heavily used in this study. This is because various government farm policies which result in different constraints on competition might have affected resource use under different contracts in different ways.

[15]. These averages are computed from data in Andrew J. Grajdanzev, "Statistics of Japanese Agriculture", mimeographed (New York: Institute of Pacific Relations,1941), table 17, p.32. The findings are based on about ninety-five sample farms in each category. The same pattern is observed if "able-bodied man-units" are used instead of "number of members per family". For evidence from Taiwan, see chapter 7 of this study.

Earl O. Heady and Earl W. Kehrberg conducted a survey (Iowa, 1949) with seventy-four selected pairs of cash-lease and share-lease farms. They found "no significant differences between share and cash leases" (p.661) for the input intensities of farming. The authors, however, refused to accept the conflict between their theory and findings, and resorted to other factors which "might account for this lack of difference" (pp.661-62). With the tax approach, they had concluded that, in theory, farming intensity under share rent is necessarily lower than under cash rent (at pp.658-60). See Heady and Kehrberg, "Relationship of Crop Share and Cash Leasing Systems to Farming Efficiency", Research Bulletin 386 (Iowa State College Agricultural Experiment Station, May,1952), pp.635-83.

[16]. Buck, Chinese Farm Economy, p.156. No numerical data are provided.

[18]. This is shown earlier in this section. For some indirect evidence from Taiwan, see chapters 7 and 8.

[19]. Computed from data in Sino-American Joint Commission on Rural Reconstruction "JCRR Annual Reports on Land Reform in the Republic of China", mimeographed (1965), p.35.

[20]. Only a small amount of numerical data has been provided, and examples of landowners holding over one thousand leases are perhaps cases selected to emphasize the concentration of landownership. According to information provided in J.P. Gittinger, "Vietnamese Land Transfer Program", Land Economics (May,1957), "before the land transfer program in Vietnam, 2,170 persons…… have declared total holdings amounting to 976,602 hectares". Given that "the average tenant holding is approximately 2 hectares or perhaps slightly more", the average number of leases for each of these landowners was over two hundred.

## 4. Transaction Costs, Risk Aversion, and the Choice of Contractual Arrangements

The analysis thus far has been primarily based on the condition that transaction costs, and in particular the costs of contractual negotiation and enforcement, are zero. The theory of share tenancy thereupon derived shows that, subject to the constraint of private property rights, economic efficiency is the same under various land tenure arrangements. Although transaction costs exist in the real world, the theory succeeds in explaining a class of observations. But the presence of a variety of contractual arrangements under the same constraint of competition poses the question: Why are different contractual arrangements chosen under the same system of private property rights? The purpose of this chapter is to advance, in a rather informal manner, a choice-theoretic approach based on nonzero transaction cost and risk aversion to explain the observed contractual behavior in agriculture. The observations used are largely drawn from the Chinese experience.

If a firm can increase efficiency in production by employing productive resources of more than one resource owner, a contract to combine the resources will obtain. The formation of the contract involves partial transfers of property rights in one form or another, such as leasing, hiring, or mortgaging.[1] These transfers, and the associated coordination of inputs of various factors in production, are costly events.[2] There are costs of negotiating and of enforcing the stipulations of the contract.

Given the state of personal wealth distribution and the portfolios of assets held as private property by resource owners, some owners will seek contractual arrangements with others in combining resources for production.[3] There is a variety of arrangements under which this can be done. At least two reasons may be offered for the existence of different types of contractual arrangements. First is the existence of natural risk, defined here as the contribution by nature or the state of the world to the variance (or standard deviation) of the product value.[4] Given a nonzero variance for the expected output yield (the total income for the contracting parties), different contractual arrangements allow different distributions of income variances among the contracting parties. Under the postulate of risk aversion, an individual will seek to avoid risk if the cost of doing so is less than the gain from the risk averted. He may avert risk by searching for information about the future (which may not be attainable even at infinitely high cost), by choosing less risky options when investing (which options include portfolio diversification), or by choosing among arrangements with which his burden of risk can be dispersed to other individuals - such as insurance and various contractual arrangements. The last is one of our concerns in this chapter. A second reason for the existence of different contractual arrangements is the different transaction costs that are associated with each. Transaction costs differ because the physical attributes of input and output differ, because institutional arrangements differ, and because different sets of stipulations require varying efforts in enforcement and negotiation.[5]

Let me advance the following hypothesis: the choice of contractual arrangement is made so as to maximize the gain from risk dispersion subject to the constraint of transaction costs. I shall develop this hypothesis and apply it to some observations later.

For any resource, a number of individuals compete for ownership. Each potential buyer or user possesses some knowledge not only of alternative uses of the resource, but also of different transaction costs associated with different arrangements by which the resource may enter into production. Assume away information problems that may exist in competitive trading in the market place;[6] the resource will find that owner whose use of the resource yields the highest value. Competition for and transferability of the ownership right in the marketplace thus perform two main functions for contracting. First, competition conglomerates knowledge from all potential owners - the knowledge of alternative contractual arrangements and uses of the resource; and transferability of property rights ensures that the most valuable knowledge will be utilized. Second, competition among potential contract participants and a resource owner's ability to transfer the right to use his resource reduce the cost of enforcing the stipulated terms in a contract. This is because competing parties will stand by to offer or accept similar terms. In sum, competition in the marketplace reduces the costs of finding and pursuing the most valuable option in which a resource may be contracted for production. While transaction cost determines, it is also determined.

In the absence of transaction costs, that state of resource allocation under which it is no longer possible to reallocate resources to benefit one individual without another's losing (that is, the Pareto condition) implies the familiar set of marginal equalities in resource use. With transaction costs included, however, resource allocation in conformity with the Pareto condition need not satisfy the same set of marginal equalities.[7] Although it is difficult, if not impossible, to separate one type of transaction cost from another, it may nonetheless be useful to distinguish two types of marginal inequalities generated by the presence of transaction costs.

The first may be regarded as being among firms; that is, the same factor input yields different marginal productivities in different firms or in different uses. For example, with transaction costs there may not exist a uniform factor price in the market, and the buyer's price may differ from the seller's price. These price differentials will lead to different marginal productivities of the same factor input in different firms. To the extent that transaction costs are so high as to prohibit transfers of resource rights, the resource value can only be expressed in terms of a nonpecuniary measure.

A second type of marginal inequality may be regarded as being within the firm; that is, the marginal product of a factor employed by a firm may diverge from its marginal factor cost owing to transaction costs. Let me clarify this. Consider a contract involving a lump-sum payment for the use of a resource, wherein the quantity of the resource is not stipulated at all. For example, a landowner, in letting his scarce water resource to a tenant, may charge only a flat fee and allow the tenant to use the quantity of water freely. This arrangement is chosen because the transaction or enforcement cost of quantification (through metering or other devices) is so high (owing to, say, the physical attributes of water) as to make alternative arrangements inefficient.[8] Under this form of contractual payment, the water resource will be used by the tenant until its marginal product is zero, even though the marginal factor cost of the water is positive. But if the marginal inequality of resource use is due to different costs of contracting, it does not imply inefficient resource use. Indeed, if the gain from choosing another contractual arrangement (e.g., a contract with a unit price charged on water) had been greater than the cost of quantifying the use of water, some metering device would have been adopted and the contract with the lump-sum payment abandoned.[9]

Efficient allocation, therefore, requires that each resource be used in the highest-valued option subject to the added constraint of transaction costs. The option value may be measured in utility or in wealth, depending on the existence of market prices. But since transaction costs may also depend on alternative legal arrangements, the "highest-valued option" is not always clear. I shall try to say more on this in the concluding section of this chapter.

The existence of transaction costs appears to have at least three predictable effects. First, they tend to reduce the volume of transactions, thus impairing economic specialization in production and the employment of resources. Second, they may affect marginal equalities (and intensities) of resource use. And third, they will affect the choice of contractual arrangements. The last is our main concern here.

### A. The Choice of Contracts in Agriculture

Consider the three main forms of contracts in agriculture; namely, a fixed-rent contract (rent per acre stated in cash or in crop), a share contract, and a wage contract. Under private property rights, the contracting parties are free to choose among these forms. The observed patterns of contractual choices vary from place to place. For example, share contracts were more frequent than fixed rents in Taiwan and Southeast Asia before the agrarian reforms; in China fixed rents were more frequent than share rents in the 1930s; in Japan, fixed rents predominated; and in general, wage contracts (farm hands) have been infrequent, occurring in about 1 to 5 percent of the farming households in various localities.[10] Why do the patterns of contractual choices differ? What determines the choice of contracts?

Any contract combining resources from different owners for production involves, in addition to negotiation costs, the enforcement costs of controlling inputs and distributing output, according to the terms of the contract. Contracting on a share basis appears to involve higher transaction costs as a whole (the sum of negotiation and enforcement costs) than a fixed-rent or a wage contract. The terms in a share contract, among other things, include the rental percentage, the ratio of nonland input to land, and the types of crops to be grown.[11] These are mutually decided by the landowner and the tenant. For fixed-rent and wage contracts, however, given the market prices, one party alone can decide how much of the other party's resources he shall employ and what crops shall be grown. And since in a share contract the sharing of output is based on the actual yield, efforts must be made by the landowner to ascertain the harvest yield. Thus negotiation and enforcement are more complex for a share contract than for a fixed-rent or a wage contract.

The ranking of transaction costs of fixed-rent and wage contracts appears uncertain. The physical attributes of land are such that the cost of enforcing the contracted amount of input is lower than with labor. That is, the "shirking" of labor input, which may exist in a wage contract (also in a share contract) without either enforcing the input or checking the output, is costly to prevent.[12] But while this "shirking" problem does not appear significant for land input in a fixed-rent contract, policing (or enforcing) the maintenance of soil and other assets owned by the landlord is more costly for a fixed-rent or a share contract than for a wage contract.[13] If we accept the above reasoning, pending empirical confirmation, and if transaction cost is the only consideration, then the minimization of transaction cost implies that share contracts will never be chosen. Why, then, are share contracts chosen?

Suppose the transaction cost is zero or the same for all forms of contract. Let us employ a behavioral postulate of risk aversion, defined here to mean that an individual, given the same expected average income, prefers a lower to a higher variance. In agriculture, variables exogenous to the production function, such as weather conditions and pests, are risk factors which are difficult to forecast and which may significantly affect the variance of the value of output. Under a fixed-rent contract, the tenant bears most, if not all, of the risk; under a wage contract, the landowner bears most, if not all, of the risk. Share tenancy may then be regarded as a device for risk sharing (or risk dispersion); that is, the variance of the output yield is distributed among the contracting parties. Given the postulate of risk aversion, a share contract will be mutually preferred by the landowner and the tenant.[14] Risk, however, exists in varying degrees in any tenancy. Why, then, are fixed-rent and wage contracts chosen at all?

I suggest that the choice of contract be analyzed by employing both the differences in transaction costs and the postulate of risk aversion. Given the state of risk associated with a particular output, a higher transaction cost will lead to lower returns to the productive assets. On the other hand, given the transaction cost, risk aversion implies that asset values and the variances of income are negatively related.[15] While in itself the dispersion of risk under a share contract will lead to higher values for the contracted resources, the higher associated transaction cost will lead to lower asset values. Wealth maximization (or utility maximization, depending on the relevant measurement problem) implies that the contractual arrangement chosen will be the one which yields the highest values for the contracted resources.

Given the variance of output value and the rental percentage, a share contract prescribes a specific distribution of income variances for the contracting parties. The associated state of risk dispersion may not conform to the most preferable state according to the parties' preference functions. Since some dispersion of risk is preferred to no dispersion at all, however, a share contract will be chosen rather than a fixed rent or a wage contract if the higher transaction cost is at least compensated for by the gain from risk dispersion. There exist, of course, still other arrangements under which the dispersion of risk can be tailored to fit each case. But as I shall discuss in the next section, the transaction cost of an arrangement for risk dispersion more flexible than a share contract may be so high as to make it undesirable.

Evidence is available to support the applicability of this kind of analysis:

1. Since transaction cost is asserted to be higher for share than for fixed rents, there would be room for a third party to insure the amount of crop yield. That is, if a third party (an insurance company) were to insure the expected mean yield, the contracting parties would choose a fixed-rent contract and would be willing to pay the insuring party an amount no higher than the saving from a lower transaction cost plus a premium for the almost certain income now obtained as compared with the variable income in a share contract. Yet we seldom find such a crop insurance without government's taking an active role. The reason, perhaps, is that the cost of handling insurance transactions may be so high as to be prohibitive: the insuring agent would have to check not only the actual crop yields but also the amount of nonland inputs. For the French metayage (sharecropping), however, Constantia Maxwell observed:

The usual procedure for French seigneurs was, while retaining the chateau and its immediate neighborhood for their own use, to let out their lands in gros to middlemen or fermiers (to be distinguished from fermiers exploitants), who paid a fixed sum to the proprietor and gathered the rents from metayage or censitaires at their own risk for a personal profit. Some of these middlemen, like the landlords, were absentees and worked the estate through sub-agents.[16]

In this case we see the fermiers, a third party, interposing between landlords and tenants to provide a more certain income for the former.[17] To my knowledge, no similar arrangement existed in China, though another practice prevailed (see next section). In Japan, share tenancy has been rare; and, at the same time, a compulsory crop insurance system has been enforced.[18]

1. In China, share tenancy reportedly was more frequent in the wheat region than in the rice region. Taking the hectare yield data of wheat and rice crops in Taiwan, we find significantly higher proportional variances for wheat than for rice. This is shown in table 1. Owing to the lack of price data, only the variances of physical output are computed, although value of output would be a more appropriate measure.

In table 1, , where Xi is the hectare yield in kg., and n the number of years. The higher frequency of share contracts among wheat crops appears to be a universal phenomenon.[19]

1. According to three independent surveys conducted in China (1930-35), share rent is generally slightly higher than fixed (crop) rent,[20] and this premium may be regarded as a return to the landowner for risk bearing.

Let me summarize. The postulate of general risk aversion or the minimization of transaction costs, taken separately, do not explain well the observed coexistence of several forms of contracts. For this reason I use both, and the choice of contracts is determined by weighing the gains from risk dispersion and the transaction costs associated with different contracts. Two factors appear to be important in explaining different patterns of contractual choices in different localities. First, different physical attributes of crops and types of climate often result in different variances of outputs in different agricultural areas. Second, different legal arrangements, such as compulsory or subsidized crop insurance, affect the variances of income as well as affecting transaction costs for the contracting parties. An examination of some contractual details in the next section will suggest a third factor: different market arrangements also affect the choice of contractual forms.

[1]. If only outright transfers exist for all resources, "owner" production will exist for all firms. Contracting for outright transfers does not concern us here.

[2]. See Ronald H. Coase, "The Nature of the Firm", Economica (November,1937).

[3]. Portfolio selection is a complicated subject. The two major theses that have been advanced center on anticipated changes in the general price level and on the aversion of risk. Transaction costs may imply a third.

[4]. While this concept has the advantage of treating risk as a measurable quantity that can be conveniently applied to observations, it also has some theoretical difficulties. See, for example, Jack Hirshleifer, "Investment Decision under Uncertainty: Choice-Theoretic Approaches", Quarterly Journal of Economics (November,1965).

[5]. Transaction costs may also depend on other factors, such as the number of participants and transactions, which I shall not explore here. Changes in prices and innovations will also affect the costs of transactions. See, for example, Theodore W. Schultz, Transforming Traditional Agriculture (New Haven: Yale University Press,1964), chap.11.

[6]. An analysis of market information is available in George J. Stigler, "The Economics of Information", Journal of Political Economy (June,1961).

[7]. See Harold Demsetz, "The Exchange and Enforcement of Property Rights". Journal of Law and Economics (October,1964). Demsetz's work constitutes an important reinterpretation of Francis M. Bator, "The Anatomy of Market Failure", Quarterly Journal of Economics (August,1958).

[8]. Similar arrangements are found in consumption. For example, in apartment rentals the cost of utilities is frequently "paid" by an amount added to the apartment rent; restaurants serving buffets allow customers to eat as much as they please after paying a lump-sum cover charge.

[9]. Marginal inequality within a firm, as in the case of lump-sum charges, may also result in marginal inequalities among firms. For example, if the same water resource in other uses is under a contract that stipulates a unit price, the marginal product of water under the lump-sum charges will be lower than that elsewhere.

The marginal quantities under discussion are those which would be equal at every margin should transaction cost be zero. If we ignore corner solutions (as implied in the example of lump-sum charges), it is possible to define a different set of marginal quantities by incorporating transaction costs in such a way that a different set of marginal equalities could be obtained.

[10]. For the situation in China, see J.L. Buck, Land Utilization in China (Chicago: University of Chicago Press,1938), p.198. For Japan, see R.P. Dore, Land Reform in Japan (London: Oxford University Press,1959); for other parts of Asia, see sources cited in chapter 1, notes 10 and 14.

[11]. See chapter 2. Samples of share contracts obtained from China (see next section) are consistent with this statement.

[12]. For the tenant's incentive to use an amount of input less than that stipulated in a share contract, see chapter 3, section C. The adoption of different forms of contractual payment for labor alone owing to "shirking" problems and enforcement costs appears to constitute an important subject which has not been explored. For example, a piece-rate contract will be preferred to a wage contract on an hourly basis if checking output costs less than enforcing input. However, with piece rates the worker is inclined to be "sloppy" and produce products of lower quality. Thus, a piece-rate contract will be less preferable if the physical attributes of the product are such that it is relatively costly to police a specified standard. Similarly, commission payments (as with insurance salesmen) are preferred to other forms when the value of output depends on the intensity of work per sale; "tipping" payments (as in the case of waitresses) are preferred to other forms when the quality of services is significant - in either case, the costs of enforcing "intensity" and "quality" of work appear to be relatively high.

[13]. In horticulture, for example, the usual contracts other than owner cultivation are wage or piece-rate contracts. This may imply that in horticulture, owner management involves a lower cost of policing the orchard assets than fixed-rent contracts. On the other hand, one expects that wage contracts would be infrequent when the landholding is large, for high costs of labor supervision would be incurred.

[14]. This result is implied in William F. Sharpe, "Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk", Journal of Finance (September,1964); Jack Hirshleifer, "Investment Decision: Choice-Theoretic Approaches"; and idem, "Investment Decision under Uncertainty: Applications of the State-Preference Approach," Quarterly Journal of Economics (May,1966). "Risk-exchange" models derived from the current state-preference and mean-variance approaches, with the aid of an Edgeworth-Bowley box, suggest that risk sharing is preferred - if we ignore transaction costs.

[15]. For a theoretical treatment of asset prices and risk premiums as determined in the marketplace, see Sharpe, "Capital Asset Prices".

[16]. Arthur Young, Travels in France during the Years 1787,1788, and 1789, ed.C. Maxwell. (Cambridge University Press,1929), p.395, editor's note.

[17]. To interpret the existence of the fermiers on grounds of risk aversion alone seems inconclusive. Ronald Coase has pointed out to me that the fermiers resembled the "farmers" in England, who served to collect taxes and postal revenues for the Crown. Coase's explanation for the existence of the English "farmers" is as follows: a collecting agent who is allowed to take the difference between what he can collect and what he has to pay the Crown has a greater incentive to maximize receipts than if he is paid a wage rate for his service. This argument, I believe, is correct, and can be expressed alternatively: transaction costs differ, among other things, because different sets of stipulations require varying efforts in enforcement and negotiation; and for collection a "farming" contract involves a lower cost of enforcement than a wage contract. The fermier of France may therefore be viewed as a "farming" agent as well as an "insuring" agent.

[18]. See Takekazu Ogura, ed., Agricultural Development in Modern Japan (Tokyo: Japan FAO Association,1963), chap.13. I have been unable to find the frequency of share contracts in Japan before the introduction of the compulsory crop insurance.

[19]. See, for example, J.L. Buck, Land Utilization in China, p.198; and James O. Bray, "Farm Tenancy and Productivity in Agriculture: The Case of the United States", Food Research Institute Studies, vol.4, no.1 (1963).

[20]. See Appendix B; Department of Internal Affairs, Public Reports of Internal Affairs, vol.2 (1932); Legislative Yuan, Statistical Monthly 2,5 (1930); and Shu-Ching Lee, "The Heart of China's Problem, the Land Tenure Systems", Journal of Farm Economics (October,1946).

### B. Characteristics of Fixed and Share Contracts (China,1925-40)

In this and the following section I shall analyze in some detail the observed stipulations of fixed and share contracts. This will serve not only to clarify the hypothesis that contractual arrangements are chosen to disperse risk bearing and minimize transaction cost, but also to further confirm the theory of share tenancy derived in the last two chapters. I turn to some information from China, roughly from 1925 to 1940. This choice of data is based not only on the availability of information, but also on the fact that during this period in China, some 93 percent of the farm land was held under private ownership.[1] Let me begin by translating a few sample contracts of fixed rent.

Sample a - fixed (crop) rent contract with definite lease duration (Shantung Province):

Tenant A now leases from landowner B [so many acres] of land at location C. We hereby stipulate, with the presence of referee D, that the annual rent per acre includes [so many catties] of wheat, and [so many catties] of millet, soybeans, and Indian corn. The payment in wheat will be one month after the wheat harvest, and autumn crops two months after the autumn harvest. In a famine year, rental payments shall be adjusted [downward] according to local customs. The duration of the lease is [so many years].[2]

Sample b - fixed (crop) rent contract with indefinite lease duration (Kiangsi Province):

We contractually establish an iron-sheet [firmly fixed] rent…… Regardless of good or bad years, not a fraction of rent can be reduced…… In the event that the rental payment is reduced or delayed, the landowner is free to take back the land, together with all existing crops, and to contract a new tenant for cultivation…… Furthermore, the landowner shall pay the tenant 20 copper coins for the delivery of every 100 catties of grains.[3]

Sample c - fixed (crop) rent contract with landowner providing nonland farming inputs (Tsinghai Province):

The landowner will furnish [so many catties] of seed, together with [so many pairs] of water buffalo,[so many head] of donkeys, and all essential farming equipment. The durable assets are for use purposes only, and shall not be damaged or lost [by the tenant]…… and they must be returned to the landowner without delay at the termination of the lease. The [aforementioned] rental rate is subject to adjustment according to local customs in a famine year.[4]

The above samples of fixed (crop) rent contracts are the most representative I could find. They are identical to cash rent contracts in all aspects except that in the latter rental payments are stated in monetary units.[5] According to observations collected by the Department of Real Estates, covering twenty-two provinces in China, cash rents are generally slightly lower than crop rents.[6] This differential can be explained by landowners' sharing in the product selling cost undertaken by tenants. We may also note that with inflation occurring in 1938, owing to the Sino-Japanese War, beginning in 1937, 13.3 percent of cash rents were converted into crop rents and 15.3 percent were converted into share rents.[7] This observation, of course, is consistent with minimizing transaction cost. Under inflation, renegotiation of cash rent contracts becomes more frequent and thus more costly.

The characteristics of fixed-rent contracts are not of special interest except for one feature which we single out to elaborate on here - the frequent inclusion of the provision for rental reduction according to "local customs" in a "famine" year (see samples a and c), a provision which is absent under an "iron-sheet" rent (see sample b). Let us call this provision an escape clause for the tenant, the inclusion of which in a fixed-rent contract imposes a risk burden on the landowner.

We may interpret "local customs" as a set of market prices for "famine" adjustments, even though the exact magnitude of the possible reduction of rent is not stated when the contract is signed. The escape clause comes into play only in a year so "bad" that the market considers it to be a "famine". Given a sufficiently large number of fixed-rent contracts which include the escape clause, competition among landowners to keep their tenants will yield certain market rates of rental reduction which each land-owner will follow. Other things being equal, the increased risk burden on the landowner associated with the inclusion of the escape clause implies that a premium will be added to the "fixed" rent over the "iron-sheet".[8]

Although shifting the risk burden by including the escape clause in a fixed-rent contract is not quite the same as the risk dispersion in a share contract, we may imagine the formation of share contracts via the escape clause. Suppose "famine" is defined as occurring when the actual harvest is reduced to a certain percent of the expected mean yield owing to natural causes. The tenant under fixed rent has the option to choose between agreeing to an "iron-sheet" contract or buying an "escape" right by paying an "insurance" premium to the landowner - such that in the event of "famine", rental payment will be reduced by a certain percentage according to a market rule.

To further the argument: there could exist in the marketplace not just one escape clause as observed, but a wide range of similar clauses each associated with a different level of "famine", such that the tenant could obtain any or several of them by paying different premiums to the landowner. As such, the risk burden could be dispersed between the contracting parties in an infinite number of ways, each with slightly different arrangements. This hypothetical world would perhaps exist if the costs of negotiating and marketing all the different escape clauses were zero. But with increasing transaction cost associated with additional escape clauses - in particular, the cost of defining different levels of "famine" in the marketplace, and the cost of negotiating the rental reduction for each - the incremental gains of having them may be so small that no further "custom" is developed by the market. Instead, an alternative device chosen is a share contract, under which multiple "escape" provisions for the tenant will be implicit, and within which the rental payment is no longer fixed.[9]

From the above one may deduce two implications with respect to transaction cost and risk aversion. First, I have argued that the transaction cost is higher for share rent than for fixed rent, pending empirical confirmation. Observed contractual arrangements in China suggest that the transaction cost for a wide range of escape clauses is higher than for share rent. The reason is that a wide range of escape clauses would allow a greater variety of choice for risk dispersion than a share contract, and yet only one escape clause is observed as available. Thus, the range of contractual choices is constrained by transaction cost. Second, since, as noted earlier, evidence indicates that share rent is slightly higher than fixed rent owing to the added risk burden imposed on the landowner, I conjecture that the landowner's income would be higher than with a share contract if an escape clause were adopted to the effect that the tenant's income variance is reduced to zero. Imaginative as it may seem, we find that such an escape clause exists in the real world, disguised under the name of wage contract.

Available data on the frequencies of escape clause adoptions under different contracts do not refute my suggestions. A survey conducted by the University of Nanking, covering four provinces in China in 1935, reveals that the escape clause (as in contract samples a and c) was adopted in 83 percent of the crop (fixed) rent contracts, 63 percent of the cash (fixed) rent contract, and not at all in share contracts.[10] The higher frequency of adoption for crop rent than cash rent is what we would expect. In the event of a generally poor harvest, the market price of agricultural yield will rise, and with cash rent the tenant's income will be compensated by the rise in price more than with crop rent; hence, the escape clause will be less preferable to the tenant.

The existence of the escape clause in the market implies, other things being equal, a more frequent choice of fixed-rent contracts than of share contracts. Outside China in Southeast Asia, before the agrarian reforms, the escape clause was unpopular. There existed, however, some guaranteed minimum rents or wages associated with share contracts. These guarantees could be similarly analyzed with the suggested choice-theoretic approach if more information were available. The different market practices explain, in part, the higher frequency of share contracts in Southeast Asia than in China. Indeed, the fermiers of the French metayage, the escape clause associated with fixed rents in China, and the minimum guarantees associated with share rents elsewhere are market practices that serve as intermediate arrangements between pure fixed rents and pure share rents. Each of them has different risk distributions and transaction costs, thus widening the range of contractual choices. Why these intermediate arrangements differ in different markets is a question I do not seek to answer.

Turning to sample contracts of share rents in China, we find that their stipulations are more complex than those of fixed rents, owing to the added stipulations on tenant inputs and crops to be grown.

Sample d - share contract with uniform sharing percentages for all crops (Shantung Province):

Tenant A agrees to cultivate [so many mows] of land for landowner B. We hereby stipulate that tenant A provides [so many head] of water buffalo, [so many bodies] of men; and every year the tenant must cultivate wheat once, Indian corn three times, and soybeans twice. Fertilizer expenses are to be shared [in certain proportions]. The yields of all crops are to be shared [in certain proportions]. The lease may terminate only after the autumn harvest.[11]

Sample e - share contract with varying sharing percentages (Honan Province):

[Stipulations of land size and nonland inputs]…… We hereby stipulate that the wheat yield will be split 20-80; millet, yellow beans, sesame, green bean-all will be split 30-70; cotton and sweet potatoes split 50-50……; millet straws, and bean and sesame stalks split 30-70.[12]

Sample f - share contract with some products unshared (Honan Province):

Tenant A…… voluntarily agrees to furnish [so many] men, [so many head] of water buffalo and donkeys…… and all plowing equipment…… We clearly stipulate that seeds of major crops are to be provided by the landowner, and seeds of minor crops by the tenant. All crop yields will be split equally, in dry and clean form…… But the straws go to the [tenant's] water buffalo entirely; the droppings go to the [landowner's] soil; …… and all fertilizer expenses are to be borne equally by both parties. All grinding equipment and living rooms are provided [by the landowner], which the tenant shall repair for his own use. These assets must be returned to the landowner at lease cancellation.[13]

For share contracts, several things should be noted. First, the explicit stipulations of tenant inputs and crop plantations are implied by the theory of share tenancy (see chapter 2).[14] Evidence indicates, however, that only the actual yields are inspected, for by comparison with adjacent farms or past experience the landowner will be able to decide whether the contracted terms have been fulfilled:

The absentee landlords send their agents, or go themselves, to the fields and estimate the yield of the crop and the share given by the tenant is based on this estimate. Such men are very expert in approximating the true yield…… [The tenant] commonly cheats by skilfully hiding some of the threshed grain before division takes place and also by giving the landlord inferior crops. On the other hand, the landlord or his agent often uses a large measure. When the agent collects rent the tenant has to treat the agent very well and often has to bribe him in order to keep the land for cultivation another year.[15]

Exaggerated as this quotation might be, an intramarginal tenant with specific farming knowledge (hence, with higher yields than marginal tenants) can "hide" as much as the rent imputed to his special skill and still retain his tenancy; an agent can collect enforcement cost in "bribes" from both the landowner and the tenant as much as other competing agents allow. Nonetheless, this justifies my claim that transaction cost is higher for a share contract than for fixed rent.

A second characteristic of share contracts is that the precise and at times complex delineation of resource rights between the contracting parties suggests that the sharing of investment inputs can be adjusted along with the rental percentage to use resources efficiently. This is consistent with a conclusion reached earlier, in chapter 2: The landowner may either require the tenant to invest more in land and charge a lower rental percentage, or the landowner may invest in land himself and charge a higher rental percentage. The investment will be made in one way or the other if it leads to a higher rental annuity.

A third characteristic of share contracts is that the rental percentage may vary among different crops in one contract (see sample e). As is implied by the theory of share tenancy, the rental percentage is dependent upon the cost of tenant inputs and the relative fertility of land. Since different crops usually require different ratios of tenant inputs to land, the sharing percentages for different crops should be expected to differ within a single contract. But any set of different rental percentages for different crops can also be expressed in terms of a single (weighted time average) rental percentage, uniform for all crops, to yield the same present value of the rental return. It appears that the latter option of a uniform rental percentage (see contract sample d) would be more convenient. However, if a tenant is subject to dismissal at any time in the event of poor performance, the use of one uniform rental percentage for crops harvested at different seasons would be likely to lead to disputes or renegotiation should tenancy dismissal be in effect. We usually find a uniform rental percentage being used in a share lease with specified duration, and that, when multiple percentages are found in a lease with indefinite duration, a uniform percentage is usually used for different crops harvested in the same season (see contract sample e).

We may summarize the characteristics of share contracts by quoting the observation made by two writers - who were critical of tenant farming in China:

Under the system of share rent, the yields after each harvest are to be shared according to certain mutually stipulated percentages between the landowner and the tenant. With the exception of some land used for farmstead purposes, the tenant is required to cultivate almost all the assigned fields for the production of crops. Sometimes, the tenant is even required to furnish farming equipment…… and other expenses. The landowner and the tenant mutually decide the area to be used for each crop…… Besides the above, the only affair of management over which the landowner exercises control is confined to permanent improvements of land assets. This last characteristic is identical with fixed-rent contracts.[16]

[1]. See J.L. Buck, Farm Ownership and Tenancy in China (Shanghai: National Christian Council,1927).

[2]. National Government, Statistics Department, Statistical Analysis of Tenancy Systems in China (China: Cheng Chung Book Store,1942), pp.52-53.

[3]. Pe-Yu Chang and Yin-Yuen Wang, Questions of Farm Tenancy in China (Chungking: Commercial Press,1943), p.68.

[4]. National Government, Statistical Analysis, pp.54-55.

[5]. See ibid., pp.53-54; and Chang and Wang, Questions of Farm Tenancy, pp.67-70.

[6]. Department of Real Estates, China Economic Yearbook (China,1936), pp.G62-83.

[7]. Executive Yuan, Changes in Land Rights (China,1942), no.2. The data were obtained from sample contracts in fourteen provinces in 1938.

[8]. Unfortunately, I have not been able to find data that would confirm or refute this statement.

[9]. Note that with a share contract the landowner not only shares the possible loss in a bad year, but also the gain of a good harvest which will reduce the risk premium by a fraction.

[10]. See the University of Nanking, Rental Systems in Four Provinces (Nanking,1936), pp.65-67.

[11]. National Government, Statistical Analysis, pp.54-55.

[12]. Chang and Wang, Questions of Farm Tenancy, p.63.

[13]. Ibid., pp.63-64.

[14]. Sample share contracts had not been available to me when chapter 2 was written. When I predicted that, according to economic theory, tenant inputs are stipulated in a share contract, I did not mention the stipulation of multiple crop plantations because only one mutually agreed product was assumed.

[15].J.L. Buck, Chinese Farm Economy (Chicago: University of Chicago Press,1930), pp.149-50.

[16]. Chang and Wang, Questions of Farm Tenancy, p.49. For similar observations see Ching-Moh Chen, Land Rents of Various Provinces in China (Shanghai: Commercial Press,1936); Chi-Ming Chiao, A Social and Economic Study of Farm Villages in China (Nanking: University of Nanking,1938), chap.9; and China Economic Research Department, Source Materials of Recent Chinese Agricultural History,1912-27 (Peking: United Book Store,1957), pp.89-95.

### C. The Duration of Lease Contracts

An investigation conducted in China (1934), covering a total of ninety-three prefectures in eight provinces, shows that the distribution of lease durations was as follows: 29 percent of the tenant contracts were indefinite (that is, unspecified and usually terminable after every harvest), 25 percent annual leases, 27 percent from three to ten years, 8 percent from ten to twenty years, and 11 percent were perpetual leases.[1] Two things should be noted. First, a stipulated lease duration means only that tenancy may not be terminated as long as the contracted terms are fulfilled by each party. That the duration of the lease is specified does not prohibit mutual renegotiations within the lease duration. Second, as the frequency of short-term leases has been used to illustrate the turnover rate of tenancy, it should be pointed out that lease termination is not the same as tenancy dismissal. Available data reveal that the frequency of tenancy dismissal was not high.[2]

In the literature on land tenure, two arguments have been commonly used in support of the alleged inefficiency of lease durations of less than ten years.[3] One of them claims that short durations impose insecurity on the tenant and thus impair his incentive to farm. But insecurity, although undesirable for the tenant, may provide a stimulus to farming activity.[4] Another argument is that the short-term lease discourages investment in land. But this is refuted by the fact that yields per acre on tenant farms are not lower than on owner farms; nor has any evidence been offered to show that, in China, productivity under tenancy varies with the duration of a leasing contract.

The right to each privately owned resource is, by definition, transferable and exclusively delineated. Rights to resources invested in land and other assets are no exception. In the formation of a lease contract, the participating resource owners are free to accept or reject the contractual terms being negotiated. Again, the choice of the duration of the contract is no exception. Thus the relevant question here is not whether a "short-term" lease is inefficient; the relevant question is why different lease durations are chosen.

In a world uncomplicated by transaction costs and risks, in which the right to the income generated by private investment could be costlessly secured and transferred, and in which changes in contractual stipulations could be costlessly negotiated at any time, the duration of the lease becomes irrelevant and its explicit stipulation superfluous. With transaction costs included, I argue that the lease duration will be chosen to minimize these costs. To do so, it is convenient to separate the cost advantages of "long" and "short" lease durations.

The Choice of Relatively Long Lease Duration

A relatively long lease duration is chosen to reduce the cost of transferring (transacting) tenant assets attached to land. There exist differences in physical attributes of capital assets which involve different moving costs at lease dismissal. For example, a water buffalo owned by a tenant for grain grinding is easier to move at lease dismissal than an improvement in water irrigation made by him. Of course, the landowner could have invested in the water irrigation himself, or he could purchase the tenant's committed improvement outright.[5] But when assets attached to land are owned by the tenant, disputes may arise in the event of tenancy dismissal. A lease with a sufficiently long duration may become the preferred option.

However, the cost of moving the physical asset is not necessarily the relevant cost to consider. The tenant's property right to his committed investment may be transferred, either to a third party or to the landowner, at a market price. The problem is that such a price may not exist, or cannot be obtained in a short period of time, due to transaction costs. One need only point out that the depreciated value of a used asset is costly to evaluate; the landowner may choose to select his new tenant instead of allowing any party who purchases the asset to take over the lease. Also, other information problems exist in the marketplace. An appropriately long lease duration will thus reduce disputes and the anticipated cost of transferring the tenant's property right. This choice, however, can be made only at the expense of some cost advantage which a shorter lease duration provides.

The foregoing discussion can be supported by observations on the perpetual lease in China:

Under perpetual leases the landowner holds ownership right to the [bottom of] land, and the tenant owns the right to the soil…… These two rights are separate. The occurrences of perpetual leases are confined to the following: (1) the tenant exploited [privately owned] wasteland and developed it into farm land, thus gaining a perpetual [ownership] right to the soil from the landowner. (2) Permanent improvements in land made by the tenant…… such as building up water-conserving devices in otherwise sandy fields…… (3)…… where labor is scarce and land plentiful, the landowners attracted tenants from afar by offering the perpetual right to till [the soil]…… (4) The tenant had paid a lump-sum payment to obtain the perpetual right to till…… And (5) the peasant, when in need of money, sold the ownership right to the land bottom but retained the right to till the soil. Since ownership rights to the bottom and surface of land are separate, both the landowner and the tenant can sell their rights freely, without the consent from each other.[6]

In every instance, the tenant's asset attached to land (for example, the right to the soil) is physically "permanent". With the perpetual lease and the contracted terms in effect, the landowner may not arbitrarily raise the "bottom" rent (or use other devices) to drive the tenant away. Yet such a lease duration would not be necessary if transaction costs were zero: If the "bottom" and "surface" rights were clearly delineated and costlessly enforced as private, and if these rights could be costlessly transferred, there would exist market prices for these rights at which transfers could be executed at any time.[7] Thus there would be no need for long lease durations to protect the "immobile" investments of the tenant. The same can be said for other assets attached to land.

The Choice of Relatively Short Lease Duration

The adoption of a relatively long lease duration involves forgoing some cost advantage which a shorter duration provides. When assets attached to land owned by the tenant are to be exhausted in a short period of time, or when the landowner provides all the "permanent" assets, a relatively short lease duration reduces the costs of enforcing the contracted terms and of renegotiating these terms.

When a contract is formed, the contracting parties may lack sufficient information on each other's reliability. Within a specified lease duration, the violation of the contracted terms by either party may call for increasing enforcement efforts, or for revoking the contract before its termination date through court action or other means - all to be done at some cost. The choice of a shorter lease duration, which facilitates tenancy dismissal, will reduce these costs. As was noted at the beginning of this section, however, the frequency of tenancy dismissals was far less than that of short-term leases, suggesting that most terminated leases were renewed. Available data show that the frequency of lease dismissals caused by rental disputes was low.[8] I conjecture, therefore, that short-term leases are chosen more as a device to facilitate contractual renegotiation than as a device to reduce the costs of enforcing the contracted terms.

It is useful to distinguish two types of contractual renegotiation (revision), though at times one relates to the other. The stipulated terms in any tenure contract in essence specify two things: (a) the state of resource use, or allocation, mutually agreed upon by the contracting parties, and (b) the contracted distribution of income for the parties. To revise (a) through renegotiation for more efficient resource use may benefit all parties to a contract, that is, all parties may gain or lose less. To revise (b), however, one party must lose.

Consider the contractual renegotiation which entails mainly a reallocation of resources; for example, changes in relative product prices which call for shifts to different crops, or innovations which call for the adoption of new seeds or new methods of cultivation. Renegotiations of this type are largely confined to share contracts, since under fixed rents the tenants are left to make their own decisions on resource use except improvements in land and maintenance of the landowner's assets. In principle, since all contracting parties expect to benefit from the revision, renegotiation can take place at any time and lease termination becomes unnecessary. But different individual knowledge of the market may give rise to difference in opinions as to whether the revision is desirable. A relatively short lease duration is a convenient device which allows resource reallocation in the event of unsuccessful renegotiation.[9] This, together with the more complicated contractual enforcement required for share contracts, explains why durations of share leases are generally shorter than those of fixed rent.[10]

Consider further the contractual renegotiation which entails the revision of income distribution, when one party gains at the expense of the other. It applies to fixed and share contracts alike. Resource allocation may also be affected. For example, changes in relative asset prices of the contracted resources, a cash-rent contract with unanticipated inflation, or decision errors made in the initial contract - which call for a revision of the rental rates - are examples in point. Since some party must lose when the initial distributional terms are revised, that is, the gainer either cannot or will not fully compensate the loser in making the revision, lease termination (hence, the choice of an appropriately short duration) is essential.[11] Again, this would not be necessary if transaction cost were zero (even if unanticipated events occur independently). In the absence of transaction cost, a contract would be designed to allow day to day changes in rental payments; within any lease duration, the distribution of income would not be held fixed throughout.

[1]. These percentages are computed from Department of Real Estates, China Economic Yearbook (Shanghai: Commercial Press,1935), pp.G101-4. A similar investigation conducted in the same localities ten years earlier yielded an almost identical distribution (ibid.).

[2]. According to a survey conducted by the Executive Yuan (Changes in Land Rights), covering fourteen provinces in China (1937), 7.5 percent of the lease contracts were dismissed in that year. But, since inflation began in the same year, the cited percentage might be higher than that of preceding years. For the rise in prices owing to the Sino-Japanese war, see Chang and Wang, Questions of Farm Tenancy, chap.9.

[3]. See sources cited in chapter 1, notes 10 and 13.

[4]. Armen A. Alchian has argued that the desire for security leads to "long-term" contracts. But his analysis is based on a property right system with is not private, where the private cost of acquiring security is relatively low. See his "Private Property and the Relative Cost of Tenure", in The Public Stake in Union Power, ed. Philip Bradley (University of Virginia,1959), pp.350-71.

[5]. Two independent surveys (China,1921-24 and 1935) reveal that, among tenant farms, landowners owned about 60 to 70 percent of the housing assets; tenants owned about 75 percent of the draft animals and 95 percent of the farming equipment. The total values of nonland assets on owner and tenant farms were roughly the same. See National Government, Statistical Analysis, pp.99-116.

[6]. Chiao, Study of Farm Villages, p.261. For similar observations see National Government, Statistical Analysis, pp.56-58; and China Economic Research Department, Source Materials, pp.84-89.

[7]. I apply here the thinking in Ronald H. Coase, "The Problem of Social Cost", Journal of Law and Economics (1960).

[8]. Legal records which cover fifty-six prefectures in six provinces (China,1934-35) reveal a total of 124 tenancy disputes (mostly in rental payments) over a one-year period. Even though the total number of tenant contracts is not available, the number of disputes brought to court appears to be so small that one suspects many more never reached court. Over two-thirds of these recorded cases ended in tenancy dismissals, together with payment settlements. See Department of Real Estates, China Economic Yearbook, (1935), pp.G118-20; and (1936), pp.G143-44.

[9]. With lease termination, for example, a share tenant who alone wants changes in the production plan can request a fixed-rent contract, purchase the land outright, or seek tenancy with another landowner. Without lease termination, further negotiation may still take place if one party who wants the revision pays the reluctant party an amount to make the revision "convincing".

[10]. Localities with higher frequencies of share leases (China,1934) were associated with higher frequencies of short-term leases. See National Government, Statistical Analysis, p.43, tables 20 and 21; and p.59, table 26.

[11]. Given an unexpired lease which fixes the rental rate, changing economic conditions may lead to a redistribution of income. But the efficiency of resource allocation may not thereby be hindered.

### D. Concluding Remarks

Every transaction involves a contract. The transactions conducted in the marketplace entail outright or partial transfers of property rights among individual contracting parties. These transfers may be negotiated through many different contractual arrangements.

For generations economists and land tenure writers have sought to rank the relative efficiency of resource use under different leasing arrangements. But their inquiries were undertaken without explicit reference to the property right constraint involved. And in many instances, the characteristics of various lease contracts had not been carefully examined. Different contractual arrangements do not imply different efficiencies of resource allocation as long as property rights are exclusive and transferable. The characteristics of lease contracts presented above also confirm this statement.

In this chapter I have asked: Why are different contractual arrangements chosen under the same system of private property rights? To answer this question I have introduced transaction costs and risks. The attempt to formulate a choice-theoretic approach to explain the observed contractual behavior in agriculture has perhaps raised more questions than it has answered. And I have been unable to piece together fragments of analysis into a formal theory: the problems in the theory of choice involving transaction costs and risks are still formidable.

Although the presence of transaction costs or risks may result in different intensities of resource use, as was noted earlier in this chapter, available data do not reveal notable differences in farming intensities for various leasing arrangements under private property rights.[1] The reasoning is as follows. The main forms of tenure arrangements considered thus far include owner cultivation, wage contracts, fixed rents, and share contracts. For these arrangements, economic theory implies a tendency toward the same set of marginal equalities of resource use even if transaction costs exist. To obtain a notable tendency toward marginal inequalities, transaction costs would have to be so high as to yield, for example, the choice of a contract with a lump-sum charge and without quantifying and pricing resource units. Lump-sum contracts, however, have not been important in Asian agriculture and can, therefore, be ignored. Instead, the effects of transaction costs and risks are revealed mainly in the observed choice of contractual arrangements and, less notably, the risk premiums distributed among the contracting parties.

Among some related problems that I have explicitly avoided, the following are significant. First, with respect to risk aversion, a more general analysis would include all risky choices, not contractual choice alone. The analysis would be less difficult if transaction costs were not involved. Second, with respect to transaction costs, a more general analysis would derive some specific and well-behaved cost function of transactions. This step is essential to the development of a model of general equilibrium including transaction costs.

Still other problems I have avoided implicitly. In particular, some level of law enforcement by legal authorities is taken for granted. One may well ask: What will happen to the choice of contracts if the government changes its enforcement efforts? To what extent will these efforts be consistent with the Pareto condition? What set of legal institutions is consistent with the operation of the marketplace? With these questions unanswered, the conditions defining efficiency with transaction costs are not all clear. Let me explain.

In production, cost minimization requires not only the fulfillment of the familiar set of marginal equalities, but also the choice of the lowest-cost production method available. In transactions, one relevant consideration is the cost of alternative contractual arrangements, which I have discussed at some length. One might think that, as a cost constraint, efficiency will be attained when, other things equal, the set of arrangements with the lowest transaction cost is chosen. But transaction costs also depend on alternative legal arrangements. For example, the varying effectiveness of law enforcement, or the varying corruptibility of courts, will affect the costs of transactions in the marketplace. Given the existing legal institutions, I have attempted to explain the observed contractual arrangements. But insofar as I have ignored the choice and development of the legal institutions, the Pareto condition with transaction costs is ambiguous.

I have also not explored the contractual behavior associated with different property right constraints. Various restrictions on the transfer of property rights, or various methods of attenuating the right of a resource owner to obtain income from his resource, will affect the leasing arrangements as well as resource allocation. In the remaining chapters I analyze one important constraint, namely, the legal limitation on the landowner's rental income to a maximum percentage of the annual yield.

[1]. One exception is that we find that housing assets in tenant farms were generally of lower value than in owner farms (National Government, Statistical Analysis, pp.100-1). It may be because housing is also a consumption item and tenants were typically poorer.

## 5. Transfer Effects of Rental Share Restriction: Hypothesis and Evidence of Offsetting Contractual Rearrangements

In this and the following three chapters, I will turn to the first phase of Taiwan land reform. Beginning April 1949, a set of regulations on agricultural land lease was promulgated and enforced by the Taiwan National Government. Under this phase of the reform, the rental percentage was reduced from an estimated mean of 56.8 percent of the annual yield to 37.5 percent.[1] This was applied to all farm rentals (fixed and share rents alike) except horticulture. The magnitude of farming resources affected by this enactment is shown in table 4, chapter 8. The relevant legal provisions will be discussed in the proper context in this and the next chapter.

The difficulty of analysis seems to stem from the problems of (1) identifying various legal constraints, both those that are vaguely implied and those explicitly stated in law; (2) specifying the options of choice; and (3) ascertaining the effectiveness of various legal constraints. The same control device under different legal authorities, for example, may give rise to different outcomes simply because the interpretations of law differ, or the efforts of enforcement differ. The above conditions must be specified before one can analyze the observable consequences theoretically and empirically. This I shall try to do.

I begin the analysis by presuming that the rental percentage alone is legally and effectively reduced. Other restrictions will be considered later. It is useful to classify the predictable events as implied by economic theory into two main groups. The first group are offsetting contractual rearrangements, which include compensating payments and changes in tenure arrangements through the transfer of rights. The second are the resource reallocations. In this chapter I shall concentrate on the first.

### A. The Hypothesis of Offsetting Contractual Rearrangements

An offsetting contractual rearrangement is defined as a contractual revision which, in the absence of transaction costs and risks, produces no effect on the initially contracted resource allocation and income distribution. A legal restriction on the rental percentage will induce a variety of contractual rearrangements between the contracting parties to restore the equilibrium that had been arrived at by the market terms of contract, unless these rearrangements are prohibited by law. These rearrangements may be regarded as transfer effects resulting from the rental share restriction.

Suppose an unrestrained market rental percentage of 70 percent of the annual yield is legally restricted to 40 percent. If no adjustment whatsoever is made, the result of this percentage reduction is a redistribution of income from the landowner to the tenant, an outcome which Taiwan reformers had intended. Without reallocating resources, what will the contracting parties do?

Compensating Payments

As a device to attenuate private land rights, the restriction on rental percentage alone affects little, if anything at all. With the ownership of land exclusively defined, there are several revisions in the form of payment that the landowner could use to protect his wealth position without violating the restriction. (1) The simplest of these is to charge lump-sum annual payments offsetting the reduction in rental percentage. This can be done in a variety of ways. One way is to charge a full lump-sum in the form of "key" money or in the name of a security deposit. If the duration of lease is contracted for several years, the landowner may require the tenant to pay a present lump-sum equal to the discounted value of the reduced annual rental income for the contracted years. (2) If, before the rental share restriction, part of the nonland farming cost had been borne by the landowner, he may simply require the tenant to pay the full cost. (3) It had been a practice in some share contracts that a landowner allowed the tenant to use part of the land to grow some "cereals" for the tenant's personal consumption, which were not required to be shared; instead, the owner would charge a higher rental percentage of the stipulated crops. Under the rental share restriction, the landowner may share in all crops alike. Depending on the magnitude of the percentage restriction and the original terms of the contract, it is clear that the last two methods combined may not be sufficient to restore the initial situation.

Tenure Rearrangements

Suppose compensating payments were also legally prohibited. The landowner could protect his wealth position by changing tenure arrangements through a transfer of property rights. This can also be done in a variety of ways. (1) He may simply repossess the tenant's holding and farm the land himself, or hire laborers to do the tilling. (2) Suppose the lease duration contracted is such that the landowner cannot cancel the lease for land repossession. He will be willing to offer a fee to the tenant to buy back the lease right. (3) Alternatively, the landowner may sell his holdings outright to his tenants, at a price per acre equal to the discounted present value of rent per acre before the share restriction. (4) If a tenant does not have sufficient funds to pay the land price immediately, an installment plan may be made whereby he pays the landowner no more than the rental annuity. The transfer of the ownership of land to the tenant, however, is a clear restoration of the initial wealth condition only if the price of land does not fall, that is, if there is no specialization in risk bearing or management, or if there is no decline in speculative demand to hold land as a result of the land reform.

Sufficient adoption of the above offsetting contractual rearrangements, independently or jointly, will restore the resource allocation and income distribution under unregulated tenancy. The change in tenure arrangements is sufficient, for its occurrence requires no other forms of revision. On the other hand, compensating payments may or may not be sufficient. Not only may the three forms of compensating payments take place jointly, but they may take place along with adjustments in production. In a world complicated by transaction costs and risks, these contractual rearrangements will yield lower values for the contracted resources than previously. This is so because if the new arrangements involve lower transaction costs or more preferable risk distributions, they would have been chosen before the rental share restriction. To what extent the higher transaction costs or less preferable risk distributions will affect resource allocation is a question which I do not pursue here.

It is clear that the occurrence of events other than offsetting contractual rearrangements, namely, resource reallocation, depends upon (1) tenure rearrangements not taking place in all lease contracts which are affected by the rental share restriction, and (2) compensating payments not taking place in all contracts to the extent of restoring the initial equilibrium. After the enforcement of a uniform 37.5 percent maximum rental percentage in Taiwan in 1949, it is evident that these rearrangements occurred. According to the evidence presented in a later section, however, they occurred only in a small portion of the tenant contracts. This is due to other legal restrictions, and perhaps more important, as we shall see in the next chapter, due to adjustments in resource allocation.

[1]. It is not clear whether the estimated mean of 56.8 percent is weighted.

### B. Legal Restrictions on Offsetting Contractual Rearrangements

Rental share restriction in Taiwan involved legal provisions set at three different dates: the prereform land law before 1949; the regulation of rental percentage of 1949; and a new rent reduction act of 1951. Some of the prereform provisions governing land lease have already been discussed, in chapter 1. The reduction of the rental percentage to a maximum 37.5 percent of the annual yield, uniform for all tenant contracts, was promulgated and enforced in Taiwan on April 14, 1949. The regulations of that date, together with the existing land law, limit the possible scope of the offsetting contractual rearrangements discussed earlier. The intention of Taiwan officials and their efforts to enforce the 1949 rental reduction are evident.

However, the 1949 regulations must not be confused with the Farm Rent Reduction Act promulgated on June 7, 1951. But the rules for enforcing the new 1951 act were not established until February 2, 1952, and after this date there exists little evidence that the additional provisions in the new act were really enforced.[1] It is interesting to point out that literature published after 1951 which discusses the 1949 farm rent reduction usually refers to the act of 1951 as if it were the 1949 regulations.[2]

In the present study we are primarily concerned with the regulations as of 1949, which include those of the prereform land law and the regulations of 1949. The changes in regulations made in the act of 1951 only interest us here as indirect evidence for hypothesis testing. Thus in every instance the specific date of the regulation will be indicated.

Restrictions on Compensating Payments

The regulations established prior to and including 1949 restricting compensating payments encompass the following:

(1) The lessor of farm land shall not collect farm rent in advance.… Security deposit shall not exceed one-fourth of the annual rent… which (together with interest) shall be deemed to be part of the farm rent [3] [prereform]. (2) If, according to arrangements originally made, the lessor is to supply the lessee with draft animals, seeds, or other implements of production, the lessor shall, after the enforcement [of the rental percentage reduction], continue to make such supply [4] [1949].

In (1) above, payments in the form of security deposit or "key" money are prohibited. In (2), payments exacted by requiring the tenant to pay the full nonland farming cost —in the event that the landowner had contracted to supply a part —are restricted. But the term "other implements of production" is ambiguous. So an appended provision in 1951 states that:

The farmhouse of the lessee which has been originally provided unconditionally by the lessor shall continue to be used by the lessee… and the lessor shall not… charge any fee therefor.[5]

And we shall see in the next section, disputes also arose as to who should bear the water fee in certain cases. In spite of this, it is clear that compensating payments were restricted by (1) and (2). The third possible type of compensating payment, where the landowner shared in all crops alike in instances when certain "cereals" were originally contracted to be consumed entirely by the tenant, was, as we shall see in the next chapter, only implicitly restricted. Nonetheless, the potential extent of this form of compensating payment is very small, and insignificant compared to the reduction in rental percentage.[6]

Restrictions on Tenure Rearrangements

The landowner's right of cancelling the lease and repossessing the land was conditionally restricted:

The lessor shall not terminate the lease contract except in accordance with the provisions of… the (pre-reform) Land Law.[7] [1949]

And in the prereform land law, we find:[8]

Art. 109: Any farm lease contract for a definite period shall, unless the lessor takes back the land for his own cultivation on the expiration of the contractual period, be deemed to have been renewed.

Art. 114: Any farm lease contract for an indefinite period may be terminated… where the lessor takes back the farm for his own cultivation.

However, the term "own cultivation" is vague and may be interpreted to include the hiring of farm hands for commercial farming. Vigorous enforcement followed immediately in 1949 to prevent the repossession of land for commercial farming. Furthermore, new contracts were later required to be no shorter than three years. This minimum lease period was extended to six years in the new act of 1951.[9]

The other possible form of tenure rearrangement, that the landlord sell his holdings outright, was not significantly restricted in all provisions prior to and including 1949. Not only could the owner sell his land to his tenant, but he could also sell to a third party. Although a tenant's preferential right to purchase from his landlord was expressed in the prereform land law, the tenant had to be willing to accept "the same terms as are offered to any other person".[10]

From the above we arrive at the following conclusions: (a) Compensating payments were restricted in regulations up to 1949, and the alternative that the landowner might share in "cereals" after the percentage rent reduction, though feasible, was not substantial. And, (b) tenure rearrangements were conditionally restricted if the landowner repossessed land from the tenant. But the owner's right to sell his holdings outright was not restricted.

Two additional remarks are called for here. First, although compensating payments to a landowner in the form of the tenant's paying the full nonland farming cost were restricted (in the event that the landowner had contracted to pay a part), there existed no restriction on the landowner's requiring the tenant to commit additional farming inputs. Second, although repossession of land from a tenant through lease cancellation was conditionally restricted and subsequently vigorously prohibited, partial repossession of land which entailed no lease cancellation was not restricted in the period under study. I shall discuss these points more fully in the next two chapters.

[1]. The Land-to-the-Tiller program was at this time in the making and drew the main attention of the land reformers.

[2]. A fuller explanation for this will be provided in the next chapter.

[3]. Art. 112, in Cheng Chen, Land Reform in Taiwan (Taipei: China Publishing Co., 1961), p. 155.

[4]. Art. 5, in Hui-Sun Tang, Land Reform in Free China (Taipei: JCRR, 1954), p. 222. It is under "Regulations Governing the Lease of Private Farm Lands in Taiwan Province".

[5]. Art. 12, in Chen, Land Reform in Taiwan, p. 193.

[6]. Only some tenants had been contractually allowed to grow such "cereals" for home consumption, and only on marginal land. Note that in none of the rental disputes on record under the share restriction is this third type of compensating payment mentioned at all.

[7]. Art. 9, in Tang, Land Reform in Free China, p. 222.

[8]. Chen, Land Reform in Taiwan, pp. 154-55. These two articles were cited in chapter 1 of this study.

[9]. Art. 5, ibid., p. 191.

[10]. Art. 107, ibid., p. 154.

### C. Evidence of Illegal Compensating Payments and Tenure Rearrangements

Since under the share restriction most offsetting contractual rearrangements were prohibited by law, the hypothesis that compensating payments and tenure rearrangements would follow from the maximum share control will be tested by a review of the evidence as to various illegal practices and law enforcement. It is not easy to determine just how comprehensive the enforcement was, or how arbitrary. We know the following facts. (1) The 37.5 percent rental rule was widely publicized in every town and village in 1949, and official contract forms were provided [1]. (2) The enforcement was intensified as illegal practices increased:

The inspection work was confined to harvest times and random sampling of tenant families and their farm leases was made to find out whether rent was paid according to the reduced scale. About 126 county workers and 26 provincial inspectors covered the Province in two months after each harvest. This inspection pattern… became ineffective when the number of rent disputes became alarmingly great. JCRR and Provincial Land Bureau, therefore, designed a new supervision pattern using full time employees with full pay.[2]

(3) A total of 7,325 field workers, as well as other resources, were employed for the enforcement.[3] And (4) by 1951, the inspection work was no longer based on random sampling but covered the whole population affected by the share restriction, and most of the illegal practices were soon suppressed.[4]

Compensating Payments and Rent Disputes

Higher water fees were one form of compensating payment in 1949:

Though tenant farmers were obligated to supply labor for improvement of water conservancy, they were now asked to share in the payment fee after the improvement had been completed. This was one device by which landlords sought to reduce their loss from the rent reduction.[5]

This occurred because the responsibility for payment of water rights fees had not been clearly defined:

According to the old regulations… , landowners paid a special water fee which was used for farm improvements and tenants paid an ordinary water fee to cover irrigation costs to the local water conservancy associations which were composed exclusively of landowners. However, no clear demarcation and definite percentage for these two fees were made. Landowners therefore frequently shifted the burden of special water fee to the tenants.[6]

The resulting water-fee disputes, however, were soon settled by the Provincial Land Bureau and Water Conservancy Bureau. Under the new regulation, water charges by landowners in excess of the stipulated terms were reported in 107 cases between June 1949 and March 1950.[7]

Inspections —initially by random sampling and later of the affected population in total — conducted for a two-year period up to June 1951 revealed other forms of compensating payments: forty-two cases in which landowners charged deposit money with no refund, forty-seven cases in which the landowners had collected rent on farm houses over and above the original stipulations, and 537 cases of black-market rent payments. A more frequent cause of rental disputes (6,328 cases) was recorded in an item called "harvest loss", which resulted mainly from disagreement between the contracting parties on the "actual" harvest yield. Nearly all these cases were settled by the Provincial Land Bureau.[8]

Tenure Rearrangements

Lease cancellations by landowners constituted the most serious problem confronting the reform officials. In the words of Cheng Chen:

After the reduction of the rental percentage… a minority of landowners… threatened the tenants with lease cancellation.… Usually with cash, goods, or part of the cultivated land, the landowners bribed the tenants to voluntarily terminate the lease. This bribe money is known as the "fee for lease right". The tenants, in fear of the owners' power, were obliged to acquiesce. After the repossessions of land, the landowners seldom cultivated the lands themselves. They usually leased them at black-market rates, sold them, or hired farm hands to do the tilling, so as to protect their illegal unscrupulous profits.[9]

In the two-year period up to June 1951, disputes arising from lease cancellations (dismissals) amounted to 16,349 cases, of which 8,877 cases involved actual lease dismissal.[10] The official settlements of lease cancellations were troublesome, since according to the law the landowners were allowed to cancel leases upon contractual termination and for owner cultivation. The result was that 1,971 lease cancellations were officially granted, and the remaining leases were restored to the tenants.[11] The area of land associated with these granted cases amounted to 1,376 chias, or about 0.04 percent of the total area of land under tenancy before the share restriction.[12]

During the same period (June 1949 to June 1951), the area of farm land purchased by tenant farmers from landowners amounted to only 2.2 percent of the total area of land under tenancy.[13] This form of transfer was allowed by reform authorities in the expectation that it would lead to more equal distribution of land. It is not easy to understand why outright transfers of landowner-ships to tenants were so few. The answer to this seems to lie in the drastic fall of land prices. From 1948 to 1951, land prices measured in rice for various land grades fell by a range of 46 to 62 percent [14] —proportional decreases greater than the rental share reduction. This means that the rental income to the landowner under the share restriction, even in the absence of compensating payment or resource reallocation, would be higher than the interest return from selling outright. The drastic fall in land prices may be partly explained by the specialization in risk bearing or management associated with alternative tenure arrangements. It may have been due also to a fall in the speculative demand to hold land, generated by the expectation of other reform measures forthcoming.

Conclusions

The evidence of compensating payments and tenure rearrangements presented in this section illustrates the offsetting contractual rearrangements derived in section A. These are revisions through which the initial equilibrium of a lease contract may be restored under the rental share reduction. In spite of all the claims that these rearrangements (mostly illegal) were "alarming", available sources indicate that they occurred in no more than 10 percent of the tenant contracts affected by the rental restriction, and the greater part of these illegal practices was suppressed and corrected according to law. In view of the fact that the number of farm leases affected by the rental share restriction totaled 380,473, and considering that the rental percentage was reduced from an estimated mean of 56.8 percent of the annual yield to 37.5 percent, one wonders why compensating payments and tenure rearrangements were so few. Effective enforcement of law is one explanation, but it hardly seems sufficient. Indeed, the enforcement of the share restriction in 1949 was regarded by the lawmaker as "incredibly smooth".[15] Why? Suppose we ask a seemingly unrelated question: If compensating payments and tenure rearrangements do not take place to the extent of restoring the initial resource use, what adjustments will we observe in resource allocation? As it turns out, the answer to this helps, in part, to explain why the illegal practices were infrequent: the landowner could protect his rental income from land after the percentage reduction, without violating the law, through adjusting farming intensity.

[1]. See Cheng Chen, Records of Taiwan Land Reform (Taipei: Chung Hwa Book Co., 1961), chap. 2.

[2]. Sino-American Joint Commission on Rural Reconstruction, "JCRR Annual Reports on Land Reform in the Republic of China", Composite volume, mimeographed (Taipei: Farmer's Service Division, 1965), p. 9.

[3]. Ibid., chap 2.

[4]. See Chen, Records, chap. 2; JCRR, "Annual Reports", chap. 2; and Taiwan Provincial Land Bureau, "Implementation of the 37.5 per cent Farm Rent Limitation Program", mimeographed (Taipei: JCRR Translation, 1950).

[5]. JCRR, "Annual Reports", p. 9.

[6]. Ibid., p. 10

[7]. TPLB, "Implementation", pp. 24-25, table 5.

[8]. JCRR, "Annual Reports", p. 33. As of June 30, 1951, 184 cases were listed under "To Be Settled". It is impossible to determine the exact number of illegal compensating payments which occurred. Divergences exist in different sources. Closer inspection, however, reveals that their differences are largely superficial. They differ in their criteria of counting: one source counts all clerical errors in lease contracts as illegal; another disregards them as immaterial. The cases cited above were those in which compensating payments had actually occurred. The greatest difference is seen in comparing JCRR, "Annual Reports", p. 33, and TPLB, "Implementation", pp. 24-25, table 5. In the latter, illegal practices (including tenure rearrangements) amounted to 41,994 cases (the highest number on record), but over 80 percent of these cases were due to improper wordings or signatures used in the contracts. Furthermore, if several errors were found in one contract, all were counted. These clerical errors are understandable because most of the tenants were illiterates and were prone to make errors on the written contracts now officially required. The JCRR, on the other hand, did not consider these as substantial and disregarded them.

[9]. Chen, Records, p. 30.

[10]. See ibid., and JCRR, "Annual Reports", pp. 29, 32, table 6.

[11]. JCRR, "Annual Reports", p. 32, table 6. Some of these granted cases were a result of official errors (ibid., p. 28).

[12]. Ibid. The total area of private land amounted to 672,360 hectares. Before the share restriction, 56 percent of these were under tenancy. See chapter 7 of this study.

[13]. TPLB, "Implementation", pp. 24-25, table 5.

[14]. See Chen, Records, p. 41; and JCRR, "Annual Reports", p. 27.

[15]. See Cheng Chen, An Approach to China's Land Reform (Taipei: Cheng Chung Book Co., 1951), p. 24.

## 6. Allocation Effects of Rental Share Restriction: The Hypothesis of Increased Farming Intensity

In the preceding chapter we concluded that offsetting contractual rearrangements through which the initially contracted resource allocation and income distribution may be restored occurred only sparingly in Taiwan after the rental percentage reduction in 1949. One explanation is the comprehensive legal restrictions and enforcement of law. Another explanation, which will be explored in this chapter, is an associated increase in output under the share restriction which partially compensates the landowner for reduced income from land. In practice, the landowner may be compensated both by an illegal compensating payment and by an increase in output.

As we derive here the hypothesis of increased farming intensity, for analytical convenience let us suppose that under the reduction of the permissible rental percentage, no illegal compensating payment is made. The implications of this hypothesis can be summarized. Under the share restriction, nonland resources will be directed to tenant farms from all other uses. The marginal product of land in these farms will then be higher, and the marginal products of tenant inputs will be lower, than those of similar resources employed elsewhere. In other words, the marginal products of tenant inputs will be lower than the associated marginal opportunity cost.

Suppose the market (unrestricted) rental percentage of 70 percent (r=0.7) is restricted by regulation to 40 percent ( = 0.4). The landowner now in the true sense receives 40 percent, r, of the annual yield with no other form of compensation. The tenant is thus getting from the total annual product twice his initial share of 30 percent. Will an existing tenant, therefore, offer to commit more farming inputs? And if so, will this lead to more efficient resource use?

The unrestricted market rental percentage, recall, is reached when the marginal tenant (nonland) cost of farming equals the marginal product of tenant (nonland) input. Thus, any further increase in nonland farming input will not pay the tenant. To illustrate, suppose, with the restricted = 0.4, the tenant commits an additional ﹩10 of farming input, the corresponding gain in product will be less than ﹩10. Therefore, the tenant will receive an additional income of less than ﹩6 with the reduced rental percentage (which is obtained by using the amount of less than ﹩10 times 1 — ). Indeed, the tenant would not voluntarily offer to commit additional farming input to the area of land he already is using even if the land were given to him.

By the same reasoning, the landowner will not commit more farming input himself a fortiori. Under the share restriction (constraint), however, competition will require the tenant to do so. For example, suppose before the restriction the total yield is ﹩200, and with r — 0.7 the tenant's income is ﹩60, an amount which equals his alternative earning. With the imposition of = 0.4, the current tenant will receive ﹩120, or ﹩60 over his alternative earning. This tenant, in order to keep or get land will offer to commit an additional farming cost of at least ﹩60. To illustrate, suppose the tenant agrees to commit an additional nonland farming cost of ﹩60, the amount over his alternative earning resulting from the share restriction. The corresponding value of marginal product will be less than ﹩60 —say ﹩40. Of this ﹩40 the owner will receive ﹩40 × , or ﹩16, and the tenant will receive ﹩40 × (1 - ), or ﹩24. To the tenant, this ﹩24 is an amount over his alternative earning, which the owner can again induce the tenant to invest in farming.

It follows that to maximize his income from land subject to the legal constraint of rental share reduction, (a) the owner can successfully induce the tenant to commit more farming inputs as long as the tenant receives a farming income greater than his alternative earning (since other potential tenants will offer to do so); and (b) provided that condition (a) holds, the landowner will induce more intensive farming as long as the marginal return to additional tenant inputs is greater than zero, for the rental income under the share constraint will then be higher than without additional tenant inputs. In condition (a), we have the non-land cost constraint; in (b), we have the limit of the physical constraint, namely, zero marginal returns for tenant inputs. The maximization of rental income for the landowner will be subject to whichever limit comes first. The result will be an increase in farming intensity, simply defined here as an increase in f/h, the amount of nonland farming inputs (or nonland farming cost, f) per land area. The economic implications of this are important, and we shall analyze them at some length.

But first certain assumptions must be specified. (1) After the imposition of , assume that there is no compensating (offsetting) payment of any kind. In other words, the owner's income from land is confined strictly to of the annual yield. (2) For analytical convenience, assume that all farming costs other than land are financed by the tenant. The relaxation of this assumption will not affect the implied allocation. And (3), assume zero transaction costs.

### A. Partial Land Repossession —Illustrated with Input Adjustments for One Tenant Farm

The increase of farming intensity under the share constraint may be attained in two general ways. One is to hold the tenant inputs constant, while competition allows the landowner to reduce tenant landholding through partial repossession of land. Another way is to keep the tenant landholding unchanged, while competition induces more tenant inputs over the given land, which is also to the interest of the landowner. of course, the landowner may choose a combination of both. In any event, the result is an increase in tenant inputs on land in tenant farms. The first is discussed in this section; the second is discussed in section B. Both are different views of the same phenomenon.

Figure 6 involves manipulations of figure 3 in chapter 2. A brief review of its notations is in order. The curve f/h is a rectangular hyperbola, representing the contractually stipulated (fixed) cost of tenant inputs, f, divided by the landholding, h. Given the average product of land, associated with the constant tenant inputs, the rent per unit of land, (q-f)/h, is defined. The initial tenant landholding is oa, where (q-f)/h is at a maximum, and the open market rental percentage equals ar/ap.

Under the additional constraint of , however, the (q-f)/h curve is no longer relevant for decision making. Instead of maximizing rent per acre of land, (q-f)/h, the owner will now maximize (q/h) subject to the constraint of f/h. To illustrate, suppose the initial rental percentage of 70 percent (ar/ap) is legally reduced to of 40 percent. The landowner's share constraint is thus represented by the curve , which is at 40 percent of the average product of land. This will be the case as long as the rent is reduced and stated according to percentage, regardless of whether the initial contract is for fixed or share rent.[1] The corresponding share for the tenant will be .

Without resource reallocation, the rent per unit of land will now be ae instead of ar, and the tenant's income will be at times oa, which is higher than his alternative earning. To maximize (q/h) , the landowner can reduce the tenant's landholding to ob without reducing tenant inputs, subject to the nonland cost constraint, f/h. Given f/h as shown, competitive equilibrium is reached at point h, where the tenant income from farming equals the tenant (nonland) cost. The rent per unit of land for the landowner will now be bg, which is higher than ae. To utilize the withdrawn land, the landowner can parcel it to other tenants in a similar manner, cultivate it himself, or hire laborers to till it (this last possibility was restricted in Taiwan). The result, under the strictly enforced share constraint, is a resource reallocation to a higher f/h ratio in tenant farms, or an increase in tenant farming intensity over that in an unrestrained market.

The economic implications of increasing f/h are several. Let us assume that there are only two factors of production, land, h, and tenant labor, t. Under this assumption, total nonland farming cost, f, is the prevailing wage, W, times t. Thus, given W, an increase in f/h means an increase in t/h. At the initial land size division T1, the marginal product of land, , is ar. With the higher t/h ratio associated with landholding ob after the adjustment, the marginal product of land becomes higher. This means that the marginal product of land would be higher in adjusted tenant farms than in owner farms. On the other hand, the marginal product of tenant input will be lower than in other occupations. In other words, the marginal cost of tenant input is higher than its marginal product. Thus, nonland inputs invested in tenant farms as a result of the share restriction will yield returns at a rate less than the interest rate.

Two side remarks can be made. First, if the tenant wage rate is low enough to form f*/h (dotted line in fig. 6), and given the constraints and assumptions specified, competitive equilibrium becomes undefined. In this case, the landowner will reduce land size to oc, where (q/h) is at a maximum and where the marginal product of tenant labor (the "other" factor) will be zero.[2] The tenant will then be receiving a residual income (which equals ij times oc), an amount which the landowner could extract in a lump sum (a compensating payment which we assume away).

Second, the farming input which defined f does not have to be one factor alone. The economic implications of increasing the f/h ratio for several tenant inputs will be essentially the same. If several tenant inputs are increased while holding land constant, the meticulous reader may rightly point out that the marginal product of a particular factor may increase if factor complementarity is strong enough. Diminishing returns must operate for all increased tenant inputs together, however, resulting in a marginal farming cost (which includes all tenant inputs) being higher than the marginal return.

[1]. Note also that under of the annual yield the landowner cannot convert a share contract to a fixed rent contract to protect his wealth.

[2]. Linear homogeneous production function is assumed here.

### B. Increased Tenant Inputs —Illustrated with Input Adjustments for Multiple Tenants

The implication of increased farming intensity is so important that it deserves still another exposition. Treating tenant input as an explicit variable, we at the same time expand our analysis to several tenants. A few simplifying assumptions will help. Let us assume that (a) the landowner employs a large number of tenants on his landholdings; (b) the tenants are cultivating homogeneous land, producing the same product with identical production functions, and thus the initial equilibrium rental percentage for each tenant before the share restriction is the same; and (c) there is only one tenant input, say labor.

In figure 7, total product is measured along the vertical axis, and the number of tenants or tenant workers, t, is measured along the horizontal axis. The curve is the total product of tenant workers, with the total landholding of a landowner held constant (with assumption a). Its shape exhibits diminishing marginal returns to tenant labor. The curve Wt represents the total tenant cost of farming, with W representing the wage rate and t the number of workers (with assumption c). Under a competitive tenant labor market, Wt is a straight line. The curve is the total rent curve given the landowner's total landholdings. It is derived by subtracting the total tenant cost, Wt, from the total product, . Without legal share restriction, the equilibrium number of tenants employed will be ot, where the total rent is at a maximum. With ot of tenant workers employed, the total rent will be tb (= ta — ti), and the rental percentage charged for each tenant equal to tb/ta (with assumption b). In equilibrium, the marginal tenant cost equals the marginal product, i.e., = W = As a result of the percentage rent reduction (say from 70 percent to 40 percent), the share constraint to the landowner is represented by the curve , where Q is the total product and is the restricted rental of 40 percent. That is, is 40 percent of at every point. The curve , on the other hand, represents the total share for the tenants as their labor input varies. Under the constraint of and with no adjustment in farming intensity, the landowner's total share will be td, and the tenants' share tc (= ta — td). Given the tenant cost constraint of Wt, however, the landowner will increase tenant input to t', where both and are maximized subject to the constraints of and Wt; that is, Wt = . Employing tenants ot' on the given land, the landowner's share will be t' g, and the tenants' share t' e. To the landowner, the total rent curve subject to the constraints of and Wt will be the heavy line , which rises with from o to g and kinks downward along with The portion of from g to k measures the difference of and Wt, which means that the constraint of Wt exceeds the constraint of in this portion. Thus, has a discontinuous derivative, and the marginal receipt, , is undefined at g, where is at a maximum.[1] If the limit of the physical constraint (point j in figure) is not reached, maximum is attained when Wt=[123b], or the alternative earning of tenant labor equals its income as a share of (see point e in figure). Resting at g (ore), therefore, is a new equilibrium based on the premise of rental maximization subject to the additional legal constraint of . At this equilibrium, the marginal tenant cost, = W, is greater than the marginal product, .

If the tenants' alternative earnings are lower, with wage rate W', the tenant cost constraint will be represented by the dotted line W't. Given this, to maximize income the landowner will only allow the increase in tenant input to t", where is at a maximum and where the marginal product of tenant labor is zero. With the limit of the physical constraint coming first, the constraint W' t is no longer relevant for the farming intensity decision. The peak of will now be at j, where we have In this case, however, the tenants will be receiving residuals over their alternative earnings; and with compensating payments to the landowner assumed away, the new equilibrium under the constraint of becomes undefined. Imagine the situation: inducing tenant input to t", the landowner wants to pause; however, other potential tenants come knocking at his door, offering him larger shares and bribes for the tenancy. Nonetheless, the condition with W't is not important to our analysis, for at any rate farming intensity is increased under the constraint of .

Though the marginal product of land cannot be conveniently derived in figure 7 the economic implications of the legal share restriction for resource allocation are exactly the same as those derived in the preceding section. Under the share constraint, resources are directed to tenant farms from owner farms, from tenant farms unaffected by the share restriction, and from enterprises other than agriculture. Alternatively and concurrently, the tenant now works longer hours, cultivates the land more intensively, and applies more costly fertilizers. The extra resources allocated to land by tenants on tenant farms yield lower returns than similar resources employed elsewhere.

If the aggregate supply of tenant labor to farming is upward sloped, the increased farming intensity will bid up the wage rate, making the curve Wt in figure 7 swing upward. The predicted resource reallocation will thereby be somewhat dampened, but not entirely cancelled. To the extent that the tenants work harder or for longer hours, the tenants' incomes from farming will rise at the expense of leisure and comfort. To the extent that the landowners' incomes from land are confined strictly to of the annual yield, their incomes will fall. To the extent that resources are directed from other industries to agriculture as a whole, the total agricultural output will rise.[2] But the discrepancies in the marginal products of resources in the various sectors imply economically inefficient use of the existing resources of the society.

[1]. Note that as long as the law of diminishing returns to tenant labor holds, rental income for the landowner under the share constraint will never be as high as it was initially.

[2]. This implies that the prices of farm products will fall relative to those of other goods, which further reduces the value of the marginal products of tenant inputs. The general implications of resource reallocation remain unchanged, however.

### C. Fixed Total or Percentage Rent Restriction? A Confusion between the 1949 and 1951 Provisions

Before we proceed to test the implication of increased farming intensity in the next two chapters, it is essential to point out that, under the share restriction in Taiwan from 1949 to 1951, there existed no law prohibiting the landowners from reducing land size per tenant or accepting more tenant inputs competitively offered by tenants. According to several writers, however, the rental restriction in Taiwan was a fixed and not a percentage restriction, which would mean that the landowner's incentive for increasing tenant inputs over land was eliminated.[1] This confusion in law requires clarification. To cite a typical example, let us turn to Sidney Klein:

The incentives among tenants to invest their hard-earned precious incomes were great.… Under the rent reduction statutes, their rent was fixed at 37.5 per cent of the standard or expected yield and not the actual yield harvested, so that when they sold the harvest above the standard yield they could keep all the proceeds for themselves.[2]

Klein is essentially saying that the landowner's rental receipt was fixed according to some standard after the share reduction, and therefore the tenant's incentive was high because he could keep the residual. But as we saw at the beginning of this chapter, any increase in tenant input will not pay, since the initial rental percentage implies that the marginal tenant cost equals the marginal product of tenant input. Thus, if rent were reduced to a fixed total, the tenant would not invest more in the existing farm; and even though he is willing to commit more input to acquire the lease right, the landowner would be indifferent to any incremental input. Only if the rental percentage ceiling is set lower will the tenant commit more inputs under competition, since it will be to the landowner's interest that he do so.

Klein, like several others, confuses two different laws: (1) the new 1951 Rent Reduction Act, versus (2) the enforcement of at 37.5 percent which was carried out in April 1949. Under the 1949 provisions, the owner's share was reduced to 37.5 percent of the annual yield, not of a standard yield.

Before we turn to the law proper, let me quickly dismiss Klein's interpretation by referring to the judgment of the lawmaker. Writing in 1951 and speaking of the 37.5 percent maximum rental share regulations of 1949, Cheng Chen stated:

The enforcement of the program took only a period of three months, from April to June 1949, when new contracts were signed. ... It was almost incredible that the program was so smoothly put into execution.… The most important factor was that rice production increased by 20 per cent in 1949 compared with the previous year, thereby in most cases preventing the landlords from suffering great losses as they received approximately the similar amount of foodstuffs as rent as before.[3]

Furthermore, according to the official report, in a comparison of the first phase of Taiwan land reform with a similar reform measure which began earlier in a few provinces in mainland China, it is discerned that the basic difference lies in the percentage rate of reduction: In mainland China it was a 25 percent reduction of the original contractual rent rates; in Taiwan the rental maximum was a flat 37.5 percent of the yield "actually harvested".[4] Also, the enforcement of the share restriction in Taiwan was conducted by inspecting the yield at harvest time.[5] If the rental restriction were indeed on a fixed and not a percentage basis, the yield need not be checked at all.

Let us trace carefully how this rental share restriction has been erroneously interpreted as a rental restriction at 37.5 percent of a fixed standard. Turning to the actual provisions of 1949, we find:

Before the land value is assessed according to law, the lease of all private farm land shall conform to the order… dated March 29, 1947, which provides that the amount of farm rent shall not exceed 37.5 per cent of the total annual yield of the main crops. If the rent originally agreed exceeds 37.5 per cent, it shall be reduced to 37.5 per cent; and if it is less than 37.5 per cent, it shall remain unchanged.

The main crops referred to in the preceding paragraph shall be chiefly the crops customarily accepted as payment of rent.[6]

Several things should be clarified. First, an assessment of the annual yield of land (not land value) was made two years later, but primarily at the request of the landowners, who apparently demanded such assessments so that they would have a fixed rental receipt as 37.5 percent of the assessed yield, in the expectation that high enough assessments would give them rental earnings as high as before the percentage reduction. The landowners should have known better, for the eventual official assessments were very much in the tenants' favor.[7] At any rate, these assessments were officially adopted in February 1952, and afterward there was little evidence of their being enforced.

Second, by "the order dated March 29, 1947" was meant the order of share reduction which had formally begun in a few provinces in mainland China. Since the nationalist government retreated to Taiwan, gradually from 1948 to early 1949, they have regarded Taiwan as one province of China proper. In April 1949, therefore, the promulgation and enforcement of the share reduction in Taiwan was regarded as a continuation of what they had started in mainland China. When the share restriction was first practiced in mainland China, it was largely confined to the provinces of Hupeh and Szechwan, and "the method adopted for determining the amount of the total yield of the main crops was based upon a joint report by the landlord and tenant".[8] This method was basically unchanged after the restriction was introduced to Taiwan in 1949, with share disputes investigated and settled by government officials.

And third, "The crops customarily accepted as payment of rent" obviously means any crop, provided that it is contractually specified to be planted or rotated. The word "main" in the cited article is apparently intended to restrict the form of compensating payment in which the landowner may share in all crops alike under the share restriction, since in some share contracts certain "cereals" cultivated by the tenant had not been shared by the landowner.[9] But we find no evidence whatsoever that this had been enforced. What in one farm had been contractually a "cereal" might be "main" in another. Furthermore, in farms adopting two to six years rotations with few repetitions of the same crop, all regularly rotated crops were "customarily accepted as payment of rent".[10]

From the above it is clear that the landowners' total rental receipts were not fixed, and therefore increases in tenant inputs are implied. What actually was fixed by law was the rental percentage of the annual yield.

But the conditions of sharing are quite different in the new act promulgated in June 1951:[11]

Art. 2: The amount of farm rent shall not exceed 37.5 percent of the total annual yield of the principal products of the main crops… The term "main crops" shall mean the crops most commonly grown or the rotation crops actually grown according to local farming practices, and "principal product" shall mean the chief article for which the crop is grown.

Art. 4: The standard amount of the total annual yield of the principal article of the main crop of a farm land shall be appraised, with reference to the grade to which it belongs.

Note that (a) the substance in Art. 4 was entirely absent in the regulations of 1949, and even in 1951 the future tense "shall be appraised" was used; and (b) the so-called main crops in this new act are equally difficult, if not impossible, to define, since any crop can be rotated. In fact, none of the additional provisions in this new act was put into effect, since, as was noted in chapter 1, the enforcement rules for the 1951 act were not set until February 1952, at which time another program of land reform was in the making.

What interests us here is why the changes in the official sharing method came about. The answer is clear. Chen's primary intention in reducing the rental percentage was to enforce a redistribution of income in favor of the tenant, which would serve as a political response to the exploitation-of-tenant claim which Mao Tse-tung used in his rise to power. But the landowners, under the share restriction, managed partially to restore their income by inducing more intensive farming, nullifying tenant gains. In early 1951, Chen had contemplated prohibiting the owners from requiring tenants to commit additional inputs, but this was never put into the law.[12] It had been expected then, that the owners would rush to sell their holdings to tenants at low prices because of the share restriction, thus resulting in more equal distribution of landholdings. But as was seen in the last chapter, outright transfers of lands to tenants under the share restriction were few. Thus, in June 1951, further restrictions were put on, intended to be enforced early in 1952. But it was soon realized that the new provisions would be overly costly to enforce, if it could be done at all. There remained one guaranteed way of redistributing income in favor of the tenants: to compulsorily purchase land from the landowners at low prices and sell it to tenants at the same low prices. This is known as the Land-to-the-Tiller program, elaborately drafted in 1952 and enforced in 1953, under which no private transfer of land is allowed.

The act of 1951, though not enforced, has since been publicized as if it had started in 1949, and indeed in most official sources the 1949 regulations are completely deleted. Writing in 1961, ten years after his own claim that the landowners' rental incomes were almost as high as before the share restriction, owing to the increase in output after the restriction, Chen changed his tone and asserted that the redistribution of income had been exactly as intended — a claim understandable as political propaganda.[13]

Turning to legal regulations directly restricting the landowners' inducing more intensive farming, the only relevant provision I can find in the regulations as of 1949 reads:

A limitation shall be set on the total area of both the leased farm land that may be taken back by the lessor according to the law and the farm land originally owned and cultivated by himself.[14]

This provision, intended to restrict partial repossession of land by the landowner, was apparently drafted from the brief experience of share restriction conducted earlier in mainland China. However, the promised limitation was not imposed until the Land-to-the-Tiller program began in 1953. Furthermore, when, in April 1949, the government urged the renegotiation of contracts to lower rental percentages the landowners could have reduced land size for tenant farms at will. Finally, the landowner could increase tenant inputs without reducing land size to attain the same result.

To conclude, it is evident that the theory of share tenancy implies increased farming intensity as a result of the share restriction's being enforced in 1949. Because of the new legislation's coming into effect in 1952, the hypothesis testing in the following chapters will be confined mainly to four years. The year 1948 will be used as the base year before the legal share constraint, and the three years from 1949 to 1951 will represent the years after. It is in these three years that the enforcement of the share restriction was strong and evident, that compensating payments and right transfers did not take place to any significant extent to restore the initial resource use under farm tenancy, and that we are certain that adjustments of farming intensity were not restricted.

[1]. Without compensating payments or other devices such as licensing the right to lease land, the equilibrium of this fixed rent control would be unclear; and the possible outcomes would be several and so varied that we could be certain of none.

[2]. See Sidney Klein, The Pattern of Land Tenure Reform in East Asia after World War II (New York: Bookman Associates, 1958), p. 77.

[3]. See Cheng Chen, An Approach to China's Land Reform (Taipei: Cheng Chung Book Company, 1951), pp. 24-25.

[4]. See Sino-American Joint Commission on Rural Reconstruction, "JCRR Annual Reports on Land Reform in the Republic of China", mimeographed (Taipei: Farmers' Service Division, 1965), p. 9. Italics added. See also JCRR, General Report, vols. 1 and 2 (Taipei: 1950 and 1952).

[5]. See chapter 5 of this study.

[6]. Art. 2, in Hui-sun Tang, Land Reform in Free China (Taipei: JCRR, 1954), p. 221. In Tang's translation, one error of substance has been found and corrected by the present author, that is, "main crop" should read "main crops". In Chinese, when the exact number is not specified, plural is implied, and there is no numerical specification in the original version. As will become clear later, the lawmaker meant to say one "main" crop per acre of land per period of time.

[7]. See JCRR, "Annual Reports", p. 20-26.

[8]. Chen, An Approach to China's Land Reform, p. 14. In other sources it is evident that the procedure of "reporting" the actual yield was frequently omitted if share disputes did not arise.

[9]. See chapter 5 of this study.

[10]. Perhaps the translation from the crops "chiefly wanted" to "main" is somewhat misleading.

[11]. Cheng Chen, Records of Taiwan Land Reform (Taipei: Chung Hwa Book Company, 1961), p. 181.

[12]. See Chen, An Approach to China's Land Reform, p. 29.

[13]. See Cheng Chen, Land Reform in Taiwan (Taipei: China Publishing Co., 1961), p. 309.

[14]. Art. 2, in Tang, Land Reform in Free China, p. 222. Note that this article was later deleted in the act of 1951.

### D. A Digression on Property Right Assignment and Resource Allocation

In this section I seek to discuss more fundamentally the theoretical results of the rental share restriction in terms of property rights.

Let me introduce an operational concept for the present discussion: an individual's right to use a resource exclusively and his right to derive unattenuated income (or rental annuity) from the use are mutually implied. In other words, given an exclusive right to use a resource (that is, the property right held by an individual is independent of other people's action), the right to derive income therefrom is also exclusive. "Income" is here used as a comprehensive term to include the flows of both pecuniary and nonpecuniary benefits. Transferability of the property right in the marketplace is essential in order to realize the maximum value of the resource. Assume away transaction cost problems, and competition for the ownership right will reveal that maximum value.

An attenuation of the right to obtain income from a private property resource by a legal authority may be enforced on a percentage basis or on a fixed price basis. In the latter case, available economic theory yields an equilibrium (and hence a set of predictable behavior with respect to resource use) only if there are other associated devices such as licensing or rationing. On the other hand, income attenuation on a percentage basis usually yields a specifiable set of constraints which permits a theoretical solution, because the dimensions restricted allow changes in choices among options predictable by the theory of choice.

Attenuating the right to derive income from resource use on a percentage basis, with the diminishing portion of income unappropriated and unassigned to some specific individual(s), will produce the same effects as attenuating the exclusive right to use the resource.[1] Consider the rental share restriction in Taiwan. Under a free market, a landowner is entitled to receive a portion, r, of the annual yield as income for his exclusive right over the use of his land. Under the share constraint, , and without input adjustment, the landowner's income from his resource ownership is reduced by (r— )/r percent. We ask: to whom is assigned the right to collect this portion of rent, that is, (r— )/r percent of the rent under a free market? Inasmuch as we can interpret the existing provisions and enforcement of law in Taiwan, under the share restriction the right to this portion was not assigned exclusively to an individual tenant. This unassigned portion of income from land became a residual, and all potential tenants were allowed to compete for it.

Consider an alternative. If, under the share restriction, the (r—)/r portion of rent were assigned exclusively to an individual tenant (or tenants), say, through the issuance of stocks against the market value of the given land, joint ownership of the land would be established. With the entire return to the ownership right assigned to private parties, each joint owner would thereby be granted authority to make decisions concerning his share of the resource. With the right to derive the rent from land wholly exclusive or unattenuated, the right to use the resource is also wholly exclusive. Various forms of transfers would insure that the value of each share be determined in the market. The implied resource use would therefore be exactly the same as before the share restriction. The implication is that the inefficient allocation of resources under the rental percentage restriction lies not in the landowner's share being reduced, but in that the right to derive income from land is not wholly exclusively assigned. In other words, under the share restriction in Taiwan, the right to land ownership was no longer wholly exclusive.

Let me carry this argument further. Under the share restriction, it will be recalled, the marginal product of tenant labor falls and the marginal product of land rises. But as long as the rental percentage is greater than zero, the marginal product of tenant labor will never be negative. (This is implied by the maximizing behavior of the landowner under the added share constraint.) Suppose both the rental percentage and the nonpecuniary benefits of owning land are effectively reduced to zero; that is, private landownership is reduced to a mere title, devoid of any form of income. Return momentarily to figure 7; competition among tenants will push labor input to point k, where the wage rate equals the average product of labor, or where the entire unassigned income from land is exhausted. Depending on the production function and the wage rate, the marginal product of labor in this case may be negative.[2] That is to say, if the entire income from land is unassigned to any individual or private party, land becomes a common property resource (in our case, with a private title), and competition among users implies that nonland resources will be allocated to land until the income to land (rent) is zero.[3]

From the above discussion we may generalize: For any production function requiring resource inputs h and t, if the right to a portion of the income from h, however small, is not appropriated or exclusively assigned, the ratio t/h will rise under competition, implying a fall in the marginal product of t and a rise in the marginal product of h; as the unassigned income from h increases, the marginal product of t will accordingly be lower, and may become negative when the entire income from h is not assigned to any individual party. We may call this "the production theorem of property right assignment".

[1]. This should be distinguished from an income tax, where the tax receipt is appropriated by the government. As implied in the following discussion, the misallocation generated by a tax on resource use stems not from the tax itself, but from the fact that no government official is thereby granted authority to make decisions on resource use.

[2]. In this and my earlier statement that the marginal product of tenant labor will never be negative if is greater than zero, the argument would perhaps be more convincing if the marginal product of labor in the relevant range were always positive, so that when = 0 it is still positive (with the wage rate equaling the average product of labor). In this case the increases in tenant input will be gradual and continuous in response to gradual decreases of . Theoretically the possibility of the existence of subzero marginal product of labor requires not only = 0, but also the supposition that the title holder cannot deny tenant farming on a competitive basis.

[3]. With a somewhat different approach, this condition is reached in H. Scott Gordon, "The Economic Theory of a Common-Property Resource: The Fishery", Journal of Political Economy (August, 1954). And it is implicit in Professor Pigou's imaginative example of two roads, which is expounded by Professor Knight. See A. C. Pigou, The Economics of Welfare (London: Macmillan, & Co., 1920), p. 194; and F. H. Knight, "Some Fallacies in the Interpretation of Social Cost", Quarterly Journal of Economics, vol. 38 (1924).

The correct analysis explaining the dissipation of rent, to my mind, is analogous to Cournot's duopoly solution with free entry: with rental income replacing monopoly rent, the aggregate average product of labor replacing market demand for product, and a positive wage instead of a zero cost of production.

## 7. Evidence of Input Adjustments under the Share Restriction

The effects on resource reallocation under the share restriction as implied by the theory of share tenancy can be summarized as follows. The imposition of a maximum rental percentage restrains the market-determined rental percentage. In the absence of sufficient compensating payments and tenure rearrangements through which the initial resource use may be restored, the tenants would receive earnings from farming in excess of their alternative opportunities; but competition among tenants offering to increase nonland farming inputs eliminates the residual earnings (unless the marginal product of nonland inputs goes to zero). Under effective share restriction, emphasis will be placed more on the maximization of gross crop value than on the crop value net of nonland costs. The yield per acre of land will rise, and with a lower rental percentage the landowner's rental income will be partially restored by a higher output.

Resources will be directed to tenant farms from all other uses. As more farming resources are directed to tenant farms, the marginal product of land in these farms will be higher — and the marginal products of tenant inputs will be lower — than those of similar resources employed elsewhere. This implies that the marginal products of tenant inputs will be lower than the associated marginal opportunity costs; similarly, resources invested by tenants will yield returns at a rate lower than the interest rate.

There are several circumstantial and technical difficulties in confirming these implications empirically. First, owing to the war years preceding 1947 and the multiple governmental reform programs conducted in Taiwan agriculture since 1953, the data I use are primarily from 1948 to 1951, though on occasion I use data of 1947 and 1952 also.[1] Cross-section data, when available, will be emphasized. Second, inflation and new currency issues during the period under study rule out the use of price data. With different crops harvested at different dates, for example, inflation would render the relative prices of crops useless. Third, for the greater part of the available data, we cannot separate owner farms from tenant farms. Thus, in the next chapter I shall attempt to rank the prefectures in Taiwan according to the respective magnitudes of resources affected by the share restriction. Finally, for technical reasons I shall not attempt to measure numerically the changes in marginal productivity. Only implications of these changes will be tested.

Observations from Taiwan will be presented in two chapters. In this chapter, I focus on the reallocation of input resources. In the next chapter, with more data available on crop yields, I shall undertake more delicate treatment of output responses. As much as possible, I choose data from primary sources.

After contracts were renegotiated in April 1949, under which a maximum flat rental rate of 37.5 percent of the actual annual yield was put into effect, adjustments in input intensity occurred in a variety of ways.

### A. Adjustments in the Intensity of Labor Input

Although the population in Taiwan increased during the period under study, the civilian labor force declined. The population increase was largely juvenile, and there was an intensification of national defense.[2]

In table 2, column 1 contains index numbers for the farming population during the period of share restriction. This reveals the reallocation of labor resources from other sectors to the agricultural industry as a whole. With negligible changes in the area of cultivated land in the same period,[3] the result was decreasing land holding per farmer (column 6). Columns 2 and 3 reveal that the population of both owner and tenant farmers increased, with the former rising at a faster rate than the latter after 1950. This is why the ratio of tenant farmers to owner farmers, computed from the actual population, declines in 1950 and 1951 (column 4).

However, the ratio of land under tenancy to owner land declined even more significantly. Before the share restriction, 56.3 percent of the private land had been under tenancy. In December 1949 —some six months after the share control —only 44.5 percent was under tenant cultivation.[4] This indicates that 20.9 percent of land under tenant cultivation was converted to owner cultivation. Of this portion of land converted, only about 2 to 3 percent could be accounted for by lease dismissals and land purchases by tenants.[5] The major factor, therefore, was partial repossession of land from the tenants by the landowners, which involved no lease dismissal.

The reduction of land size per lease contract through partial land repossession was not restricted by law. The landowners could themselves cultivate the land repossessed in this manner, hire farm hands to do the tilling, or sell it to parties other than the tenants. This, together with land purchased by tenants, accounted for the increase in owner farmers. On the other hand, the owners could lease the repossessed portions to new tenants, resulting in the rise in tenant farmers and lease contracts shown in column 3.

All this leads to the values in columns 7 and 8, which may be regarded as results of "critical experiments" suggesting confirmation of the hypothesis of increased farming intensity. Measured in hectares of land per farmer, the holding per owner farmer increased, and the holding per tenant farmer decreased under the share restriction. The data confirm not only that resource reallocation took place but, more specifically, that under the share restriction the intensity of labor input to land was significantly higher in tenant farms than in owner farms.

Several remarks need to be made here. First, the values given in columns 7 and 8 indicate that before the share restriction the landholding per tenant farmer was slightly larger than per owner farmer. This was so largely because a small portion of the cultivated land was rented to tenants under state ownership, and the labor intensity for this portion of land was reportedly low.[6] But we lack information separating this type of tenant from those cultivating private land. Thus, instead of furnishing rough estimates, we let this bias remain throughout. The correction of this would lead to slight decreases for all values in column 8. Second, since the rent limitation was a flat rental rate of 37.5 percent for all tenant contracts, some tenant farms were unaffected by this control (see chapter 8, table 4). If the unaffected tenant farmers were transferred from column 8 to column 7, which is the proper procedure for our present purpose, the difference between these columns after 1949 would be even greater.[7] Finally, it is clear that the major adjustments in labor intensity occurred quickly, and further changes after 1949 were minor. This is also seen in columns 9 and 10, where the average land sizes for each lease and tract declined only slightly from December 1949 to June 1951.

The decrease in land size per tenant farmer is not the only measure by which to detect more intensive labor input. Given the number of tenant farmers working on a fixed area of land, labor input may be increased by working more days in a year and longer hours per day. Before the share restriction it was reported that the "income from farming was often insufficient to cover living expenses and the tenant farmers depended on additional income from off-farm employment".[8] Off-farm employment was also frequent after the Land-to-the-Tiller program in 1953, but by then the tone had changed to "numerous changes have occurred in farmers' social status, ... to shift labor off farm or use more labor in sideline works".[9] During the period 1949-51, however, it was reported that "the rent reduction… greatly encouraged the tenants, to the extent that not a single one did not work his best to increase production".[10] And in a village with 140 tenant households, "seven of them built new houses, twenty of them took time to repair the fields…".[11]

[1]. For the erratic crop yields from 1942 to 1946, see Sino-American Joint Commission on Rural Reconstruction, Taiwan Agricultural Statistics, 1901-1955 (Taipei: 1961), notably the rice yields on pp. 20-21. It was not until 1948 that the prewar level of production was generally restored. As is indicated in the last chapter, for the year 1952 we are not certain of the actual method of crop sharing. Since 1953, not only has the massive Land-to-the-Tiller program been enforced, but fertilizers and irrigation have been heavily subsidized by the government. See Tsung-han Shen, Agricultural Development on Taiwan since World War II (New York: Comstock Publishing Associates, 1964); Office of Economic Policy, "The Taiwan Fertilizer Industry", mimeographed (U.S. AID Mission to China,1962); and E. L. Rada and T. H. Lee, Irrigation Investment in Taiwan (Taipei: JCRR, 1963).

[2]. See Shen, Agricultural Development on Taiwan, chap. 19.

[3]. See chapter 8, table 3, column 6.

[4]. See chapter 8, table 4, column 1. See also sources for table 2.

[5]. See chapter 5, section C.

[6]. See JCRR, "JCRR Annual Reports on Land Reform in the Republic of China", mimeographed (1965), p. 47. During 1948-50, about 6 percent of the total cultivated land was leased to tenants under state ownership, and the rental charge was reported to be one-fourth of the annual yield. But the determination of land size and the number of tenants involved are not on record. Part of this public land was sold to tenants in 1951, contributing partially to the notable jump in owner farmers (table 2, column 2) and the further decrease in tenant landholding from 1950 to 1951 (column 8). Of the lands under private ownership, sporadic evidence in various locations indicates that tenant farms had been slightly more labor-intensive than owner farms owing to the higher number of paddy fields under tenancy. See Department of Agriculture and Forestry, Taiwan Agricultural Yearbook, 1951 Edition (Taiwan: Provincial Government, 1951), sees. 1 and 2.

[7]. Information available includes only the number of tenant households affected, not the number of tenant farmers in these households.

[8]. T. H. Lee, "Impact of Land Reform on Agricultural Development and Rural Employment in Taiwan", (JCRR mimeograph 63-RED-M-176), p. 2.

[9]. Ibid., p. 6. After 1953, various reports indicate that a farmer rarely used more than two hundred days in a year for farm work, depending on the size of the farm assigned by the government. See Arthur W. Peterson, "An Economic Study of Land Use in Taichung Hsien and City, 1960", mimeographed (Tai-chung: Research Institute of Agricultural Economics, Chung Hsing University, 1961); Committee on Census of Agriculture, General Report on the 1961 Census of Agriculture (Taipei: 1963); and Nien-tsing Lu, An Analysis of Farm Family Economy of Owner-Operators under the Land-to-the-Tiller Program in Taiwan (Taipei: The Research Department of the Land Bank of Taiwan, 1965).

[10]. Cheng Chen, Records of Taiwan Land Reform (Taipei: Chung Hwa Book Company, 1961), pp. 37-38.

[11]. Ibid., pp. 39-40. For other similar examples see sources cited in chapter 1, section B.

### B. Qualitative and Quantitative Changes in Fertilizers and Other Inputs

Observations regarding fertilizer inputs under the share restriction noted the conversion from farm-produced fertilizers to commercial fertilizers. The latter, admittedly, cost more but are more effective.[1] Under a free market, the quantity and quality of fertilizers applied —whether committed by the landowner or by the tenant —must be consistent with the maximization of rental annuity. Under the share restriction, however, it is to the landowner's interest, and competition will require, that the tenant use higher-cost fertilizers if a higher annual yield is thereby obtained.

Measured in total quantities consumed, with 1948 as the base year, commercial fertilizers increased by 18.6 percent in 1949 and 120.5 percent in 1950, with the quantity consumed in 1951 approximating that of 1950.[2] On the other hand, farm-produced fertilizers decreased by 8.4 percent in 1949 and 13.5 percent in 1950, with the quantity consumed in 1951 approximating that in 1950.[3] Among commercial fertilizers, mineral fertilizers increased by more than two and one-half times for the same period; vegetable fertilizers increased by about one-third in 1950 and decreased thereafter; and animal fertilizers, which had been used only in small quantities, declined.[4] For the same quantity, mineral (chemical) fertilizers are generally the most costly of all, and are regarded as the most effective.[5]

In the period 1948-50, the most significant decrease in farm-produced fertilizers occurred in burned soil, which dropped by 50.1 percent. This was paralleled by a correspondingly large increase in commercial mineral fertilizers. Another notable decrease occurred in green manure, which declined by 45.1 percent, and was replaced by commercial vegetable fertilizers.

Besides burned soil and green manure, other farm-produced fertilizers such as animal manure, rice hull, ash, straw, and human manure decreased slightly.[6]

The qualitative change in fertilizers consumed is most interestingly illustrated in the commercialization of green manure. Although the use of green manure as fertilizer decreased as indicated above, the production of green manure crops did not.[7] Before the share restriction, green manure crops were seldom harvested, but were plowed under for fertilization. Under the restriction, however, the direct consumption of green manure declined. Instead, part of the crops were now harvested, processed in cake form, and sold in the market as commercial vegetable fertilizers. Besides a higher processing cost and more effectiveness in use, another reason may be offered for the marketing of green manure. The harvesting of the crops released the land earlier for other crops, a condition essential for the faster crop rotation in tenant farms as implied by the hypothesis of increased farming intensity.

The aforementioned changes in fertilizer consumption conform with a report provided by the Taiwan Provincial Land Bureau which indicates that the greater part of the increase in tenant income was invested in farming.[8] It was reported that, in 1950, 39.8 percent of the increase in tenant income was spent on fertilizers, 14.2 percent on draft animals, and 12.8 percent on farm implements. Aside from housing, health, and miscellaneous expenditures, which are both consumption and investment spending, the increase in pure consumption as a percentage of the increase in tenant income included 7.3 percent in food and clothing and 2.3 percent in amusement.

[1]. See Shen, Agricultural Development on Taiwan, pp. 136-39, 187-88 and elsewhere. The usual practice was that fertilizers and seeds were provided by the tenants (housing was usually provided by the landowners). The qualitative changes in fertilizers represented increasing farming intensity and not compensating payments.

[2]. DAF, Yearbook 1952, sec. 5. C. Note that there did not exist a well-established trend toward the increasing use of commercial fertilizers before 1948. Ignoring the war years before 1947, there was a decrease in commercial fertilizer consumption of 4.5 percent from 1947 to 1948, and afterward the trend of increasing commercial fertilizer consumption is clear. We consider these qualitative changes in fertilizers consumed as evidence confirming the hypothesis of increased farming intensity, which must not be confused with further increases in commercial fertilizers after 1953, when their production was subsidized by the government. See Office of Economic Policy, "The Taiwan Fertilizer Industry", p. 3.

[3]. DAF, Yearbook 1952, sec. 5. D. In absolute weight, which does not appear to be a relevant measure, the decrease in farm-produced fertilizers was slightly greater than the increase in commercial fertilizers.

[4]. Ibid., sec. 5. C.

[5]. Shen, Agricultural Development on Taiwan, pp. 136-39, 187-88. See also Office of Economic Policy, "The Taiwan Fertilizer Industry".

[6]. DAF, Yearbook 1952, sec. 5. D.

[7]. Ibid., sec. 3. V.

[8]. Taiwan Provincial Land Bureau, "Implementation of the 37.5 per cent Farm Rent Limitation Program", mimeographed (1950), pp. 37-38. The report was compiled from questionnaires answered by tenants, the accuracy of which is questionable. Various estimates of the average increase in tenant income are available in many sources, but all are computed simply by using an arbitrary constant as a measure of the increase in the tenant's income. They are overstatements for political propaganda purposes.

### C. Conclusions

Admittedly, the data presented in this chapter are inadequate. With respect to reallocation of labor resources, information is lacking on how the farming population data were compiled. One may question whether farming population is a correct measure of actual working units. Various reports indicating that tenants worked longer hours under the share restriction are based on only a few samples. The significant reduction in tenant landholdings, however, renders stronger evidence. With respect to adjustments in fertilizer inputs, we lack information on whether the increases in fertilizer expenditures occurred primarily in tenant farms. One may conclude, however, that available evidence related to resource reallocation is consistent with the implications of the hypothesis of increased farming intensity, and I could find no evidence which is inconsistent with my hypothesis. Reinforcing evidence is available when we investigate the behavior of various crop yields in the next chapter.

But the increase in tenant input intensity under the share restriction in Taiwan has been regarded by reform officials and Taiwan economists as "significant". Several hypotheses have been advanced to explain this observation. These hypotheses have two themes in common.[1] Accepting the evidence on input adjustments as presented, we may dismiss the arguments with which Taiwan reform officials have claimed "success".

Their first line of argument rests on the assertion that tenant cultivation had been inefficient initially, and that the share restriction led to more efficient land use. Under share tenancy, they argue, the tenant's incentive to farm is reduced, since part of the output is taken away by the landowner. According to this thesis, tenant inputs committed to land had been "insufficient" under a free market, with the marginal products of tenant inputs higher than the marginal costs, so that the reduced rental percentage encouraged greater tenant inputs. Indeed, this argument implies that as long as the rental percentage is greater than zero, inputs committed in tenant farms will never be as high as in owner farms. According to my theory, however, resource use in owner farms and in tenant farms is equally efficient under unrestrained private property rights. Under the share restriction, the hypothesis of increased farming intensity implies "excessive" resources allocated to tenant farms. Which, then, is the useful hypothesis?

According to evidence obtained from Asian agriculture (see chapters 3 and 4, section A of this chapter, and Appendix B), we find no notable difference in input intensities among farms with different land tenure arrangements under unrestrained private property rights. But, as is shown in columns 7 and 8 in table 2, under the share restriction in 1951, farming population (admittedly a rough measure of labor input) per unit of land was 47 percent higher in tenant farms than in owner farms.[2] And this does not include increases in other tenant inputs as discerned. The significantly higher labor-land ratio in tenant farms than in owner farms implies that the marginal product of land in tenant farms was higher, and the marginal product of tenant labor was lower, than similar resources employed elsewhere.

Their second line of argument rests on the so-called ability to invest. A higher tenant income under the share restriction, they claim, allowed the tenant a greater ability to invest, which led to higher investment in farming and more efficient resource use. If investment in farming yields returns at a rate higher than the interest rate, however, incremental investments will be committed by either the landowner or the tenant. For example, if more costly (effective) fertilizers led to a higher rental annuity, they would have been used without rental share control. The claim of insufficient investment owing to low income of the tenant ignores the possibilities of borrowing through the market,[3] and that the landowner may invest in land and raise the rental percentage.

The period 1949-51 reportedly had the highest interest rates on record in Taiwan.[4] The prevailing inflation does not seem to explain fully why these rates were considerably higher than those of later years.[5] In the same period, the greatest increase in tenant farming intensity was recorded. These seemingly contradictory phenomena (since the existing political instability and high interest rates tend to discourage investment) are consistent with our hypothesis of increased farming intensity. Finally, an increase in investment does not necessarily imply more efficient resource use. The rate of interest measures the alternative rate of return, and efficient allocation requires the equalization of marginal returns everywhere.

[1]. See sources cited in chapter 1, section B.

[2]. This is obtained by dividing the reciprocal of 0.155 by the reciprocal of 0.228 and subtracting one. It is interesting to note that to support their argument, Chinese writers emphasized only the increased tenant inputs. The year 1951 perhaps yields a better approximation than 1949 and 1950, since in 1951 part of the public-tenant lands was sold to tenants (recall that a small portion of the tenant lands was under state ownership). The labor-land ratio in tenant farms would be slightly higher if the unaffected tenant farms were transferred to column 7. On the other hand, a small margin should be allowed for more intensive tenant farming, since a greater portion of paddy fields were under tenancy (see table 4 in chapter 8).

[3]. See Irving Fisher, The Theory of Interest (New York: Macmillan Co., 1930); and Jack Hirshleifer, "On the Theory of Optimal Investment Decision", Journal of Political Economy (August, 1958).

[4]. See S. M. Yeh, Rice Marketing in Taiwan (Taipei: JCRR, 1955), pp. 73-74. The commercial bank rate of interest was about 4 percent per month.

However, "the interest rates charged by these banks are uniformly regulated by the government, which are much lower than prevailing rates in local money markets,...rice merchants were not all benefited ...because most of them could hardly obtain the needed credit from the banks".

[5]. As was seen in the preceding footnote, the money rate of interest during 1949-51 was in the neighborhood of 50 percent per year, as compared to around 20 percent after 1953. For two reasons one is led to believe that the real rate of interest was also higher during 1949-51. First, after the issuance of a new currency in 1949, the rise in the price of rice in 1950-51 had been less than 15 percent. Thus the anticipated rate of inflation would have had to be considerably higher than the actual rate to iron out the differences in real interest rates compared with later years. Second, inflation continued in Taiwan, though at a lower rate, well after 1953.

## 8. Patterns of Crop Choice, Hectare Yields and Marginal Productivities Implied by Increased Farming Intensity

Under the share restriction, recall, the hypothesis of increased farming intensity states that there exists a tendency toward the maximization of gross crop value per unit of land per period of time. Since different values are associated with different crops, and since different physical attributes of crops require different input intensities for their cultivation, certain patterns of crop choice are implied. That is, with the added constraint on rental sharing, there exist predictable pattern changes in crop plantations. The conformity of the observed behavior of various crop yields to the predicted patterns of crop choice in turn renders evidence confirming that in tenant farms the marginal product of land had increased-and the marginal products of tenant inputs had decreased —under the share restriction.

In this chapter, therefore, I shall try to do two things. One is to detect patterns of crop choice in Taiwan under the share restriction. Another is to relate the associated changes in the observable average crop acre yields to the implied changes in marginal products of farming inputs.[1]

### A. The Measurement

Let me begin by clarifying the meaning of some terms which I shall employ.

Cultivated Land and Crop Area

"Land area" in our theoretical discussions means cultivated land. With the small portion of the cultivated land in Taiwan owned by the state excluded, all privately owned land used for farming purposes is included under cultivated land, regardless of how many crops are grown each year or whether no crop is grown. In the analysis of land rent or land area assigned to each tenant under a share contract, this spatial (physical) area is the relevant measure.[2] During the period under study in Taiwan, partly because the cultivation of the land at external margins (arable but not privately owned or cultivated) would yield zero or low returns and mainly because the external margins were owned by the state,[3] the area of cultivated land increased only slightly. This is shown in table 3, column 6.

Whereas the area of cultivated land is quite inflexible over time, that is not the case for crop area.[4] To illustrate, suppose that over a one-year period, only one crop of rice is grown on one hectare of cultivated land; then the crop area is one hectare. If, over the same physical area for the same (one-year) period, the farmer rotates one crop of rice, one crop of sweet potatoes, and one-half crop of vegetables, then the crop area will be counted as two and one-half hectares.[5] In short, crop area represents the rate of crop rotation over a given area of cultivated land per period of time, measured in hectares of crops actually cultivated. An increase in crop area without a corresponding increase in cultivated land means that farming intensity has increased through an increase in the rate of crop rotation.

As is seen in table 3, using 1948 as the base year, the crop area of rice and common crops after the share restriction increased by roughly the same percentage. Special crops increased more, reaching a peak of 21.5 percent increase in 1950. Vegetables increased the most in crop area, reaching a peak of 31.7 percent increase in 1951; however, the crop area for horticulture (the area harvested) decreased.[6] These varying changes in crop areas represent changes in the frequencies of crops chosen. As we shall see later, they conform to the predicted patterns of crop choice under the share restriction.

The Density and Rate of Planting

There are two general ways by which the increase of farming inputs for a fixed area of land can be committed. One way is to hold the crop area unchanged, but to increase the yield per crop by using better seeds, fertilizers, and pesticides, improving irrigation, repairing fields, and perhaps planting the plants closer to each other. This method we shall call increasing the density of planting. Another way to increase the output on a given area of land is to speed up the rate of crop rotation, thus increasing the crop area. This can be done by planting additional crops during an otherwise idle season, by "shortening" the growing time required for each crop by relaying, interplanting, or earlier harvesting. Given an additional amount of tenant inputs to be committed under the share restriction, a portion of the incremental inputs will be allocated to the density of planting and a portion will be allocated to the rate of planting.

An increase in the density of planting implies a higher cost of planting for the same crop area. Provided that the marginal product of tenant input is above zero, the yield per crop area will rise. It is also clear that an increase in the rate of planting, with a corresponding increase in crop area, implies a higher planting cost per crop area for the same crop. At least two reasons can be offered. The first is the technical problem of faster rotation; for example, it costs more to relay or interplant (to squeeze in) an extra crop than to plant the same crop without relaying. And the second, which perhaps includes the first, is that haste makes a higher cost.[7] Unlike increasing density, where the yield per crop area will rise, the increase in planting rate may lead to a lower yield per crop area.

Intramarginal and Marginal Crops

The critical test of the implication that marginal products of tenant inputs decrease is the response of the marginal crops. Imagine a tenant farm which had grown one crop of rice per year before the share restriction, and suppose that this rice crop continued to be grown afterwards but with higher density of planting. The rice yield per crop area would rise. Here, we call rice the intramarginal crop because it was grown on the farm before the share restriction as well as after. Under the share restriction, the density of planting of this rice crop would be increased, and the corresponding rise in crop hectare yield implies that both the average and marginal product of land increased. This increased yield per crop area is also consistent with decreasing marginal returns to tenant inputs. Without incremental cost data, however, confirmation of our conclusion, that the marginal cost of tenant inputs is necessarily greater than the associated marginal product, is not possible.

Suppose, on another owner-cultivated farm, rice had also been an intramarginal crop. Under the share restriction, the reallocation of nonland resources from owner farms to tenant farms would tend to lead to a decrease in the density of planting on this owner farm. The crop area yield for rice on the owner farm would decrease. In the aggregated data available, the average crop area (hectare) yield for rice might rise or fall. The aggregated average would rise if (1) resources were drawn into agriculture to the extent that the increase in rice yield on the tenant farm outweighed the decrease in yield on the owner farm, or (2) rice was dominantly the intramarginal crop of tenant farms.

Suppose, however, that this tenant farm continued to squeeze in additional crops for annual rotation over a fixed area of land. There would eventually be a crop which, regardless of planting density, would have a considerably lower yield per crop area than elsewhere. This is so because the better land would be exhausted, the soil depleted, the season unfit, and the planting congested. The last crop which the tenant farm squeezed into the annual rotation would be the marginal crop. This does not mean, of course, that all tenant farms would go to this extreme under the share restriction. The general idea is that a crop which was added at the margin would have a lower yield per crop area than the same crop on another farm where it had been intramarginal. To the extent that the crop hectare yield of this marginal crop in the tenant farm pulled down the aggregated average yield per crop area, the relatively low marginal returns of tenant inputs would be confirmed.

If we are able to identify the marginal crop as described above, and if a large number of tenant farmers simultaneously chose the same crop to add into their crop rotation, the aggregated crop hectare yield for this crop would fall, regardless of the fact that the same crop might have been intramarginal in some farms. It follows that we have here a problem of crop identification. Any crop might indeed be added at the margin, and what in one tenant farm was the last crop to be added might not be the last in another. We shall attempt, therefore, to identify a class of crops whose physical attributes are such that they are likely to be added at the margin. That is, the physical attributes of planting them are such that their being added is consistent with the theory of choice under the share constraint. In still other words, there exists a class of crops which had not been commonly rotated under a free market, whereas under the share restriction, competition among tenants squeezed them in. This class of crops we shall refer to as marginal crops, in spite of the fact that they might not actually be added at the margin on some farms.

From the above it should be clear that under more intensive nonland input commitments, the yield per crop area (hectare) for an intramarginal crop is likely to rise, with a relatively small increase in crop area. On the other hand, for a marginal crop the crop hectare yield is likely to fall, with a relatively large increase in crop area.

[1]. Since Taiwan Agricultural Yearbook will be cited heavily hereafter, its method of data collection deserves attention. According to a letter from Yen-Tien Chang, Commissioner of Taiwan Provincial Department of Agriculture and Forestry, the following information is provided: (1). The three-level crop-reporters are responsible for collecting all basic agricultural data.… There are 720 township reporters who make field surveys and submit data collected to hsien (county) government. Each hsien (county) government has two crop-reporting inspectors, all together 44 of them, who lead and inspect the township reporters under them. After checking and processing, the county reporters submit the data to the Provincial Department of Agriculture and Forestry. (2). Per hectare yield is calculated by sample-cutting. Usually one sample is taken from every 100 hectares. Surveys are made four times a year to record changes.… (3). Other data, such as cultivated area and population, are obtained from the Provincial Land Bureau and the Provincial Department of Civil Affairs respectively.

[2]. Cultivated land is not the same as arable land. The latter area is estimated by the land bureau according to certain geological criteria. Strictly speaking, all land is "arable", but whether or not it is economically worthwhile to cultivate is something else.

[3]. See Sino-American Joint Commission on Rural Reconstruction, "JCRR Annual Reports on Land Reform in the Republic of China", mimeographed (Taipei, 1965), chap. 8.

[4]. Crop area is sometimes called cropping area, crop hectares, the area of plantation, etc. See Arthur W. Peterson, "An Economic Study of Land Use in Taichung Hsien and City, 1960", mimeographed (Taichung: Research Institute of Agricultural Economics, Chung Hsing University, 1961); and Nien-tsing Lu, An Analysis of Farm Family Economy of Owner-Operators under the Land-to-the-Tiller Program in Taiwan (Taipei: Research Department of the Land Bank of Taiwan, 1965).

[5]. The one-half crop of vegetable may mean that only one-half of the hectare of land is being utilized, or that the entire hectare is utilized but only half of the growing time is within the one-year period. The crop area varies greatly depending on land grades and the intensities of nonland inputs.

[6]. Since horticultural crops are "perennial", they are not rotated like "annual" crops. The significance of the decrease in their harvested area will be discussed later.

[7]. Note that we are speaking here of the cost of planting one hectare of the same crop, not the cost per unit of output. To convert one into the other, a production function is required.

### B. Output Responses of Intramarginal Crops: Rice

The identification of crops which were dominantly intramarginal or dominantly marginal is not difficult. Between these extremes, however, there exist a number of crops which we are not able to identify, in terms of position, with the available information. In general, rice and common crops are intramarginal. "Common" crops are those which have been commonly grown under a free-market.[1] If we are to single out one of these as clearly intramarginal for our present discussion, rice is the obvious choice. Three reasons can be offered. First, most tenant farms had grown rice before the share restriction.[2] The high frequency of tenancy in paddy and upland rice fields is evident. Second, in Taiwan rice plantations could be single or double cropping, but it was physically impossible to rotate three crops of rice in one year over the same acre of land. This means that for rice crops the flexibility of adjustment through increasing the rate of planting was rather limited.[3] And third, rice crop yield, for upland rice in particular, is highly responsive to increasing the density of planting. With better fertilizers and irrigation, the yield per crop area might well double.[4]

Increasing Rice Crop Area

As is seen in table 3, using 1948 as the base year, the crop area of rice had increased 9.9 percent by 1951. The increase for paddy rice amounted to 9.5 percent, compared with a 16.5 percent increase for upland rice.[5] The reason for the difference is that, for paddy rice, double cropping had been more common before the share restriction, and the opportunity for adding additional crops was more limited.[6] For rice crops in general, the best growing season in Taiwan runs roughly from July to November.[7] For single cropping, there were more farmers choosing to plant the "second" (July to November) crop than the "first" (March to July), and thus the crop planted first or earlier in the calendar year should be the more frequent crop to add if a single-cropping farm now turned to double cropping. For paddy rice, however, we find that the first and second crops increased by the same percentage in crop area (9.5 percent and 9.4 percent). The reason is that the official counting of the first and second crop is based not on the date of planting but on the date of harvesting (before or after August 15).[8] For paddy rice, this dividing line cut off some of the earlier planted crops and counted them as "second". The same dividing line did not affect the counting of upland rice, for the first crop of upland rice was harvested earlier. Thus, for upland rice we find a 44 percent increase in crop area for the first crop, compared with only 3.9 percent increase for the second.[9]

Increasing Rice Yield per Crop Area

Whereas the increase in crop area as a result of increasing the rate of crop rotation is a discretionary variable determined in the market, the change in yield per crop hectare is dependent not only on the density of planting, but also on certain exogenous variables such as weather and pests. Some judgment, however arbitrary, must be made as to whether a particular season is good, bad, or normal. For the period under study (1948-51), let me point out that in Taiwan there was no major agricultural disaster of any kind. That is, in the annual reports of agricultural losses due to drought, flood, and typhoon, we do not find notable annual differences.[10] Thus, the years under study may be regarded as "normal" —or equally "abnormal" —and nature was not a significant factor affecting crop yields.

Compared with 1948 (the base year), the rice yield per crop hectare was higher by 9.3 percent in 1949, 23.7 percent in 1950, and 26.4 percent in 1951.[11] The relatively small increase in 1949 might have been due to the fact that the enforcement of the share restriction began in April-June of that year. For 1949, we would expect the crop hectare yield for the second crop to increase by more than the first, since the share restriction was not enforced during the planting of some of the first crop. This is evident in upland rice, where for the first crop we find a 10 percent increase, as compared to a 41 percent increase for the second.[12] The increase in crop hectare yield was more significant percentagewise for upland rice than for paddy rice. The lower-grade soil quality of upland fields apparently allowed a greater margin of improvement through increasing the input density of planting. Of the 26.4 percent increase in 1951 over 1948, for example, paddy rice yield increased by 23.2 percent, compared with a 76.5 percent increase for upland rice.[13]

Conclusions

Taking rice as an intramarginal crop does not mean that the rate of planting (crop area) could not be increased. Extra rice crops were added, and the increase in crop area was greatest in the first crops of what had been single-cropping farms. The general increase in rice crop hectare yield was due to (a) the predominance of rice-growing in tenant farms and (b) the response of rice yield to the increase in the density of planting.

We measure the aggregate increase in annual rice yield as equal to the product of the index number of crop area multiplied by crop area yield. Using 1948 as the base year, we find a 13.7 percent increase in 1949, 33 percent in 1950 and 38.9 percent in 1951.[14] These percentages were increases in output from rice crops only. If we were able to separate rice outputs of tenant farms from those of owner farms, the increases might be greater for tenant farms than indicated.[15] At any rate, under the share restriction the landowners' rental incomes from land were partly restored by the increase in annual rice yield.

As is shown in table 3, the increases in crop area for rice and common crops are lower than for special crops and vegetables.

Among other common crops, we find that the increase in yield per crop area generally is less than for rice, and for a few common crops there is no increase.[16] But the more crucial test of the hypothesis of increased farming intensity lies not in rice and common crops, but in crops which were added at the margin.

[1]. All rice crops have been classified as common crops, and because they are predominant in Chinese agriculture, a separation is usually made between rice and other common crops in statistical sources. See the footnote attached to table 3.

[2]. Of the forty-eight common types of crop rotation listed in Lu, Analysis of Farm Family Economy, pp. 124-34, rice exists in most, frequently with various common crops. Special crops and vegetables exist in only a few.

[3]. Note, however, that the margin for the increase in the rate of rice planting may be found not only in farms where one crop of rice was grown per year, but also in some farms where one crop of rice was grown in two or three years.

[4]. See Tsung-han Shen, Agricultural Development on Taiwan since World War II (New York: Comstock Publishing Associates, 1964), pp. 174-95. See also D. H. Grist, Rice (London: Longmans, Green and Co., 1953).

[5]. Department of Agriculture and Forestry, Taiwan Agricultural Yearbook, 1952 Edition (Taiwan: Provincial Government, 1952), sec. 3. 1. Comparing these two percentages with the total increase of 9.9 percent, it becomes clear that paddy rice was by far the predominant rice crop.

[6]. See JCRR, Taiwan Agricultural Statistics, 1901-1955 (Taipei, 1961), p. 12. Not all paddy rice had been double-cropping, or there would have been no increase at all in crop area for paddy rice.

[7]. Shen, Agricultural Development on Taiwan, pp. 174-95.

[8]. Ibid.

[9]. DAF, Taiwan Agricultural Yearbook, 1952, sec. 3. 1.

[10]. See DAF, Yearbook 1949, 1950, 1951 and 1952, sec. 4. In particular, the area actually damaged was roughly the same every year from 1948 to 1951.

[11]. DAF, Yearbook 1952, sec. 3. 1.

[12]. Ibid., sec. 3. 1. D. For paddy rice, recall that the date of counting mixed up the first and the second crops.

[13]. Ibid., sec. 3. 1.

[14]. Ibid.

[15]. Such a separation is not possible with the available information. For more detailed treatments of the marginal crops, an attempt will be made to rank prefectures.

[16]. DAF, Yearbook 1952, Sec. 3. 2.

### C. Output Responses of Marginal Crops

Before we turn our attention to the behavior of three types of marginal crops, let us set the stage on which the responses of these crops may be observed in more convincing detail. The lack of output data separated for tenant and owner farms makes it difficult to confirm the implication that output responses in the two types of farms move in opposite directions. One method I employ to measure the relative responses is to rank the prefectures in Taiwan according to the degree to which their farming resources were affected by the share restriction, so that a prefecture of the higher order is expected to respond with greater proportional changes in the marginal crops, according to my theoretical predictions.

The Prefecture Ranking

It should be clear that the degree to which a prefecture was affected by the share restriction depends on two factors: (a) the proportion of farm land affected, and (b) the initial market rental percentage, since the flat rate control led to a greater rental share reduction if the original percentage had been higher. Our ranking problem is greatly simplified by the fact that (a) and (b) are positively related. In table 4, columns 4 and 6 reveal that tenancy occurred more frequently in paddy fields. The exception, Hwalien, can be explained by the predominance of upland rice crops in this prefecture. Columns 4 and 6, together with the proportionate acreage of paddy fields and dry land (not shown) determine the percentage values of column 2.

A comparison of columns 2 and 3 reveals that, since the reduced rental was a maximum uniform 37.5 percent of the annual yield for all tenant contracts, not all tenant farms were affected. The differences in these two columns depend primarily on the proportions of paddy fields and dry land under tenancy. Paddy fields are generally more fertile and had higher market rental percentages than dry land.[1] As is seen in columns 4 through 7, paddy fields were far more vulnerable to the flat rate control than was dry land. Columns 5 and 7, together with the proportionate acreage of paddy fields and dry land affected (not shown), determine the percentage values of column 3.

Since prefectures with proportionally more paddy fields were associated with a higher frequency of tenancy, and since paddy fields were associated with higher market rental percentages, the values in column 3 render reliable criteria for the ranking of prefectures in order of the degree to which they were affected. Such a ranking, however, is not without ambiguity. According to most sources, Sinchu and Taipei had the highest rental percentages; according to another, Taichung was the most fertile.[2] The disagreement stems from the fact that no source is speaking of a weighted average of the rental percentages, and heterogeneous land with different rental percentages in each prefecture makes comparisons difficult. Our solution here is to classify Sinchu, Taipei, and Taichung in the same group. Thus, in group I we have prefectures where we are certain that higher rentals had prevailed, and that private land resources were affected in the greatest proportions.[3] In similar fashion, we place Kaoshiung and Tainan in group II, and Hwalien, Taitung, and Penghu in group III. The Five Cities, which occupied small areas in Taiwan, will be excluded because of the massive computations required.[4] Ranking by the criteria of market rental percentages and proportions of private land affected is also in accordance with the proportion of farming households affected. The latter is shown in column 8. We do not, however, have information on the number of farmers affected, which would be the more appropriate measure of labor resources.[5]

Three Types of Marginal Crops

Recall that under the rental share restriction a higher tenant input cost tends to come about, and that the crops chosen to be rotated will be selected more according to their gross value than according to their net worth. Equilibrium under the share restriction no longer implies the maximization of rental annuity. Instead, it will tend toward the maximization of yield value per acre of land per period of time.[6] With this reasoning, three types of crops added at the margin are detected under the share restriction, and I discuss each in turn below.

The Crop of the Internal Land Margin

The term "land margin" as it is used in land economics usually refers to "arable" land which is not cultivated. The internal land margin used here refers to the bits and pieces of land within a privately owned farm, the quality of which is so low that under a free market they are seldom utilized. In Taiwan, these margins include hill slopes and badly eroded fields, upon which something of value could be grown. Under the share restriction, the crop chosen for this marginal land was citronella, a wild perfume grass which was now widely introduced for organized cultivation.

As is shown in table 5, the crop area for citronella increased from 2.36 thousand hectares in 1948 to a peak of 16.87 thousand hectares in 1951.[7] In prefecture ranking, the response in crop area increase is clearly more significant for the prefectures in group I than for those in groups II and III. The yield per crop hectare increased slightly from 1948 to 1949, but generally declined rapidly in 1950 and 1951. The large decline in crop hectare yield after 1949 as the crop area continued to increase may be interpreted as the result of a gradual use of less suitable land margins for citronella cultivation, where the very low yield of marginal planting pulled down the average crop hectare yield.

Presumably, the decrease in crop hectare yield in 1950 cannot be explained by the exhaustion of better land margins alone. Take Sinchu, for example, where the total yield in 1949 (5.06 × 10.4) was only a fraction lower than that in 1950 (8.3 × 6.61). The 64 percent increase in crop area in 1950 led to only a 4.3 percent increase in total yield, suggesting that the yields from incremental crop area were so low that one doubts whether the land-margin-exhaustion interpretation would suffice. Planting at a rapid rate might have led to generally poor organization for citronella cultivation. Of course, there is also the possibility of random fluctuations owing to weather, etc. But, the further decline in average crop hectare yield in 1951, with further increase in crop area (hectare), again confirms the very low yield of marginal planting owing to the exhaustion of better internal land margins.

To the extent that the total yield on the given (physical) land increased as a result of increasing crop area (planting rate), as is clearly the case when we compare 1948 with any of the following three years (table 5),[8] the average product of land had increased (i.e., shifted upward). This implies that the marginal product curve of cultivated land also increased, owing to the increase in tenant inputs. Judging from the fact that the total output (crop area times the hectare yield) increased significantly more in prefectures of the higher rankings (ranked according to proportions of farms affected by the share restriction), the marginal product of land in tenant farms increased relative to farms unaffected by the restriction. To the extent that the average yield per crop hectare sharply decreased, as was clearly the case after 1949, the marginal yield of increasing tenant inputs in cultivating incremental crop hectare (area) must have fallen to a very low level. If the cost of tenant inputs required to cultivate each additional crop hectare remained constant or increased, the very low yield of tilling the incremental crop area confirms that the marginal cost of tenant inputs was higher than the associated marginal product. Relating this to the relative prefecture responses as predicted in accordance with their ranking, the evidence also confirms that the marginal product of farming resources other than land yielded lower returns in tenant farms than in owner farms. All this is implied by the hypothesis of increased farming intensity derived from my theory of share tenancy.

Crops of the Interplanting Margin

Other identifiable marginal crops behaved similarly. Here let me turn to the crops of the interplanting margin. Between the rows of a planted crop something else could be grown. If another crop is added in the space between the rows to grow at the same time, it is called interplanting. If another crop is added in between, but continues to grow alone after the initial crop is harvested, it is called relaying. To relay is to interplant partially.

Depending on their physical attributes, some crops are more suitable for interplanting than others. Any crop which is suitable for interplanting may, of course, be planted singly at a lower planting cost per crop area. Meticulous care must be provided when planting or harvesting a crop between rows. On the other hand, any crop may be interplanted, but we are interested only in those which had been interplanted in Taiwan. Not all the "interplantable" crops were clearly marginal crops added for faster crop rotation. For the present purpose, I eliminate sweet potatoes, peanuts, barley, and Indian corn. Though these crops were interplanted on some farms, they were largely planted singly and were commonly grown.[9] Next, we eliminate sugarcane and tobacco. These two crops were largely interplanted; but the Taiwan Sugar Corporation, a government enterprise, grew the greatest portion of the cane crop, and tobacco planting was regulated by the government.[10] Remaining are five other interplanting crops — oriental pickling melons, cucumbers, watermelons, potatoes, and eggplants. With a higher tenant cost constraint allowed under the share restriction, the higher total yield obtainable from the given land through interplanting made these crops an "ideal" choice as marginal crops on tenant farms.

The five crops cited above also serve to illustrate the utilization of space in interplanting. First, the air space above the ground is utilized by combining cucumbers and eggplants:

Cucumbers are usually interplanted with eggplants on the ridge of the field. Between every two ridges lies a ditch for drainage. Bamboo sticks are placed on each side of the ridge and are tied together at the top. The vines of cucumbers grow on the bamboo sticks, thus making use of the space between ridges.[11]

Second, the utilization of ground space is seen in the relaying of watermelons and oriental pickling melons. Before the initial crop is harvested, the crawling vines of these melons are such that the fruit can be strategically placed. And third, the unoccupied soil space underground can also be utilized. Potatoes, though uncommon in the Chinese diet, have the advantage that they can be planted six inches deep.

These five crops are listed in table 6. In this table, the prefecture groups are listed in column 2, with their percentage changes in crop areas in column 5, and the percentage changes in yield per crop hectare in column 8. In general, we find that (a) increases in crop areas occurred more significantly in prefecture groups I and II than in III, (b) the yield per crop hectare declined in groups I and II, and (c) the yield per crop hectare increased in group III.

Interpretations for (a) and (b) above are the same as those for citronella. To repeat, the increases in crop area were more significant for marginal than for intramarginal crops. The exhaustion of suitable interplanting land margins, and perhaps the generally poor planting organization generated by the high rotation rate, contributed to the decrease in crop hectare yield.[12] The implications for the marginal products of farming resources are the same as before. Since in group III there were mainly farms unaffected by the sharing restriction, farming resources other than land were drawn from this group to groups I and II. Thus, a crop which had previously been interplanted might now be grown singly, resulting in a rise in yield per crop area as discerned in (c). This also confirms our prediction that output responses in owner and tenant farms move in opposite directions under the share restriction.

Crops of the Seasonal Margin

Perhaps the most powerful confirmation of my hypothesis is found in a third type of marginal crop: vegetables. The relatively large crop variety on record allows us to detect some definite patterns of choice under the share restriction. Owing to the tropical climate in Taiwan, most vegetables are grown in winter —not that vegetables cannot be grown in other seasons, but in other seasons farmers find more lucrative crops to grow.[13] Under a free market, in the short winter season (December to March) farmers cultivate small plots of land as vegetable "gardens". Only a small portion of the better land (usually paddy fields) is chosen for vegetables because the cost required to cultivate them is exceptionally high.[14] The greater portion of land is either left idle, used to grow green manure at very low planting cost,[15] or used to grow wheat and Indian corn in the appropriate fields.

The relatively high cost of planting, associated with the relatively high market value of vegetables,[16] together with the idle time margin in which they could be grown, made them among the preferred choices as marginal crops under the share restriction. As is shown in table 3, the crop area of vegetables increased the most of all crop classes, reaching a maximum increase of 31.7 percent in 1951.[17]

In table 7, the names of all vegetables on record are listed in column 1. They are ranked in descending order according to the percentage changes in crop area (column 4). The major prefectures contributing to the percentage changes in crop area (increase or decrease) of each vegetable are listed in column 8. They are all prefectures belonging to groups I and II, with Taichung and Taipei (group I prefectures) being the overall major contributors. With the exceptions of vegetable items 1, 9, 13, and 14, for which the crop hectare yield could not be computed, the percentage changes in crop hectare yield are listed in column 7. As is shown, the crop hectare yields for all but two crops (garlic and wax gourd) decreased.[18] Our interpretation is the same as before: the gradual use of less suitable land, the urgency of the time margin which must be released for common crop plantations, and perhaps for some the unfit season, all contributed to the diminishing returns to tenant input resources in cultivating additional crop hectares.[19] Over the same area of (physical) land, the higher total yield generated by a faster rate of planting implies that the marginal product of land under tenant cultivation was higher than elsewhere.

The ranking of vegetable crops in terms of percentage increases in crop area allows us to detect two patterns of choice implied by the hypothesis of increased farming intensity. First, we find the more frequent choice of a crop (higher percentage increase in crop area) which takes a shorter time to grow. This pattern is particularly clear in comparing items 1,9, 13, and 14. In general, stem (item 1) and leaf (item 9) vegetables grow faster than fruit (item 13) and root (item 14) vegetables. With a few exceptions which may be explained by another detected pattern, the growing time as a criterion of choice is also clearly shown by comparing vegetables of the same type: for example, among the listed root vegetables the highest ranked, radish (item 4), is noted as the fastest growing, whereas ginger (item 19) and dasheen (item 20) are "notoriously" slow; among fruit vegetables the highest ranked, tomato (item 6), perhaps grows the fastest; and among stem vegetables the lowest ranked, water convolvulus (item 17), is notably slow growing.[20]

Second, we find the more frequent choice of a crop which costs more to cultivate. Take Chinese cabbage and cabbage, for example; both crops require roughly the same growing time (about 95 days). But it is estimated that Chinese cabbage (item 2) requires 354 labor days for the cultivation of one acre, compared with 256 labor days for cabbage (item 5).[21]

The "time" and "cost" criteria of vegetable crop choice can be resolved into one: the maximization of income from land subject to a higher tenant cost constraint under the share restriction.

Under a competitive market, and provided that a crop is grown, a higher cost of production is associated with a higher value. Given the same growing time for different crops, those added more frequently at the margin under the share restriction were those which had a higher planting cost. On the other hand, given the same cost requirement of cultivation, the marginal crops more frequently added were those which required a shorter time, so that the given land could be released sooner for other crops. Both choices are implied by my theory, the results of a tendency toward the maximization of gross crop value rather than the crop value net of nonland costs.

[1]. See, for example, Cheng Chen, Record of Taiwan Land Reform, chaps. 1 and 2; JCRR, "JCRR Annual Reports", chaps. 1 and 2. Note that within paddy fields or dry land, the land prices vary according to soil grades and locations. In general, prefectures with higher rental percentages were associated with higher land prices. As was discerned in chap. 2, however, the market rental percentage depends not only on the fertility of land and the cost of farming, but also on the input commitments shared by the contracting parties. Although the price of land is the more appropriate measure for land fertility, for our present purpose the initial rental percentage is the relevant measure.

[2]. See Peterson, "Economic Study of Land Use".

[3]. The market rentals reported for these three prefectures usually range from 50 to 70 percent of the annual yield, though on some tenant farms a 90 percent rental share was observed. See Chen, Records of Taiwan Land Reform, chaps. 1 and 2; and JCRR, "Annual Reports", chaps. 1 and 2.

[4]. See the footnote attached to table 4 for prefecture delineation,

[5]. In 1948, the number of workers per tenant household was as follows: Sinchu, 7.28; Taipei, 6.97; Taichung, 6.21; Kaoshiung, 6.03; Tainan, 6.05; Hwalien, 6.32; Taitung, 5.21; and Penghu, 3.28. See DAF, Yearbook 1949, sec. 2.

[6]. Theoretically, the maximization of yield value per acre of land implies that the marginal product of tenant input has fallen to zero. Here we are only concerned with the tendency toward this condition.

[7]. It is reported that some tenants grew citronella illegally on the external land margin owned by the state. The crop area for citronella has declined since 1951, and particularly after the Land-to-the-Tiller program in 1953, so much so that the JCRR considered subsidizing its cultivation in 1960. See DAF, Yearbook 1958, sec. 3. 1. 18; and Shen, Agricultural Development on Taiwan, p. 108.

[8]. That is, during the share restriction the increase in crop area of citronella planting (faster rate) was proportionately greater than the decrease in crop area (hectare) yield, which led to a higher total output on the given land.

[9]. See Lu, Analysis of Farm Family Economy, p. 124; Shen, Agricultural Development on Taiwan, chap. 8 and particularly pp. 155-64; DAF, Yearbook 1948-52, sec. 3, 2; and Young-chi Tsui, A Study of Peanuts in Taiwan (Taipei: JCRR, 1954).

[10]. See Shen, Agricultural Development on Taiwan.

[11]. Tsung-han Shen, Agricultural Resources of China (New York: Cornell University Press, 1951), p. 220. This particular paragraph refers to peasants in Canton, China.

[12]. Unlike citronella, the hectare yield for the marginal crops listed in tables 6 and 7 usually started to decline in 1949, with the lowest hectare yield realized either in 1950 or in 1951. See DAF, Yearbook 1952, Sec. 3. 4. B. Both for clarity of presentation and for ironing out occasionally "erratic" output changes inherent in agriculture, the averages of a three-year period (1949-51) are used. Since the share restriction was imposed in April-June 1949, the confirmation of my hypothesis would be more critically displayed if these averages were taken for 1950-51.

[13]. Shen, Agricultural Resources of China, chap. 24.

[14]. On the average, the labor required to plant one crop hectare of vegetables is about seven times greater than for a common crop, and about five times greater than for a special crop. See Lu, Analysis of Farm Family Economy, p. 142, table 121.

[15]. Ibid. According to Lu, the labor required to cultivate green manure is about one-fiftieth of that required for vegetables.

[16]. A condition implied by a competitive market. For the market values of vegetables and other crops, see DAF, Yearbook 1952, sec. 3. B.

[17]. Note that there had been no trend toward increasing crop area for marginal crops in general before 1949. See DAF, Yearbook 1950, sec. 3. 4. B. The crucial evidence here, however, is that the crop area for marginal crops had increased relatively more than that for intramarginal crops under the share restriction. The three-year annual averages (1949-51) of the crop area for marginal crops used hereafter are less dramatic than two-year annual averages (1950-51), since the share restriction began in mid-1949. See table 3 for annual changes.

[18]. A "t" test was run on the observations of hectare yields before and after the share restriction for all marginal crops in tables 6 and 7. The statistics computed were the standardized mean difference of the twenty-one paired observations. The null hypothesis, that there was no systematic and significant difference between the before and after hectare yields, was rejected at the a = .01 level of significance. The hectare yields for the marginal crops under the share restriction were significantly lower, as is implied by the hypothesis of increased farming intensity.

[19]. Note that for items 16 to 20 (table 7), the decreases in crop area were associated with decreases in crop hectare yield. This is not difficult to understand. The exhaustion of better land and time and the higher cost of planting generated by a faster rotation rate combined to inflict "harmful" spillovers on crops of the same seasonal margin.

[20]. Information on the growing time required for most vegetables is usually in ambiguous terms such as "two to three months". That is why we cite here only the extreme cases for which the demarcation of growing time is clear.

[21]. See S. C. Hsieh and T. H. Lee, "The Effects of Population Pressure and Seasonal Labor Surplus on the Pattern and Intensity of Agriculture in Taiwan", (mimeographed, 1964), p. 5. A complete list of labor requirements for each vegetable is not available.

### D. Crops Unaffected by the Share Restriction: Horticulture

The method I have employed to cope with the lack of separate output data for tenant and owner farms is to rank the prefectures according to the magnitudes of land resources affected by the control. The conclusions stemming from this method may be confirmed further by investigating a class of crops not included under the share restriction. We find this in horticultural crops:

Horticultural crops grow on dry land, but they differ significantly from other crops grown on dry land. They take several years of cultivation before the fruits can be harvested… , that is why the rules of rental share restriction could not apply to them.… Our decision for tenancy in horticultural crops was to leave the contracts as they had been.…[1]

As shown in table 3, the harvested area for horticultural crops declined. In table 8, we list all the horticultural crops on record. Of the twenty-three crops listed, the plant yield for all but four crops decreased.[2] Both observations may be interpreted as follows: Under the share restriction imposed on other crops and the corresponding reallocation of resources, farming inputs committed to horticulture declined. Some plants which had previously been harvested at the margin were now not harvested at all, which led to the decrease in the area harvested. By the same reasoning, the lower inputs committed to tree-climbing and annual upkeep of the plants led to the decrease in plant yield for the plants actually harvested.

For the marginal crops discussed in the last section, the increase in crop area associated with a smaller proportionate decrease in crop hectare yield implies a rise in the marginal product of land and a decrease in the marginal product of tenant input. In horticulture, however, the decrease in the land area harvested implies that the marginal product of land decreased (i.e., shifted downward). This, together with the decrease in plant yield, implies that nonland resource inputs declined, confirming rises in the marginal products of these inputs. Thus, the observations are consistent with the implication that the marginal product of land in horticulture was lower than that of other crops under the share restriction, and that the marginal products of farming resources other than land were lower in these other crops than in horticulture.

[1]. Chen, Records of Taiwan Land Reform, p. 37. Perhaps the fact that tenancy seldom exists in horticulture was the main reason Chen made this decision. I simply fail to see why the rules could not apply.

[2]. Apparently, the number of plants per hectare is not fixed, but we lack information on the yield per hectare. A "t" test was run on the observations of plant yields before and after the share restriction, by using the standardized mean difference of the twenty-three paired observations in table 8. The null hypothesis, that there was no systematic and significant difference between the before and after plant yields, was rejected at the a = .01 level of significance. The plant yields for horticulture under the share restriction were significantly lower.

### E. Conclusions

Observable evidence obtained from the first phase of Taiwan land reform fails to falsify the implications of the hypothesis of increased farming intensity. These implications, derived from the theory of share tenancy with the added share constraint, are inconsistent with efficient allocation of resources. My findings confirm that under the rental share reduction the marginal product of land in tenant farms was higher, and the marginal products of tenant inputs were lower, than those of similar resources employed elsewhere. Combining the findings of chapters 7 and 8, let me summarize:

1. In the reallocation of resources, the changes in nonland input intensity responded in a variety of ways. More resources were directed to agriculture as a whole. While there was a small increase in owner-operated farms under the share restriction, we find nonland input intensity increasing in tenant farms and declining in owner farms. Specifically, we find (a) the land-farmer ratio declined for tenant farms and increased for owner farms, to the extent that the former was significantly lower than the latter; (b) tenants were reported to have worked longer hours and more days in a year, (c) fertilizers used were of higher quality; and (d) the increase in tenant income was reported to have been largely invested in land.
2. The implication that the marginal product of land in tenant farms was higher than elsewhere is confirmed by the following findings: (a) the reallocation of resources as described in paragraph 1 above; (b) the rise in crop area and crop hectare yield of intramarginal crops in tenant farms; (c) the significant rise in crop area for marginal crops added in tenant farms at a rate proportionately greater than the decrease in crop hectare yield, thus leading to a rise of total yield per acre of cultivated (physical) land in tenant farms; and (d) the production responses of horticulture.
3. The implication that the marginal products of resources other than land in tenant farms were lower than elsewhere is confirmed by the following findings: (a) the reallocation of resources as described in paragraph 1 above; (b) the general decrease in crop hectare yields for marginal crops added in tenant farms; and (c) the output responses of horticulture.
4. For paragraphs 2 and 3 above, the detected patterns of marginal crop choice also point to the same implications. The more frequent marginal crops chosen, as is demonstrated among vegetable crops, were those that require (a) a shorter growing time, or (b) a higher cost of planting. These patterns of choice are implied by the theory of share tenancy with rental share restriction.

## 9. Conclusions

This study has analyzed the interplay of property laws, economic theory, and observations of land use in Asian agriculture. This interplay, though complex, is perhaps the only approach through which the economics of land tenure can be properly understood.

Of the various aspects of the economics of property rights, I have concentrated on the theory and implications of leasing arrangements in agriculture. The share contract has been the main concern, for among tenure arrangements, sharecropping has been the main target of condemnation. Two sets of property right constraints were analyzed.

A preliminary investigation of property laws governing farm land ownership in China and Taiwan before 1949 concluded that a system of private property rights had existed in Chinese agriculture. The constraint of private property rights, therefore, was combined with standard economic analysis to derive a theory of resource allocation under share tenancy. Contrary to previous analyses of the subject, the theoretical results showed that different contractual arrangements do not imply different efficiencies of resource use. Implications of alternative theories were tested against observations chosen from periods and locations where the existing system of property rights agreed with the constraint of private property rights. Not only did the observations confirm my theory of share tenancy, but they refuted the traditional tax-equivalent approach. Indeed, the tax approach failed the trial of logic as well as the trial of fact.

The term "economic efficiency" used in this study has a simple meaning. It is a condition of market equilibrium logically deduced from the theory of choice, subject to the constraint of private property rights under a freely competitive market. It is a positive term, devoid of welfare implications. Viewing it as such, one wonders why the inefficiency argument of share contracts has prevailed for so long. Perhaps the illusion produced by the prima facie similarity between an excise tax and a share contract offers a marginal equality which appears attractive. Or perhaps the frequent condemnations of sharecropping in the land tenure literature provided a convincing impression of misallocation.

But the inefficiency argument against share leases is only one of several which have been used against the general system of tenant farming. For example, a high rent is taken as "exploitation" that dampens the tenant's incentive, and short-term leases are regarded as "insecure" tenure that inevitably reduces investment in land. Unfortunately, these assertions often have been taken as facts. And measures of agrarian reforms have been undertaken by governments to rule sharecropping illegal, to intervene in the terms of the contract, or to abolish farm tenancy altogether.

While I argued that the theoretical results of the tax-equivalent analysis are erroneous, I did not claim that unattenuated private property rights necessarily lead to efficient resource use in the real world. Certainly, decision errors alone may be wasteful. But under the specified constrained maximization, assuming zero transaction costs, the theory of share tenancy derived here shows that share tenancy satisfies the Pareto condition. Even though transaction costs in fact exist, the theory succeeds in explaining much of the observed farming behavior.

Since a variety of contractual arrangements exist under private ownerships of resources, I further questioned why different arrangements are chosen. By introducing transaction costs and risks, a choice-theoretic approach to this question was advanced. The general hypothesis —that contractual choices are made to attain the preferred distribution of risk subject to the constraint of transaction costs —was expounded against the factual background of the Chinese experience from 1925 to 1940. In the same way, it was argued that lease durations are chosen to minimize transaction costs.

The theory of share tenancy was also applied to a situation in which the rental percentage is reduced to a uniform legal maximum, a frequent practice under various agrarian reforms in Asia. With Taiwan chosen as an example, the pertinent laws of the rental share restriction were examined and interpreted. Equilibrium analysis with the addition of a legal constraint on shares was performed. Two hypotheses stemmed from this analysis and tests were constructed for both. The first hypothesis, that there would be compensating payments and tenure rearrangements, was confirmed by the existence of law prohibiting them and of prosecutions of violators of the laws. It served as a prerequisite for the second hypothesis, that of increased farming intensity.

The implications of the second hypothesis, that there would be marginal inequalities of resource allocation as a result of the rental share restriction, was confirmed by observations of both resource reallocation and patterns of output responses. The share restriction in the first phase of Taiwan land reform led to inefficient allocation of resources.

It has frequently been asserted that standard economic theory cannot be applied to conditions in underdeveloped countries. Exceptions have been alleged, "irrational" behavior has been discerned, and new postulates have been advanced to explain whatever observations are regarded as "mysterious". But a useful theory of choice can be derived only if the constraints involved are specified. In the literature of "underdeveloped" countries, recognition and proper treatment of the relevant property right constraints have been rare.

With the many facets of property laws governing resource use, it is, of course, not always easy to define the set of constraints with which to identify the options of choice. But to develop special theories to interpret some allegedly "mysterious" observations without investigating the pertinent property laws is unwarranted. "Disguised unemployment" and the "dual economy" are examples of such special theories. I argue, in Appendix A, that the increase in total yield owing to the share restriction refuted any argument that disguised unemployment had existed in Taiwan agriculture. Furthermore, under the share restriction, reallocation of nonland resources to agriculture as a whole produced a "dual economy". Yet the hypothesis of increased farming intensity was derived entirely from standard economic theory.

A number of issues related to this study could be investigated further. In conjunction with the agrarian reforms in Asia, two important topics call for analysis. The first is a comparative study of the various property laws underlying these reforms. An understanding of their origins, similarities and dissimilarities, the relative costs of enforcing them, and, perhaps more difficult, the economic forces which brought them into being, will throw light on the formation of property rights. A second topic is the derivation of what one might call a theory of fixed tenure, aimed at explaining resource allocation under the condition where the right to farm is exclusively assigned to individual farmers and transfers of this right are prohibited. This theory is needed to understand the Land-to-the-Tiller programs in Taiwan since 1953 as well as similar programs in several Asian countries during the past decade. The same theory may throw light on resource allocation in feudal Europe and Tokugawa Japan.

The theory of share tenancy may also be extended to other industries. Not only are share contracts observed in agriculture, but they are also common among retail stores, beauty salons, gasoline stations, amusement-park rentals, and even the much regulated oil and fishery industries. Of course, when the constraints of competition differ, modifications need to be made in the theory of share tenancy to interpret resource allocation in these industries. One might also apply the theory to a situation in which some resource in a share contract belongs to the state. For example, in Taiwan before 1951, a portion of farmland owned by the government was leased to tenants on a share basis; in mainland China in the 1950s, a popular form of business firm involved state-private cooperation with sharing arrangements.

Finally, more formal analyses of the choice of contractual arrangements and of transaction costs are needed. As was noted in chapter 4, I failed to arrive at general equilibrium solutions, owing to an inability to derive some specific transaction-cost functions and to disentangle some problems of choice theory involving risk. Whatever the contractual arrangements, the proper approach to analyzing land tenure is to investigate the nature of the property laws which define the constraint of competition, and not merely, as has frequently been done, to condemn what may appear to be defective leasing arrangements and to advocate a change in the laws.

## APPENDIXES

### APPENDIX A: Some Comments on the Hypotheses of Disguised Unemployment and the Dual Economy

A number of writers have attributed the small landholdings in Asia (and particularly in China) to family and social structure. Others have taken the crowded farming condition as evidence of disguised unemployment (where the marginal productivity of peasants is said to be zero or negative). Still others have claimed that whether or not the marginal productivity of peasants is zero, the productivity of labor is lower in agriculture than elsewhere. Various hypotheses of disguised unemployment and the dual economy have been developed in terms of family structure, of unlimited labor supply, of peculiar fixed-factor coefficient production functions owing to the peasants' ignorance of farming methods, and of a "rock-bottom" subsistence theory. Some insist that it is the average product and not the marginal product which underlies farming decisions in underdeveloped areas.[1]

The hypotheses constructed on these premises are incorrect. First, to accept the phenomenon of crowded farming as social is to leave an economic question unanswered. Second, the existence of unlimited labor supply or labor "surplus" is an assertion which in fact has no empirical foundation. Third, it is presumptuous to say that the peasants are ignorant of farming methods, for competition will induce sophistication. It is far closer to the truth to say that it is the economic theorist who does not know. Fourth, the "rock-bottom" is soft and variable. Finally, lacking an explicit behavioral postulate, the average-product argument is inconsistent with wealth maximization under private ownership of resources.

The theory of share tenancy derived in chapter 2 provides a different explanation for higher ratios of labor to land in Asian agriculture: the peasants' landholdings are small because their alternative earnings are low. And their low earnings are due to the small area of arable land relative to labor force, together with the fact that farming skill and knowledge are not highly valued in other industries. Under private ownership of land, it is to the landowner's interest that no negative marginal effort is "disguised". Given the existing resources, crowded farming is the result of wealth maximization, not of "irrationality". It is not difficult to show that "original" theorems or hypotheses intended to explain resource allocation in "underdeveloped" agriculture are unnecessary. For Asian agriculture, the aforementioned hypotheses can be readily dismissed in light of observations on land use and the analysts' neglect of the pertinent property right constraints.

First, let us take a closer look at the so-called overcrowded farming in Asia, which "overcrowding" cannot be denied by Western standards:

The growing seasons of rice and cane crops overlap. At the time cane should be planted, the preceding rice crop is not yet ripe. This problem is solved by planting cane among the ripening rice plants one month before the latter are to be harvested. Because the distance between [cane] rows is 1.39 meters and between [cane] plants 0.4 meter, farmers often plant peanuts, sweet potatoes, cotton and soybeans between the rows of canes in summer. These interplanted crops are harvested in November or December before the canes grow tall.[2]

Examples as intricate as this are abundant.[3] But as J. L. Buck points out:

It is evident from this maze of detail on the actual use of land by crops in China, that, in spite of the intensive use of crops for human utilization directly rather than indirectly by first producing animal products, still greater production could be obtained.[4]

While Buck is amazed by the various methods of intensive farming and soil conservation in China, the flexibility of land use is frequently ignored by development economists. An unimaginative theorist, unfamiliar with the actual situation, might easily consider the crowded tilling wasteful, and hastily develop fancy theorems and policies to slay the dragon.

Evidence suggests that, before the agrarian reforms in Asia, the marginal product of agricultural labor was not only positive but also nowhere near zero. Take Taiwan, for example, where in 1948 the landholding per person in farming was about as small as one could find. As was shown in chapters 7 and 8, however, increases in labor and other inputs on tenant farms under the rental share restriction led to significant increases in outputs.

Indeed, the fact that common crops are grown confirms that the marginal product of labor is positive. This is so because the same land can be used to cultivate other crops, for example, vegetables. Vegetable crops have considerably higher market values than common crops, and they generally require eight times as much labor to cultivate.[5] Forgoing some common crops for vegetable planting will lead to increasing labor input, and a higher gross income. Yet only a small portion of cultivated fields has been used for vegetables in Asia.[6]

Zero marginal productivity of labor implies a condition where it is no longer possible to choose a more labor-intensive crop to obtain a higher income, which is refuted by the facts. It further implies that there exists no idle land margin or that it is impossible to adopt a faster crop rotation rate with increasing labor input and income, which is also refuted by the facts. It implies, too, that most dry fields are converted into paddy fields through the use of labor. And the list of refuting evidence goes on.

A second reason for rejecting the hypotheses of disguised unemployment and the dual economy lies in their neglect of the existing system of property rights. Under private ownership of land, disguised unemployment cannot be derived from standard economic theory, regardless of how crowded farming may be. Under common ownership, however, zero or negative marginal product of labor is consistent with the general body of economic theory.[7] This is so because competition among users will reduce the rental value of land to zero, thus equating the average product of labor with the wage rate (or alternative earning).

Suppose that, as development economists have taken for granted, empirical evidence did confirm that the marginal product of labor is lower in agriculture than in other industries. If their hypotheses are intended to interpret Asian experience in the past twenty years, as many of them are, then they are still incorrect. They are incorrect because private farm land ownership with free markets, conditions which their hypotheses have implicitly presupposed, have been rare in Asia since 1950. The various agrarian reforms in Asia, as was mentioned in chapter 1, may in fact have led to "dual" economies. In particular, as analyzed in the second part of this study, the rental share restriction yields a condition where the marginal product of labor is lower in tenant farms than elsewhere. This result is derived entirely from standard economic theory.

The above discussion should not, however, be interpreted to mean that under private ownership of resources the marginal productivities of homogeneous labor must be equal everywhere at all times. The costs of information and of migration, and differentials in nonpecuniary gains and in the costs of living associated with different jobs, are sufficient to produce unequal marginal products of labor. These factors, of course, can be incorporated into the general body of economic theory. I object to the hypotheses of disguised unemployment and the dual economy on the grounds, rather, of their neglect of the flexibility of land use and of the pertinent property right constraints.

[1]. The literature sharing these views is enormous. But see W. A. Lewis, "Economic Development with Unlimited Supplies of Labor". Manchester School of Economic and Social Studies (May, 1954). For general discussion of these hypotheses, see Benjamin Higgins, Economic Development (New York: W. W. Norton & Company, 1959), chaps. 11-17; C. H. C. Kao et al, "Disguised Unemployment in Agriculture", in Agriculture in Economic Development, ed. C. K. Eicher and L. Witt (New York: McGraw-Hill Book Co., 1964); H. Myint, The Economics of the Underdeveloped Countries (London: Hutchinson & Co., 1964); and A. K. Sen, "Peasants and Dualism with or without Surplus Labor", Journal of Political Economy (October, 1966).

[2]. Tsung-han Shen, Agricultural Development on Taiwan since World War II (New York: Comstock Pub. Associates, 1964), pp. 198-99.

[3]. See, for example, F. H. King, Farmers of Forty Centuries (Emmaus: Organic Gardening Press, 1900). As an observer in 1900, King, like J. L. Buck after him, took pains to understand farming techniques in Asia, though he mainly emphasized the intensive farming in his photograph-illustrated work. Oriental writers, however, are far less impressed with their farming methods. See, for example, Shen, Agricultural Development on Taiwan; idem, Agricultural Resources of China (New York: Cornell University Press, 1951); and Nien-tsing Lu, An Analysis of Farm Family Economy of Owner-Operators under the Land-to-the-Tiller Program in Taiwan (Taipei: The Research Department of the Bank of Taiwan, 1965). See also chapter 8 of this study.

[4]. J. L. Buck, Land Utilization in China (Chicago: University of Chicago Press, 1937), p. 242.

[5]. Estimates of labor intensity required for different crops are available in Lu, Analysis of Farm Family Economy, pp. 142-44; and in S. C. Hsieh and T. H. Lee, "The Effects of Population Pressure and Seasonal Labor Surplus on the Pattern and Intensity of Agriculture in Taiwan", mimeographed (1964). The market values for vegetables and other crops are available in Shen, Agricultural Resources of China, chap. 24; and Department of Agriculture and Forestry, Taiwan Agricultural Yearbook.

[6]. See chapter 8, table 3; and Shen, Agricultural Resources of China, chap. 24. See also the forty-eight observed systems of crop rotation listed in Lu, Analysis of Farm Family Economy, pp. 124-34.

[7]. See, for example, H. Scott Gordon, "The Economic Theory of a Common-Property Resource: The Fishery", Journal of Political Economy (August, 1954). See also Anthony Bottomley, "The Effect of Common Ownership of Land upon Resource Allocation in Tripolitania", Land Economics (February, 1963).

### APPENDIX B: Rental Payments of Fixed and Share Contracts in China

In this appendix I seek to show two things: that rental percentages vary with land grades; and that share rents are generally slightly higher than fixed (crop) rents. The data, obtained from Chinese sources published in the 1930s, leave a great deal to be desired. I have been unable to determine in detail the methods by which the data were compiled, or even the exact numbers of samples. But my confidence in their reliability is enhanced by the fact that observations in different sources consistently exhibit the same patterns (see chap. 3).

1. In table 9, rental percentages of seven different land grades in twenty-two provinces (China, 1932) are shown. They represent percentages of the main crops, obtained exclusively from share contracts. With a few exceptions, lower-grade lands are associated with lower rental percentages. According to another survey with similar details (twenty-three provinces and six grades of land, China, 1930), only two exceptions are found in a total of 124 observations.[1]

The lack of fuller information stands in the way of interpreting these exceptions. For example, the rental percentage depends not only on the fertility of land, but also on the amount of other nonland inputs provided by each of the contracting parties. Furthermore, some lands might grow more "minor" crops than others, and their sharing does not seem to have been appropriately included.

1. In table 10, I intend to show that share rents are generally higher than crop rents because of the risk sharing. The figures are expressed as fixed monetary values. Not only do we find a few exceptions, but some of the differences appear unduly large. According to another survey of similar comparison (China, 1932),[2] in which fixed and share rents are expressed in percentages of the output yields, we find a few exceptions also (fixed rents being higher than share rents), but the differences are generally small.

In addition to the lack of fuller information on data compilation by which the exceptions and "erratic" differences may be explained, natural reasons are important. If fixed and share rents are to be compared, they must be expressed in the same dimensions, either both in fixed values or both in percentage values. However, share rents calculated and expressed in terms of fixed (monetary) values (as in table 10) will exhibit lower absolute rents with a bad harvest, and the converse with a good harvest. On the other hand, fixed (crop) rents expressed in percentage values will exhibit higher percentage shares with a bad harvest, and the converse with a good harvest. Only in a normal year, ceteris paribus, can the risk premiums for share rents be accurately revealed. At any rate, the implication of the tax-equivalent approach, that the rental receipt under a share contract is necessarily lower than that under a fixed-rent contract, is refuted by evidence.

It is hoped that the risk premiums of different contractual arrangements will be investigated further.

[1]. See Legislative Yuan, Statistical Monthly 2.5 (1930).

[2]. Department of Internal Affairs, Public Reports of Internal Affairs 2; vols. 1 and 2 (1932). See also Legislative Yuan, Statistical Monthly, 2.5 (1930).

## Bibliography

Note: An asterisk indicates that the work is available only in Chinese; quotations from these sources are my translations.

Adams, Warren. "Reflections of Recent Land Reform Experience in Iraq". Land Economics, May, 1963.

Alchian, Armen A. "Private Property and the Relative Cost of Tenure". In The Public Stake in Union Power, ed. Philip Bradley. University of Virginia, 1959. __"Some Economics of Property Rights". Mimeographed. Santa Monica: The Rand Corporation, 1963.

Bator, Francis M. "The Anatomy of Market Failure". QuarterlyJournal of Economics, August, 1958.

Betham-Edwards, ed., Arthur Young's Travels in France during the Years 1787, 1788, and 1789. London: George Bell and Sons, 1892.

Bottomley, Anthony. "The Effect of Common Ownership of Land upon Resource Allocation in Tripolitania". Land Economics, February, 1963.

Bray, J. O. "Farm Tenancy and Productivity in Agriculture: The Case of the United States". Food Research Institute Studies, vol. 4, no. 1 (1963).

Buck, John Lossing. Chinese Farm Economy. Chicago: University of Chicago Press, 1930. __An Economic and Social Survey of 102 Farms near Wuhu, Anwhei, China. Nanking, 1923. __An Economic and Social Survey of 150 Farms, Yehshan County, Hopei, China. Nanking, 1926. __Farm Ownership and Tenancy in China. Shanghai: National Christian Council, 1927. __Land Utilization in China. Chicago: University of Chicago Press, 1938. __, ed. Land Utilization in China —Statistics-A Study of16,786 Farms in 168 Localities and 38,256 Farm Families in Twenty-Two Provinces in China, 1929-1933. Nanking: University of Nanking, 1938.

Bull, Frate. "Philippine Land Reform, 1950-1958". International Cooperation Administration, USOM/Philippines. Manila, 1958.

*Chang, Pe-Yu, and Wang, Yin-Yuen. Questions of Farm Tenancy in China. Chungking: Commercial Press, 1943.

Chang, Yen-Tien. "Land Reform in Taiwan". Mimeographed report no. 1. Taichung: Department of Agricultural Economics, 1954.

Chen, Cheng. An Approach to China's Land Reform. Taipei: Cheng Chung Book Company, 1951. __Land Reform in Free China. Taipei: Free China Review,1953. __Land Reform in Taiwan. (Taipei: China Publishing Co.,1961. *__Records of Taiwan Land Reform. Taipei: Chung Hwa Book Company, 1961.

*Chen, Ching-Moh. Land Rents of Various Provinces in China. Shanghai: Commercial Press, 1936.

Chen, Han-Sheng. Landlord and Peasant in China. Shanghai: Kelly and Walsh, 1936.

*Chen, Tang-Yuen. Systems of Land Tenure in China. Shanghai: Commercial Press, 1932.

Cheng, Hsiao. The Theory and Practice of Land Reform in China. Taipei: Chinese Research Institute of Land Economics,1953.

Cheung, Steven N. S. "Private Property Rights and Sharecropping". Journal of Political Economy, December, 1968. __"Transaction Costs, Risk Aversion, and the Choice of Contractual Arrangements", Journal of Law & Economics, April, 1969.

*Chiao, Chi-Ming. A Social and Economic Study of Farm Villages in China. Nanking: University of Nanking, 1938.

*China Economic Research Department. Source Materials of Recent Chinese Agricultural History, 1912-1927. Peking: United Book Store, 1957.

*China Land Bureau. Rules and Regulations of Land Administration. China: National Government, 1947.

*Chinese National Government. Shengching Route Economic Report. 1931.

*Chung Sen Department of Education. Papers on Land Rent Problems in China. Shanghai: Commercial Press, 1937. Coase, Ronald H. "The Nature of the Firm". Economica, November, 1937. __"The Problem of Social Cost". The Journal of Law and Economics, October, 1960.

Committee on Census of Agriculture. General Report on the 1961 Census of Agriculture. Taipei, 1963.

Demsetz, Harold. "The Exchange and Enforcement of Property Rights". Journal of Law and Economics, October, 1964.

Department of Agriculture and Forestry. Taiwan Agricultural Yearbook, 1948 Edition. Taiwan: Provincial Government,1948. __Taiwan Agricultural Yearbook, 1949 Edition. Taiwan:　Provincial Government, 1949. __Taiwan Agricultural Yearbook, 1950 Edition. Taiwan:　Provincial Government, 1950. __Taiwan Agricultural Yearbook, 1951 Edition. Taiwan:　Provincial Government, 1951. __Taiwan Agricultural Yearbook, 1952 Edition. Taiwan:　Provincial Government, 1952. __Taiwan Agricultural Yearbook, 1953 Edition. Taiwan: Provincial Government, 1953. __Taiwan Agricultural Yearbook, 1958 Edition. Taiwan:　Provincial Government, 1958.

*Department of Internal Affairs. Public Reports of Internal Affairs, 2.1. China, 1932. *__Public Reports of Internal Affairs. 5.1 and 2. China,1932.

*Department of Real Estates. China Economic Yearbook. Shanghai: Commercial Press, 1935. *_____China Economic Yearbook. Shanghai: Commercial Press, 1936.

Dore, R. P. Land Reform in Japan. London: Oxford University Press, 1959.

Eicher, C. K., and Witt, L., eds. Agriculture in Economic Development. New York: McGraw-Hill Book Co., 1964.

*Executive Yuan. Changes in Land Rights. China, 1942.

Fisher, Irving. The Theory of Interest. New York: Macmillan Co., 1930.

Froehlich, W., ed. Land Tenure, Industrialization and Social Stability: Experience and Prospects in Asia. Milwaukee: Marquette University Press, 1961.

Georgescu-Roegen, N. "Economic Theory and Agrarian Reforms". Oxford Economic Papers, February, 1960.

Gittinger, J. P. "United States Policy toward Agrarian Reform in Underdeveloped Nations". Land Economics, August, 1951. _____".Vietnamese Land Transfer Program". Land Economics, May, 1957.

Gordon, H. Scott. "The Economic Theory of a Common-Property Resource: The Fishery". Journal of Political Economy, August, 1954.

Grajdanzev, Andrew J. Modern Korea. New York: Institute of Pacific Relations, 1944. __"Statistics of Japanese Agriculture". Mimeographed. New York: Institute of Pacific Relations, 1941.

Grist, D. H. Rice. London: Longmans, Green and Co. 1953.

Heady, Earl. "Economics of Farm Leasing Systems". Journal of Farm Economics, August, 1947.

Heady, Earl, and Kehrberg, Earl. "Relationships of Crop-Share and Cash Leasing Systems to Farming Efficiency". Research Bulletin 386. Iowa State College Agricultural Experiment Station, May, 1952.

Higgins, Benjamin. Economic Development. New York: W. W. Norton & Company, 1959.

Higgs, Henry. "Metayage in Western France". The Economic Journal. March, 1894. Hirshleifer, Jack. "Investment Decision under Uncertainty: Applications of the State-Preference Approach". Quarterly Journal of Economics, May 1966. __"Investment Decision under Uncertainty: Choice-Theoretic Approaches". Quarterly Journal of Economics, November, 1965. __"On the Theory of Optimal Investment Decision". Journal of Political Economy, August, 1958.

Hsieh, S. C, and Lee, T. H. "The Effects of Population Pressure and Seasonal Labor Surplus on the Pattern and Intensity of Agriculture in Taiwan". Mimeographed. 1964.

Institute of Pacific Relations. Agrarian China, Selected Source Materials from Chinese Authors. Shanghai: Kelly and Walsh, 1938.

Issawi, Charles. "Farm Output under Fixed Rents and Share Tenancy". Land Economics, February, 1957.

Johnson, D. Gale, "Resource Allocation under Share Contracts". Journal of Political Economy, April, 1950.

Johnson, V. Webster. "Agriculture in the Economic Development of Iran". Land Economics, November, 1960.

Jones, Richard. An Essay on the Distribution of Wealth and on the Sources of Taxation. Part 1 —Rent. London: John Murray, 1831.

Kao, C. H. C. et al. "Disguised Unemployment in Agriculture". In Agriculture in Economic Development, ed. C. K. Eicher and L. Witt. New York: McGraw-Hill Book Co., 1964.

*Kim, Lin-Ho. A Study of the Economics of Chinese Farming Villages. Shanghai: Chung Wah Book Store, 1937.

King, F. H. Farmers of Forty Centuries. Emmaus: Organic Gardening Press, 1900.

Klein, Sidney. The Pattern of Land Tenure Reform in East Asia after World War II. New York: Bookman Associates, 1958.

Knight, F. H. "Some Fallacies in the Interpretation of Social Cost". Quarterly Journal of Economics, vol. 38 (1924).

Kristjanson, B. H. "The Agrarian-Based Development of Iran". Land Economics, February, 1960.

Lee, Shu-Ching. "The Heart of China's Problem, the Land Tenure Systems". Journal of Farm Economics, October, 1946.

Lee, T. H. "Impact of Land Reform on Agricultural Development and Rural Employment in Taiwan". Mimeographed paper 63-RED-M-176. Taipei: Sino-American Joint Commission on Rural Reconstruction.

*Legislative Yuan. Statistical Monthly, 2.5, China, 1930.

Lewis, W. A. "Economic Development with Unlimited Supplies of Labor". Manchester School of Economic and Social Studies, May 1954.

Long, Ervin J. "The Economic Basis of Land Reform in Underdeveloped Countries". Land Economics, May, 1961.

*Lu, Nien-Tsing, An Analysis of Farm Family Economy of Owner-Operators under the Land-to-the-Tiller Program in Taiwan. Taipei: The Research Department of the Land Bank of Taiwan, 1965.

McCulloch, J. R. Principles of Political Economy. Edinburgh, 1843.

Marshall, Alfred. Principles of Economics. 8th ed., 1920; London: Macmillan & Co., 1956.

Mill, John S. Principles of Political Economy. 4th ed.; London: John W. Parker and Son, 1857.

Miller, Walter G.; Chryst, Walter E.; and Ottoson, Howard W. "Relative Efficiencies of Farm Tenure Classes in Intrafirm Resource Allocation". Research Bulletin 461. Iowa Agricultural and Home Economics Experiment Station, November, 1958.

Ministry of Agriculture and Forestry. Agricultural Land Reform Legislation. Tokyo: 1949. __Supreme Commander Allied Powers, Natural Resources Section Report no. 79. Farm Tenancy in Japan. Tokyo, 1947. __Report no. 127. Japanese Land Reform Program. Tokyo, 1950.

Mitchell, C. C. "Land Reform in South Korea". Pacific Affairs, vol. 22, no. 2 (June, 1949). Myint, H. The Economics of the Underdeveloped Countries. London: Hutchinson & Co., 1964.

*Nagamu, Ro. A Study of Land Systems in China. Chinese translation; Shanghai: Kuo Kwong, 1934.

*National Government, Statistics Department. Statistical Analysis of Tenancy Systems in China. China: Cheng Chung Book Store, 1942.

Office of Economic Policy. "The Taiwan Fertilizer Industry".Mimeographed. U. S. AID Mission to China, 1962.

Ogura, Takekazu. ed. Agricultural Development in Modern Japan. Tokyo: Japan FAO Association, 1963.

Parsons, Kenneth H. "Land Reform in the Postwar Era". Land Economics, August, 1957.

Peterson, Arthur W. "An Economic Study of Land Use In Taichung Hsien and City. 1960" Mimeographed. Taichung: Research Institute of Agricultural Economics, Chung Hsing University, 1960.

Pigou, A. C. The Economics of Welfare. London: Macmillan &Co. 1920.

*Pun, Lin-Fong. New Ideas of China Land Reform. Taipei: Chung Wah Cultural Publishers Association, 1957.

Rada, E. L., and Lee, T. H. Irrigation Investment in Taiwan. Taipei: JCRR, 1963.

Rasool, M. H. Hashimi, and Edwards, Alfred L. "Land Reform in Iraq: Economic and Social Implications". Land Economics, February, 1961.

Schickele, Rainer. "Effect of Tenure Systems on Agricultural Efficiency". Journal of Farm Economics, February, 1941.

Schultz, Theodore W. Transforming Traditional Agriculture. New Haven: Yale University Press, 1964.

Sen, Amartya K. "Peasants and Dualism with or without Surplus Labor". Journal of Political Economy, October, 1966.

Sharpe, William F. "Captial Asset Prices: A Theory of Market Equilibrium under Conditions of Risk". Journal of Finance, September, 1964.

Shen, Tsung-han. Agricultural Development on Taiwan since World War II. New York: Comstock Publishing Associates, 1964. __Agricultural Resources of China. New York: Cornell University Press, 1951.

Sino-American Joint Commission on Rural Reconstruction. General Report, vol. 1. Taipei: 1950. __General Report, Vol. 2. Taipei, 1952. __"JCRR Annual Reports on Land Reform in the Republic of China". Composite volume, mimeographed. 1965. __Taiwan Agricultural Statistics, 1901 -1955. Taipei, 1961.

Sismondi, J. C. L. Simonde de. Political Economy. 1814; New York: Augustus M. Kelley, 1966.

Smith, Adam. Wealth of Nations. 1776; New York: Modern Library edition, 1937.

Stigler, George J. "The Economics of Information". Journal of Political Economy, June, 1961.

Taiwan Provincial Land Bureau. "Implementation of the 37.5% Farm Rent Limitation Program". Mimeographed. Taipei: JCRR Translation, 1950.

Tang, Hui-sun. Land Reform in Free China. Taipei: Sino-American Joint Commission on Rural Reconstruction, 1954.

Tang, H. S., and Hsieh, S. C. "Land Reform and Agricultural Development in Taiwan". In Land Tenure, Industrialization and Social Stability: Experience and Prospects in Asia, ed. W. Froehlich. Milwaukee: Marquette University Press, 1961.

Tawney, Richard H. Land and Labour in China. London: G. Allen and Unwin, 1937.

Tsui, Young-chi. A Study of Peanuts in Taiwan. Taipei: JCRR, 1954. ______"Land-Use Improvement: A Key to the Economic Development of Taiwan". Journal of Farm Economics, May, 1962.

Tuma, Elias H. "The Agrarian-Based Development Policy in Land Reform". Land Economics, August, 1963.

*University of Nanking. Rental Systems in Four Provinces. Nanking, 1936.

Warriner, Doreen. Land Reform and Economic Development. Cairo: National Bank of Egypt, 1955.

Yeh, S. M. Rice Marketing in Taiwan, Taipei: JCRR, 1955. Young, Arthur. Travels in France during the Years 1787, 1788,1789. Vol. 2. Dublin reprint, 1793. __Travels in France during the Years 1787,1788, and 1789. Ed. Constania Maxwell. Cambridge University Press, 1929.

*Yu, Quan-Chung. Land Systems and Regulations in China. Vols. 1 and 2. Szechwan: National Szechwan University, 1944.

## INDEX

Agrarian reforms, in Asia, 5-15, 169

Air space, 146

Alchian, Armen A., 80n

Allocation of resources. See Resource allocation

Alternative earning: as constraint on rental share, 16-29; and increased farming intensity, 101-8; and disguised unemployment, 166

Arable land, 130n, 142

Assessment of annual yield, 111

Assets: transfer of rights to, 81; in tenant farms, 81n. See also Investments; Portfolio selection

Average product argument: with common ownership, 117; with disguised unemployment, 165, 166

Average rent: defined, 23; maximization of, 23-24; equality between fixed and share rent, 27-28, 50-51

Bentham-Edwards, Miss, 34-35 "Bottom", right to, 82-83 Bray, James O., 59 Buck, John Lossing, 55-61, 167 "Cereals": as compensating payment, 90; sharing of restricted, 93

Chen, Cheng: on enforcement of share restriction, 110; on intention of rental share restriction, 113-14

Chen, Ching-Moh, 57

China: land use in, 55-61; patterns of contract in, 66; characteristics of fixed and share contracts in, 72-79; land reform in, 110, 114

Chinese writers, 55, 79

Citronella, 142-45, 149n

Coase, Ronald H., 70n; 83n

Common crops, 135, 149n, 167; listed, 13In; effect of share restriction on, 132

Common ownership, 117, 168

Compensating payments: 4, 108, 109n, 160; cereals as, 90, 93, 112; restrictions on, 92-93; evidence of illegal, 95-97

Competition: among potential owners, 16n; ensures economic efficiency, 28, 159; conglomerates knowledge, 64; reduces enforcement cost, 64; and increased farming intensity, 101; and maximum value of resource, 115; and common ownership, 117

Contract: choice of, 4; as mutual agreement, 16, 31, 49, 50; as partial transfer of property rights, inflation, 8n; contractually determined, 84. See also Contractual terms; Rental share restriction 60. See also Contractual arrangements; Share contract

Contractual arrangements, 159; risk distribution in, 63-87 passim; transaction costs, 63-87 passim; choice of, 64, 69, 71, 161; patterns of, 66-67; evidence supporting hypothesis, 72-87 passim; effect of inflation on, 73-74; transfer of rights in, 85; effects of legal arrangements on, 86-87

Contractual renegotiation, 84-85

Contractual terms: in share contract, 3, 16-29, 52-55; government intervention in, under Taiwan land reform, 7; in fixed-rent contract, 27; neglected in tax-equivalent analysis, 31; in metayage, 40-41; in wage contract, 42-43; sample contracts, 72-73, 76-77

Cournot approach, 117n

Crop area, 150n; defined, 130-32; effect of share restriction on, 132-57; of citronella, 142; of interplanting crops, 146; of vegetables, 150

Crop choice: stipulated in share contract, 23-28, 76-77; patterns of, 129-57

Cucumbers, 146

Cultivated land. See Land area "Customs": and rental shares, 39-48; and famine adjustments, 74-76

Demsetz, Harold, 65n

Density of planting, 132-57 passim

Dien: in prereform land law, 11; defined, 12 Disguised unemployment, 160,165-69

Dismissal, 80, 83, 95, 97-98, 120 Distribution of income. See Income distribution Double cropping, 136-37 Dry fields: crop area increase in, 136-37; and yield increase under share restriction, 138; frequency of tenancy in, 139; effect of rental restriction on, 141; rental percentage of, 141 Dual economy, 160, 165-69 Duration of lease contracts: choice of, 79-85; distribution of, 80; with zero transaction cost, 81, 85

Efficiency: in Smith's time, 33; and transaction cost, 64-66, 87; with legal arrangements, 86-87; under Taiwan land reform, 116, 156-57, 160; and investment, 128; economic, 159. See also Pareto condition

Eggplants, 146

Elasticity of output, 21,26

Enforcement cost: reduced by competition, 63-68; varies with lease duration, 83-85; of Farm Rent Reduction Act, 67n, 68n, 70n, 113. See also Transaction cost

Escape clause, 74-76

Exclusivity of right: and limited authority, l0n; to use resource, 115-17. See also Property rights

Exploitation: in tenancy, 8, 159; claimed by Mao Tse-tung, 113

External land margin, 130 "Famine", 74

Farmers (tax collectors in England) 70n

Farm Rent Reduction Act of 1951, 92, 110, 112, 114; and minimum lease period, 94; enforcement cost of, 113

Fertilizers, 123-25

Fixed rent: compared with share contract, 27; frequency of in Asian agriculture, 66; transaction costs of, 67; sample contracts of, 72-73; escape clause and risk dispersion with, 74-76; lease duration with, 84

Fixed tenure, 161

Freehold, 34

Green manure, 124-25, 148n Gross crop value, 118, 142n, 154 Ground space, 146 "Harvest loss", 97

Higgs, Henry, 44

Horticulture: area harvested in, 132; effect of share restriction on, 154 Hwalien, 139, 141, 142

Improvements in land. See

Investment Income distribution: redistribution by land reform, 7; redistribution by

Increased farming intensity, 5, 156, 160; hypothesis of, 100-117; farming intensity defined, 102; effect of on landowner's income, 107n, 114; and resource allocation, 108, 116; restriction on, 114; supporting evidence for, 118-28 passim; labor-land ratio in tenant and owner farms, 122; and nonland inputs, 123n, effect on tenant income, 125n; and rotation, 132-33

Inflation, 119; effect on contractual choice, 73-74; tenancy dismissal, 80n; in Taiwan, 127-28

Information, in market, 63, 64

Insurance, 69-70, 74

Interest rates in Taiwan, 127-28

Interplanting, 145-49, 166

Intramarginal crop: defined, 133-34; output responses under share restriction, 135-39

Intramarginal tenant, 25, 78

Investment, 127-28; and share contract, 8, 159; and prereform land law, 13-15; and rental annuity, 26; classical and modern concepts of, 37-39; inputs shared in, 78; under share restriction, 125

Iron sheet rent, 73-74

Issawii, Charles, 48

Japan: land reform in, 9; rental share restriction in, 57n; labor-land ratio in, 60; pattern of contractual arrangements in, 66; crop insurance in, 70

Johnson, D. Gale, 33n, 49-51, 55 Jones, Richard, 39, 40, 40n; on land size division, 36; on metayage, 36-38; and ranking land tenure, 46

Kaoshiung, 141, 142n

Kehrberg, Earl W., 60n

Key-money. See Lump-sum payment; Compensating payments King, F. H., 167 Klein, Sidney, 109-10

Korea: land reform in, 9; farm sizes and prices in, 58 Kweichow province, 56

Labor "surplus", 166 Land area, 130

Land law, in China and Taiwan,9-15, 93-94

Land prices: determined, 24n; under different forms of tenure, 60; under share restriction, 98

Land reform, 5. See also Agrarian reforms, in Asia; Taiwan land reform

Land size, per tenant: and share contract, 16-29, 58; in France, 36; in Italy, 36; Jones on, 36; and "subsistence" income, 40; under share restriction, 120

Land-to-the-Tiller Act, 7, 9, 92n, 113-14, 119n, 123, 143n

Lease: in Britain, 33-34; freehold, 34; perpetual, 44; efficiency of, 46, 85; number of, 61; duration of, 79-85; termination of, 80, 83; in prereform land law, 93-94; in 1951 land law, 94; cancellations of, 95, 97-98. See also Dismissal

Legal arrangements: effect of on income variance, 72; effect of on contractual choice and efficiency, 86-88

Legislative Yuan, 57

Long lease duration, 81-83

Lump-sum payment: as alternative to unit pricing, 65-66, 86; as compensating payment, 89-90; restrictions on, 92-93, evidence of, 96; and residual tenant income, 105

Main crop, 111, 112-13

Mao Tse-tung, 113

Marginal contract rent curve, 17-19

Marginal crops: defined, 134; output responses of, 139-54, 150n; of internal land margin, 142-45; of interplanting margin, 145-49, of seasonal margin, 149-54

Marginal inequalities, 64-66. See also Disguised unemployment; Dual economy; Increased farming intensity

Marginal product of labor: equals wage rate, 21; equals marginal tenant cost, 54-55; on commonly owned resource, 117, 168; and theories of disguised unemployment and dual economy, 165-69. See also Increased farming intensity

Marginal product of land: reduced to zero, 17, 54n; in share contract, 21, 25; under share restriction, 104-8, 156-57

Marginal product of nonland inputs: under share restriction, 5, 133-35, 156-57. See also Increased farming intensity

Marshall, Alfred, 42-51

Maximization, of wealth or utility: in share contract, 16-29; opposed to custom, 48; tax-equivalent approach to, 48; and hypothesis of contractual choice, 69

Maxwell, Constantia, 35, 70, 76

Metayage: Smith on, 32; Young on, 34; Maxwell on, 35; Bentham-Edwards on, 35; Jones on, 36-38; Sismondi on, 38; English authorities on, 38-39; McCulloch on, 39n; and custom, 39-41; Mill on, 41-42

Mill, John Stuart, 36; on rent as production cost, 24n, 42n; on share tenancy, 38-42; on custom, 39, 40; on "improvements", 39-40

Minimum rent and wages, 76

Mortgage, 11

Negotiation cost, 63, 67, 75, 83-85. See also Transaction cost Neoclassical view on share tenancy,42-51

Nonland inputs: ratio to land, 16-28, 57-58, 61; Heady's argument on, 47; and dispersion of tenant inputs, 53-54; labor-land ratio in different forms of tenure, 60; as compensating payment, 90, 96; legally restricted, 92-93, 94; diminishing returns to, 105, 107; under rental share restriction, 108-9

Offsetting contractual rearrangements 88-99; defined, 89; and compensating payments, 89-90; and tenure rearrangements, 90-91; effect on resource value, 91

Ordinary lease, 12-15

Oriental pickling melon, 146

Paddy fields, 168; increase in crop area of, 136-37; increase in yield of, 138; frequency of tenancy of, 139; rental percentage of, 141; effect of share restriction on, 141

Penghu, 141, 142n

Perpetual lease: yungtien, 11-12; frequency in China, 80; observations on, 82-83. See also Freehold

Philippines, land reform in, 5n, 9

Piece-rate contract, 67n, 68n

Portfolio selection, 63

Potatoes, 146

Prefectures, impact of share restrictions on, 139-42

Present value: maximization of, 23, 79; of land, 24n

Property rights: and resource allocation, 4, 50; transferability of, 7, 115-17; in prereform land law, 10, 15; and constraint in theory of share tenancy, 16; and ranking of different contractual arrangements, 31, 33, 46, 158, 160; and constraint in contractual choice, 85; attenuation of, 89, 115-17; and analysis of rental share restriction, 115-17; exclusivity of, 115-17; production theorem of assignment, 117; and economic efficiency, 159; and agrarian reforms, 161

Purchase of land by tenants, 120

Rate of planting, 132-33

Regulations governing the lease of private farm lands in Taiwan Province, 1949: See Rental share restriction; Taiwan land reform

Relaying, 145-46

Rent: maximization of, 20; as a cost of production, 24, 42; when is land commonly owned, 117. See also Rental percentage

Rental annuity: maximized in share contract, 23-24; defined, 24n

Rental percentage: as variable determined in a share contract, 16-28, 53; in France and England, 35, 36; variation in, 56-58, 171-73; in share and fixed-rent contracts, 60, 171-73; of cash and crop rents in fixed-rent contract, 76; for different crops, 77-78

Rental share restriction: effects of on contractual arrangements, 88-99; and attenuation of land rights, 89; and sale of land, 98, 120; number of farms affected by, 99; effect of on landowner's income, 107n, 114; aims of, 113-14; effect of on resource allocation, 118, 132, 15657; and tenant income, 125n; in different prefectures, 139-42

Rent disputes: and choice of lease duration, 83-85; under Taiwan land reform, 96-97

Repossession of land, 102, 120

Residual: in tax-equivalent approach, 44; under rental share restriction, 108-9; under common resource ownership, 115-17

Resource allocation: with transaction costs, 66; contractually determined, 84; and increased farming intensity 108, 151; misallocation due to attenuation of rights, 116; effect of share restrictions on, 118, 132, 156-57; in feudal Europe and Tokugawa Japan, 161. See also Contractual arrangements; Contractual terms; Efficiency

Rice: crop area of under share restriction, 132; and output response, 135-39

Risk: and contractual choice, 63-87; natural risk defined, 63; risk aversion, 63, 68, 159; share contract a device for risk dispersion, 68-71; risk-exchange models, 68n; and fermiers, 70n; and escape clauses, 74-76; and price of land, 90, 98 "Rock-bottom" subsistence theory, 165-66

Schickele, Rainer, 47-48

Share contract: exploitation of, 8, 113; frequency of in Asian agriculture, 10, 66; and property rights, 16-28, 30-31; compared with fixed rent, 27; and custom, 39-48; under state ownership, 54n; transaction costs of, 67; as alternative to multiple escape clauses, 74-75; samples of, 76-79; short duration, 84; in other industries, 161. See also Contractual arrangements; Share tenancy

Share restriction. See Rental share restriction

Share tenancy: defined, 3; inefficiency of, 3, 7-8, 30, 126, 159; theory of resource allocation derived, 16-29; J. L. Buck on, 29, 78; in England, 32, 34; classical view of, 32-42; Neoclassical view of, 42-51; and tax-equivalent approach, 51-55; criticized by Chinese writers, 55; and tests of implications of alternative theories, 55-61; in Asia, 66-67. See also Share contract

Shirking of labor, 67

Short-term lease: inefficiency of, 8, 26n, 80; choice of, 83-85; frequency of, 84n

Sinchu, 141, 142n, 143

Single cropping, 136-37

Sino-Japanese War, 55, 73

Sismondi, J. C. L. Simonde de: on metayage in Italy, 38; on rental percentage, 40-41

Smith, Adam: on tax-equivalent approach, 32-34; on land tenure arrangements, 46

Soil space, 146

Special crops: listed, 131n; under share restriction, 132, 136n

State ownership of land: and equilibrium condition, 54n; in Taiwan, 122, 161

Stigler, George J., 42n "Surface", right to, 82-83

Taichung, 141, 142n, 150

Tainan, 141, 142n

Taipei, 141, 142n, 150

Taitung, 141, 142n

Taiwan Agricultural Yearbook, 129n

Taiwan land reform, 88, 91-92, 160; terms of, 5-6; intentions and justifications of, 7; timing of, 8n; and paddy and dry fields, 57; tenure rearrangements under, 90-91; restrictions on tenure rearrangements under, 93-95; and illegal tenure rearrangements, 95-98; enforcement of, 95, 99, 110, 114; effect on land prices, 98; percentage, not absolute restriction, 109-15; attenuation of of resource rights, 115-17; misallocation effects, 116, 156-57, 160; and "dual economy", 169

Taiwan Sugar Corporation, 145

Tax: misallocation effects, 115n

Tax-equivalent approach: analysis of share contract under, 30, 39, 42-51, 159, 172; inefficiency of, 42-51; compared with standard theory of share tenancy, 51-55; tests of implications, 58-61

Tenure rearrangements, 90-91, 160

Transaction cost: prevents dispersion of tenant inputs, 55; and contractual arrangements, 63-87; and portfolio selection, 63n; and efficient allocation of resources, 64-66; effect of legal arrangements on, 66, 86-87; and ranking of contracts, 67-68, 78; and escape clauses, 75; and lease duration, 81-85; and used asset prices, 82; and intensity of resource use, 86

Transactions, 85

Transferability of rights: partial transfer, 62-63, 85; under Land-to-the-Tiller Act, 7, 113; under prereform land law, 11-15; outright transfer and owner production, 11, 63; and exclusivity in use, 46n, 115; ensures utilization of knowledge, 64; reduces enforcement costs, 64, 81; and maximum value of resource, 115

United States, and Asian land reforms, 9

Variable proportions, law of, 54n, 105 Vegetables, 132, 136n, 149-54, 167 Vietnam, 61n

Wage contracts: frequency of in Asian agriculture, 66; transaction cost of, 67. See also Contractual arrangements

Water fees, 96

Watermelon, 146

Weather: as exogenous variable, 68; in Taiwan, 137

Yields: on tenant and owner farms, 59; and marginal product of farming inputs, 129-39; effect of increased density and rate of rotation on, 133; of rice, 137-39; maximization of, 142; of citronella, 143; of interplanting crops, 146, 149n; of vegetables, 150

Young, Arthur, 34-36

Yungtien, 11