Microeconometrics

{Cameron book}

• GMM estimation,
• nonparametric regression,
• simulation-based estimation,
• bootstrap methods,
• Bayesian methods,
• stratified and clustered samples,

Topics and methods:

1. NONLINEAR MODELING
   • nonlinear models
2. inference under minimal DISTRIBUTIONAL ASSUMPTIONS
   • robust inference
3. identifying and measuring CAUSATION rather than mere association
   • simultaneous equations
   • panel data fixed effects
   • treatment evaluation
     • differences-in-differences
     • instrumental variables
4. correcting problems of complex SURVEY methodology
   • departures from simple random sampling
   • sample selection
   • measurement errors
   • incomplete and/or missing data

Important chapters:

• key methods chapter (Chapter 5)
• GMM estimation (Chapter 6)

Part I

The essential components of microeconometric analysis:

1. an economic specification,
2. a statistical model,
3. a data set.

Chap 1

Characteristics of microeconometric problems, in contrast to macroeconometrics which models market and aggregate data:

1. Disaggregation
   • at individual, household or firm level
2. Nonlinear relationships
   • Limited dependent variables: censored, truncated, and discrete variable
3. Heterogeneity (of individuals, firms, and organizations)

Highly parametric models are sufficiently detailed to capture the complexities of data, but these models can be challenging to estimate. Alternatively, statistical inference can be based on standard errors that are “robust” to complications such as heteroskedasticity and clustering.

Handling unobserved heterogeneity:

1. Assume that unobserved heterogeneity is uncorrelated with observed heterogeneity, and the outcome being studied has no intertemporal dependence.
2. fixed effect, i.e. an individual specific constant
3. random effects
   1. one or more regression parameters varies randomly across the cross section.
   2. individual specific random component.

Chap 2: Model

Exogeneity Identification

Identifiability of causal economic relations.

Structural models:

1. Linear simultaneous equations models (SEM)
2. Single-Equation Models
3. potential outcome model

If the structural approach is implemented with aggregated data, it will yield estimates of the fundamental parameters only under very stringent (and possibly unrealistic) conditions.

Reduced form models:

Reduced form analysis does not always take into account all causal inter-dependencies

In general, the parameters of the reduced form model are functions of structural parameters. They may not be interpretable without some information about the structural parameters.

Chap 3: Data

three main types of data:

1. observational data
   • cross-section and longitudinal survey and census data
   • electronic records
2. data from social experiments
3. data from natural experiments

Data source:

aggregate time-series data collected by government agencies