Generalized Method Of Moments
- Estimation method that uses sample moments (expectations) rather than a specified functional form.
- Flexible: can combine unconditional and conditional moments, and handle cross-sectional or time-series moments.
- Can incorporate additional information (for example, exogenous variables) into the moment conditions to improve estimation.
Definition
Section titled “Definition”Generalized Method of Moments (GMM) is a statistical estimation technique that uses a set of sample moments to estimate the parameters of a model. Unlike other methods such as the least squares method, GMM does not require the assumption of a specific functional form for the model.
Explanation
Section titled “Explanation”GMM forms parameter estimates by matching sample moments to the theoretical moments implied by a model. It permits using a variety of moment conditions — including unconditional and conditional moments and moments drawn from cross-sectional or time-series data — which increases flexibility and potential efficiency. GMM also allows additional information, such as exogenous variables, to be incorporated into the moment conditions to help identify and more accurately estimate model parameters.
Examples
Section titled “Examples”Income and education example
Section titled “Income and education example”Suppose we are interested in estimating the relationship between income and education. Using GMM, one can incorporate a variety of moment conditions that capture different aspects of this relationship. This can include unconditional moments that capture the overall relationship between income and education, as well as conditional moments that capture the relationship within different subgroups of the population. Additional variables such as age and gender can be incorporated as exogenous variables into the moment conditions to control for their effects.
Use cases
Section titled “Use cases”- Applied to a wide range of data and well-suited for a wide range of applications, particularly when flexible moment conditions or additional information (e.g., exogenous variables) are available.
Related terms
Section titled “Related terms”- Least squares method
- Moment conditions (unconditional and conditional)
- Exogenous variables
- Cross-sectional data
- Time-series data