Summit recently hosted an Econometrics Seminar Program presentation on Factor Analysis. Factor Analysis is a statistical method that uses regression techniques to model the variability among correlated variables as a function of latent or unobserved variables called "factors." Specifically, Factor Analysis models the observed variables as linear combinations of the potential factors, plus error terms. This modeling approach yields useful information about the interdependencies between observed variables, which can be used later to reduce the set of variables in a dataset.
Factor analysis originated in psychometrics, but has become frequently used in the social sciences, particularly in economics, that deal with large quantities of data. Factor Analysis is useful for reducing highly correlated data as well as for constructing indices that capture information from the correlation of key variables. It is useful in many settings involving large quantities of data and also has applications in the forecasting literature. In addition, there is a growing literature in Econometrics that uses factors as instrumental variables for observed variables that may be correlated with the error term in the underlying outcome equation of interest. Factors are valuable as instruments since they are posited to be driving the underlying correlations in observables only, and are presumed to be uncorrelated with the outcome equation error term.
Summit works on many projects where latent variable modeling is a useful approach to assessing and modeling what causes observed correlations in the data. To this end, the Econometrics Seminar Program presented a two-part series on factor models, with both theoretical and empirical applications. For more information, please see this PowerPoint presentation, as well as the Stata log file, which details how to implement these models using Stata statistical software.
