Challenges in Econometrics
Guido W. Imbens
Stanford Graduate School of Business
American Economic Association, Ten Years and Beyond: Economists Answer NSF's Call for Long-Term Research Agendas
To frame what is in my view of the main challenges facing researchers in econometrics, let me set the stage by describing the current state of research. Much of the traditional research in econometrics can be divided into two branches, the first comprising cross-section and panel data econometrics and the second time series analysis. In the cross-section branch of econometrics researchers have data on a large number of units, often individuals, or groups of individuals, firms, or markets. For each unit there is information on a relatively small number of variables, sometimes measured at a single point in time, sometimes with repeated measures as in panel data. The units are viewed as exchangeable, or independent in the sense that there is no interaction between the units: what happens to one unit does not affect other units. In time series analysis the typical setting is one with observations on a small number of variables, at many points in time, with relatively unrestricted dependencies between the different variables. For models designed for data configurations of these two types we have learned much in the last few decades. In fully parametric models, as well as in the more flexible semi and non parametric models we have gained an impressive understanding of the appropriate ways of analyzing such data, and the properties of many estimators and methods for inference.
In my view the biggest challenges faced by economists in terms of analyzing economic data concern fundamentally different configurations of the data, with complex, largely unknown, dependence patterns and a relatively large numbers variables per unit. In such cases the current methods to do approximate inference based on large sample results, which are specifically designed to exploit laws of large numbers and central limit theorems, are likely to be inadequate. Moreover, trying to fit these more complex data configurations into the old methods would be unlikely to lead to much progress. In some cases econometricians and statisticians have made some progress on such alternative data configurations, but for the most these are unexplored areas for research.
Number of Pages in PDF File: 6
Date posted: August 12, 2011
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 0.203 seconds