Econometrics Group

The Group will emphasize the development of econometric techniques that are particularly suited to the analysis of micro panels (households and firms), as well as regional and cross- country data sets. Particular emphasis will be placed on how best to allow for dynamics, heterogeneity, cross section dependence and non-linearities that prevail in the analysis of economic and financial series. The Group plan to develop simple tests of cross section dependence, model cross section dependence using residual factor structures, and allow for heterogeneity by means of random coefficient models.

Macroeconometric Applications (using regional and cross-country data sets):

• Analysis of contagion and financial crises : This focuses on the development of techniques that allow contagion, herding and interdependence to be empirically distinguished, and investigate conditions under which such identification could be achieved. This provides a close link with the topic of “globalization and financial crises” included in the research agenda of the International Finance and Development Group.
• Spatial and temporal influences on property prices : There is considerable interest in modelling the spatial as well as the temporal nature of the economic activities. An important application would be to the analysis of house prices across different time periods and counties in California . This research could be of interest to the School of Public and Regional Studies.

Microeconometrics Applications

• Evaluation of social interventions : The estimation of the effect of interventions is a central problem in econometrics and statistics. Randomized experiments are the gold standard of evaluation research in the social and life sciences, but in many situations they may not be feasible or the randomization may be compromised due to non-compliance. If randomization is not possible, we must rely on other methods to obtain a reliable estimate of the intervention effect. Recently there have been new developments that clarify what can and what cannot be done with experimental methods. These new procedures are also at the cutting edge of another development, the emergence of semi-parametric procedures that require fewer assumptions and are therefore more robust. The new developments have led and will lead to applications in economics, other social sciences, but also in clinical research. This is a truly interdisciplinary venture, because subject matter knowledge is essential for successful application of the new methods for non-experimental evaluation of interventions . The methodology could also be of use to USC pharmaceutics economics group who are interested in the measurement of the effects of variable treatments or new drugs.
• Selective attrition in panel data : A prominent example relates to labor market histories. Another example is the problem of survivorship bias associated with financial panel data sets, with the additional complication of new entries (this research could be particularly of interest to the Finance Group in the Business School).