Thursday, October 29, 2009

ANU Economics Showcase

Here is my presentation for the ANU Economics Showcase - a two day event that presents the range of ANU research to potential students, members of the public service, other ANU economists, and the general public (i.e. anyone who wants to come along :)). The showcase is on 25th and 26th of November. I'm scheduled for 26th November in the session from 1:30-3:00pm.

Signals or Noise? What Does Meta-Analysis Tell Us About the Reliability of Individual Econometric Studies?

David I. Stern

Arndt-Corden Division of Economics
Environmental Economics Research Hub
Centre for Applied Macroeconomic Analysis

Econometric studies can appear to be well conducted by the usual standards of the economics profession but still not provide results that can be relied upon for theory or policy purposes. The technique of meta-analysis can combine the results of many individual studies to provide more reliable results. I illustrate these points using a meta-analysis of more than forty studies of inter-fuel elasticities of substitution – a measure of the difficulty of switching between different fuels and electricity in production. These studies show a very wide range of values for any given elasticity. This dispersion of values reduces as the sample size of the original studies increases. This shows that to get reliable results we either need a very large sample or conduct a meta-analysis of existing studies. The results also show that the type of data – time series, cross section, or panel data - and the type of regression model used have a very significant effect on the estimated elasticities. This shows that we also need to have a good understanding of econometric theory in order to select an appropriate estimator.

No comments:

Post a Comment