I don't know if DiNardo is watching this or not, but another issue that should be dealt with in the JEP piece regards the "panel data revolution" that has swept a lot of the policy-oriented micro econometric research in recent years. This combines both cross-section and time-series approaches. Barkley Rosser -----Original Message----- From: Peter Dorman <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] <[EMAIL PROTECTED]> Date: Tuesday, May 11, 1999 1:22 AM Subject: [PEN-L:6630] Re: In response to Peter >Two quick responses to the issue about stability raised by Michael: > >1. I'm a cross-section rather than time series person -- that's what I >know and feel comfortable defending. In fact, on exactly Michael's >grounds I feel less persuaded by time series work in general. But, as >the next point will hopefully make clear, we are talking about a >continuum, not a dichotomy. > >2. The issue is not utter faith and belief in econometric evidence >versus corruscating doubt. I agree that there is a wide range of >problems that should cause us to question a statistical result. >Stability is one, data quality another, measurement and omitted >variables, etc. etc. Each case is different. Careful assessment of >econometrics requires two things: a case-specific analysis of the >robustness of the results and comparison with other sorts of evidence. >(Actually, my comment in response to Eban Goodstein's cross-section >study of environmental costs and trade, included in GLOBALIZATION AND >PROGRESSIVE ECONOMIC POLICY, makes exactly these points in context.) > >Incidentally, the central role of the hypothesis testing paradigm in the >pen-l discussion is also problematic. I like what Dierdre McCloskey has >to say about this. We need to get away from the holy alter of >Significance. > >Peter > >Michael Perelman wrote: >> >> Peter took exception to my dismissal of econometrics. I guess that I >> side with Keynes in this regard. People on PKT are familiar with the >> problems of statistical analysis. In particular, you have to assume >> that the underlying system is stable. In addition, even if it is >> stable, we still have to contend with the problem of brittleness. >> >> Jim Devine mentioned Robert Gordon. I think of Gordon, as well as a lot >> of MIT economists, as being back of the envelope theorists. Although >> they generally do not rely on super-sophisticated models or >> econometrics, I tend to trust their results more than I would more than >> those of most mainstream economists. >> >> -- >> Michael Perelman >> Economics Department >> California State University >> Chico, CA 95929 >> >> Tel. 530-898-5321 >> E-Mail [EMAIL PROTECTED] > >