Session Topic: The Flow of Funds and Interest Rates-I: U.S. Financial Models: Discussion

Econometric model building of the type represented in the Duesenberry-Bosworth paper was at best an attempt to see whether systems of, say more than 25 equations, add much to our knowledge of the economy and the response to policy changes. They have not, I believe, at least to this point, and I am not optimistic about a change in the next few years. Instead of commenting on the details of the particular models presented at this session, I will explain some of the reasoning that brings me to that conclusion. Before doing so, let me make clear that my comments are not directed at either "theory" or empirical science. There is very little of either in current econometric practice. I view my comments as a criticism of large models, models of twenty-five or more equations, not as a criticism of modelling or testing economic hypotheses.