I was reading some of the essays in the volume onedited by Daniel Hausman, and was struck by the echo in a terrific comment by Herbert Simon of something Dani Rodrik says in . (The Simon paper was originally in the AER P&P volume for 1963, Vol 53 (1963): 229-231.)
Simon, like Rodrik, points out the logical fallacy of using an empirical observation to validate the assumptions of a theoretical model devised to explain – or at least for consistency with – that empirical observation. The assumptions must also be empirically valid, or validated, Simon argues. “The remedy for the difficulty is straightforward, although it may involve more empirical work at the level of the individual actors than ost conventionally-trained economists find comfortable.”
Market theories, and macro theories, do need micro-foundations, but empirical ones, foundations based on how people or firms behave. Do businesses maximize profits? Of course not, or at least, economists do not test the assumption before they build models on it. In my years on the Competition Commission taught me, many businesses can be blithely unaware of which of their activities make the most profit. As I’ve complained about macro models before, they might indeed have theoretically rigorous micro foundations but are ad hoc with respect to reality.
[amazon_image id=”0521709849″ link=”true” target=”_blank” size=”medium” ]The Philosophy of Economics: An Anthology[/amazon_image] [amazon_image id=”0393246418″ link=”true” target=”_blank” size=”medium” ]Economics Rules: The Rights and Wrongs of the Dismal Science[/amazon_image] [amazon_image id=”0472050079″ link=”true” target=”_blank” size=”medium” ]The Cult of Statistical Significance: How the Standard Error Costs Us Jobs, Justice, and Lives (Economics, Cognition & Society)[/amazon_image]
Simon goes on to suggest that economics adopts the scientific practice of making sure assumptions simplify but approximate sufficiently closely to the real world. He, like, bemoans the tyranny of statistical significance rather than maningfully significant: in hypothesis testing, “We do not primarily want to know whether there are deviations of observation from theory which are ‘significan’ in this [statistical] sense. It is far more important to know whether they are significant in the sense that the approximation of theory to reality is beyond the limits of our tolerance.” Unfortunately, it is much easier to read a t-statistic from a software package than to think (and apparently also too hard to consider the statistical power of regression results) so the tyranny of statistical significance continues.