Running Regressions: Some Psychological Constraints

Two effects interact to limit the validity of econometrics: cognitive dissonance and empirical arbitrariness. With cognitive dissonance, there is a psychological motivation inappropriately to manipulate what are seen as ‘wrong’ answers. This is possible when many data/modelling possibilities exist allowing econometricians ‘room to manoeuvre’ in their research strategies. Examples from investment analysis illustrate that, under such conditions, econometrics does not necessarily facilitate the empirical separation of competing theories. One solution is for econometricians to estimate a number of theoretical alternatives rather than just one preferred theory, e.g. by using the Hendry and Mizon encompassing approach and the associated non-nested tests.