Reporting Econometric Results: Believe It or Not?

Econometric results have come under increasing suspicion because of the use of the data-dependent approach to econometric modeling. Moreover, the currently used provincial approach to the problem suffers from: 1. statistical problems arising from repeated experimentation with the data, and 2. the inability to interpret the validity of results produced by a subjectively selected final model. Alternative solutions exist in the classical shrinkage estimators and in the Bayesian paradigm. However, these approaches are likely to be beyond the interest of the average econometrician, and access to the necessary computer programs is limited. A more viable alternative is an operational compromise in which there is more complete reporting of numerical particulars from an adopted search procedure, other than just the final model. Such an approach would generate more credibility and acceptance of econometric results.