Yes, Econometrics is Alchemy

Econometrics is considered scientific under two notions: it applies statistical method for analyzing economic data; and it is empirical. This paper refutes these two notions. As regression just calculates mathematically, instead of statistically, coefficients which are unique to a specific model and dataset, there is nonexistence of t-statistic and therefore noninvolvement of statistical method in econometrics. Econometrics is not empirical because it neither investigates particular situations nor determines kinds. On the other hand, econometrics, by letting an uncontrollable error term in regression models, allows researchers to try any models for finding the results they want. This in turn easily makes econometrics a kind of con job. For all multi regression models, econometrics is essentially working on abstract numbers of no existential reference. These defects determine that econometrics cannot be saved. Preliminary evidences show that econometrics is just alchemy. Application of econometrics to verifying asset pricing models is one of the typical illustrations of the assertion