The Use of Econometric Models as a Guide to Economic Policy

IT IS desirable to provide tools of analysis suited for public economic policy that are, as much as possible, independent of the personal judgments of a particular investigator. Econometric models are put forth in this scientific spirit, because these models, if fully developed and properly used, eventually should lead all investigators to the same conclusions, independent of their personal whims. The usual experience in the field of economic policy is that there are about as many types of advice as there are advisors (sometimes even more!). Statistical models of the working of the economy are not proposed as magic formulas which divulge all the secrets of the complex real world in a single equation. The statistical models attempt to provide as much information about future or other unknown phenomena as can be gleaned from the historical records of observable and measurable facts. To the extent to which people maintain their past behavior patterns in the future, the statistical models provide information about the quantitative properties of economic variables in the future. However, econometricians do not operate in a vacuum; their methods are not purely mechanical in the sense that they do nothing but substitute in formulas. Any information of a qualitative nature that is available should be used by the econometrician in drawing inferences about the real world from his models. For example, suppose that an econometrician is called upon to forecast next year's level of employment and suppose further that this econometrician knows that war will break out next year. Would the econometrician merely substitute into his equations of peacetime behavior patterns in order to forecast employment in a period during which there will be war? Obviously, any qualitative information (e.g., the outbreak of war next year) must be taken into account in order to make a proper forecast. The nonstatistical economist has only qualitative information from which to make judgments. The statistical economist has this same qualitative information plus a thorough knowledge of historically developed behavior patterns; hence it may be said that the latter is better equipped.