Econometrics

Econometrics is the field of economics that is concerned with the application of mathematical statistics and the tools of statistical inference to the empirical measurement of relationships postulated by economic theory. That is, econometrics (hopefully) uses some clever combination of economic theory and mathematical statistics. Typically, application of econometric methods involves the following elements:

  • formulating an economic model appropriate to the questions to be answered;
  • reviewing the available statistical models and the assumptions underlying these models, and selecting the form most suitable for the problem at hand;
  • obtaining appropriate data, properly defined and matching the concepts of the economic model;
  • finding suitable computer software to enable the calculations necessary for estimating and testing the econometric model.

    The ultimate goal of an econometric exercise is to see whether an economic model is consistent with empirical (observed) behavior as reflected in the data. Note that econometrics is mostly based on large samples, i.e., on observing economic relationships over a long period of time or for a large number of individuals at the same time (or both, as in the case of longitudinal or panel data). Note also that econometricians usually have to use data that were not created in a controlled experiment (as in natural and some other social sciences). An important aspect of applied work is therefore to assess whether the sample used for estimation is actually a random sample drawn from the population for which the underlying model is supposed to be appropriate – in other words, whether the relationship of interest is empirically identified. For example, this might not be the case if there are selection problems.

    Classical vs. Bayesian methods: Note that the last paragraph is actually a description of classical econometrics. As in statistics, classical (or frequentist) methods concentrate on testing hypotheses that are derived from theory, using the data available. Bayesian econometrics (and statistics), on the other hand, stresses the role of the data itself in both development and testing of economic theories. While most empirical applications of econometrics use classical methods, Bayesian econometrics has gained importance for applied work in recent years, a fact that is partly due to the increased computer power available for computationally intensive applied work. The classical vs. Bayesian controversy extends to other social sciences as well; for an appreciation of its relevance in psychology, see e.g. Gigerenzer (1987).

    See also: identification, selection problem

    Literature: Poirier (1995) is a text that covers both the classical and Bayesian approaches to econometrics; Pudney (1989) discusses the econometric analysis of individual choice.

    Entry by: Joachim Winter

    June 17, 1999
    Direct questions and comments to: Glossary master