The value of econometric estimation lies mainly in the fact that economists want to estimate a « ceteris paribus » effect, i.e., the effect of a particular factor taken in isolation on a variable. Suppose, for example, that an economist wants to test the impact of an increase in the average wage of a population on consumption. If the economist had a laboratory, things would be fairly simple: they would only need to change the factor they wanted to test (average wages) to see the impact on the variable in question (consumption). Unfortunately, however, the data on which economists rely do not usually come from laboratories (with the exception of experimental economics): if the average wage in a country increases, it is highly likely that other factors affecting consumption will have changed at the same time, so that the effect of the wage increase alone is not apparent in the data. Similarly, it is possible that the increase in consumption has led to an increase in average wages (this is known as the problem of simultaneity, or, more generally, endogeneity). It is therefore the aim of econometrics to develop methods that provide a « ceteris paribus » estimate of the factor being tested.
Julien P.
