1 The Nature of Econometrics

Economic theory makes statements about how variables are related. The demand for a good falls as its price rises; consumption rises with income, but by less than income does; output depends on labour and capital. Statements of this kind give the direction of an effect and sometimes its form. They almost never give its size. By how much does food consumption fall when its price rises by 10%? Does a 1% rise in capital raise output by more or less than a 1% rise in labour? Are returns to scale constant? These are quantitative questions, and they can only be answered from data.

Econometrics is the discipline that answers them. Ragnar Frisch, who coined the word and founded the journal Econometrica in 1933, described it as the unification of three things: economic theory, which says what to look for; mathematics, which states the theory precisely; and statistics, which confronts the precise statement with data. None of the three is econometrics on its own. Theory without measurement cannot say how large an effect is. Statistics without theory produces correlations that nobody knows how to interpret.

The defining difficulty of the subject is that economic data are almost never produced by an experiment. A chemist who wants to know the effect of temperature on a reaction can hold everything else fixed and vary the temperature. An economist who wants to know the effect of a price on demand must take the prices and quantities that markets happened to produce, in which price, income, tastes and a hundred other things all moved at once. Much of this course is about the conditions under which such observational data can still answer a causal question, and about what goes wrong when those conditions fail.

Questions on this section

Stuck on something here? Ask below and it stays attached to this topic.