An interesting question in economics is the proper role of statistics. Ludwig von Mises is my favourite economist, yet he held that statistics have no valid role in formulating or validating economic theory. It’s interesting to explore his reasons.
First, Mises held that for economics, unlike the physical sciences, there is no laboratory where one can perform repeated experiments while holding various aspects constant. Each day brings a new and unrepeatable constellation of innumerable economic causes and effects, each in varying degree. Thus, historical statistics alone cannot give us economic principles. In fact, it’s the other way around: economic principles must be brought to bear in order to analyze and understand economic statistics and history.
Second, Mises argued that the economist is in the opposite position to the physical scientist. The physical scientist observes effects, and must use experiments and measurement to reason back to the causes that are producing those effects. In contrast, the economist knows from the start the fundamental causes of all economic phenomena: individuals acting to pursue their self-interest. What’s not (initially) known is how exactly these causes interact and play out to result in various economic phenomena. This is what the economist must reason towards.
This point leads to an important methodological principle of Austrian economics, what they call “methodological individualism”. This principle holds that every economic concept and principle must be reducible (and reduced) to the choices and actions of individual people. (This is in contrast to e.g. Keynesian theories, where floating “aggregates” act upon and “cause” one another.)
Granted, Mises’ economic theories are set upon his untenable theory of “praxeology”. But I am of the firm opinion that most, if not all, of his economics can (and should) be reset upon Objectivist foundations. For example, the principle of “methodological individualism” is, in my view, the correct application in economics of the Objectivist principle of concept-reduction.
Economic theory aside, I do think statistics is enormously valuable in the practice of business and the financial markets. While it’s true that humans have free will, and might choose differently even facing the same circumstances, usually there are broad patterns of behavior that can be counted on to endure, within a certain degree of variation, for a certain length of time. Statistics can help to quantify these patterns. If this were not the case, no business could rationally plan what to produce for the future.
As an example, while it’s possible for every American to suddenly choose never to shop at Wal-mart again, it’s false to assert that Wal-mart’s sales next year could just as easily be down 95%, as up 5%.