The world would be a happier place if official economic statistics had never been created. They are often inaccurate or otherwise flawed and misleading. An even more serious consideration: official statistics help to provide rationales for pernicious policy making. FULL ARTICLE by Robert Higgs

Source link: http://archive.mises.org/15406/the-trouble-with-economic-statistics-2/

# The Trouble with Economic Statistics

Previous post: The Nightmare of 1923 and Its Cause

Next post: My Reply to Krugman on Austrian Business-Cycle Theory

{ 2 comments }

An interesting article, but I have one critique. In commenting on poverty measures, and the concept of the “distribution of income” Higgs states that:

I think this is quite an unfair criticism of statisticians (full disclosure: I am one) and their analysis. Higgs suggests that it was wrong for statisticians to have constructed the “distribution of income” and that this information is unnecessary, serving solely to “feed the fires of envy”. Both of these assertions are false.

Firstly, if one is studying macroeconomics, and wants empirical information pertaining to this subject, it is perfectly reasonable to be interested in information on people’s incomes. Higgs seems to suggest that this information should be off-limits, even at an aggregated level. Such information may well be unnecessary for the conduct of a just government, as Higgs states, but it is certainly useful for

something, and hence, it is a valid object of study. (The fact that Higgs admits to having used statistical information on these kinds of topic in his own work demonstrates the point.)Secondly, once one concedes that incomes are a useful object of study, the notion of the “distribution of Income” is a perfectly legitimate statistical term. The word “distribution” in this context, refers to the “empirical distribution function” or “probability distribution” of the relevant quantity (roughly speaking, this is the shape of the histogram of that quantity). Probability distributions are the basis of probability theory, which underlies statistical science, and have been around as a statistical and mathematical term for at least a few hundred years. The concept of a “distribution” is central to statistical science. To my knowledge, the concept predates the modern drive for official statistics on income, by quite a margin.

If one concedes that statistical science is useful in empirical analysis (if not, then presumably the critic has an alternative theory of quantitative inference that they can demonstrate is superior) then it is perfectly sensible to use standard statistical theory and statistical tools to analyze a subject of interest, such as incomes. The “distribution of income” (presuming it is calculated correctly) is a very useful piece of information: it shows what proportion of the population fall within any given income bracket, which is indispensable to its analysis.

The problem with the use of the “distribution of income” statistic is nothing to do with any problem in statistical science, or anything done by statisticians. It is to do with the fact that people who are unaware of the meaning of the term may incorrectly take the word “distribution” to imply that the income is not properly earned or rightfully owned, but is just “distributed” to people arbitrarily. This is a problem of

equivocation, not a problem of statistics or economics. The solution, as with all problems of equivocation, is to make clear, when using the term, that the word “distribution” is solely a technical term referring (roughly) to the shape of the data histogram.In analyzing income quantities, statisticians are perfectly correct to use standard statistical tools such as empirical distribution functions and probability distributions to look at these quantities. In technical statistical work on the subject, the concept of the “distribution of income” would be indispensable to any empirical analysis on income quantities. (Since Dr Higgs has done data analysis before, I suspect he is all too familiar with concepts like the “probability distribution”, “empirical distribution function”, etc.)

It would be rather a tall order to overturn standard terminology in probability theory and statistical science (which as been around for hundreds of years) in order to try to avoid a minor problem of equivocation in a specific type of empirical analysis (i.e., the analysis of incomes). Thus, the most that can be said in criticism of the concept of the “distribution of income” is that it is a technical term that may give false impressions to the layperson, if not properly explained.

Whatever critiques can be leveled at statisticians (and I do not see any sensible critique in Higgs’ discussion of the distribution of income), they certainly do not calculate their concepts to “feed the fires of envy and political rapacity”.

Also, I note Higgs’ criticism of the fact that official statistics may

notbe calculated correctly. This is certainly true, and a valid critique, but is not relevant to the above statement.Comments on this entry are closed.