≡ Menu

Quotation of the Day…

… is from pages 210-211 of the 5th edition (2015) of Thomas Sowell’s Basic Economics:

While preferences for some groups and reluctance or unwillingness to hire others have often been described as due to “bias,” “prejudice,” or “stereotypes,” third-party observers cannot so easily dismiss the first-hand knowledge of those who are backing their beliefs by risking their own money.  Even in the absence of different beliefs about different groups, application of the same employment criteria to different groups can result in very different proportions of these groups being hired, fired, or promoted.  Distinguishing discrimination from differences in qualifications and performances is not easy in practice, though the distinction is fundamental in principle.  Seldom do statistical data contain sufficiently detailed information on skills, experience, performance, or absenteeism, much less work habits and attitudes, to make possible comparisons between truly comparable individuals from different groups.

It is essential always to remember that no amount of statistical data about the economy – no matter how carefully and objectively these data are gathered, processed, and interpreted – can in practice ever capture the full amount of relevant information that is ceaselessly being gathered, processed, interpreted, and acted upon in markets.  That statisticians and econometricians seldom stake anything material of their own on their academic empirical investigations is not the only, or even the chief, source of this problem.  The chief source of this problem is the inherent limitations of statistical data themselves: the market constantly processes unfathomably nuanced and (to use a fashionable word) “granular” pieces of information – most of which are small, many of which are unobservable to (and, hence, unquantifiable by) third-parties, and all of which are subject to unexpected change; there is simply no practical way for all of these relevant details to be captured reliably in statistical data even though they are captured and acted upon routinely in markets.

Does this reality mean that statistics are useless and should be shunned?  Emphatically not.  What this reality does mean is that much more caution and humility should be exercised before using econometric findings as justification for government intervention.  The presumption that good econometric studies capture reality sufficiently to justify government intervention is merely that: a presumption – and it is too-often an unjustified one.


Next post:

Previous post: