Goodhart’s law

In the early 1970s central banks increasingly began to adopt monetary targets as an intermediate, and potentially manageable, variable in pursuit of their final objective of controlling inflation. Naturally each country that did so, including the United Kingdom, tended to choose that particular monetary aggregate that, up to the date of choosing, appeared to have the most stable relationship with nominal incomes, and hence inflation. By 1975, however, these econometric relationships had in many cases broken down, not only for most demand-for-money or velocity relationships, but particularly so in most countries for that aggregate chosen as the monetary target. While some decline in (predic- tive) relationship might have been expected in light of the disturbances of 1973-74, for instance the oil shock, sharp rise in inflation, house/property boom/bust, sharply varying interest rates, and so on, what was remarkable was that it was in the case of the chosen targets where the breakdowns seemed most extreme. As Governor Bouey of the Bank of Canada is reputed to have said, "We did not leave the monetary targets; rather they left us."

It was that observation that led me, at a Reserve Bank of Australia conference in Sydney in 1975, to the comment that the breakdown of such relationships accorded with "Goodhart's Law, that any observed statistical regularity will tend to collapse once pres- sure is placed upon it for control purposes". The best source to find this quote now is Goodhart (1984). It was intended as a humorous, throwaway line, and, unlike the Lucas critique, was not based on some deeper underlying analysis, just some limited empirical observation.

Some ways of describing relationships catch on, whereas others do not. Although I had never expected this semi-jocular statement to become regularly used, and moreover used seriously, it was taken on in a broad range of cases, mainly in the social sciences and mainly in the United Kingdom, as an explanation why the translation of prior statistical relationships into control targets so often led to the breakdown of the prior relationship. Goodhart's law, of course, encapsulates the core of the Lucas critique, which is that a change to the control mechanism will elicit a change in the behaviour of the controlled, as they will usually now have an incentive to adjust their behaviour so as to meet the target. But it is also rather more general; in particular the authorities having set a target often feel an incentive to show that their targets are met, and can either change their own behaviour or forebear when they perceive the regulated is manipulating the outcome, so as to be able to claim a "success".

Anyhow, the common validity of the concept was clear, and the presentation of Goodhart's law seemed simpler than that of the Lucas critique, and so was widely taken on, and became elevated, again by others (not by me), into a serious component of the social sciences, particularly in the United Kingdom. It was dignified, for example, in the paper by Chrystal and Mizen (2003) on "Goodhart's law: its origins, meaning and impli- cations for monetary policy'", and has been extended into other social sciences. Thus, as noted in Wikipedia, Keith Hoskin (1996) has illustrated its broader applicability. See also Strathern (1997), who restated the same concept noting that "When a measure becomes a target, it ceases to be a good measure".

See also:

Monetary aggregates; Monetary targeting.