Tuesday, 4 July 2017

Getting our facts right

A few weeks ago I was involved in a debate with a young analyst who refused to believe that exchange rates are driven by factors other than trade deficits (not current accounts, simply the flow of trade in goods). After fruitless attempts to try and engage in some form of intellectual debate, only to be met each time with the stock response “I disagree,” I simply shut down the conversation. This is not my preferred mode of interaction – far from it. We learn from discourse and I like to think I am open to changing my mind on various issues if the facts prove I was wrong.

It was in this vein that I read with interest a blog piece by Noah Smith entitled “Is economics a science?” "Real" scientists would treat the question with contempt and indeed I never try to claim that it is. But what economics tries to do is measure and draw inference from observation. In that respect it employs scientific methods even if it does not always result in scientific conclusions. One reason why the theory and practice differ so much is that the logical economic answer is not always politically acceptable. Economics also has deep philosophical roots which colour the prior beliefs of many practitioners. Indeed, one of Adam Smith’s noted works - admired by many on the political right - was a Theory of Moral Sentiments published 17 years before the Wealth of Nations. It is perhaps these philosophical underpinnings which explain why adherents to the Austrian school of economic thought, which also derives from a branch of philosophy, eschew empiricism in favour of a priori deduction in order to reach a conclusion.

I could not help thinking during the Brexit debate last year that many of the leading Brexiteers were adherents of free market economics of the kind espoused by the Austrian school. It therefore does not surprise me that many of their arguments were not backed up by empirical analysis. I have also been struck by the apparent shift in tone of those who 12 months ago supported Brexit. Only today, the campaign director of Vote Leave, Dominic Cummings, admitted that “in some possible branches of the future leaving will be an error”  (let me correct you there, Dominic. In pretty much all branches of the future leaving will be an error). Cummings appears to be directing much of the blame for this on the way it has been handled by Downing Street. Personally, I prefer the explanation that those responsible for promoting the cause did not do their homework and failed to think through the implications of their actions.  In other words, they adopted a very unscientific approach.

However, we also have to be very careful when making arguments based on data alone. One of the issues which the academic world is currently very concerned about is the accuracy and replicability of much (non-economic) scientific work. Only last week, the president of the Royal Statistical Society, Professor Sir David Spiegelhalter, pointed out that public trust in scientific conclusions is being undermined by a “failure to adhere to good scientific practice and the desperation to publish or perish.” As Spiegelhalter points out, most scientists do not overtly falsify their data, but they sometimes play fast and loose with statistical inference (credit should also go to The Economist for having made this point repeatedly in recent years).

Aside from problems arising from the accuracy of results, economics suffers from another problem due to the quality of the underlying data. Although I do believe that economic statisticians are free from political bias, economic data often suffer from sample bias due to the fact that it is constructed by drawing population inferences from a relatively small sample. It is often an approximation to reality at best. A case in point is UK labour force data, where a tightening of the criteria for benefit eligibility means that many people whose fitness for work is questionable, have been reclassified as part of the labour force. UK immigration data are also not fit for purpose either, despite the fact that they form a key element in the government’s Brexit strategy (amongst other reasons, because the UK does not require migrants to register after arrival, the figures are compiled from the International Passenger Survey, which has numerous methodological shortcomings).

But for all that, a debate based on some form of data is always more informed than one based purely on belief and supposition. As the Canadian academic Marshall McLuhan pointed out, “a point of view can be a dangerous luxury when substituted for insight and understanding.” A year on from the Brexit referendum, that rings all the more true.

No comments:

Post a Comment