Concepts like understanding and meaning are usually associated with a particular view of the Social Sciences. Social life produces and reproduces symbolic meaning. Social scientists need to acquire an understanding of the inherent symbolic meaning in social life. They do this, it is said, by adopting the viewpoint of a passive participant observer. In this view, the role of the social scientist is seen as distinctly different from that of the natural scientist. The object of study of the social scientist is society, the network of social interactions. Society does not exist outside the bracket of social interactions. The social sciences deal with the pre-interpreted world of the social participants. The social scientist interprets a social world, which already carries symbolic meaning. The symbolic meaning of the social world is produced and reproduced by the social actors. The study of the social world by social scientists is a matter of human subjects studying other human subjects. It is a matter of symbolic dimensions meeting other symbolic dimensions, a subject-subject relation.— Friedel Weinert (2004, 75) The Scientist as Philosopher. Springer-Verlag.
MUST WE SCRAP ECONOMETRICS?
(….) Keynes, of course, was scathing in his criticism of econometric modelling; a field which first emerged in the late 1930s as his theories were gaining traction. He likened it to ‘those puzzles for children where you write down your age, multiply, add this and that, subtract something else, and eventually end up with the number of the Beast in Revelation’ (Keynes 1939, p. 562). (….) Estimating models, whether to test the models themselves or make predictions about the future, is an awful and embarrassing game, and it is high time that economists gave it up. It is also a desperate waste of time. There is so much real policy work to be done; so many real issues to be examined and studied; but with the current impetus to do macroeconomic modelling, many economists are literally contractually obliged to engage in make-work. The most unfortunate and cynical thing is that many of those who are seasoned from working in this particular field know just how bogus it is. (Pilkington 2016, 300-302)
Does this mean that all econometrics should be scrapped? Not really. Keynes’ pointed criticisms of the field have been roughly felt by the discipline but those that read the paper often miss a comment at the end. Keynes writes:
This does not mean that economic material may not supply more elementary cases where the method will be fruitful. Take, for instance, Prof. Tinbergen’s third example—namely, the influence on net investment in railway rolling-stock of the rate of increase in traffic, the rate of profit earned by the railways, the price of pig iron and the rate of interest. Here there seems a reasonable prima facie case for expecting that some of the necessary conditions are satisfied. (Ibid., pp. 567-568)
Keynes had spent an awful lot of his life putting together statistics. He had done a lot of what would today be considered the ‘dirty work’ of economics. He had also written a great deal on the philosophy and methodology of statistics (Keynes 1921). He knew that there were some relationships within economic statistics that met the criteria required to use them in an econometric study. But these were extremely limited. In order to fit the bill, there had to be an immediate relationship known basically before the fact. Keynes laid this out explicitly in response to a letter from a statistician called Szeliski who worked on the problem of demand for automobiles. (Pilkington 2016, 302)
You have chosen just the sort of problem where multiple correlation methods may be useful. You are dealing with details of a specific problem where the main causes are pretty well known a priori, and where the statistics are definite and precise. The method is always full of danger, but, in my opinion, it is the kind of problem to which you have applied it rather than in those to which Tinbergen has applied it that the method is properly in place. (Cited in Garrone and Marchionatti 2004) (Pilkington 2016, 302)
‘What then’, the reader will ask, ‘is the point of running regressions? If we already know that a very immediate relationship exists, then why use econometrics?’ The answer is: because econometrics should be used less to establish causality and more so to present statistics supporting a causal argument in a clear and concise manner. Thus, econometrics is less a manner of doing empirical work and more so a means of clearly presenting statistical relationships that are basically know in advance. (Pilkington 2016, 302-303)
Take a very simple example. We know for a fact that, at the time of writing, Scotland is heavily reliant on oil exports. We know this because, among other reasons, the oil revenues are included in the Scottish national accounts and make up part of the overall trade statistics (Pilkington 2014c). We can then use regression techniques to estimate how reliant Scottish oil and gas exports are in the price of oil. (…) (Pilkington 2016, 303)
Note that the regression here is not being used to verify or falsify a truth-claim that I am making. Rather, it is used as a means to present statistical data. ‘We know’, I say, ‘that Scotland is heavily reliant on oil revenues for its trade surpluses. Now here is a number showing in a neat way just how dependent it is on the price changes in oil.’ Nor are we making a prediction using the regression techniques. Rather than making concrete numerical predictions, we might say: ‘Now that we are aware of how dependent the country’s trade is on changes in the price of oil we can discuss the dangers that there might be if the price of oil were to decline in the future.’ Again not that we are not making forecasts as to what such a future price decline may be. Nor are we making forecasts about what a given price decline will have on the trade balance (while not completely outlandish, we are already moving into murky territory here). Rather, we are just presenting the statistics and warning the Scottish that they had better keep a close eye on how they are structuring their economy because a shock to the price of oil might lead to a serious deterioration of their trade balance. This is the direction in which the usage of econometric techniques should be moving. Right now, driven by a silly need to give off an air of false precision, the profession is engaged in nothing but what Keynes referred to as ‘black magic’ and ‘statistical alchemy’. And it is far better to be roughly right than precisely talking nonsense. (Pilkington 2016, 303-304)