Philip Pilkington: Economics as Machine – The Nature and Folly of the Forecasters

By Philip Pilkington, a writer and research assistant at Kingston University in London. You can follow him on Twitter @pilkingtonphil

Too large a proportion of recent “mathematical” economics are mere concoctions, as imprecise as the initial assumptions they rest on, which allow the author to lose sight of the complexities and interdependencies of the real world in a maze of pretentious and unhelpful symbols.

– John Maynard Keynes

As children we look up to adults – both literally and figuratively. Literally, of course, these creatures are physically taller and stronger than us and seem able to navigate a world in which we ourselves sometimes find overwhelming. Figuratively, we look up to adults in the sense that they are authority figures who lay out rules that dictate what we should and should not do.

The Subject Supposed to Know

It is probably true that we all navigate our adult lives with reference to our early relationships with adults. How we interpret these relationships and how they are coloured is, of course, dependent upon contingencies – we may take a hostile stance toward their rules, or we may strive to embody them – but the reference point is already there and probably, if we are honest with ourselves, quite inescapable.

This raises all sorts of questions related to freedom and necessity, but these do not here concern us. What we are more interested in is that this developmental experience leaves upon all of us a very specific stamp: namely, what the French philosopher and psychoanalyst called a “subject supposed to know”. What he meant by that was that we all carry around with us an implicit notion that somewhere out there is a being who “knows” that which we do not. We might formulate that being consciously, as is the case in many religions or we may actively deny it as in the case of atheism, (although the latter often fetishise science, which then plays the same role); but again, the reference point is always already there and any attempt to escape it will just entangle us in it all the more.

The reference point itself has a very important social role in that allows humans to form into groups based upon structures of authority. Those above us in any given non-political structure are assumed to “know” what we do not and this is why, ultimately, we give them power over our lives. Thus the scientist in his capacity as scientist is but an echo of our childhood relations with those big masterful creatures to whom we submitted for those years when our psyches formed and took shape.

The scientist “knows”. What does he or she “know”? If we knew that then we would also be scientists and we would be “in the know”! Suffice it to say that the scientist “knows” in a rather abstract sense as the subject supposed to know and that we assume that we should perk up our ears when he or she speaks on their area of research. So too with the lawyer, the engineer and the doctor – although it should be said that the latter is coming increasingly under attack in our modern age for reasons that we need not get into here.

The Economist as Subject Supposed to Know

In a recent paper, the economist Jamie Morgan seeks to explore from what basis the economist can be said to “know”, insofar as he or she pontificates on future events and the policies that should be implemented in the present to deal with these events.

Morgan starts with a discussion of the subject supposed to know – which he refers to as “the Sherlock”, after the omnipotent Sherlock Holmes character of the Conan Doyle novels. Morgan then goes on to point to a number of peculiar points upon which society entrusts the subject supposed to know in any given field. He notes first of all that although the economist in his/her capacity as subject supposed to know makes pronouncements based upon probability – i.e. they try to forecast based on probabilities rather than absolute pronouncements – society always focuses on the statements they make rather than the probability of it being true. But, Morgan notes, society is only placing trust in the statement itself because they assume that the calculations used in deriving it had something to them. It is the calculations that the economist wears as their symbolic garb that allows them to adopt the position of subject supposed to know, much like a priest might wear a collar or a shaman might hold a wand.

This means, as Morgan notes carefully, that forecasts by economists could thus be acts of construction rather than of prediction and that they might be geared to reshaping the world rather than simply guessing the likely outcome – that is, the priest may have altogether more impure and worldly motives. So, Morgan argues, a particular forecaster might be concerned about a particular problem – indeed, we should add, for purely idiosyncratic or even self-interested reasons – and that they might forecast in such a way that they emphasize this particular problem ad absurdum to the point where it becomes blown out of all proportion. They will then offer a remedy to an apocalypse that they themselves have created. Such is the case, for example, in the “forecasts” made by those who want to privatise the social security system in the US at the moment.

All well and good. While we should be aware of these problems they are likely harder to protect ourselves from in the case of economics than in the case of, say, medicine. We cannot do a whole lot about this, unfortunately, except to keep on our guard against ideologues. Rather more curious, however, is that the economics profession seems allowed to make what often appear nakedly constructive “forecasts” that turn out to be completely wrong most of the time. As Morgan notes:

Successful description can actually be a basic illusion, trading on the perpetual adjustment to the modeled process of forecasting as and when new data emerges that confounds the initial basis of the initial forecast. This has been a particular feature of IMF forecasting over the last 4 years. Each new significant publication (World Economic Outlook etc.) updates prior forecasts for growth; the use of the term ‘update’ masks the basic fact that the models have been consistently wildly inaccurate. All too often the curious defense has been that the unexpected occurred!

In modern societies, in contrast to those dominated by outright superstition, our subjects supposed to know – from doctors to lawyers – would not consistently get away with this. Now, we have every right to take a cynical attitude toward this and simply say that the IMF and her sister institutions are just ideologues for the powers-that-be and that all their so-called forecasts are just so many constructions to ensure that power relations as they are in the world economy continue apace. There is certainly some truth in this, as can be amply seen in that the IMF changed its tune in a fundamental way after 2008 when the rich countries had suffered a financial crisis, while it had always remained hard-line when developing countries had suffered the same fate.

But this does not fully explain our conundrum. For one, I have no doubt that many IMF economists have good intentions, even if they do absorb their opinions from the power structure. In addition to this, Morgan discusses another example that seems to prove categorically that the problem is not as simple as political power-play. He cites the fact that many hedge funds also fall into the same trap of generating nonsense predictions that do not work. Despite the fact that this loses them money they carry on regardless and it is those that maintain more distance from grandiose predictions that come out on top:

Hedge funds and various other financial organizations have been employing physicists and mathematicians to construct [these predictive] models for quite some time; there is no evidence that their use gave hedge funds in general any advantage in forecasting the global financial crisis or in generating consistent returns (though there are hedge funds that use undisclosed proprietary trading models in fabulously profitable ways, such as Renaissance Technologies’ Medallion Fund). Those hedge funds that did profit from the crisis (such as Paulson & Co), did so based on simple scepticism regarding an ever-expanding housing market and a timely introduction to a new innovation in credit default swaps, which allowed them to short the market based on AIGs naivety as a counterparty.

We simply cannot be so cynical here as we were with the IMF. Remember these hedge funds are not generally bailed out if they take losses, so we cannot say that they use bogus forecasting techniques on purpose. By all rights it should be in their own self-interest to make better predictions and yet they continue, largely, to follow the same path. And it is those that take a sceptical view of the whole mathematical forecasting game that seem to come out on top. (Indeed, we may conjecture that it is due to the very fact that they eschew the blindfold that forecasting provides that they are successful).

A Maze of Pretentious and Unhelpful Symbols

Now let us turn to the forecasting itself and see if we can shed a little light on what is wrong with it. Economic forecasting usually comes in two strains. One involves postulating a strict model of the economy; testing it against the data over a long period and then if it seems to fit using it to predict the future. There are a number of problems with this approach.

The first, and most obvious, is, as the economist Paul Davidson has eloquently argued on this site before, that the future is not the past and it is unlikely to follow its trajectory neatly. When we homogenise past data in order to use it to prove a model we strip out everything that makes each moment in time that the data was collected different. We literally take out variety in the data as a matter of principle to get an “average”. Is it any surprise then that when something new or different happens in the future the model fails? What we are essentially doing by homogenising the data is ensuring that any anomalies, which are precisely as forecasters we should be interested in, are buried within a trend line or a mathematical average. Thus we lose sight of potential future changes not by accident, but by design. This is a very simple point but one lost on many economists, financiers and hedge fund managers.

The second problem is simply that the economist will likely squeeze the data to fit the model. If you have ever worked with modern econometric tools you will know exactly why this might occur. Working with data at such a level is frustrating and you want to make the whole thing fit together – perhaps just so that you can go on your lunch break and have done with the damn ill-fitting data! Thus, there is a very strong incentive to squeeze the data that is not related to ideology or anything else and this has powerful effects on the results. (It is also why econometric analyses are notoriously unreliable, with one analyst often getting entirely the opposite result to another analyst; thereby violating the scientific rule that a real experiment should be repeatable).

The other approach to such analyses is basically to wing it. This involves choosing data in a slightly arbitrary fashion and then seeking out patterns using advanced econometric tools. Whereas the model-oriented approach was at least consistent and open to criticism, this approach (which, so far as I can see, is far more popular) is a complete mess. We not only encounter all the problems laid out above in relation to the modelling approach, but we generate even more. On top of the above mentioned problems personal idiosyncrasies inevitably slip into the analysis. The analyst, having no real professional hypothesis to test against, track patterns based upon their own interests (which are likely influenced by whatever propaganda is being spouted in the newspaper at the time), while meanwhile ignoring those potential patterns which do not interest them. The result is often confirmation bias in its most naked form.

The Art of Economic Thinking

So, is there any alternative? While I do not feel able or obliged to provide a solid framework within such a limited space, I believe that there is indeed another approach. In the Keynesian era after the Second World War economists had a more intuitive approach to both forecasting and teasing out policy problems. They had a firm and consistent idea of how the economy operated inside their heads – but they treated their understanding as being more so knowledge of an art rather than the certainties of a science.

Instead of having restrictive models which they thought they needed to “prove” against “reality” they instead learned broad principles that they then applied in ways that they saw fit – but ways that were also quite explicit in how they were constructed and could thus be criticised. Highly complex econometric techniques were not necessary as most of the data was allowed to speak for itself and those that were applied were done so in a pragmatic rather than a dogmatic fashion.

Reading papers from this era it is obvious that this approach was far better and led to better forecasts and policy advice. But, of course, when the models that these economists – who were mostly neoclassical-Keynesians – were absorbing broke down in the 1970s inflation the whole approach fell apart quite quickly. Treating economics as an art, then, still requires a good understanding of how the economic system works and if that understanding is lacking or dogmatic in any way – as it was in both the old neo-Keynesians and their modern New Keynesian protégés – such an approach will not function because the moment something does not fit their dogma they will fall apart.

In order to analyse and forecast in a correct manner economists need to be sceptical, vigilant and open to changing their minds should the facts or the argument so dictate. In addition to this the approach should be largely based on negation rather than construction. A story should be constructed, but each narrative point should be subject to empirical scrutiny and if even one doesn’t hold up the whole story should be dropped and another one sought. Economists should understand that their theories are not rigid doctrines, but narrative frameworks and if some aspects of them come up lacking in a given situation there is no need to take this as a bruise to one’s ego. Keynes knew this, of course, and we now expand the quote with which we opened this piece:

The object of our analysis is, not to provide a machine, or method of blind manipulation, which will furnish an infallible answer, but to provide ourselves with an organised and orderly method of thinking out particular problems; and, after we have reached a provisional conclusion by isolating the complicating factors one by one, we then have to go back on ourselves and allow, as well as we can, for the probable interactions of the factors amongst themselves. This is the nature of economic thinking. Any other way of applying our formal principles of thought (without which, however, we shall be lost in the wood) will lead us into error. It is a great fault of symbolic pseudo-mathematical methods of formalising a system of economic analysis that they expressly assume strict independence between the factors involved and lose all their cogency and authority if this hypothesis is disallowed; whereas, in ordinary discourse, where we are not blindly manipulating but know all the time what we are doing and what the words mean, we can keep “at the back of our heads” the necessary reserves and qualifications and the adjustments which we shall have to make later on, in a way in which we cannot keep complicated partial differentials “at the back” of several pages of algebra which assume that they all vanish. Too large a proportion of recent “mathematical” economics are mere concoctions, as imprecise as the initial assumptions they rest on, which allow the author to lose sight of the complexities and interdependencies of the real world in a maze of pretentious and unhelpful symbols.

History, Irony and a Forecast of the Future

Alas, it is not likely that such an approach be adopted by the profession formally in the coming future. The profession has completely congealed and hardened in this regard – trapped in a cage of pretentious symbols that they themselves have built. Anything that is not thrown into an econometric black box is subject to suspicion. Thus the institutions of the profession have walled themselves in behind their own needlessly complex constructions – much like the old Soviet economic engineers in the post-Brezhnev era. The hope, if anywhere, lies paradoxically with Big Finance which, incentivised as they are by monetary gains, seem to be becoming more sceptical of the old ways and are consequently starting to accept that a new approach is needed.

And so the fate of the profession lies sealed within the great Frankenstein that the economists themselves helped construct; the monster which ultimately led to their profession’s own growing disrepute. So we might say that while history is certainly not a fixed entity easily subject to cold econometric regressions, she nevertheless has a sense of irony. Perhaps, then, it is history that gazes – coolly, ironically – on the economists and not they that gaze upon her.