:: Article

Prediction and the flagpole

By Emran Mian.

The sun is in the sky. There is a pole in the ground. It casts a shadow. If I know the position of the sun and the height of the pole, then I can calculate the length of the shadow. Equally, if I know the length of the shadow and the height of the pole, then I can calculate the position of the sun.

Economists might say that there is an association between the three elements. They like the term ‘associated’. Here’s Olivier Blanchard, Chief Economist at the International Monetary Fund, writing about the effects of austerity since the financial crisis. “Stronger planned fiscal consolidation has been associated with lower growth than expected,” he says in presenting the results of a new estimation model. Less technically, commenting on his research tracking the concentration of income and wealth over something like three centuries, Thomas Piketty suggests that “the democratic ideal has always been related to a moderate level of inequality”.

Yet ‘related’ is just as imprecise a term as ‘associated’. To notice what it obscures, let’s go back to the pole in the ground. While having any two of the three data points – position of the sun, height of the pole, length of the shadow – allows me to calculate the third, this ‘association’ doesn’t help me to explain what’s going on. What I mean is that the position of the sun in the sky can’t be explained by the height of the pole and the length of the shadow. The explanation only works in one direction. In other words, the length of the shadow can be explained by the position of the sun and the height of the pole, but that’s it. To describe any other relationship between the three elements as an ‘association’ is more or less unilluminating and may, in the absence of other information, be actively misleading.

The study of economics is faced with this problem – let’s call it the flagpole problem (I’m borrowing it from the philosopher Julian Reiss) – all the time. The IMF says there is an association between austerity and lower growth. Yet a range of other phenomena were associated with both austerity and lower growth during the study period. Interest rates were at or near zero. The financial system was broken. Households had high levels of debt.

If we say that lower growth is the shadow, then the problem for Blanchard is working out whether it is austerity or one of these other phenomena that is the sun. Austerity might just as easily be the flagpole. Or the sky after the financial crisis might have a couple of different suns. Or what we think of as austerity – government spending cuts – might be both a sun and a flagpole. For example, the reduction in government spending might have a direct effect on reducing economic activity; and we might respond to knowing that government spending is going down by cutting our own spending. If the government isn’t confident about spending, then perhaps we shouldn’t be either. In that sense, the height of the flagpole might be changing at the same time as the sun is moving.

That’s probably enough to make my point. I’m playing up the complexity because often the study of economics does not. The association between austerity and lower growth is the ‘result’ even though there are a range of difficult questions about whether this result conveys any knowledge about the world or not. Yet it’s the result that gets written into the article abstract and because we live in a peculiarly economics-friendly public sphere it’s the result that shapes news reporting and policy making.

The contrast with how the claims of political philosophy are treated couldn’t be sharper. A follower of John Rawls, for example, might say that the problem with austerity is that it violates the difference principle. That is, in a setting where austerity changes the outcomes for a range of people at different income levels, it doesn’t provide the highest level of benefit to those on the lowest incomes. Rawls might say that as such austerity goes against the requirements of justice as fairness.

Intuitively, for many people the claim that austerity is unfair may have just as much salience as the claim that it is associated with lower economic growth. But try saying it and immediately you have to define what you mean by fairness. When you’ve done that, you probably have to say why your idea about fairness should matter more than someone else’s idea about fairness. Isn’t it just a matter of opinion? At the end of that discussion, it’s likely you have to explain why fairness should even be a constraint on the exercise of power in a democracy. If an elected government has the votes to implement austerity, then what’s the problem, it can’t be unfair if we voted for it.

Don’t get me wrong, a political philosopher should be expected to respond to these questions but my point is that the epistemological conditions for stating the result of an economic analysis of austerity are glossed over whereas the result of the philosophical analysis is contestable in all the fundamental ways.

A friend of economics should say at this point that my critique is about how economic analysis is read by others. Economists themselves don’t shirk the difficult questions about whether they’ve discovered truth or not. I run a largely economics-driven think tank. I work with economists every day. They don’t shirk, it’s true. Blanchard, for example, checks whether his results are “robust to controlling for additional variables that could plausibly have triggered both planned fiscal consolidation and lower-than-expected growth”. In other words, he tries to figure out whether he’s mistaken the flagpole for the sun or if there are a couple of suns in the sky.

What Blanchard finds on applying controls is that austerity is associated with lower growth across a range of countries even when those countries have different starting debt levels or varying levels of damage to their banking sector. But this leaves three questions: (a) does one of those other differences have an association to lower growth that is closer than or comparable in closeness to that of austerity; (b) is the association of a combination of differences closer than that of austerity or any other single difference; and (c) what’s the explanation – or as economists like to say, the mechanism – by which the association arises? In other words, the first two questions ask what if and the third asks why.

From my reading of his results Blanchard is on solid ground about the what ifs. This is why I picked his paper as my key exhibit. It is exceptionally thorough, more so than the vast majority of economics papers. To give a sense of the contrast, Piketty for example doesn’t control. The relationship he suggests between democracy and inequality might just as easily be a relationship between the division of labour and inequality or between human nature and inequality. Or it might be that the division of labour leads to democracy. Or that human nature does. Or that the natural inequalities between us – intelligence, physical dexterity, family – lead to everything else.

So Blanchard is unusual but he even he doesn’t have a lot to say on the mechanism, or the explanation for why one thing is associated with the other. This isn’t necessarily so much of a problem when it comes to austerity. Perhaps the mechanism is common sense to most people. Though, revealingly, it is contested within economics. In other areas of economic analysis, the understanding of the mechanism is far weaker and yet economists can be remarkably incurious about discovering it. Some advocates of quantitative easing and other unconventional monetary policy even get annoyed if you press them too hard to explain the mechanism by which their monetarism leads to the promised outcomes – they think it’s more or less irrelevant. The Bank of England has taken a different approach, presenting as many as twelve different mechanisms by which QE might work.

In some sense this problem of the mechanism is present in the physical sciences too. As Thomas Nagel has shown in his book Mind and Cosmos the mechanism by which natural life evolves is pretty mysterious and its implications for metaphysics are profound. Okay so there’s no proof for God in nature but survival of the fittest may not be the only mechanism either. As a simple matter humans don’t have to be anything like as smart or possessed of reason as we are merely in order to survive nor do many of our extras – such as consciousness – provide much in the way of natural advantage.

Some physical scientists refuse to engage with these arguments too. But the main difference between economics and other scientific disciplines is that the others are far less implicated in prediction. It’s the predictive claims of economics in particular which bring into focus the lack of epistemological rigour and the all too typical failure to understand the mechanism. The point isn’t merely that prediction is risky when we don’t know the mechanism. This is obvious. For example, Blanchard and the IMF seem to have been wrong – too sanguine – in predicting the effect of austerity on growth shortly after the financial crisis, before they looked again and Blanchard produced the paper I’ve been quoting. Nor is this a trivial form of risk. The predictions of economics frequently have large real-world consequences when they are used for policy-making.

But the problem of prediction in economics is even worse than this suggests. Prediction without resolving the flagpole problem or an understanding of the mechanism is like looking at the sequence of prime numbers that begins ‘1, 2, 3 . . . ‘ then, without having an explanation of prime numbers, predicting that the next figure is 4. This isn’t merely wrong, it’s incoherent. Luckily, it doesn’t happen all the time in economics, though it does happen with depressing regularity.


Emran Mian is a retired philosopher of law who runs a think tank and writes novels.

First published in 3:AM Magazine: Saturday, November 29th, 2014.