News and brain candy for the philosophy community
So the big news is the Eurozone crisis and what to do about it. This obscures the bigger question, which is what to do about the system of international finance. I have an idea. Let’s get rid of it. Something seems simply wrong with the idea of a system of giant, closely integrated lending firms, backed up by nationally-owned central banks. New regimes of regulation, or the ‘utility model’ – where credit institutions are treated like nationalised water or electricity suppliers – are weasly halfway houses. Let’s go eliminativist. Why not? Think Distributism, without the anti-Semitism, the leanings towards theocracy, and the social conservatism. Ok, don’t think Distributism. Just think very, very different from the way things are now.
But I’m just a historian of philosophy. I have no idea what I’m talking about. I just feel like that would be the right thing to do. Before you accuse me of being naïve, however, consider what you’re accusing me of not knowing. Is the accusation that I don’t understand economics? The problem with that is that there is no reason to think that if I did understand economics I’d be any better placed to legislate for the future. At the risk of raining on a great ongoing parade, we don’t have a social science with predictive power. We don’t even, as Jerry Fodor said in a different context, know what it would be like to have a predictive social science. If I don’t know what the consequences of a policy will be, I take comfort in the fact that nobody else does either.
Remember that this is a question of knowledge. In the Meno, Socrates reminded Anytus that Pericles himself could not teach his sons to be statesmen. That meant that Pericles did not himself know how to be a statesmen. He had true opinion with respect to leadership. But to have had knowledge he would have to have known why his opinions were true. And if he’d known that he could have taught it to his sons. Now, many governments have managed to guide their countries into periods of high growth and low inflation. Supposing that that really is what we all want, how did they do it? If we knew, we’d be doing it too. And we’re not. Ergo, there is true opinion about these things but no knowledge.
Glancing at the Times opinion pages yesterday, I saw one piece by Matt Ridley, arguing that the strategies of the rich democracies has been the opposite of what it should have been: regulating the finance sector and deregulating ‘the market’ (meaning the markets in everything besides credit). That feels plausible enough, but so does my idea. Ridley doesn’t know his plan would help anyone. I then saw another piece, by Ed Balls, arguing that the announced VAT rise by the British government weakened consumer confidence in a way that undermined its plans for deficit reduction. Plausible again. But he doesn’t know it. Neither – perhaps especially not – does the government he attacked know that they are right in rejecting his attacks or anybody else’s. They blame unforeseen contingencies for the failures of their policy so far, but this is precisely the point: when a model can’t take into account the relevant contingencies, there is the telltale sign that the theory driving the model has no predictive power.
When any of these figures defend their predictions and associated recommendations, they do so, inevitably, in terms of neoclassical economic theory. I don’t mean here to criticise neoclassical theory as such; there are many things we know that we wouldn’t know without it. But as a predictive social science, it faces a great difficulty, which is the threat of triviality. It tells us that people will, on the whole, adopt the most rational course of action in a given economic scenario. But, for that very reason, it must regard the course of action actually taken by economic agents, whatever that may be, as the rational one. What you would need to make neoclassical theory predictive would be some local criteria, in each case, determining what should count as the rational course of action, beyond the vacuous criterion of being the course which most economic agents will take. But here neoclassical theory is stuck with either making predictions that are so generic they are useless for guiding policy (imposing a price ceiling on a market may cause some kind of shortage), or so specific they are inevitably false (the prospective shortage is represented by this demand curve, over this exact period, calculated using these exact PED values, etc.).
I am not making the behavioural economist’s criticism, that people are irrational (though predictably so) in many situations. What I am proposing is that even granting the standard neoclassical assumptions like perfect rationality and symmetric information the theory would have no predictive power. Indeed the theory has rather impressive retrodictive power (Ben Bernanke would have been a great person to have around during the Great Depression). But this is actually a bad sign regarding its potential predictive power: the possibility of retrodiction shows that the problem is not that the theory doesn’t handle the data properly, but rather that we can’t access it ahead of time. And this may be for a variety of reasons, not least the clichéd problem that the conditions that provide the environment determining rational decisions are continually modified by those decisions. Moreover, taking Alex Rosenberg’s line, the rationality assumptions themselves predict an inevitable arms race between social scientists and those agents who stand to gain from not having their behaviour predicted (think hedge funds, then generalise).
Now bear in mind that it is not only governments that have to build predictive models based on neoclassical theory (or neoclassical theory kitted up with behavioural and experimental innovations). Banks need to as well. Banks cannot simply respond to prices; they have to make guesses about the future of the economy. And every time they leverage up risk, they layer on more guesses. In fact governments in the rich democracies pretty much are banks; the aim of their strategies is to profit (in votes rather than in money) by correctly predicting behaviour in the general economy. (This, incidentally, is one argument against treating banks as public utilities.) The reason they fail so often is because they guess so much. Wouldn’t it be a nicer world if we didn’t all have to live with the consequences of these astronomically scaled guesses?
I suppose this all sounds a little Austrian, so let me do some distancing. Ludwig Von Mises’ 1949 book, Human Action, sets out a case against the idea of a predictive social science. But his conclusion is that economics ought to be regarded as an a priori science, comprised of analytic judgments following a set of intuitions about human rationality that are taken as axiomatic. Here I can’t follow, because I think that if people were made in accordance with my intuitions about rationality, economics wouldn’t exist. ‘All men are mad’, A.J.P. Taylor once wrote, ‘who pursue power when they could be fishing, or painting pictures, or just sitting in the sun.’ Replace ‘power’ with ‘profit’ and the point, to my mind, stands even firmer. If economics exists at all it must be an a posteriori science, and as such it faces the difficulties Von Mises noted (as did Hayek in his classic paper ‘The Use of Knowledge in Society’). Anyway, the Austrian position is too strong; I don’t think there are insurmountable epistemological barriers to a predictive social science. I just claim we haven’t surmounted them, and we can’t imagine what things would look like on the other side.
Readers of my last post on this blog, if there were any, would complain that I’ve contradicted myself. I don’t feel compelled to reply to criticisms from such a distant possible world, but if I did I would say this: There I supported top-down social policies based on large-scale aggregative models, as opposed to localist, ‘Big Society’ ideology. I still think the former comes out better in the comparison. I haven’t even changed my mind about the ‘based on aggregative models’ part. I just no longer think we have a clue how to build them.
But remember that the primary impulse behind social reforms has never been social science. The great reforms of the late-nineteenth and early-twentieth century did not come out of Utilitarianism or welfare economics. They came out of general instincts, such as those which told people that making children work in factories is pretty demented when you think about it, or that Dickens seemed to have a point, or that women don’t really seem very stupid at all, or that watching your children die slowly seems an excessive punishment for not being able to find a job. Of course the justification behind rolling back so many social reforms today lies in the belief that pursuing increased prosperity will make all measures to protect the vulnerable unnecessary. It would be wonderful if anybody had the vaguest idea how to ensure prosperity. Until we do, I don’t think it would be a terrible idea to switch back from ‘outcome-oriented’ approaches to doing what we feel is right. If somebody points out here that there is no reason to trust my instincts as reliable guides, I reply that this is something I really do know – better than you ever could. But at least when I guess I know I’m guessing.
When a body of theory is full of elegant models and formidable equations it doesn’t follow that it explains anything, let alone predicts anything. If governments and banks were setting the course of our lives according to some favourite theory of quantum gravity, we would, I think, regard them as being somewhat irresponsible. And yet what is ‘border science’ in physics is far closer to complete than what is guiding orthodoxy in the social sciences. The arguments are as strong as ever that our social science still has the status Kant ascribed to metaphysics: that of blind groping in the dark. To be fair, that sounds like a great party. But I think the party is over.