Reinhart-Rogoff and the Philosophy of Macroeconomics

Paul Krugman has reacted to the Reinhart-Rogoff scandal:

You can already see quite a few people reacting to this affair by declaring that macro is humbug, we don’t know anything, and we should just ignore economists’ pronouncements. Some of the people saying this are economists themselves!

But the truth is that basic macroeconomics — IS-LM type macro, the stuff that’s in Econ 101 textbooks — has performed spectacularly well in the crisis.

It is good for philosophers of economics when an influential economist sets out to defend his discipline against methodological criticisms. (My main points about this are in the last three paragraphs – if you’re not interested in IS-LM please skip to those.)

But a return to IS-LM? Is that really the main moral we ought to draw?

First of all, in what sense has ‘IS-LM type macro’ performed spectacularly well in this crisis? Krugman lists the crucial predictions the IS-LM framework has made:

Unprecedented budget deficits, the model said, would not drive up interest rates. A tripling of the monetary base would not cause runaway inflation. Sharp government spending cuts wouldn’t free up resources for the private sector, they would depress the economy more than one-for-one, so that private spending as well as public would fall.

But one problem with the claim that IS-LM has such impressive predictive power is the sheer amount of possibilities for which the model allows. John Hicks first shared his invention, the IS-LM model, in a paper called ‘Mr. Keynes and the Classics’. One reason he gave it that name is that it allows for both Keynes and classic, Marshallian economics to be right.

Take a look at the basic IS-LM model (anybody wanting an explanation of it should click on the link to Krugman’s explanation in his quotation above). The IS curve represents the various possible combinations of interest rate and level of GDP at which the rate of investment and of savings (the two rates held, under equilibrium conditions, to be equal) can be preserved. The LM curve (Hicks called it LL – that’s not relevant) represents the possible combinations at which a single level of liquidity preference and money in circulation, again, in equilibrium, assumed to be equal, can be preserved. The idea is that under normal circumstances the curves will be shaped like this (I’ve pinched these diagrams from Krugman’s blog – I hope that’s ok…):

So far the model doesn’t make the predictions Krugman wants, particularly the prediction that (as Hicks put it) ‘increasing the inducement to invest [will] not [raise] the rate of interest.’ Thus Hicks added his ‘liquidity trap’. The LM curve is basically flat at one end, since once people are indifferent between holding cash and holding bonds, there is no more effect that a lowering interest rate can have (Hicks, for reasons I think are important but won’t go into here, didn’t give that explanation; he just said that ‘there is some minimum below which the rate of interest is unlikely to go’). When we’re up against that lower bound, the situation looks like this:

I don’t mean, of course, that Krugman’s ‘IS-LMentary’ blog (his pun, I’m afraid, not mine) is meant to express everything he means by ‘IS-LM type macro’. But if we simplify the story to this extent, we can see clearly how questionable it is to treat this framework as having real predictive power. It predicts a situation where increasing the inducement to invest will raise the rate of interest (even close to the lower bound – look at the ‘elbow’ of the LM curve), and it predicts a situation in which this won’t be the case. Thus the predictions begin to look trivial; it’s easy to make correct predictions when you’re allowed to predict two basically opposing results.

Moreover, the IS-LM framework, if it can be used for policy guidance, makes false predictions. If the IS-LM equilibrium determines the interest rate and the level of output, then the government shouldn’t be able to shift the curves around, for then it is no longer true that the equilibrium determines the result. But all the predictions Krugman lists involve the results of the government shifting curves around: shifting the IS curve eastward by deficit spending, for example, or moving LM around with monetary policy. We could, of course, treat the government’s action as an exogenous factor not represented in the model. But then why can’t the actions of firms and households be similarly exogenous? If the government can push the IS curve eastwards by deficit spending, why can’t a group of individuals do the same thing by collectively spending beyond their means? Hicks acknowledged this problem in a later repentant article:

There can be no change of policy if everything is to go on as expected – if the economy is to remain in what (however approximately) may be regarded as its existing equilibrium.

Hicks made another self-criticism that is relevant here. His model doesn’t explain why there should be a liquidity preference at all. ‘[T]here is’, he wrote, ‘no sense in liquidity, unless expectations are uncertain.’ This is a point that Keynes, whom Hicks originally pretended to be formalizing, had made very clearly: liquidity preference emerges out of people’s uncertainty about the outcome of investments. But now the problem is that if expectations are uncertain, the whole equilibrium analysis provided by IS-LM is obviously wrong. If there were an assured equilibrium, there would be nothing to be uncertain about.

Of course there could be some mysterious reason why an analysis like IS-LM predicts macroeconomic phenomena, even when the equilibrium it assumes doesn’t really exist. Milton Friedman famously argued that it is the predictive power of a theory that matters, not the realism of its underlying assumptions. But the biggest problem with that argument is now well known. We can’t really assess the predictive power of a theory without knowing the accuracy of the causal story grounding the predictions. To use Daniel Hausman’s analogy, you have evidence that a used car has a good engine when you drive it a few times and it goes well, but you’d reserve your final judgment until you’d opened up the bonnet and had a look. When you open up the bonnet of IS-LM you see an explanatory engine that can’t possibly be telling the right causal story, since part of its story involves an equilibrium that doesn’t really seem to exist – and can’t exist unless one of its curves doesn’t actually represent what it’s meant to represent. But if it’s not tracking the real causes, its predictive power might rest on temporary correlations and local invariances; not something to bet your life on.

Moreover, it isn’t as though the predictions of IS-LM can’t be made using other frameworks. As I hinted above, Keynes (the real Keynes – not the Keynes of the Big Lie) explained the fact that increased inducement to invest doesn’t always raise the rate of interest in terms of the uncertainty of expectations. Indeed, there are any number of informal, behavioural ‘animal spirits’ type explanations, requiring no shaky equilibrium assumptions, that can be offered to explain recently observed macroeconomic phenomena with a good deal of psychological plausibility.

The main lesson here is that if macroeconomics is to draw the right lessons from the crisis and develop in the right way then the question shouldn’t be which available model is the best. It should be what the hell is going on. IS-LM doesn’t seem to get the right answer to that, even if it is a nice irony that it in some ways it does better than the high-tech DSGE models meant to replace it or supposedly empirical studies like Reinhart-Rogoff.

But – this is the important point – Krugman is right to reject the counsel of despair that some have taken as the moral of all this. Economics remains the only real tool we have for understanding the complex set of situations we’re in. It is very tempting these days to say that macroeconomic models are mere mathematical masturbation, and that the causes of the crisis are obvious – greed, irresponsibility, political ineptness, not fixing the roof (what roof?) when the sun (what sun?) was shining, etc. But that is not a scientific analysis; nor is it a scientific analysis to appeal to some vague Marxist theory, heavy on consequences and sparing on details, that capitalism is just inherently flawed.

Nobody has ever understood something as complicated as the globalized macroeconomy using such rudimentary theoretical tools as knee-jerk reactions and a general sense of cultural feel. Suppose somebody said: ‘Aeronautical engineers are spouting meaningless maths; stick a couple of wings on her and away she goes’. Suppose, when the plane crashed, somebody said: ‘Of course it crashed; the thing is riddled with internal contradictions – stop trying to bamboozle us with your mumbo-jumbo’. I know that macroeconomics is bound up with political vested interests, but so, I suspect, is aeronautical engineering, all of which is paid for either by the military or by an oligopoly of giant firms. Impatience should never overpower curiosity; this might be what happened when Krugman preferred to reach back for an old framework rather than thinking about how to build a new one. But his main point – don’t despair – is salutary.


One thought on “Reinhart-Rogoff and the Philosophy of Macroeconomics

  1. Pingback: Why Philosophers of Economics Should Study History of Economics | Origin of Specious

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s