Paul characterises me as saying that macro has nothing useful to say, and that justifies doing nothing about anything. This greatly misstates my meaning. My post is about how reasonable people could cite differing theories and evidence to back prescriptions for the appropriate policy response to the crisis that embrace not only Paul and Simon’s own [UK and US should have stimulated even more than they did] but also what actually happened. [Both countries engaged in huge fiscal stimuli anyway].
Modern macro offers lots of advice, but it doesn’t confer certainty on policymakers. (Though it even offers lots of useful advice about what to do about that uncertainty). The way to respond to that is, in my view to try to be Bayesian about it. Put weights on the competing explanations of what you see, and the corresponding policy prescriptions, and what you do will be, crudely, a weighted sum of what you would have done in each of the competing world-views. Paul and Simon put all their probability mass on the IS/LM/New Keynesian explanation that there was far more demand deficiency than was responded to by the increase in the deficit. I don’t. In fact, I think it’s perverse to put all your weight on an explanation that has no role for the financial sector; that ignores the rough constancy of inflation, or at least explanations that include the stimulus having been large enough, roughly; that sweeps aside important questions about the value of money (surely a big no-no at the zero bound), and many more (see my long paragraph, third from bottom, in the original post, for all the questions up for grabs, none of which are dealt with by either Paul or Simon).
Paul concludes with a worry about my motives. Is my reluctance to come down on his side ‘distaste’ that economics should ever have to confront the messy task of doing something useful? Not at all. It’s part of a debate about how to translate what we know into sound advice. Sound advice being advice that communicates a fair sense of the range of plausible competing views and corresponding prescriptions. If we don’t do that, the policymaking clients may listen for a while, but they may conclude at some point that the other views are being left out of the picture for political reasons.
David Andolfatto defends me, spotting that Paul and Simon incorrectly translate my post into a message of ‘do nothing because it’s all very uncertain’. Simon responded to this with his post ‘humility and chameleons’. The nub of that seemed to be that we should be wary of letting politicians do what they want just because we have a few thought-provoking but whacky models on the table that give opposite prescriptions to those we feel are right.
On its own that’s a fair point. But hang on. You don’t have to put chameleon models on the table to overturn Paul and Simon’s view that fiscal policy was drastically too tight in the US and the UK. You could, for example (though I don’t) put all your weight on the unadulterated New Keynesian model that Paul and Simon use to justify even more stimulus. That model, to repeat, says: if inflation is roughly on target, you are done. On-target inflation means that whatever other reduction in activity you are observing and feel you should be correcting you should be doing with some other policy (like – if you modify the model to attempt to explain the key thing that happened in the crisis, namely the contraction in credit – intervening to expand credit (presumably more aggressively than either the Fed or the BoE did with its credit easing policies)).
Simon points out that you don’t have to take the models instructions lock stock and barrel. In particular, you could question whether it’s instruction that inflation fluctuations are an order of magnitude more concerning than fluctuations in real activity is valid. True. (Although Simon’s reference to the heterogeneous agent literature won’t work, I suspect, for reasons that would require another post to explain). However, your Bayesian policy maker could reserve SOME weight for taking the model as it comes.
Digging into the particular issue Simon raises about the concern for inflation helps to make my main point that this sticky price model can’t be the be all and end all policy model. So… Dispense with that relative weight on inflation versus output, and you are querying whether you have got agents’ utility functions messed up. Or whether indeed they are utility maximising at all. You are also implicitly worried that the assumption that firms and workers would have to respond to the volatility in relative prices caused by inflation by meeting all the attendant fluctuations in demand is not right. (And rightly so). And you are saying to yourself: hang on, should we really think of firms as worrying that there is some chance they will NEVER EVER change their prices? All these things would be legitimate concerns. Responding to them would probably not get rid of the desirability of counter-cyclical fiscal policy but it could easily lead to arguing for less stimulus than we saw, or more. When you look at this family of models closely, it’s pretty whacky itself isn’t it?
But, even if you ignored all of this, what if you said to yourself: ok, I’m taking the model lock stock and barrel, except the bit about how important inflation fluctuations are, and I’m going to estimate it, and see what those estimates imply for explanations of the crisis, and policy prescriptions. An unfortunate fact of life is that this model, and most others like it are, in the jargon, badly identified. Macro data are going to leave you with a lot of uncertainty about all the parameters in the model, and the model’s explanations of which shock caused what, and that is going to translate into a LOT of uncertainty about policy, easily enough to embrace the competing calls for looser or tighter policy in the UK. Not enough to embrace the ‘get rid of the state altogether’ view of the Tea Party. But still.