Paul Krugman produced a couple good posts yesterday. Here are some of his comments on forecasting:
So as I see it, we should first of all be evaluating models, not individuals; obviously we need people to interpret those
entrailsmodels, but we’re looking for the right economic framework, not the dismal Nostradamus. Second, we should be evaluating models and the individuals who claim to have these models based on broad performance, not single events; if your approach (say) predicted the housing crash but then also predicted runaway inflation from Fed expansion — I assume everyone knows who we’re talking about — it’s not a good approach. Finally, I think we’re looking for conditional predictions — what happens given events that are themselves not part of the model — not absolute predictions. It was, for example, very hard in the fall of 2011 to know how the ECB would respond to the escalating financial crisis in Europe; failing to predict that Mario Draghi would find a way to funnel vast sums to debtor nations through discounting would have lost you a lot of money, but wasn’t really a failure of the economic model.
Of course I prefer market forecasts, which are conditional on current expectations of the future path of policy. If policymakers don’t like those forecasts, they need to change the policy.
Here’s Krugman on AS/AD:
So why do AS-AD? First, you do want a quick introduction to the notion that supply shocks and demand shocks are different, that 1979-80 and 2008-2009 are different kinds of slump, and AS-AD gets you to that notion in a quick and dirty, back of the envelope way.
Second — and this plays a surprisingly big role in my own pedagogical thinking — we do want, somewhere along the way, to get across the notion of the self-correcting economy, the notion that in the long run, we may all be dead, but that we also have a tendency to return to full employment via price flexibility. Or to put it differently, you do want somehow to make clear the notion (which even fairly Keynesian guys like me share) that money is neutral in the long run. That’s a relatively easy case to make in AS-AD; it raises all kinds of expositional problems if you replace the AD curve with a Taylor rule, which is, as I said, essentially a model of Bernanke’s mind.
So there is a place for AS-AD, although it’s an awkward one, and the transition to IS curve plus Taylor rule plus Phillips curve, which is the model you really want to use for America right now, is a moment that fills me with dread every time we take it on in a new edition.
I mostly agree, but I’d prefer to stick to AS/AD over any interest rate-oriented model. Keynesians view interest rates as the indicator of monetary policy, and also the transmission mechanism. In my view NGDP expectations are the best indicator, and the most important transmission mechanism. Another 2 Krugman posts from a few days back illustrate my point. Here’s Krugman on May 29:
And while day by day there are variations, basically what you see over the last month or so is line 3: falling bond prices accompanied by rising stocks and a rising dollar. So this looks like a story about macroeconomic optimism.
In other words, interest rates are rising because things are getting better. And here’s Krugman on June 1, just three days later:
With stocks down and the dollar up, this looks like a market that has upgraded its estimate of the chances that the Fed will tighten too soon. And yes, I mean too soon, for sure.
Sorry Krugman bashers, I’m not going to accuse him of inconsistency. I think he was right on May 29th (growth expectations were driving up interest rates) and he was right on June 1st (fears of tighter money were driving down stocks and pushing the dollar higher.) My only quibble is the previous paragraph from the same post:
Still, a rise in bond rates is not helpful just as there are signs the economy is gaining momentum despite the best efforts of politicians. So what is happening?
These two statements (please read original post for context) create the impression that the Fed has reduced stock prices by driving up long term rates. But long term rates are positively correlated with stock prices. The stock market was pretty strong in the period up to May 28, and long term rates were rising. Long rates have been flat since the 28th, and stocks have fallen. Krugman’s right that monetary policy expectations are driving the various asset markets (no liquidity trap), but overstates the usefulness of long term interest rates as a monetary indicator. For long term rates, the inflation and income effects often dominate the expected liquidity effect. In other words, a higher expected future path of short rates might reflect tighter money, but it might also reflect faster growth generated by a successful Fed policy of managing expectations.
Since he looks to stocks and forex markets to figure out what the various interest rate movements actually mean, why not just cut to the chase and focus on direct indicators of monetary policy? In other words, why not have the Fed create and subsidize trading in an NGDP futures market? And if we don’t have that, look at the asset prices most strongly correlated with expected NGDP growth. (TIPS spreads, and, for high frequency changes—stocks.)