[Readers already convinced should skip this long post.]
David Andolfatto offers the following challenge:
Proponents of NGDP targeting, however, like Scott Sumner and David Beckworth, for example, seem to believe very strongly in the vast superiority of a NGDP target–not just as a policy that would mitigate the effects of future business cycles–but also as a policy that should be adopted right now by the Fed to cure (what they and many others perceive to be) an ongoing “aggregate demand deficiency.”
What I am curious about is not that they believe this, but how strongly they believe in it. I respect both of these writers a lot, so naturally I am led to ask myself how they came to hold such a strong belief in the matter. What is the theoretical underpinning for NGDP targeting? And what is the empirical evidence that leads them to believe that an NGDP target right now is a cure for whatever ails us right now?
One way to seek answers to these questions is to spend hours perusing their past blog posts. I’m sure they must have answered these questions somewhere. But I figure it will be more efficient for me to just state my questions and have them (or somebody else) point me in the right direction for answers.
My approach to economics is very different from that of my thesis adviser (who was Robert Lucas.) I think one does actually have to peruse those thousands of old blog posts to get a sense of why I am so confident. That’s because I think the situation is very complicated, and must be examined from multiple perspectives. For instance, I believe that more NGDP right now would modestly lower the unemployment rate, which would cause Congress to shorten the maximum number of weeks people can collect unemployment insurance, which would lower the natural rate of unemployment. I don’t recall seeing that argument in any Lucas models. It’s not that I think there is any single overwhelmingly persuasive case for faster NGDP growth, but rather that there are many arguments, of varying persuasiveness, all pointing in the same direction.
But let’s start with Lucas, as I imagine David (wisely) respect his views more than mine. Lucas says the 50% crash in NGDP between 1929 and 1933 was the proximate cause of the Great Contraction, and he says Friedman and Schwartz are the people that convinced him of that fact. Lucas taught us (back in the 1970s) that he saw his job as providing a model with micro-foundations for the sorts of stylized facts explained by Friedman and Schwartz. Recall that F&S thought that unstable monetary policy produced fluctuations in NGDP. In the short to medium run these NGDP changes were partitioned between changes in prices and changes in output, but only prices changed in the long run. So I believe NGDP is important for basically the same reason that Friedman, Schwartz, and Lucas do—there’s lots of empirical evidence that it’s important in the short run, but not the long run. Do I have a model? No, but I can point to dozens of different NK models that all reach the same general conclusion—minimize nominal shocks. We don’t know which one is right; my hunch is that there are multiple reasons why nominal shocks matter, so most of the models are catching some aspect of reality.
The recent (2008-09) NGDP crash was the largest since the 1930s, and Lucas has argued that the Friedman and Schwartz story applies to the steepest part of that crash, in late 2008 and early 2009. However he also believes that the slow recovery is better seen as an example of the sort of stagnation that hit Europe after the 1970s, when natural rates of unemployment rose to a much higher plateau. That’s not an entirely unreasonable hypothesis; after all, inflation has fallen by less than most NK models would have predicted. So it does look like we have some supply-side issues (even more so in Britain.)
David’s query breaks down into two separate questions:. First, does one accept the basic “nominal shocks have real short run effects” approach of everyone from Krugman to Lucas. I do, but won’t defend that here. And second, the much harder question of whether the original (late 2008) nominal shock has now morphed into a real (or “supply-side”) problem. As I indicated with the UI example, I think demand and supply-side factors can get entangled, but I have other reasons for continuing to favor faster NGDP growth. I’ll name just a few, and this list is by no means comprehensive:
1. I favor level targeting much more than I favor NGDP targeting. Recent macro models (Woodford, etc.) suggest that current NGDP is strongly impacted by future expected NGDP. In that case level targeting offers two big advantages. If the Fed had promised in late 2008 to eventually return NGDP to the original trend line, then the short term crash would have been much smaller. This goes against the common sense of elite macroeconomists like Jim Hamilton. It seemed like things were falling off a cliff in late 2008 and that there was nothing the Fed could do about it. But one of the major factors causing that perception was crashing asset prices, which were making the financial crisis much worse. With level targeting people would expect asset prices to be higher in the future, which would tend to make current asset prices higher, and that would lessen the severity of the financial crisis. One reason I favor more NGDP now is that even a belated shift toward level targeting would help to restore Fed credibility in future recessions, which is especially important if they again run up again the zero bound (as Woodford has emphasized.) It’s easier to prevent NGDP from plunging in the short run if markets expect it to return to trend in the not-too-distant future. The second advantage is that it makes policymakers more accountable. They can’t keep sweeping past errors under the rug (as the BOJ does.)
2. But does it still make sense to go back to the pre-2008 trend line? Probably not, recently I’ve been calling on the Fed to go about 1/3 of the way back to that trend line, and then start a new policy trajectory (hopefully explicit in this case.) Obviously that’s a subjective judgment, which weighs the costs and benefits of a bit more stimulus now. In a perfect world the Fed would have an explicit NGDP target trajectory, and we wouldn’t have to make those sorts of discretionary decisions. But the fact is that the Fed currently uses discretion, and hence ANY ADVICE that any macroeconomist gives the Fed is inherently discretionary.
3. Are there any other factors besides inefficient government policy (40% higher minimum wages, extended UI benefits, etc) that can explain the fact that inflation has fallen less than expected? Yes, energy prices have been quite high. Most macro models suggest that NGDP and price level targeting are identical in the presence of demand (or velocity) shocks, but NGDP does better when there are supply shocks. However, even from the perspective of price level targeting, one can make a strong case for more monetary stimulus, faster NGDP growth. The inflation rate since July 2008 is the lowest since the mid-1950s, well under the Fed’s 2% goal. Admittedly the Fed does inflation rate targeting, not level targeting, but economists like Woodford and Bernanke have argued that level targeting does better at the zero bound. So we need some extra inflation (and hence NGDP) just to catch up to a 2% inflation trajectory from July 2008. Doesn’t it seem slightly crazy to undershoot our inflation target in the midst of the biggest global debt crisis ever? What purpose could that possible serve, even if we didn’t face an elevated rate of unemployment?
4. Some argue that the problems are “structural,” related to Obama’s big government policies. I oppose those policies, but see no evidence they have much impact on the business cycle. I’m old enough to remember Johnson’s Great Society, which was really big government. That had no cyclical effect at all, indeed the economy boomed throughout the entire 5 and 1/2 years LBJ was in power. I do accept the studies that the 40% jump in minimum wages and the extended UI has slightly boosted the natural rate, but the huge NGDP crash caused the UI extension, and made the ratio of the minimum wage to NGDP much higher that Congress had anticipated.
5. I put much more weight on market reactions to policy than most other macroeconomists. Liberals often assume markets are irrational, and conservatives love the EMH except when it tells them their pet macro theories are wrong (as when old monetarists predict high inflation despite the TIPS markets telling them it won’t happen.) I recall that back in the 1970s higher inflation hurt equity prices (and that’s been confirmed in at least one study.) More recently, studies by David Glasner and others have showed that beginning around 2008 equity prices actually began to be positively correlated with expected inflation. Those two studies suggest the market doesn’t like very high NGDP growth, presumably because it results in higher inflation, and thus higher real tax rates on capital. But markets also don’t like very low NGDP growth, presumably because it depresses RGDP for a few years. I’m reassured that the markets seem to currently root for at least slightly higher NGDP growth during precisely the same time periods where market monetarist models say it’s needed.
6. Because I’m a big fan of ratex and the EMH, it is with great reluctance that I have come to accept that there is some money illusion in the labor market. In 2009 everyone at Bentley got a 0% pay raise. Why not some other number, like negative 0.5%? It could be because the zero raise was optimal, but zero raises are waaaaaay more common than negative 0.5% raises, and I can’t see any other explanation than money illusion. So I think one reason why unemployment remains elevated is that nominal wage growth slowed but then leveled off at about 2%, which reflects some workers getting 0% raises (who should get negative nominal raises) and some in healthier industries getting 4% raises. It is true that some workers do get nominal pay cuts, but that doesn’t mean that money illusion isn’t making the labor market at least slightly more inefficient at this moment. It’s a complicated world out there, and not all labor markets need be inefficient for this to be a problem.
7. In the 1990s we had lots of good luck, lower energy prices and increasingly cheap imports from China. Now Chinese wages are soaring, and Chinese growth is boosting world commodity prices. Thus it’s baked in the cake that Americans must accept a lower living standard, or at least a lower rate of increase than in the 1990s. With NGDP targeting that is effected through modestly higher inflation, and stable nominal wage growth. With inflation targeting you get stable inflation, and wage cuts. Given my read on our labor market, I think the modestly higher inflation approach is the easiest way to make the needed living standard adjustments. I understand it’s hard to model this claim, and others may disagree. But that’s how I read the empirical evidence going all the way back to my study of the Great Depression–sticky wages are a bigger problem than sticky prices.
8. Suppose it’s a coin toss as to whether my macro arguments are correct. Maybe we get a better path of real GDP, maybe we don’t. We also face a severe global debt crisis, so I think that also at least slightly tips the balance toward faster NGDP growth, especially since inflation has averaged well under 2% since mid-2008. BTW, David Beckworth has argued that foreign monetary policies are linked to US policy. Hence it’s likely that faster growth in the US would lead to at least slightly faster growth in Europe. Notice that the strongly positive reaction of US equity markets to Fed easing initiatives is echoed in foreign equity markets as well. And the effects are so strong I doubt it is merely reflecting their exposure via exports to the US.
9. The widespread perception that we have a demand shortfall has created all sorts of inefficient government policies. For instance, many economists have argued for more government spending to boost demand. I fear that will lead to lots of waste. If we target NGDP forecasts, there is no longer any argument for (wasteful) fiscal stimulus. GM was bailed out partly because many feared GM workers would not be able to find jobs elsewhere, due to the recession. If we knew that any extra spending on GM cars would mean less spending on other US produced output, and less jobs in other companies, the case for bailing out GM (which I opposed anyway) would be much weaker. Ditto for too-big-to-fail, which is often defended on the grounds that a banking panic would depress AD.
These are the arguments that I was able to immediately recall, but if you are crazy enough to pour through 1000s of my old blog posts, you’ll find dozens of others. I’m an eclectic pragmatist, not someone who believes we can build “the model” and get useful policy advice from that model.
My National Affairs article has lots of arguments against inflation targeting. Here’s two that many macroeconomists overlook:
1. The measured CPI inflation rate doesn’t measure the sort of inflation that is useful for purposes of macroeconomic stabilization. For instance, housing is 39% of the core CPI, and it actually rose, even in relative terms, between mid-2008 and mid-2009. The reason deflation is bad is because companies can no longer profitably produce output, and therefore lay off workers. We know that housing prices plunged between mid-2008 and mid-2009, and that’s why far fewer houses were built. But our wonderful CPI told us that housing prices were rising faster than other goods, hence firms should have been rushing out to build more new houses!
2. The public doesn’t understand inflation targeting, because they aren’t able to distinguish between supply-side inflation, which reduces real income, and demand-side inflation, which raises real income in the short run. Hence when Bernanke announced in 2010 that he wanted to boost core inflation back up from 0.6% to 2.0%, he had in mind higher AD which would boost both inflation and real income. But people who heard this on the news could only picture supply-side inflation. They thought Bernanke wanted to raise their “cost of living.” There was a firestorm of criticism which made the Fed’s job much harder. NGDP targeting is easier to explain, you are trying to make the incomes of Americans rise at a rate consistent with prosperity, but not excessive inflation. I think that policy could be sold to the public. I don’t think they’ll ever buy a symmetrical inflation target. They think inflation is bad. Indeed so much so that when I point out that plunging housing prices should have reduced inflation, I get nothing but puzzled looks.
Andolfatto concludes as follows:
I have not even touched upon the practical feasibility of NGDP targeting–I’ll save this for another day. But for now, I’d like to know the answers to my questions above. Who knows, I too may become one of the faithful!
I find that a very refreshing attitude.
HT: David Beckworth
Update: Nick Rowe provides three pretty good arguments for NGDP targeting, and one absolutely irrefutable argument.