About 20 years ago I published a paper arguing that the Fed should use futures markets to target a nominal hourly wage index. The intuition was as follows:
1. Recessions are basically sub-optimal employment fluctuations. They occur due to sticky wages.
2. When the aggregate wage rate falls, the wage rate rises above its equilibrium value. That’s because some wages are sticky. So when wage growth declines (as in 2009) equilibrium wage growth is falling even more sharply, but some wages are sticky. Hence (paradoxically) the average wage level is actually “too high.” The opposite is true when wage growth accelerates.
3. In a large diverse free market economy, the law of large numbers assures that a monetary policy that keeps average hourly wage growth steady is likely to keep aggregate nominal wages close to the equilibrium level (think of “equilibrium” here as being analogous to the Wicksellian equilibrium interest rate. The one that provide macro stability)
4. More recently Greg Mankiw and Ricardo Reis reached a similar conclusion, for similar reasons, but much more rigorous modeling.
I’ve never seen an hourly wage target as being politically feasible, so I gravitated toward NGDP targeting, which had already been proposed by people like Bennett McCallum, and seemed most likely to replicate the good effects of a wage target. Here is how they are similar:
1. Suppose you target nominal wage growth at 4%. Now assume that the public’s preferred growth rate of hours worked is pretty steady, say about 1%/year. Then a policy targeting 5% growth in aggregate nominal labor compensation should get you similar results. Even better, it is much easier to measure, because many jobs do not have hourly wage rates. So it might be better.
2. And a NGDP target is likely to be similar to a total labor compensation target, as labor compensation is much more than 50% of GDP, and other key components like depreciation and indirect business taxes are non-volatile.
Those who find a wage target unrealistic, or who reject sticky wage models because of a misreading of the wage cyclicality data, are likely to move in a different direction, toward inflation targeting. And yet the logic of stabilizing the stickiest prices is so powerful that they may not move all that far away. Here is Miles Kimball:
Since oil prices are flexible, the quickest way to get to the relative prices that will prevail in the medium-run is to have those flexible oil prices adjust, while the short-run price index emphasizing sticky prices stays unchanged. (This is why I am in favor of the Fed’s emphasis on “core inflation,” though I think the Fed does too little to focus on the especially important prices associated with especially interest-rate-sensitive goods.)
At first glance a core inflation target seems quite different from a NGDP target. But I see both as imperfect compromises–substitutes for wage targeting. Indeed core inflation is heavily influenced by wage inflation, as it excludes the prices that most strongly deviate from wage inflation–food and energy.
Why do I trust NGDP more than core inflation? I don’t think the inflation rate measured by the government is a particularly good proxy for the inflation rate that produces macroeconomic stability (when stabilized.) Indeed Kimball alludes to that problem with his comment on interest rate sensitive goods. Between 2006 and 2012 the core inflation rate showed housing prices rising about 10%, and housing is 39% of the core index. Over the same period Case-Shiller showed housing prices falling 35%. The official index looks at rental equivalent, not the price of newly constructed homes. Which one better explains what was happened to housing output? Or employment in housing construction?
In an imperfect world I trust NGDP much more than core inflation, partly for reasons identified by Kimball:
Actually, contrary to conventional wisdom, I am not persuaded that there are many events commonly called “recessions” that have supply-side causes, except when supply shocks led to inappropriate monetary policy responses.
I agree. NGDP includes RGDP growth, which is (by definition) highly cyclical. The NGDP signal would have definitely told the Fed that money was too tight in 2008-09. The core inflation signal? Yes, but to a lesser extent. And when you get to level targeting there is the possibility of mischief. Core inflation had previously overshot the target, and thus a passive central bank could excuse inaction with a core inflation rate that had fallen 1% below trend in 2009 much more easily that a central bank with a NGDP target. NGDP had fallen 9% below trend between 2008:2 and 2009:2. Note that the Fed’s preferred PCE inflation rate has averaged almost exactly 2% since June 2003. But only 1.2% since July 2008. And if you don’t do level targeting of core inflation, there is the possibility of a Japanese situation, where year after year they claim they tried but failed to achieve price stability.
The case for NGDP is very much a pragmatic case. You won’t find any DSGE models spitting out “NGDPLT” as the answer. It’s a good compromise target that is robust to a wide range of flaws in our models, and our monetary policy apparatus.
BTW, Miles Kimball’s post was a reply to a very good Bill Woolsey post. Here’s an excerpt from the Woolsey post:
While I wouldn’t describe this as a reason why the price level should be kept stable, I might see this as a reason why stabilizing the growth path of wage income is better than stabilizing total nominal income. Glasner and Sumner both have argued for stabilizing the growth path of a wage index for much the same reason. These are the sorts of reasons why most of us understand that nominal GDP level targeting is not perfect, just better than inflation or price level targeting.
Jeffrey Frankel also sees NGDP as a robust target:
This is yet another instance of a long-standing point: if central banks are to focus attention on a single variable, the choice of NominalGDP is more robust than the leading alternatives. A target or threshold is a far more useful way of communicating plans if one is unlikely to have to violate it or explain it away when things change later.
PS. I have a new post over at Econlog that is loosely related to this one.
PPS. Here is a presentation that I did at an IEA conference in London.