In a couple of posts (here and here) responding to some most excellent challenges from Max Sawicky (here and here, with an assist from Brad DeLong), I attempted to explain my hesitancy to apply the term "slack" to the labor market (expounded on at length over at Econoblog, but you probably already knew that). As if to rescue me from my clumsy attempts at clarification, Bob Hall takes on the same topic -- and comes to similar conclusions -- in a paper delivered at the Federal Reserve Bank of Kansas City's annual Jackson Hole conference.
Hall begins with a crystal-clear description of the "traditional" view of ups and downs in the economy:
The traditional idea is that neoclassical constructs – production functions, consumption demand functions, labor supply functions, embedded in markets that clear – describe the actual operations of the economy in the longer run. There is a *-economy that generates variables such as y*, called potential GDP, u*, called the natural rate of unemployment, r*, called the natural rate of interest, and so on…
In the early years of what Paul Samuelson called the “neoclassical synthesis,” the *-economy was viewed as generating smooth trends, as described by Solow’s growth model, the keystone of neoclassical macroeconomics. Short-run movements around the smooth trend were transitory, the result of imperfect information, delayed adjustment of prices, or other non-neoclassical features of the economy.
He then goes on to explain why this will no longer do:
An early milestone in the unfolding breakdown of the neoclassical synthesis was Kydland and Prescott (1982)’s discovery that the *-economy is anything but smooth, once the actual volatility of productivity growth is included in the model… This discovery forbids extracting the disequilibrium cyclical movements as deviations from a smooth trend. Instead, one would have to solve the *-model and calculate deviations from the volatile *-variables. I’m not sure that this lesson has fully informed the community of practical macroeconomists who try to use signal-extraction methods based on statistical characterizations of the *-variables as moving smoothly over time.
The second element in the breakdown is the high persistence of the deviations of actual from neoclassical performance. The puzzle is the most visible in unemployment… Unemployment has large low-frequency swings – low in the 1950s and 1960s, high in the 1970s and 1980s, low again in the 1990s and 2000s. The idea is unpalatable that these movements are the result of transitory cyclical forces, for they are only barely transitory. The standard view that the *-economy explains longer-run movements seems to call for adding low-frequency movements of unemployment to the *-economy – that is to create a model of the natural rate of unemployment that permits slow-moving changes...
I conclude that neoclassical principles properly applied in an environment with volatile driving forces delivers predictions rather different from the smooth growth implicit in most thinking based on the cycle over trend. Sudden movements in GDP and other variables are not necessarily part of the disequilibrium business cycle – they may reflect the neoclassical response to shifts in productivity and exogenous spending. And non-neoclassical forces in the labor market may result in long-lasting, smooth changes in unemployment that are not distinguishable from the more rapid movements that observers have earlier assigned to transitory disequilibrium.
Those "non-neoclassical forces" that Hall refers to include imperfect information and "search frictions" that are the bedrocks of any sensible approach to thinking carefully about labor market phenomena like the unemployment rate. I haven't made the following point in any of ramblings on this topic, but I wish I had:
Subtle changes in the economic environment, such as changes in the distribution of information known to one side of the unemployment bargain but not to the other, can cause large changes in unemployment. And these changes can be long-lasting – they may play an important role in the sub-cyclical movements of the labor market that are so prominent in the data but escape existing models.
Not only that, this way of thinking about labor markets has important implications for the way we think about wage deterimination and how changes in compensation to employees might or might not be useful in interpreting what is going on in labor markets.
More on that to follow, but it is worth emphasizing that Hall is no more sympathetic to constructs like the "neutral rate of interest" than he is the natural unemployment rate concept:
Wicksell's natural or normal interest rate, as distingushed from the actual market rate, is not a feature of modern macro-finance models. Each of the many real interest rates in the economy moves differently -- they do not obey even the relatively unrestrictive principles of basic finance models. We are not equipped to judge when monetary policy is neutral in terms of interest rates.
This may all sound very nihilistic, but I don't read it that way. Theory may still be ahead of our abilities to measure what we really want to measure, but at least we know where to look, and that is a result of the vast amount we have earned in the last 25 years. And Hall, for one, is not so sure it matters anyway:
None of these conclusions stands in the way of intelligent monetary policy-making. Under Alan Greenspan's stewardship, the U.S. has acheived remarkably low levels of inflation and inflation volatility, despite the lack of real reference points. We do not need to know the GDP gap, the unemployment gap, or the neutral real interest rate, to keep the price level near constancy.
UPDATE: New Economist has the entire Jackson Hole agenda.
UPDATE, THE SEQUEL: Arnold Kling finds Hall's paper at the professor's website (Duh!), and has a few comments of his own.
EMBARASSING UPDATE: Max and PEmberton both note a typo in my original post in the last paragraph. Hall, of course, said what now reads above -- We do NOT need to know the output gap etc. to maintain price stability.