Why Wasn’t the Recession Far Worse?

Stock and Watson have a new paper that provides support for the market monetarist view of the recession:

First, a combination of visual inspection and formal tests using a DFM estimated through 2007Q3 suggest that the same six factors which explained previous postwar recessions also explain the 2007Q4 recession: no new “financial crisis” factor is needed. Moreover, the response of macro variables to these “old” factors is, for most series, the same as it was in earlier recessions. Within the context of our model, the recession was associated with exceptionally large movements in these “old” factors, to which the economy responded predictably given historical experience.

This is what we’ve been saying all along.  The massive decline in NGDP after mid-2008 was by far the worst demand-side shock since the 1930s.  A severe recession would have occurred after such a shock even if there had been no financial crisis at all.  They don’t attribute all of the decline to monetary policy, but that’s probably because of the way they identify monetary policy: the fed funds rate.  In 2003 Ben Bernanke pointed out that the fed funds rate is not a reliable indicator of monetary policy, and suggested that aggregates such as NGDP and inflation are “the only” reliable indicators.  If you average those two indicators, then 2008-09 was the tightest money since the Great Depression.  Had Stock and Watson used that indicator, they would have blamed the Fed for almost all of the Great Recession.

Stock and Watson correctly noted that the slow recovery in RGDP is partly due to a slowdown in labor force growth.  However that doesn’t explain the sharp rise in unemployment, which is the distinctive feature of the recovery.  In addition, it doesn’t really explain the slow growth in NGDP, another factor in the slow recovery.  But they are right that if the labor force was still growing at the peak rates of the 1960s to 1990s period, then the recovery would have been better in RGDP terms (although not in terms of the unemployment rate.)

Here’s Tyler Cowen:

The paper itself can be found here (pdf). By the way, for market monetarists, equity markets seem to agree.  Stock and Watson, of course, are two of the most technically accomplished macroeconometricians.  This is further evidence — perhaps the most thorough empirical paper on the topic to date — that the Great Recession has been about the interaction of cyclical and structural forces.

The second sentence had me a little confused.  Tyler had just quoted some material from Stock and Watson pointing to the structural factors in the slow recovery.  This is certainly consistent with market monetarism, as we do not think monetary stimulus can magically increase the trend rate of labor force growth.  But it has no bearing on the slow recovery in the unemployment rate.  In addition, I don’t see the link in the second sentence as providing any meaningful evidence that equity markets agree with Stock and Watson (although I’d guess they do, as Stock and Watson are probably right about the slower trend growth rate.)

Tyler’s link is to a Sober Look post, which shows that current P/E ratios are currently below the 10 year average.

The equity markets are pricing in a significantly slower growth for most of the world than we’ve experienced in the past decade.

A few observations:

1. What is actually being forecast is future expected profit growth, which has been only loosely correlated with economic growth during recent years.
2. FWIW, experts on stock prices such as Robert Shiller argue that stocks are hugely overvalued.  I don’t agree, but I find it odd that the experts can’t agree on whether stocks are hugely undervalued or overvalued on a P/E basis
3. The US has experienced roughly 3% trend RGDP growth for more than a century, and wildly volatile P/E ratios.  This makes me wonder whether P/E ratios are a useful predictor of changes in trend RGDP growth.  I.e., all the previous wild swings in P/E ratios have failed to predict any sustained and significant shift in trend RGDP growth.
4. Having said all of that, I do buy Tyler’s conclusion, but don’t see the connection to market monetarism.

Part 2:  Why wasn’t the recession far worse?

I recall that in the past Tyler has argued the recession was about 1/3 nominal shock and 2/3 real shocks (or perhaps monetary/structural, I don’t recall the exact terminology.)  It seems like he sees the Stock and Watson paper as providing at least some support for this view.  I don’t agree.

Let’s suppose Tyler was right, and that the recession was 2/3 real.  In that case the unemployment rate would have risen by about 2/3 as much as it did, even in the absence of any nominal shock at all.  Thus if the Fed had kept NGDP from growing at 5% throughout this period, we still would have had a pretty severe recession.  At this point some real shock proponents will rebel, insisting that NGDP fell partly as a result of the real shock.  That may be true, but it’s beside the point.  We are interested in the independent effect of each shock.

Consider a medical analogy.  Suppose someone with lung cancer gets pneumonia, and then dies of pneumonia.  How could we separate the impact of pneumonia and lung cancer, as cancer often triggers pneumonia?  The best way would be to treat a group of lung cancer patients with an antibiotic that prevents pneumonia. If (as I suspect) they would have eventually died of lung cancer in any case, then it’s reasonable to see the lung cancer as the more important factor in the cause of death.

If I’m right that NGDPLT can prevent serious declines in NGDP, then the key counterfactual for the structuralists is to come up with a plausible estimate of how bad the recession would have been with the real estate bust/banking crisis, but without the fall in NGDP.  In my view we might not have had any recession at all, or at worst a very mild recession.

Here’s the problem with the structural argument.  If 2/3 of the recession was due to real factors, then monetary factors would have caused at worst a very small increase in unemployment, maybe 1.7% points of the roughly 5 percentage point increase.  That would be milder that any post-WWII recession.  But we know for a fact that the nominal shock was by far the worst since the 1930s.  So if we are to take seriously the structural view, we’d have a bizarre situation where a massive negative nominal shock, which by itself should have caused a severe recession, miraculously failed to produce any sort of big increase in unemployment.  How did we get so lucky?

Remember, it doesn’t matter why NGDP crashes; falling AD will always reduce short term growth.  If it crashes because Mexican drug lords hoard lots of Federal Reserve notes, and the Fed doesn’t increase the monetary base to accommodate that demand, then we get a severe recession.  Note that there is no direct real effect of drug lords on our GDP, just the indirect effect of tightening monetary policy.  A financial crisis is both a monetary and a real shock.  It disrupts credit allocation, hurting credit-intensive industries, and it indirectly causes the Fed to lose control of NGDP (due to their foolish interest rate targeting approach.)  Both hurt the economy, but the part of the damage due to lower NGDP is not miraculously less bad just because there are real problems too.

If you are stabbed with a knife soaked in pneumonia bacteria, the direct damage from the knife itself isn’t mitigated by the bacteria on the blade.  Suppose the knife wound was of the sort that would normally kill someone.  Would we need to even consider the effects of the pneumonia germs in the post mortem?  The NGDP shock was a knife wound to the US economy severe enough, all by itself, to cause the current recession.  A priori, I’d expect the banking crisis to have made things even worse, but I see almost no evidence that it did.  If it did reduce RGDP, why wasn’t the recession far worse?  The NGDP decline already explains the severity.

But it’s even worse for the structuralists, far worse.  The banking crisis itself was roughly two thirds caused by the fall in NGDP expectations.  So if there’s no nominal shock, then the real shock is also much smaller.  Thus whatever small part of the recession is real, is itself 2/3rds caused by the fall in NGDP.  That makes the real part of the recession extremely small.

The Sober Look post provided one data point for the US recession (of dubious relevance.)  Fortunately, there many other stock market data points in support of market monetarism.  As David Glasner and others have discovered, the stock market “rooted” for more inflation after 2008, but much less so before 2008.  Another study found they rooted against higher inflation during the Great Inflation.  All these market responses are consistent with the argument that stocks do best with steady 5% NGDP growth.  And recently stocks have responded strongly to even small hints of modest monetary easing, which refutes those who argue monetary policy is ineffective in a liquidity trap.

The stock market hates the RBC model, and it’s equally contemptuous of the post-Keynesians who say the Fed is out of ammo.  It’s not surprising that the markets are market monetarist, as my views of macro were largely developed by watching how markets responded to the massive and hence easily identifiable monetary shocks of the interwar period.  That’s why I never lose any sleep at night worrying about whether market monetarism will ever be discredited; I know the stock market agrees with me.