Harvard Professor James Stock and Princeton Professor Mark Watson presented a very interesting paper last week at the Spring 2012 Conference for the Brookings Papers on Economic Activity. Their paper studied similarities and differences between the 2007-2009 recession and other U.S. business cycles.
Stock and Watson characterized the comovements over 1959:Q1-2007:Q3 of 198 different U.S. macroeconomic variables in terms of 6 primary factors. These factors could be calculated from the first 6 principal components of a non-redundant subset of their observed variables. This method amounts to finding 6 different summary indexes (or 6 different sets of weights to associate with each of 132 of these series) that could collectively account for as much of the variation as possible of all the data.
Their first question was whether the observed U.S. macroeconomic data continued to track those factors in the same way during the most recent recession and recovery as they had historically. Stock and Watson’s answer was, for the most part, yes. For example, the solid line in the graph below plots year-over-year real GDP growth rates (relative to trend), while the dashed line gives the values you would have expected if you’d known only what the 6 historical factors were doing and if you assumed that the relation of GDP to those factors was the same since 2007 as it had been before. GDP seems to have a similar relation to other macro variables during the most recent recession and recovery as it had historically. Statistical tests fail to reject the hypothesis of a stable relation for most of the 198 series they studied. Some of the series that did seem to exhibit some new dynamics include commodity prices, unemployment durations, and some monetary indicators.
 |
From these tests the authors conclude:
We believe that the most natural interpretation of these three findings is that the 2007Q4
recession was the result of one or more large shocks, that these shocks were simply larger
versions of ones that had been seen before, and that the response of macro variables to these shocks was almost entirely in line with historical experience. The few series for which behavior departed from historical patterns have natural explanations, in particular the DFM [dynamic factor model] predicts negative interest rates because it does not impose a zero lower bound and the DFM does not predict the Fed’s quantitative easing.
Stock and Watson then went on to try to understand the nature of the recent large shocks. The individual factors as calculated by traditional principal component analysis do not have any economic meaning or interpretation, in part because if the model were rewritten in terms of any linear rearrangement of the original six factors, it would have identical implications for the correlations and forecasts of any observed variables. Stock and Watson therefore proposed to consider six observable shocks that economists believe may be responsible for economic fluctuations, these being oil prices, monetary policy, productivity, credit spreads, uncertainty, and fiscal policy. They looked at the relation between measures that other authors had proposed for each of these structural shocks and their own estimated 6 factors, to find linear combinations of their factors most consistent with how other researchers had been summarizing the data. For example, for the oil shock, they considered using the measure proposed in my 1996 paper in the Journal of Monetary Economics, a separate measure favored by Kilian (2008) or the change in oil price itself. They construed the "oil shock factor" to be the linear combination of their six factors that has the highest correlation with one or all of these three separate measures and smallest correlation with other observed structural shocks.
The authors concluded:
the structural analysis is consistent with the recession being caused by initial large oil price shocks followed by multiple financial and uncertainty shocks….
The picture of the recession that emerges… is one of increases in oil prices through the first part of the recession, followed in the fall of 2008 by financial sector volatility, a construction crash, heightened uncertainty, and a sharp unexpected drop in wealth. Notably, there no large surprise movements of the real variables given the factors through the previous quarter.
But if the Great Recession can be interpreted as normal responses to abnormally large shocks, what about the anemic recovery? Stock and Watson attribute this to a slowdown in trend growth rates, which they infer statistically from a procedure similar to taking a 12-year average of the growth rate. Again quoting from Stock and Watson’s paper:
The explanation for this declining trend growth rate which we find the most compelling rests on changes in underlying demographic factors, primarily the plateau over the past decade in the female labor force participation rate (after rising sharply during the 1970s through 1990s) and the aging of the U.S. workforce. Because the net change in mean productivity growth over this period is small, this slower trend growth in employment corresponds directly to slowdown in trend GDP growth. These demographic changes imply continued low or even declining trend growth rates in employment, which in turn imply that future recessions will be deeper, and will have slower recoveries, than historically has been the case. In other words, jobless recoveries will be the norm.
James D. Hamilton is Professor of Economics at the University of California, San Diego. He blogs at Econbrowser
