A few days ago there was a heated debate involving, on one side Greg Mankiw and, on the other, Krugman and Brad DeLong. The spat revolved around the CEA deficit projection based on the prediction of relatively fast growth down the road. According to the CEA: “A key fact is that recessions are followed by rebounds. Indeed, if periods of lower-than-normal growth were not followed by periods of higher-than-normal growth, the unemployment rate would never return to normal”.
Implicitly, the CEA (and DeLong and Krugman) is supposing that “trend” (or “potential”) GDP and “normal” (or “natural”) unemployment are constant and that fluctuations in output (and employment) represent temporary deviations from “trend”.
Figure illustrates the concept.
What Mankiw is saying is that the “trend” itself may change. If, for example, the “trend” falls as a consequence of the recession we should not observe a strong rebound in the future exactly because “potential” GDP has fallen.
Based on his constant “trend” view of the process, Krugman asks: “How can you fail to acknowledge that there´s huge slack capacity in the economy right now? And yes, we can expect fast growth if and when that capacity comes back in to use”. The “slack capacity” is given by the distance between the level of “potential” GDP and actual GDP.
DeLong illustrates the argument for a strong rebound following a recession by showing (figure 2) that “those post recession periods of falling unemployment are also times of rapid output growth”. But figure 3 shows that if we remove points from the 1981-83 period, the positive correlation between higher unemployment and future growth disappears!
Maybe there´s something “special” about the 1981-82 recession? To find out I describe three alternative views of “potential” output and compare two periods; 1979-84 and 2002-08.
Figure 4 describes “potential” output according to the CBO estimate, figure 5 measures “potential” by applying the Hodrick-Prescott Filter (H-P) to the real GDP series and figure 6 calculates “potential” from a regression of real GDP on real consumption of non durables and services.
This last measure is based on work by John Cochrane (1994) who suggested that consumption could be used to track movements in “trend” GDP. The idea behind this measure of “trend” or “potential” is based on the Friedman´s Permanent Income Hypothesis (PIH) coupled with Rational Expectations, according to which consumption primarily reflects the expectation of private households about long term movements in income (GDP). Therefore, consumption should provide a reasonably good measure of “trend” GDP.
In the pictures the yellow shaded areas designate periods when the economy was in recession. The dotted green lines on figure 6 indicate moments when “trend” growth appears to have changed.
What is notable is that in figures 4 and 5 “potential” GDP is much smoother (“linear”) than in figure 6. Note that in figures 4 and 5, for example, “potential” GDP doesn´t budge at the time of the second (and significant) oil shock in 1979-80. Intuition and theory are more consistent with the observation on figure 6 which shows that “potential” GDP falls temporarily.
The 1981-82 recession was severe. From peak to through GDP fell by almost 3% and unemployment reached almost 11%. From figure 6, however, we see that even before the recession was officially over “potential” GDP increased so that when the economy picked up the “distance” between “potential” GDP and actual GDP had increased even more, giving rise to a robust rebound.
Figure 6 indicates that “potential” or “trend” GDP does not evolve at a constant rate. During the 1981-82 recession important structural changes were taking place. At that time Volker succeeded in controlling inflation (with gains in credibility) and Reagan convinced economic agents that economic policy (redirected towards “smaller” government) changed favorably “perceptions of the future”. These changes increased “potential” GDP which had the effect of increasing actual GDP growth. Therefore, the strong rebound in GDP growth was not the consequence of a high rate of unemployment but was more likely due to the structural changes that increased the level of “potential” GDP. This is consistent with the finding that if we ignore those points in figure 2 the positive correlation between unemployment and future growth disappears.
Another marked difference between figures 4 & 5 on the one hand and figure 6 on the other, is that in the latter we observe one break in “potential” GDP in early 2007 (when the first signs of the subprime crisis showed up) and a reversal of “trend” in mid 2008. Apparently, the “intermediation shock” and the policy reactions to it this time around worsened agents “perceptions of the future”, reducing “potential” GDP and increasing the “natural” or “normal” rate of unemployment (here also, the behavior of the stock market may be regarded as a ”blanket” indicator, with the S&P showing a decrease of around 30% since election day).
An article in the NYT (March 7) argues in favor of some kind of structural change: “… The acceleration [of unemployment] has convinced some economist that, far from an ordinary downturn after which jobs will return, the contraction under way reflects a fundamental restructuring of the American economy. In crucial industries – particularly manufacturing, financial services and retail – many companies have opted to abandon whole areas of business…”.
According to figure 6, at the moment the level of GDP is just at “potential” meaning, opposite to what Krugman argues, that there is no “slack” – large or small – in the economy as indicated by, for example, figure 4. In this situation a strong rebound, underlying the CEA predictions, is quite unlikely!
João Marcus Marinho Nunes – March 12, 2009