The Job-Filled Nonrecovery

1. Robert J. Barro

Editor’s note: The next president is in for a rough welcome to the Oval Office given the list of immediate crises and slow-burning policy challenges, both foreign and domestic. What should Washington do? Why should the average American care? We’ve set out to clearly define US strategic interests and provide actionable policy solutions to help the new administration build a 2017 agenda that strengthens American leadership abroad while bolstering prosperity at home.
What to Do: Policy Recommendations for 2017 is an ongoing project from AEI. Click here for access to the complete series, which addresses a wide range of issues from rebuilding America’s military to higher education reform to helping people find work.

Introduction
The recovery from the recent Great Recession in the United States (and many other places) has been nonexistent. The US per capita growth rate for 2009–15 was 1.3 percent per year, below the long-run rate of 2.1 percent per year (see 1869–2015 and 1949–2015 in Table 1). The growth rate during a recovery has to exceed its average to restore at least part of the cumulative loss in the level of GDP during the downturn. Table 1 also shows that the per capita growth rate of 0.8 percent per year for 1999–2009 (1.0 percent per year 1999–2015) was worse than any decade except the 1930s (0.6 percent per year).

Credit: Twenty20

Empirically, the growth rate during a recovery is positively related to the magnitude of cumulative decline during the prior downturn. The best evidence on this relation comes from major depressions. My research in this area, particularly with Jose Ursua, has used the history of rare macroeconomic disasters for 40 countries as far back as 1870 (sample of countries with annual data back at least to 1913). We isolated 185 contractions in per capita GDP of size 10 percent or more. The average size is 21 percent, with a fat tail. This historical experience was dominated by wartime destructions (especially the two world wars) and financial crises such as the Great Depression of the 1930s. Many are global events; some are for individual or a few countries. My research with Emi Nakamura and Jon Steinsson and recently with Tao Jin has used the underlying long-term data on GDP and consumption to study recoveries.
On average, an economy eventually recovers about half of the per capita GDP lost during a prior downturn. Once a disaster event is over, the recovery is typically quick; the average duration is about two years. For example, a cumulative fall of 10 percent in per capita GDP during a contraction implies subsequent recovery (toward the trend for per capita GDP) of 5 percent, implying about 2.5 percent per year higher growth during the recovery. Analogously, a more typical recession of 4 percent implies an extra 1 percent per year in the growth rate during two years of recovery. This pattern should have applied, for example, in the United States for 2009–11, but it is not in the data. However, the absence of a recovery is not that surprising because there is a lot of observed variation in the extent of recovery, including cases that exceed 100 percent (such as the post-WWII periods in Western Europe and Japan and the post-reform period in Chile) and some that are close to zero. Recovery effects are particularly hard to isolate in samples such as post-WWII United States, which encompass only comparatively mild recessions.