How “Different” is the Recovery from the Financial Crisis?

The slow recovery from the financial crisis and recession of 2007 – 2009 has become a centerpiece of the Presidential election. In last Tuesday’s debate, Mitt Romney, picking up on a theme that has been emphasized by John Taylor, contrasted the current slow recovery with the much faster recovery from the 1981 – 1982 recession. During the past two weeks, there has been an intense focus on a comparison of the current recovery with recoveries from other financial crises. On October 11, Taylor, using historical data from a paper by Michael Bordo and Joseph Haubrich, argued that the current recovery is much slower than the average of previous American recessions associated with financial crises. This was followed by an October 14 paper by Carmen Reinhart and Ken Rogoff who argue that the aftermath of the U.S. financial crisis has been typical of the recoveries from other severe financial crises, an October 15 reply by Taylor, an October 16 rejoinder by Reinhart and Rogoff, an October 17 piece by Paul Krugman, and an October 17 reply by Taylor.

Last October, we presented a paper, “The Statistical Behavior of GDP after Financial Crises and Severe Recessions,” at the Federal Reserve Bank of Boston Conference on the “Long-Term Effects of the Great Recession,” and summarized the results in an Econbrowser post. The focus of the paper was to show that, while severe recessions associated with financial crises generally do not cause permanent reductions in potential GDP, the return takes much longer than the return following recessions not associated with financial crises. We focus on five slumps, extended periods of slow growth and high unemployment, following financial crises identified by Reinhart and Rogoff in their book, “This Time is Different,” that are of sufficient magnitude and duration to qualify as comparable to the current Great Slump for the U.S. – the slumps lasting 7 ¼ years for Denmark (1989), 7 ¾ years for Australia (1990), 8 ½ years for Finland (1990), 9 ½ years for Sweden (1990), and 12 years for the U.S. (1929), with the first year of the slump in parentheses. If the path of real GDP for the U.S. following the Great Recession is typical of these historical experiences, the Great Slump will last about 9 years but will not affect potential GDP. Assuming that the Great Slump started in 2007:4, it will not end until 2016:4. This scenario is consistent with recent Congressional Budget Office projections that the economy will remain below potential until 2018, but do not allow for changes in current law to avoid the “fiscal cliff” in 2013.

Assuming that history will repeat itself (on average) and the U.S. will return to potential GDP in 2016:Q4, we can ask whether the current recovery is typical of recoveries from other severe recessions associated with financial crises. Real GDP growth has averaged 2.2 percent over the 12 quarters between the trough of the Great Recession in 2009:Q2 and 2012:Q2, and will have to average 4.04 percent over the 18 quarters between 2012:Q2 and 2016:Q4 in order for potential GDP to be restored at that date. Since 12/30 = 0.40 and 18/30 = 0.60, we call the first 40 percent of the sample the first part of the recovery and the second 60 percent of the sample the second part of the recovery. The ratio of the growth rates of the second to the first part of the recovery will need to be 1.84 in order for potential GDP to be restored in 2016:4.

How does this scenario compare with historical experience? Figure 1 depicts the first and second part growth rates for Australia, Denmark, Finland, and Sweden, as well as the projections for the U.S. Table 1 provides more detail, including the growth ratios, the dates and duration of the slumps, and the trough and magnitude of the recessions in addition to the growth rates for the two parts of the recoveries. We do not include the U.S. during the Great Depression because the recovery was interrupted by the recession of 1938, making the ratio sensitive to the exact split. The growth rate ratios provide a better comparison than the first part growth rates because of differences in the magnitude of the recessions and the pre-recession growth of potential GDP. In one respect, the scenario for the U.S. is typical, as the growth rates for the second part of the recovery are larger than the growth rates for the first part of the recovery, so that the growth ratios are greater than one, for three of the four countries. In another respect, however, the U.S. scenario is atypical, as the required growth ratio of 1.84 is higher than three of the four growth ratios and is 48 percent higher than the average historical growth ratio of 1.24.

Much of the disagreement between Reinhart and Rogoff and Taylor comes from differences in their methodologies. Reinhart and Rogoff measure a recovery as the time it takes for per capita GDP to return to its pre-crisis peak level, while Taylor measures a recovery as the average growth rate of aggregate GDP in the first 4 or 8 quarters following the trough of the recession. Because we assume that U.S. GDP will return to potential according to the historical average and compare growth ratios for each country for the two parts of the recoveries (starting from the trough) rather than comparing growth rates across countries, we are subject to neither Reinhart and Rogoff’s criticism that Taylor’s methodology exaggerates the strength of the recovery following a deep recession nor Taylor’s criticism that Reinhart and Rogoff’s methodology mixes recessions with recoveries.

Reinhart and Rogoff and Taylor also disagree about the choice of financial crises. Taylor and Bordo argue that Reinhart and Rogoff lump together countries with diverse institutions, financial structures, and economic policies while Reinhart and Rogoff argue that Taylor fails to distinguish systematic financial crises from more minor ones and from regular business cycles. We start with all episodes characterized by Reinhart and Rogoff as either mild or severe financial crises in advanced economies, and then choose the ones for which there is statistical evidence of a fall in GDP, so we are choosing episodes for which there is evidence of both a financial crisis and a recession. Among the four crises, Australia and Denmark are characterized as mild and Finland and Sweden are characterized as severe by Reinhart and Rogoff. If we restrict the comparison to Finland and Sweden, for which the magnitude of the recession and the duration of the slump are more comparable to the actual magnitude and projected duration of the Great Slump for the U.S. than Australia and Denmark, the required growth ratio of 1.84 for the U.S. is higher than both growth ratios and is 82 percent higher than the average historical growth ratio of 1.01.

Using a different metric than either Reinhart and Rogoff or Taylor, we have provided evidence that the current recovery for the U.S. has been slower than the typical recovery from severe recessions associated with financial crises. In order for potential GDP to be restored in 2016:4, the growth rate will have to be 84 percent higher between 2012:2 and 2016:4 than it was between 2009:2 and 2012:2. While it is typical for growth to be higher during the second part of the recovery from severe recessions associated with financial crises, the magnitude of the necessary difference is well above the average historical experience.

This post by David Papell and Ruxandra Prodan