Why Are Some Recoveries Short and Others Long?

Good finding from Leamer that permanent separations from manufacturing have been closely correlated with the length and depth of recent American recessions both on the national and state level, "... Using the recession recovery point equal to the month when private payrolls first exceeded their previous peak level, this paper argues that it was the negative secular trend in manufacturing jobs that was the most important determinant of the length and depth of the last three recessions/ recoveries. This negative secular trend changed the layoff/recall pattern of jobs in manufacturing into permanent displacements, a malady that lengthened the recovery periods and that is not the explicit target of either traditional monetary policy or traditional fiscal policy. Using the ideas gathered from an examination of the US two-digit sectoral data for the US overall, attention turns to the recession/recoveries of the 50 US states in the last three national recession periods. Regressions that explain the lengths and depths of the recessions in 50 US states reveal the importance of construction jobs, but the most important predictor was manufacturing jobs: the greater the share of manufacturing jobs prior to the recession, the worse was the recession/ recovery...."

His takeaway, "...The role of declining manufacturing jobs in making our recessions deeper and longer lasting has not to my knowledge had the attention of housing.... The long-term decline in manufacturing jobs from over 30% of jobs in 1950 to about 8% today has eliminated many of the best jobs for high school graduates, and devastated many communities around the United States. This decline in jobs was caused by a combination of technology, globalization and low savings. Process improvements in manufacturing have continuously increased worker productivity which means fewer manufacturing workers unless that force is offset by a combination of population growth and product innovations. Globalization which integrates high-wage countries with low-wage countries shifts the labor-intensive manufacturing work to the low-wage countries, leaving the high-wage countries with fewer manufacturing jobs.In addition, a country with a low savings rate needs a real exchange rate that is high enough to create an external deficit large enough to close the gap between savings and investment. This can shift the workforce out of the tradables sector into the nontradable service sector work: fewer manufacturing jobs and more restaurant work. The biggest public policy contributor to this outcome is probably the large deficit run by the Federal Government. Policies to increase national saving like tax breaks to encourage more savings for retirement would help out, but the fundamental technological and globalization forces cannot be reversed....In conclusion,there is substantial evidence that the behavior of job shares in construction and manufacturing predicted the length, depth and severity of the last two recessions. Theregressions that are used to support this conclusion have rather low R-squares, which means that there is a lot more than just manufacturing and construction that matter, and it remains to be seen if the results in all these regressions would be upended if other variables are included. All this is looking backward but the negative secular trend in manufacturing may be at an end, in which case what we learned about the last three recessions may not tell us much about future ones...."

The evidence, "... The unemployment data illustrated in Figure 4 reveal that the unemployment rate was always above its previous peak level when payrolls returned to their value at the previous peak. Both GDP and the unemployment rate are therefore suggesting that the recovery was not completed when payrolls returned to their previous peak level....”

“…Table 7 is a similar set of results for the Great Recession 2008/09 period with explanatory variables equal to construction, manufacturing and information employment shares in 2006, 2000 and 1990. The results on the left have a mixture of signs of coefficients and the results on the right with a reduced parameterization is again designed to make the results more understandable. All variables in the results on the right have unit standard errors and the coefficients are “beta-values.” The largest t-values and the largest beta-values are highlighted. Again manufacturing jobs contribute most noticeably to length, depth and severity of the recession, but here the manufacturing variable is not the level but the increase in employment share from 1990 to 2000 or to 2006. The change in share variables have negative coefficients which might seem hard to understand. The explanation for the negative coefficient lies in the scatter diagram at the right which compares the change in the manufacturing employment share from 2000 to 2006 with the change in the previous period from 1990 to 2000. What this scatter indicates is that a large decline in manufacturing employment in one period predicts a large decline in the subsequent period. What this suggests is that troubles in manufacturing before Again manufacturing jobs contribute most noticeably to length, depth and severity of the recession, but here the manufacturing variable is not the level but the increase in employment share from 1990 to 2000 or to 2006. The change in share variables have negative coefficients which might seem hard to understand. The explanation for the negative coefficient lies in the scatter diagram at the right which compares the change in the manufacturing employment share from 2000 to 2006 with the change in the previous period from 1990 to 2000. What this scatter indicates is that a large decline in manufacturing employment in one period predicts a large decline in the subsequent period. What this suggests is that troubles in manufacturing before…"

"...In preparation for the study of the variability of recession outcomes across US states,Figure 19 illustrates the national employment shares of manufacturing, construction and information since 1987 with the official recessions in red and the periods during which total payrolls were below their previous peak in yellow except that payrolls in the last month of each recovery exceeded the previous peak.These three sectors are thought to be “foundation” jobs on which the other jobs are constructed. The foundation sectors sell most of their output outside the state, to other states or to other nations. These are the sectors in which a state earns revenues that can be used to purchase goods and services produced elsewhere. If these foundation jobs are established in a state, then they attract support jobs in restaurants, health care, education, government and so on. The apparent exception to this statement is the construction sector which sells it’s output locally, but the finance needed to support the purchase of homes or other structures comes from the national or global bond market, bringing revenue that can be spent locally in restaurants and hospitals and schools. The declining share of manufacturing until 2010 probably contributed to both the depth and the length of the recessions, since this created large numbers of permanently displaced workers who had to find jobs in other sectors, something that probably took significantly longer than the recalls that occurred after earlier recessions. The construction sector probably contributed to the length and depth of the 1990 recession and the 2008/09 recession when the share of construction jobs fell substantially but not much in the 2001 recession when the share of construction jobs was rather constant. The Information jobs share bubbled up in the late 1990s but fell from then on. Those jobs are likely to have had adverse consequences especially for the 2001 recession but also played a role in the 2008/09 downturn....”

Explaining the Depth and Lengths of Three Recessions in 50 States Via “Foundational Jobs” Separations

“…The difference between “Layoffs and Recalls” versus “Permanent Separations” in manufacturing is made abundantly clear in Figure 9 which illustrates manufacturing jobs from cycle peak to cycle recovery point. Here we see the V-shaped patterns with a strong bounce back of manufacturing jobs except for the last three recessions: 1990, 2001, 2007. (2020 was a shutdown not a recession.) That confirms with the logic of Figure 8...."

“… The reason why the underlying long-term direction of job growth separates the labor market performance into these two categories is explained with the two illustrations in Figure 8.Each illustration begins with a cost-cutting phase with a downward stroke representing the job loss and a shorter upward stroke representing the amount of hiring that is necessary to get back to previous peak output, allowing for the improvements in productivity put in place during the recession, as was discussed above. To that cost-cutting partial V is added a secular time trend to reflect the long-term trend in employment, contrasting up from down. The first illustration shows that when positive growth in demand is added to the recession cost-cutting, the result is a happy V: layoffs followed by recalls. The lower image adds to the cost-cutting partial V a secular decline, and that is what produces the lazy L, with permanent layoffs and no recalls. The message of these images is that secular increases in jobs support a layoff and recall pattern but a secular decline creates permanently separated workers who may need to move to new locations and/or acquire new skills and/or greatly reduce their aspirations before they can find another job. That takes a lot more time than a recall….”

“…One important reason for a slow recovery is the prevalence of permanent job separations... One of the most important reasons for the slowing recoveries is illustrated in the manufacturing employment data in Figure 7 with log scales that allow straight lines to represent constant rates of growth This figure has three straight arrows, the first arrow identifying the period of generally rising employment in manufacturing and the other two identifying period of declining employment, the first labeled suggestively “Layoffs and Recalls” and the other two “Permanent Separations.”..”

“…To make clear that the last three recoveries have been unusually long,Figure 5 is a bar chart with one bar for each of the eleven expansions, with a blue segment representing the number of months from peak to trough and an orange segment representing the additional months until payrolls return to the previous peak. Until recently the recessions lasted a bit under 12 months and the recoveries occurred in another 12 months, under 24 months (2 years) in total. The previous peak was recovered in under two years in the first five recessions and the sixth was the first to exceed two-years, but by only one month. The seventh (1980) was very short-lived and then commenced a sequence of four expansions with variable length recessions but much longer recoveries....”