Two New York Fed economists, Richard Crump and Ayşegül Şahin, writing in Liberty Street Economics, have shared some interesting findings regarding developments in the labor market during the ongoing recovery. Their conclusion is that unemployed construction workers, according to several indicators, seem to be doing better than workers who lost jobs in other sectors.
Based on their research, job-finding rates for unemployed construction workers have increased more rapidly than for the overall pool of unemployed. While flows out of the labor force for unemployed construction workers have remained flat, they have increased for those who lost jobs in other sectors. Also, using the Displaced Workers Survey (DWS) conducted by the U.S. Bureau of Labor Statistics, they show that construction workers who find jobs have the same distribution of earnings as other displaced workers who find a job.
These facts, according to the authors, provide support to the hypothesis that problems in the labor market cannot be blamed on the degree of mismatch between displaced construction workers and job vacancies in other sectors.
In this post, we present an alternative view of the fate of unemployed construction workers by looking specifically at unemployed construction workers who find jobs in other industries. Our conclusion is that unemployed construction workers are generally experiencing relatively large wage declines (relative to what they earned before becoming unemployed). Except for the lowest-skilled workers, losing a job and having to take a new job in a new industry generally involves a wage decline. That effect is especially pronounced for construction workers who become unemployed.
The U.S. Census Bureau's Survey of Income and Program Participation (SIPP) followed a panel of workers from 2008 through March 2011. The SIPP asked each worker questions about his or her individual characteristics as well as that worker's labor market experiences. Using the SIPP, we investigated the wage changes workers experience before and after an unemployment spell when their new job is in a different industry. Is the wage effect of switching sectors larger for unemployed construction workers relative to those workers in other sectors? The table displays the results from this exercise looking at the last three recessions.
We divided the sample of unemployed workers according to the broad industry grouping in which they lost their job. However, given the different pool of workers in each sector, we controlled for individual characteristics to isolate the specific effect on wages earned from switching sectors. These characteristics include the level of education, gender, age, whether the worker lives in a metro or rural area, the length of the unemployment spell, and whether the worker is married.
Each cell in the table represents the relative effect of switching industries on the post-unemployment wages of workers in a given industry, having taking into account the heterogeneity in the pool of workers across sectors. For example, of those workers who lost jobs in manufacturing in the 2008 SIPP, those who became reemployed in any of the other four sectors earned 9.9 percent less than those unemployed manufacturing workers who found jobs in the manufacturing sector. For construction workers, the effect of switching sectors reduced wages by 18.8 percent. In our sample, about 50 percent of workers who lost jobs in construction found jobs elsewhere but mostly in the high- and low-skilled service industries. For comparison, we repeated the calculations for other panels in SIPP that include recessions, and the results are displayed in the columns under the 2001 and 1991 headings. It is true that industry-switching unemployed construction workers also experienced large wage declines after the 2001 recession, but the decline for the 2008 panel was considerably larger.
Our conclusion is that drawing inferences about the evolution of job finding and the unemployment rates across different sectors doesn't paint a complete picture of the situation without a comparable look at wage changes for those unemployed. That comparison should also take into account the differences between the attributes of construction workers and workers in other sectors. Our results do not necessarily contradict the facts presented by Crump and Şahin. However, using the SIPP has several advantages relative to using the DWS. It allows us to compare the initial wage after an employment spell relative to the last wage earned (as opposed to average wages in the DWS) and also to control for the length of the unemployment spell that workers experience. Our more disaggregated view of the data indicates that during the 2008 recession and recovery, unemployed construction workers who took jobs in other sectors seem to have done so at a considerable loss in income. The reason may well be a mismatch between the skills they possess and those required by their new job.
Pedro Silos, research economist and associate policy adviser, and
Lei Fang, research economist and assistant policy adviser, in the Atlanta Fed's research department