Recent Trends in Local Multipliers
Attracting good businesses to a region is important to local economic development officials, but they pay particular attention to the attraction and retention of businesses that produce tradable goods and services. A tradable good or service is one that can be sold outside of the metropolitan area in which it is produced. Tradable industries are important because they bring money into the local economy and thus can be associated with a “local multiplier” effect, that is, the possibility that adding a job in a tradable business leads to the creation of new jobs in nontradable ones (or conversely, jobs in nontradable businesses can be lost when a job in the tradable business is destroyed).
Do these “local multiplier” effects exist, and if so, how big are they? In practice, it can be hard to determine which services are tradable, so economists have developed an alternate approach to estimating local multipliers. They focus on the manufacturing industry, which produces many tradable goods, and they look at how many nonmanufacturing jobs are gained or lost in the metropolitan area in response to an increase or decrease in the number of manufacturing jobs. However, they consider only particular kinds of increases and decreases in manufacturing jobs—those that are “externally driven,” since these are less likely to also influence nonmanufacturing jobs in the metro area directly.
First the researchers identify externally driven demand and supply shocks which might manifest themselves as particularly strong employment gains or losses in a particular type of manufacturing at the national level (possibly excluding the metropolitan area under consideration). Then they estimate how much manufacturing employment would have grown in the metro area if each particular type of manufacturing present in the metro area had grown at the same rate as it did nationally. Finally, they estimate how much of the nonmanufacturing employment growth that occurred in the metro area is associated with this estimated portion of manufacturing growth. From these estimates, they calculate an elasticity, a measure of the percent change of one variable in response to a one percent change in another, and those elasticities are used to figure out the multipliers. This approach, credited to Bartik (1991), is used to calculate all of the elasticities mentioned in this article.
Using data from the 1980, 1990, and 2000 censuses, Moretti (2010) estimates an elasticity of 0.3 of nonmanufacturing jobs to manufacturing jobs. This means that a 10 percent decrease in manufacturing jobs would spur a 3 percent decrease in other jobs. However, since the base number of nonmanufacturing jobs is much larger than the base number of manufacturing jobs, a 3 percent drop in nonmanufacturing jobs compared to a 10 percent drop in manufacturing jobs translates into about 1.6 fewer nonmanufacturing jobs for each lost manufacturing job.
I re-estimate these relationships using more recent data. I find that the elasticity of nonmanufacturing jobs with respect to manufacturing jobs was near zero (0.1) for the 2000s (from 2000 to 2010). However, this figure is actually the composite of a negative elasticity (−0.4) during the housing boom period from 2000–2006 and a slightly more positive elasticity (0.2) during the housing bust and subsequent period from 2006–2011 (the most recently available data using comparable geographic definitions is from 2011).
While neither of the aforementioned elasticities is statistically distinguishable from zero, their signs are consistent with other research (Charles, Hurst, and Notowidigdo, 2013) that finds that the housing boom masked the extent of employment losses due to the decline of manufacturing employment. The negative elasticity implies a negative multiplier, which means that the declines in manufacturing jobs during the boom were associated with increases in nonmanufacturing jobs. However, the increases in nonmanufacturing jobs may have been driven by temporarily elevated demand for housing and consumption goods during the boom, which would not have been present, had the boom not occurred. Once the boom ended, the construction and retail jobs driven by the elevated demand disappeared.
So what does this portend for metropolitan areas that lost employment in the manufacturing sector during the 2000s? One possibility is that nonmanufacturing employment will adjust in a manner so that the relationship between nonmanufacturing and manufacturing employment changes observed during the 1980s and 1990s will be re-established. This would mean sharp drops in nonmanufacturing employment in many metropolitan areas. However, it is more likely that manufacturing employment has become a less useful proxy for the tradable sector as manufacturing has fallen as a share of employment in the United States. In this case, it could be that the relationship between tradable and nontradable employment from the 1980s and 1990s still holds and would be apparent in the 2000s if one were to use a better measure of the tradable sector. Research relating to such a measure is currently underway by some of my colleagues here at the Cleveland Fed (Elvery and Venkatu, 2014).
- Bartik, T.J., 1991. "Who Benefits from State and Local Economic Development Policies?" W.E. Upjohn Institute for Employment Research: Kalamazoo, Michigan.
- Charles, Kerwin Kofi, Erik Hurst, and Matthew J. Notowidigdo, 2013. “Manufacturing Decline, Housing Booms, and Non-Employment,” NBER Working Paper No. 18949.
- Elvery, Joel, and Guhan Venkatu, 2014. “New Measures of the Base Sectors of U.S. Regions,” Federal Reserve Bank of Cleveland, unpublished manuscript.
- Moretti, Enrico, 2010. “Local Multipliers,” American Economic Review.