The authors cite Lazear and Spletzer's argument that “the problem [with the U.S. economy now] is not that the labor market is underperforming; it is that the recovery has been very slow” (2012 pg. 35). Olney and Pacitti offer an explanation for the slow recovery based on a hypothesis that recoveries from downturns should be slower when services are a larger share of the economy. This comes from the simple idea that goods, and not services, can be produced ahead of an anticipated increase in demand, because only goods can be inventoried. Here is the abstract:
Do service-based economies experience slower economic recoveries than goods-based economies? We argue they do. An economy recovers from a downturn when businesses increase production. Both goods and services can be produced in response to actual demand. But only goods—and not services—can be produced in response to anticipated increases in demand, allowing optimistic forward-looking producers to inventory goods until anticipated buyers appear. Services can’t be inventoried. The more services an economy produces relative to goods, the more production is dependent upon only actual increases in demand, and the slower the recovery. We exploit variation across time and states in the share of services in output. Controlling for the depth of the downturn, the higher is the share of services, the longer is the recovery. Extending our results to the current downturn, given the depth of the downturn, the rise in services alone will make the post-2009 recovery last about 1 year longer than it would have a half-century ago.
The authors use national data for the 10 recessions in the United States from 1948 to 2001, and a panel of state data for 50 states and 5 recessions from 1969 to 2001. They use the depth of each recession as a control variable. Getting the state-level data on service sector shares and business cycles was not an easy feat, and is an important contribution of this research. They are able to document interesting, and changing, variation in the share of services by state. My home state, Kentucky, had the second lowest share of services (39%) in the nation in 1967-1969; the share has risen to 50%. Now oil-rich states Wyoming, Alaska, and Louisiana have the lowest share of services. (Take a look at figure 6 and very cool figure 7 in the paper.)
I am left wondering about the other side of the business cycle. The story in this paper is that businesses can anticipate an increase in demand and build up an inventory. What happens when a decrease in demand is anticipated? Wouldn't goods-producing businesses decrease production and start using up their inventory? This would lengthen the downturn part of the cycle by making it start sooner. If this is the case, service-oriented economies would have longer recoveries and shorter downturns. I don't know if the effect would be asymmetrical. For the state-level data, the authors measure recovery as the length of time from one peak to the next peak of the business cycle. This actually captures the length of the downturn plus recovery. I think it would be preferable to use trough-to-peak length instead of peak-to-peak length if the data allowed it, since share of services could plausibly effect peak-to-trough length and trough-to-peak length in different ways.
Overall, I think this working paper documents an intriguing empirical fact. Has anyone out there read (or written) a macro model with two firms with heterogeneous inventory costs? Please comment or send it my way if you have.Setting the inventory costs of one of the firms to infinity could represent the service sector. I think it would be useful to have a model to go along with this paper-- both to help think about my point in the last paragraph and because I am having trouble thinking about GE implications for relative prices and wages in the goods and services sectors and what role that could play. Any other comments or suggestions on the paper are of course welcome, and I will pass them along to the authors.