Creative destruction is a major driving force of modern market economies.1 Firms enter and exit the marketplace, plants are built and destroyed, and workers change jobs and occupations. In recent decades, economists have started to learn that the amount of reallocation that occurs in market economies is massive.2 It is the rule rather than the exception, and it is essential in a well-functioning market economy. The microeconomic ups and downs of reallocation allow new products to be introduced, new technologies to be put to use, and resources to be moved to productive places.
Modern market economies also experience ups and downs at the aggregate level. Booms and recessions—sometimes mild, sometimes severe—occur all the time, and stabilizing the business cycle is one of the major policy goals of many governments. But before conducting a stabilization policy, a natural question to ask is: how are macroeconomic fluctuations (business cycles) and microeconomic fluctuations (creative destruction) related? If macroeconomic fluctuations reflect the resource reallocations of a well-functioning market economy, the business cycle may not be such a problem after all.
One popular view among economists is that business cycles do in fact represent waves of creative destruction. Booms are times of heavy creation, and recessions are times of heavy destruction. If so, attempts to stabilize the business cycle could actually hamper the healthy process of resource reallocation. Recessions would not be a bad thing either, especially from a long-run perspective, because they would serve to cleanse the economy of inefficient production units.3 Not all economists agree. Some hold the opposite view and see recessions as times of slow reallocation,4 where creation and destruction decelerate. In their view, a recession is indeed a bad thing.
For policymakers then, it is important to know what actually happens to the reallocation process during the business cycle. We examine this issue for the US manufacturing sector in a recent paper (Lee and Mukoyama, 2007), using plant-level data from the US Census Bureau.5 In particular, we look closely at the entry (birth) and the exit (death) of plants over the business cycle. Overall we find that entry rates (the percentage of plants opening in a given year) are much higher in booms than in recessions. However, exit rates (the percentage of plants closing in a given year) are comparable in booms and in recessions. What is interesting is that plants entering during recessions are very different in terms of employment and productivity from those that enter during booms, whereas exiting plants are rather similar in both phases of the business cycle. On average, plants that enter during recessions are larger (they hire more workers) and are more productive than plants that enter during booms. Such differences are relatively small for plants exiting in booms or recessions.
These results cause us to rethink the relationship between business cycles and micro-level reallocations. In the literature that argues for a cleansing effect of recessions, a lot of emphasis is placed on the belief that cleansing occurs at the destruction (or exit) margin. That belief rests on earlier research, which found that job destructions at continuing plants are strongly countercyclical. In contrast, our research finds nothing special going on at the exit margin. Because closing plants are similar in booms and recessions, it suggests that recessions do not necessarily cause productive plants—those that could have survived in good times—to shut down in large numbers. During recessions, when it is hard for businesses to survive, businesses tend to reduce their workforces by firing existing workers. But recessions do not appear to be times of massive cleansing of less-efficient incumbents. Less productive plants are indeed driven out of the market, but not only in recessions. Rather, cleansing at the exit margin occurs all the time, in a similar manner, regardless of business cycle phase.
Of course, this does not mean that the business cycle and reallocation process are not related at all. Quite the contrary—entry behavior is very cyclical. The fact that entrants are very different in booms and recessions suggests that there is some very important selection going on at the entry margin over the business cycle. In booms, a small and relatively unproductive plant can enter—because times are good, an unproductive plant can still be profitable. But in recessions, only productive (and large) plants can profitably enter. Recessions might have a positive effect on average productivity by selecting only more productive plants. However, such selection does not necessarily mean that the cleansing of inefficient, existing plants is taking place. Selecting only highly productive entrants can be more important. In short, when we study the effects of business cycles, we should shift our focus from exit behavior to entry behavior: “Creation” is a more important margin than “destruction.”
We believe our findings matter for policy makers in the following respects. First, the fact that the plants that enter in booms are different from those that enter in recessions indicates that there is a much larger barrier to entry during recessions. Such a barrier may hurt the long-run growth of the economy. New plants often embody innovations, and researchers find that entry is an important source of aggregate productivity growth. For this reason, an important question is what makes entry more difficult during recessions. Perhaps the costs of the initial investment for startup are higher or financing is more difficult to obtain.
Second, our research indicates that the outcome of various stabilization policies will depend on their effect on entry and exit rates. In our paper, we build a model that is able to quantitatively replicate the features of the data, and we run several experiments with it. Imposing firing taxes could have a stabilizing effect if the policy did not affect entry and exit behavior, because the tax would induce plants to reduce the frequency of their firing and hiring.6 But we find that entry rates fluctuate more with the firing tax, which could translate into more volatile aggregate output. Entry rates fluctuate more because the firing tax discourages entry more during recessions than during booms. Because the firing tax is more likely to affect larger plants that tend to reduce their workforces in the near future, entering plants during recessions are more likely to be affected.
Our finding points to the importance of policies that are targeted to the incentive to enter. For example, an effective method for stabilizing the economy is to encourage entry during recessions. In particular, if a market inefficiency (such as financial constraints) is creating barriers to entry, then encouraging entry during recessions is indeed a good thing.
Finally, we would like to emphasize that our empirical results are based on US manufacturing data. An intriguing question for future research is to see how these results can be compared across sectors and across countries.
Barlevy, G. (2002). “The Sullying Effect of Recessions,” Review of Economic Studies 69, 41-64.
Caballero, R. J. and M. L. Hammour (1994). “The Cleansing Effect of Recessions,” American Economic Review 84, 1350-1368.
Caballero, R. J. and M. L. Hammour (2005). “The Cost of Recessions Revisited: A Reverse-Liquidationist View," Review of Economic Studies 72, 313-341.
Davis, S. J., J. C. Haltiwanger, and S. Schuh (1996). Job Creation and Destruction, Cambridge, MIT Press.
Dunne, T., M. J. Roberts, and L. Samuelson (1988). “Patterns of Firm Entry and Exit in US Manufacturing Industries,” RAND Journal of Economics 19, 495-515.
Lee, Y. and T. Mukoyama (2007). “Entry, Exit, and Plant-level Dynamics over the Business Cycle,” Federal Reserve Bank of Cleveland Working Paper 07-18.
Samaniego, R. M. (2006). “Entry, Exit and Business Cycles in a General Equilibrium Model,” mimeo. George Washington University.
Veracierto, M. L. (2004). “Firing Costs and Business Cycle Fluctuations,” mimeo. Federal Reserve Bank of Chicago.
1 The research in this article was conducted while the authors were Special Sworn Status researchers of the U.S. Census Bureau at the Michigan Census Research Data Center. Research results in this article have been screened to insure that no confidential data are revealed. The views expressed in this article are those of the authors and do not necessarily reflect the position of the Census Bureau, the Federal Reserve Bank of Cleveland or the Federal Reserve System.
2 See the pioneering work by Dunne, Roberts, and Samuelson (1989) and Davis, Haltiwanger, and Schuh (1996).
3 See, for example, Caballero and Hammour (1994) for a theoretical examination of this view.
4 See, for example, Barlevy (2002) and Caballero and Hammour (2005).
5 We use the Annual Survey of Manufactures from 1972 to 1997.
6 See Veracierto (2004) and Samaniego (2006). In Samaniego’s (2006) model, entry is endogenous but it fluctuates very little over the business cycle.