Productivity and misallocation: labor market in a volatile time


A wealth of empirical research based on longitudinal establishment data for a number of countries has consistently found large and persistent productivity differences among plants or firms producing in the same industries in any given time period.* A closely related literature has overwhelmingly demonstrated that the process of input and output reallocation that takes advantage of these productivity differences significantly contributes to aggregate productivity growth. 

There are many reasons for the existence of heterogeneity in plant-level decisions and outcomes. Differences in entrepreneurial ability and the organizational structure or the vintage of capital may all explain cross-sectional variations in productivity. Similarly, differential access to human capital, infrastructure and credit may generate variation in the manner firms invest in technology and use their resources.  Uncertainty may also underlie the observed dispersion. For instance, plant specific shocks to demand, investment opportunities, input costs and technology are the main sources of uncertainty discussed by the class of models developed in Jovanovic (1982) and Hopenhayn (1992). The development and adoption of new products or production techniques is also an uncertain process. Finally, regulation may protect some firms by discouraging entry, reallocation or innovation, through special tax exemptions, subsidies or credit priorities (Parente and Prescott, 1994; Acemoglu et al., 2001).

Some of these reasons for plant-level heterogeneity are associated to the misallocation of inputs or outputs. For instance, reallocation may be costly due to technological barriers. Also, credit constraints and special subsidies and taxes typically generate permanent differences in plant-level productivity, and may explain why largely different plants produce in the same narrowly defined industries in any given time period. Similarly, shocks and adjustment costs have transitory effects on plant productivity dispersion.

During the last years, manufacturing firms in Chile have faced a more volatile environment. Changes in the energy market –in particular cuts in the supply of natural gas imported from Argentina–as well as the increased volatility in the real exchange rate that resulted from the adoption of an inflation targeting monetary policy may have affected labor market outcomes.  In a recent paper we analyze the evolution of labor productivity dispersion over time as a proxy for labor misallocation, and its relation to the observed process of aggregate growth in Chile´s manufacturing sector as well as its relation with the rise in volatility. In particular, we use plant-level data for the manufacturing sector from 1979 until 2007 to estimate the distribution of labor productivity and its evolution over time. Based on the method of Hsieh and Klenow (2009), we estimate the extent of labor misallocation and analyze the evolution of observed dispersion over the sample period. 

We find that the dispersion of labor productivity across firms increased sharply, particularly in 2004. By 2007 –the end of our sample period– the gap had not yet been fully closed.  Using the cross-sectional decomposition first introduced by Olley and Pakes (1996) we also find that in every year the reallocation of employment has been productivity enhancing: labor is being reallocated from less to more productive plants. However, in the latest period, the relative incidence of this process in explaining overall productivity growth has diminished.

An examination of events occurring in the Chilean economy leads us to relate the rise in productivity dispersion and the reduced relevance of the reallocation process to changes in the energy market, in particular to cuts in the supply of natural gas imported from Argentina. We also argue that the increased volatility in the real exchange rate that resulted from the adoption of an inflation targeting monetary policy may too have affected the outcomes in the labor market. We also examine the role of jumps in the interest rate as a result of the Asian crisis that hit the Chilean economy by the end of the 1990s.  A simple econometric model shows that the rise in the observed macroeconomic volatility does correlate with the observed rise in the dispersion of labor productivity, in particular among those sectors that produce traded goods and/or use gas as a source of energy more intensively. 

In other words, we hypothesize that the volatility of shocks that firms face explains the observed rise in productivity dispersion. However, it is also possible that the speed at which firms adjust to these shocks has slowed down over time. Following Caballero et al. (2010) we estimate the speed of adjustment in our manufacturing data set to find that, if anything, firms tend to adjust more quickly in the recent years. We do find, however, that plants in skilled-labor intensive sectors close the gaps in employment more slowly than the mean. Again, this finding may be related to labor market regulations that mostly affect the ability of firms to adjust skilled labor. 

Finally, we estimate the aggregate effects of labor misallocation. In a simple, yet revealing exercise, we find that if half of the employees in the first quintile of plants’ labor productivity distribution were reallocated to the top quintile plants, manufacturing productivity would increase by about 25%.

*See Bartelsman and Doms (2000) and Foster et al. (2001) for surveys.
Share this