Aggregate Reallocation Shocks and the Dynamics of Occupational Mobility and Wage Inequality

This paper presents a dynamic model of structural unemployment and occupational choice in which an economy is subjected to aggregate reallocation shocks. Reallocation shocks, which change the relative labour productivity across occupations, drive variation in the distribution of workers across occupations. The wage paid to workers in a given occupation depends on its labour productivity and the number of workers employed in that occupation. Workers who wish to switch occupations in order to obtain higher wages face a fixed cost to retrain and, in addition, it is more costly to switch to occupations requiring vastly different skills relative to those of the worker’s current occupation. Thus workers may prefer to remain unemployed in occupations suffering through relatively low productivity states. Between the late-1970s and the mid-2000s the U.S. economy featured an episode during which occupational mobility rose along with an increase in wage inequality both in the top and bottom halves of the wage distribution. This was followed by an episode during which occupational mobility fell, while a rise in inequality in the top half of the wage distribution was accompanied by a fall in inequality in the bottom half. The model can produce episodes with properties similar to that of the U.S. experience and thus offers a theory of why these episodes occur. JEL CLASSIFICATION: E24, E32, J24, J31, J62 ∗School of Economics, The University of Adelaide, 3 & 4 Floor, 10 Pulteney St, Adelaide, SA 5005, Australia. Email : [email protected]. Thanks to Paul Beaudry, Chris Edmond, Aspen Gorry, William Hawkins, Viktoria Hnatkovska, Andreas Hornstein, Nir Jaimovich, Amartya Lahiri, Henry Siu, and Mark Weder for valuable comments. Especial gratitude to Lawrence Uren for insightful remarks. I also thank presentation participants at the University of Adelaide, the University of Melbourne, UNSW, the Reserve Bank of Australia, the Federal Reserve Bank of Richmond, the Australian Conference on Quantitative Macroeconomics and the 2011 SED Meetings. Thanks to Thomas Lemieux, Giuseppe Moscarini and Chris Robinson for providing useful data. Finally, this paper grew out of collaborative work with Fabrice Collard to whom I am grateful. It is now well known that in the U.S. economy there were large increases in both wage inequality and occupational mobility rates between the mid-1970s and the end of the 1980s. Interestingly, while occupational mobility rates dropped between the mid-1990s and the early2000s, wage inequality in the top half of the wage distribution, as measured by the log 90/50 ratio (i.e. the ratio of the 90th percentile wage and median wage), continued to rise while wage inequality in the bottom half of the wage distribution, measured by the log 50/10 ratio, flattened or even decreased.1 This polarization in wage inequality since the mid-1990s combined with the earlier dynamics of the wage distribution has resulted in divergent views concerning the causes of the changing dynamics of the wage structure ranging from changes in minimum wages to skill-biased technological change to offshoring of jobs. Recent work by Acemoglu and Autor (2010) and Autor, Katz, and Kearney (2008) suggest that the relative supply of workers to perform various tasks coupled with changing demands for task performance in the production process is important in determining both the distribution of wages across the aggregate economy the allocation of workers across production tasks. In this paper, I show that a simple dynamic general equilibrium model with costly occupational switching and occupational productivity shocks can be consistent with the stylized observations discussed above. Importantly, the model economy is only subjected to shocks which change the relative labour productivity across occupations. I refer to this type of shock as a reallocation shock. Furthermore, a measure of distance is introduced between occupations in order to capture the notion that switching between occupations that require the performance of vastly different tasks is more costly to workers. It turns out that the history of the distribution of workers across occupations plays a primary role in understanding the dynamics of wage inequality and mobility dynamics across both episodes. Given these simple assumptions, the model can generate episodes which are characterized by rising trends in the occupational mobility rate, as well as in wage inequality in the top and bottom halves of the wage distribution, as measured by the the log 90/50 ratio and log 50/10 ratio, respectively. The model can also generate episodes which are characterized by falling trends in the occupational mobility rate and wage inequality in the bottom half of the wage distribution along with a rising trend in wage inequality in the top half of the wage distribution. Combining specific assumptions about the initial distribution of workers across occupations along with the sequence of aggregate reallocation shocks hitting the economy, the model offers a theory to help understand the rise in both the occupational mobility rate and wage inequality from the mid-1970s through the midThe polarization in wage inequality since the mid-1990s is documented by Autor, Katz, and Kearney (2008) and Lemieux (2010). Moscarini and Thomsson (2007) and Robinson (2010) document the dynamics in occupational mobility and while the measured levels of mobility differs across their papers, the large rise and fall in trend occupational mobility is similar. Furthermore, their documentation of the rise in occupational mobility from the late-1970s through the late-1980s is consistent with the findings of Kambourov and Manovskii (2009a).