STRUCTURAL JOB TURNOVER CONCENTRATED IN CYCLICAL DOWNTURNS

While extensive work has been done to examine the impact that aggregate or transitory shocks have on turnover, the evidence is still mixed, and generalisations must be treated with some caution. In general, one would not expect job turnover to fluctuate over the economic cycle: job gains should be as likely to fall in recessions as job losses are to rise. However, Davis and Haltiwanger (1990 and 1992) and Davis Haltiwanger and Schuh (1994) found evidence of a counter-cyclical pattern in total job turnover in manufacturing in the U.S This is a result of an asymmetry, with larger increases in job losses than declines in job creation during downturns and the reverse during upswings.
Others have reached similar conclusions. For Italy, Contini and Revelli (1992) found that turnover appears to be counter cyclical6. Violante and Prat (1992) found a weakly significant correlation at the aggregate level but a more strongly significant relationship for manufacturing alone (Turin region) 1978- 1989. For Canada, Baldwin and Gorecki (1990) found evidence of counter-cyclical movements in manufacturing during the period 1970-1981. Baldwin, Dunne and Haltiwanger (1993) found a negative correlation between turnover and net employment growth for Canada but not the U.S. Konings (1993) found a negative correlation for a sample of large firms in the manufacturing sector in the United Kingdom.
By contrast, for Italy for the period 1984-1989, Gavosto and Sestito (1992) found that turnover in manufacturing was positively correlated to the business cycle7. Boeri (1994) did not find that there was a strong relationship between turnover and net employment growth for 7 OECD countries, likely as a result of intra-industry turnover which remained high through the economic cycle.The difficulty in resolving the issue of the counter-cyclicality of turnover stems partly from the lack of sufficiently long time series to take account of the magnitude of movement over the cycle. The most significant movements in turnover, when structural change may have occurred, took place at only two points -- the recessions of 1981-82 and 1990 or later. However, differences in turnover during these two periods from the remainder of the time series cannot be isolated8. Instead, most studies have made use of rank correlations between turnover and net employment growth.
There are several possible explanations for the counter-cyclical pattern in job turnover. Major allocative shocks, accompanied by sharp rises in job losses, may be at the root of recessions. Alternatively, as is widely held, recessions may still be caused by aggregate shocks but these may influence the timing of job reallocation. Blanchard and Diamond (1990) concluded that the most likely explanation is that recessions may be a time of cleaning up. Cyclical downturns may be accompanied by structural change in which the increased competition for a dwindling market produces industry restructuring. The budget constraint of firms varies over the cycle. During upswings, it typically becomes softer so that inefficiencies may arise. In downturns, when the budget constrain becomes firmer, firms must deal with these inefficiencies.