BIGGER FIRMS CUT MORE JOBS DURING RECESSIONS
Taken as a group, larger employers shrink faster than smaller ones during recessions.
They are also slower to expand afterwards, creating most of their new jobs after the economy has long been out of recession. This finding by Giuseppe Moscarini and
Fabien Postel-Vinay adds support to the idea that small businesses are the ‘engine of growth
’ during recessions because they remain relatively more active than larger firms.
Their study, presented at the Royal Economic Society ’s 2010 annual conference, uses
US data starting in the 1970s to show that, as a consequence of this different behaviour, the difference in hiring rates between large and small employers is smallest when unemployment is high during a recession, and largest once the recession has long since ended and unemployment is low.
This finding challenges the conventional wisdom of ‘Gibrat’s law’, named after the economist Robert Gibrat ’s 1931 work, which states that the growth rate of a company is independent of its initial size. Existing studies in support of this ‘law’ have typically ignored the effect of the business cycle.
The researchers find that a firm ’s employment size may actually help to predict its growth, but the sign of the relationship flips during a recession, with smaller companies growing faster – or not slowing down so quickly. While this may add weight to policies in favour of small businesses, the authors caution that the large companies still provide the most jobs because of their overall size.
More…
We present new empirical evidence that large firms or establishments are more sensitive than small ones to business cycle conditions. Taken as a group, larger employers shrink faster (or expand more slowly) during and after a typical recession, and create more of their new jobs late in the following expansion, both in gross and net terms, than smaller employers.
As a consequence, the difference in employment growth rates between large and small employers is strongly negatively correlated with the unemployment rate, and varies by about 5% over the business cycle.
We document those patterns using a variety of US datasets starting in the late 1970s and now spanning four business cycles. The pattern that we uncover is robust to different treatments of entry and exit of firms and establishments, and occurs within, not across, industries and geographical areas.
We find the same pattern in several other countries, including in longitudinal censuses of employers from Denmark, France and Brazil.
As further evidence that smaller firms fare better than large ones in bad economic times, we also establish that the cyclical pattern of relative performance in terms of employment size is also reflected in stock returns.

Specifically, we show that the difference in returns between benchmark portfolios of large cap stocks and portfolios of small cap stocks is also negatively correlated with the unemployment rate.
The firm size/growth relationship is the subject of a vast literature originating with
Gibrat ’s (1931) seminal contribution. A recurring result, known as Gibrat ’s law , states that future firm growth is independent of initial firm size, as measured by either employment (as in Gibrat
’s original work) or, more often, assets, capital or sales. This literature, however, typically ignores business cycle effects.
Our findings suggest that firm employment size may predict its growth, but the sign of this relationship flips depending on cyclical conditions: it is negative (or less positive) when unemployment is high and positive (or less negative) when it is low.
Omitting cyclical indicators may thus lead to the conclusion that, on average, these cyclical effects wash out and size does not predict subsequent growth, which is Gibrat ’s law. Interacting size with the detrended unemployment rate in a growth/size regression is likely to invalidate this conclusion.
Our results corroborate only in part the common wisdom that small businesses are the engine of (net) job creation. This wisdom has been criticised as subject to statistical fallacies: according to Gibrat ’s law, future job creation should be independent of an employer ’s initial size.
Again, while this may hold true on average over the business cycle, we establish that small firms create more jobs as a fraction of their employment only when unemployment is high (which is, arguably, when jobs are most needed).
In terms of absolute number of jobs added, however, large firms dominate at nearly all times, because they account for a larger fraction of employment to begin with.
ENDS
‘Employer Performance and Employer Size over the Business Cycle’ by Giuseppe
Moscarini (Yale University and NBER) and Fabien Postel-Vinay (University of Bristol,
Paris School of Economics and CEPR)
Contact:
Fabien Postel-Vinay, Department of Economics, University of Bristol
Email: Fabien.Postel-Vinay@bristol.ac.uk