An important recent literature examines the connection between the macro economy and uncertainty stemming from surprises at the firm level. The currently agreed-upon view, as stated in Bloom (2014), is that firms’ prospects are unambiguously more uncertain when economic growth is slow, and that uncertainty about individual firms rose massively during the Great Recession of 2007-2009.
There is no agreement about what this negative link means; does high firm-specific uncertainty cause growth to slow down or do slowdowns cause firms’ prospects to become more uncertain? One reason why uncertainty could weigh on growth is that when firms are risky borrowers, this can cause risk premia to rise and therefore slow down investment and production, as in the model of Christiano et al. (2014). Another possible reason is that when the future becomes more uncertain, firms may postpone hiring and investment decisions, which on net would mean that they temporarily produce less, as in the model of Bloom (2009). It is also possible that firm uncertainty depends on macroeconomic events, as for instance in Bachmann and Moscarini (2012).
This column focuses on uncertainty stemming from firm-specific surprises, in the sense that they are not common with any other firm. Drawing on evidence from De Veirman and Levin (forthcoming) for US firms, we argue that the negative relation between firm-specific uncertainty and the macro economy is much weaker than is commonly believed. Even during the Great Recession, firm-specific uncertainty was not particularly high. That evidence also suggests that high firm-specific uncertainty does not cause subsequent slowdowns in the macro economy.
These results for uncertainty stemming from firm-specific surprises leave open the possibility that uncertainty about future macroeconomic developments does have a negative effect on aggregate production. Bloom (2009) provides evidence in favour of that view.
How do we tell how uncertain firms’ prospects are?
We measure uncertainty due to firm-specific surprises in two steps. The first step consists of estimating firm-specific surprises as the part of firm-level sales and earnings growth which is not explained by a firm’s typical growth rate, the sector to which it belongs and how large the firm is. The second step consists of estimating the standard deviation of firm-specific surprises for a typical firm in every quarter, while controlling for the fact that the composition of the sample changes over time.
This approach is very different from the approach which led to the conclusion that uncertainty as reflected in firm-level sales data unambiguously rises in recessions. Bloom et al. (2014) proxy uncertainty about individual firms by the extent to which firms differ from each other in terms of how fast they are growing at any point of time. However, firms could differ from each other for any number of reasons besides firm-specific surprises. For instance, firms may have been growing at a different pace for some time such that the difference is hardly a surprise. Because our measure controls for firms’ typical growth rates and a number of other factors, we argue that it is a cleaner measure of firm-specific uncertainty.
Firms’ prospects uncertainty and GDP growth
Let us first look at how uncertain US firms’ prospects appear to be if one uses the interquartile range, which is a popular way to measure differences across firms. At any point of time, it is the difference in terms of sales growth between the firm that just about makes the cut for belonging to the quarter of fastest-growing firms and the firm that is just at the edge of the bottom quarter.
The top panel of Figure 1 plots the interquartile range. In both panels, grey shaded areas are recessions as dated by the National Bureau of Economic Research. The most striking feature is that the interquartile range doubled over the course of the Great Recession of 2007-2009. In the quarter before the recession started, the sales growth rate of a relatively fast-growing firm exceeded that of a slow-growing firm by 15.3 percentage points, while that difference peaked at 30.7 percentage points towards the end of the recession. While the former gap is below the sample mean, the latter is well above the level of the interquartile range at any other time on the plot (and 3.8 standard deviations above the sample mean).
On average, over the sample and in terms of deviations from trend, the interquartile range is clearly wider in recessions than at other times. It is also clearly negatively related to quarterly GDP growth.
Therefore, if one uses the interquartile range as the measure of uncertainty, one concludes that uncertainty about individual firms reached extreme levels during the Great Recession and that firm uncertainty is high when GDP growth is low.
The bottom panel of Figure 1 plots the measure of firm-specific uncertainty from De Veirman and Levin. With this measure, firms’ prospects became only moderately more uncertain during the Great Recession; expressed as a standard deviation, uncertainty rose by 17% from the quarter before the Great Recession to its peak in 2009. As uncertainty was relatively low before the Recession and as this increase is moderate, firm-specific uncertainty does not reach unusual heights during the Great Recession. It peaks at about the sample mean.
The finding that firms were not particularly uncertain during the Great Recession in part seems to reflect a tendency for firms to gradually become more stable over the course of the past several decades. During the Great Recession, firm uncertainty does rise above this longer-run trend. On average over the five most recent US recessions, however, and in deviations from trend uncertainty, firms are only a bit more uncertain than at non-recession times. This difference is so small that it is impossible to say with confidence that it applies generally or is particular to the sample we use. Similarly, firm-specific uncertainty tends to be a bit higher when quarterly GDP growth is low, but this apparently negative connection between uncertainty and quarterly growth is very weak.
These findings indicate that any negative relation between firm uncertainty and growth is much weaker than is commonly believed. Furthermore, firms were only moderately more uncertain during the Great Recession than in surrounding years.
Figure 1. Interquartile range and firm-specific uncertainty
Data source: Compustat. Sample: 1978Q1-2012Q4. The top panel is the interquartile range in four-quarter sales growth for firms with at least 100 available observations over the sample period. The bottom panel indicates the standard deviation of firm-specific shocks to four-quarter sales growth. This measure controls for changes in sample composition, so in this case we do not impose a restriction based on the number of observations per firm. Both panels are for a symmetric sales growth rate which De Veirman and Levin call ‘Davis-Haltiwanger sales growth’.
Does firm-specific uncertainty cause slow growth?
In Figure 1, note that in four out of the five most recent recessions, firm-specific uncertainty peaked at or near the end of the Recession, not at its onset or before the recession started.
This may contribute to the findings from De Veirman and Levin which suggest that when firms are uncertain, this does not have a material negative effect on growth in ensuing years. In impulse responses from small-scale Vector Autoregressions, we find that a rise in firm-specific uncertainty is followed by at most a small increase in bond spreads and at most a small decline in real GDP growth. In variance decompositions, we find that only a small fraction of changes in GDP growth are due to changes in firm-specific uncertainty. In Granger causality tests, we find that if one already knows how fast the economy has grown up to some point, information about how uncertain firms have been does not materially help in explaining how fast the economy grew from that point on.
As usual, causality is hard to pinpoint. Still, our results suggest that firm-specific uncertainty does not have a material effect on growth in ensuing years. Our interpretation is that the weak negative relation between uncertainty and growth indicates that downturns make firms somewhat more volatile.
We find that the relation between firm uncertainty and GDP growth is much weaker than is commonly believed. A stand-out finding is that firms were actually not particularly uncertain during the Great Recession. Our evidence suggests that any reason by which firm uncertainty could affect subsequent GDP growth has not been important in the US. While firms are somewhat more uncertain in a typical recession, this is more likely to be a consequence of the downturn than a cause of it.
Authors’ note: The views in this column are solely those of the author and should not be interpreted as reflecting the views of De Nederlandsche Bank.
Bachmann, R, and G Moscarini (2012). “Business Cycles and Endogenous Uncertainty,” Mimeo, Yale University.
Bloom, N (2009). “The Impact of Uncertainty Shocks,” Econometrica, 77, 623-685.
Bloom, N (2014). “Fluctuations in Uncertainty,” Journal of Economic Perspectives, 28, 153-176.
Bloom, N, M Floetotto, N Jaimovich, I Saporta-Eksten, and S J Terry (2014). “Really Uncertain Business Cycles,” Mimeo, Stanford University.
Christiano, L J, R Motto, and M Rostagno (2014),“Risk Shocks,” American Economic Review, 104, 27-65.
De Veirman, E and A Levin (forthcoming), “Cyclical Changes in Firm Volatility’’, Journal of Money, Credit and Banking.