If and how increased uncertainty may trigger recessions 1, lower the effectiveness of fiscal and monetary policy, and stifle subsequent recoveries is the main object of a new ‘uncertainty literature’ surveyed in Bloom (2014). The authors typically point to time-varying volatility of both aggregate employment growth – ‘macro volatility’ – and cross-sectional employment growth dispersion, analogously labeled ‘micro volatility’.
‘Micro volatility’ and ‘macro volatility’: The basic facts
Figure 1 illustrates the time-varying volatility of aggregate employment growth (“macro volatility”) and cross-sectional employment growth dispersion (“micro volatility”).
Macro volatility refers to the fact that aggregate employment fluctuates twice as much in contractionary episodes than in expansionary ones – so employment growth is more “jumpy” in recessions. In addition, manufacturing employment rises on average by a mere 2.1% relative to trend in expansions, but contracts by 3.3% relative to trend in downturns. In contrast, micro volatility refers to the fact that employment growth of individual firms differs more in recessions than in booms, so recessions are more disparate across firms. The firm at the top quartile grows employment by 15.7% more than the firm at the bottom quartile in expansions, but in recessions this difference increases to 20%.
Although micro and macro volatility per se reflect two separate phenomena – one describing the differences across firms in a given year, the other describing changes in the aggregate over time – their business cycle properties look strikingly similar. This correlation between micro and macro volatility has been documented in the recent literature,2 but its causes have been little discussed. Existing studies often focus on either the macro or micro level volatility in isolation, and explore the effects of exogenous micro or macro volatility shocks. Deep recessions and meek booms are often interpreted as a result of time-varying volatility of aggregate shocks (see Justiniano and Primiceri 2008, Basu and Bundick 2011, Fernández-Villaverde et al. 2011, and Gourio 2012). Likewise, countercyclical dispersion across firms has been interpreted as a result of a wider distribution of technology shocks to individual firms (see Arellano et al. 2010, Bloom et al. 2012, Schaal 2012, Berger and Vavra 2014, Christiano et al. 2014, and Vavra 2014).
We expand these literatures in two ways.
- First, we propose a mechanism that gives a unified explanation of both micro and macro volatility.
- Second, time-varying volatilities emerge endogenously. Although this does not rule out the presence of exogenous aggregate and cross-sectional uncertainty shocks, it offers a complementary explanation for changing micro and macro volatility.
Figure 1 Macro and micro volatility in US manufacturing employment
All figures are taken from Ilut et al. (2014); for detailed description of construction of data and time series, please see the paper.
Asymmetric hiring: Firms are slow to hire, but quick to fire
In this column, we consider asymmetric behavior by firms as a candidate to explain micro and macro volatility. Many model environments such as a search and matching frictions, hiring adjustment costs, or information processing of ambiguous signals will result in asymmetric firm hiring policy.3
To see how the mechanism works, suppose that a firm’s hiring/firing response to news about profitability is concave: there is less hiring after good news than there is firing after bad news. Suppose further that aggregate shocks shift the mean of all firms’ signals about future profitability: for example, a spell of bad aggregate shocks generates signals that are on average worse. With concave decision rules, the typical firm’s response to its signal during this spell of bad aggregate shocks is then stronger than during a spell of good aggregate shocks.
It follows that both macro and micro volatility of employment growth are countercyclical. Indeed, firms’ stronger responses to bad signals generate not only stronger average responses – that is, sharper movements in aggregate employment growth – but also stronger responses to idiosyncratic components in signals and hence higher cross sectional volatility. Importantly, changes in volatility here derive only from firms’ endogenous nonlinear responses to mean shifts, not from exogenous changes in volatility. Figure 2 illustrates the qualitative workings of the mechanism: firms are faced with symmetric and homoskedastic signals about productivity (bottom right panel) that are transmitted by a concave hiring rule (top right panel) to employment growth (top left panel).
Bad aggregate (red) will now transmit into a more dispersed cross-sectional employment growth distribution, while good aggregate shocks (blue) into a more compressed distribution. Similarly, aggregate employment contractions in bad times will be lower on average than aggregate employment expansions as seen from the differences between the means in the top left graph.
Figure 2 How asymmetric hiring leads to micro and macro volatility
Empirical evidence for asymmetric hiring
As we just mentioned, asymmetric hiring will lead to countercyclical aggregate volatility in employment growth (macro volatility) and countercyclical employment growth dispersion (micro volatility). Both features are borne out in the data as demonstrated in Figure 1. An additional contribution of this mechanism is that it explains micro and macro volatility simultaneously and without assuming exogenous time variation in either common or firm-specific technology shocks.
Our proposed mechanism makes sharp additional predictions that we can test in the data. Most notably, employment growth should be negatively skewed across firms in a given year, and aggregate employment growth should be negatively skewed as well. This means that contractions are on average sharper and deeper than booms, which are slow and meek. Furthermore, one should observe negative time series skewness in employment growth of an individual firm over time.
We find evidence for all three of these: Figure 3 shows that the distribution of employment growth across firms is more spread-out at the left tail (skewness = -0.5). That means that the typical shrinking firm contracts by 2.2% while the typical growing firm increases employment by only 1.5%. Similarly, the aggregate time-series skewness is -1 and the time series skewness of firm-level employment growth is -0.4.
Figure 3 Employment growth is negatively skewed
Asymmetric hiring or asymmetric shocks?
The previous analysis begs the question whether firms are asymmetrically responding to symmetric profitability signals as we propose or whether firms face asymmetric shocks and are just responding to good and bad shocks in a similar way. We use establishment-level Census data to construct profitability shocks and non-parametrically estimate the firm hiring rule. Both the distribution of profitability shocks and the non-parametric hiring rule are plotted in Figure 4.
Figure 4 Employment growth and productivity shocks
As one can see, the distribution of technology shocks is symmetric (skewness = +0.04) while employment growth (right blue scale) is asymmetric: a typical innovation to productivity (+18% output produced with the same inputs) leads to 0.7% hiring while the same negative innovation leads to 1.8% firing. The estimated hiring rule suggests that concavity in firm responses is large enough to account for a significant negative skewness in employment growth and sizable movements in employment dispersion. Table 1 compares the same moments about the employment growth distribution from the data and a simulation obtained from feeding technology shocks to the estimated hiring rule.
Table 1 Asymmetric hiring is quantitatively significant
To conclude, our theoretical and empirical results suggest that one mechanism – an asymmetric hiring response of firms to dispersed signals – has the quantitative potential to jointly explain a range of cross-sectional and time-series properties of employment growth. Making progress on understanding the underlying causes of the movements in these distributions can be fruitful for setting more informed policy responses.
Arellano, C, Y Bai, and P J Kehoe (2010) “Financial markets and fluctuations in uncertainty”, Federal Reserve Bank of Minneapolis Staff Report.
Bachmann, R, and G Moscarini, (2011) “Business cycles and endogenous uncertainty”, Working Paper.
Basu, S, and B Bundick (2011) “Uncertainty shocks in a model of effective demand”, Working Paper.
Berger, D W, and J Vavra (2014) “Consumption dynamics during recessions”, Working Paper.
Bloom, N (2014) “Fluctuations in uncertainty”, Journal of Economic Perspectives, 28(2):153–175, 2014.
Bloom, N, M Floetotto, N Jaimovich, I Saporta-Eksten, and S Terry (2012) “Really uncertain business cycles”, NBER Working Paper No. 18245.
Carvalho, V M, and X Gabaix (2013) “The great diversification and its undoing”, American Economic Review, 103(5):1697–1727.
Christiano, L J, R Motto, and M Rostagno (2014) “Risk shocks”, American Economic Review, 104(1):27–65.
Decker, R, P N D’Erasmo, and H M Boedo (2014) “Market exposure and endogenous firm volatility over the business cycle”, Working Paper.
Fernández-Villaverde, J, P Guerrón-Quintana, J F Rubio-Ramírez, and M Uribe (2011) “Risk matters: The real effects of volatility shocks”, American Economic Review, 101(6):2530–2561, October 2011.
Ferraro, D, (2013) “The asymmetric cyclical behavior of the U.S. labor market”, Working Paper.
Gourio, F (2012) “Disaster risk and business cycles”, American Economic Review, 102(6):2734–2766.
Ilut, C, M Kehrig, and M Schneider (2014) “Slow to hire, quick to fire: Employment dynamics with asymmetric responses to news”, NBER Working Paper No. 20473.
Kehrig, M, (2013) “The cyclicality of productivity dispersion”, Working Paper.
McKay, A, and R Reis (2008) “The brevity and violence of contractions and expansions”, Journal of Monetary Economics, 55(4):738–751.
Schaal, E, (2012) “Uncertainty, productivity and unemployment during the great recession”, Working Paper.
Vavra, J, (2014) “Inflation dynamics and time-varying uncertainty: New evidence and an interpretation”, Quarterly Journal of Economics, 129(1).
1 For example, the Federal Open Market Committee states in April 2008: “Several [survey] participants reported that uncertainty about the economic outlook was leading firms to defer spending projects until prospects for economic activity became clearer.” In 2009, Chief Economist of the International Monetary Fund (IMF) Olivier Blanchard wrote in The Economist: “Uncertainty is largely behind the dramatic collapse in demand”, while the Chair of the Council of Economic Advisers, Christina Romer, noted in her 2009 testimony to the US Congress Joint Economic Committee: “Uncertainty has almost surely contributed to a decline in spending.”
2 See the survey by Bloom (2014).
3 Indeed, physical adjustment costs to labour are not necessary for asymmetric adjustment. If firm decision makers are averse to Knightian uncertainty (ambiguity) and are uncertain about the quality of signals, then it is also optimal to respond more to bad news. Intuitively, ambiguity-averse firms evaluate hiring decisions as if taking a worst case assessment of future profits. With ambiguity about signal quality, the worst case then depends on what the signal says: for a good signal, the worst case interpretation is that it is noisy, whereas for a bad signal the worst case is that it is very precise. Updating from ambiguous signals thus endogenously generates asymmetric actions.