Type to search

Temporary layoffs as a measure of firms’ forecast of the future


Temporary layoffs as a measure of firms’ forecast of the future


Firms often face the need to reduce their workforce to cope with demand or productivity shocks. If the reduction in the workforce is expected to be temporary, an eventual rehiring of the same worker allows the firm to avoid hiring costs, the worker to avoid search costs, and to keep potential firm-specific human capital. However, laid-off workers might not wait. For over a century, a solution to this has been for firms to communicate their beliefs about the temporary nature of the separation, thus creating an expectation of being rehired on the side of the worker (Floud et al. 2014).

Temporary layoffs differ substantially from non-temporary or permanent layoffs. Workers  search less (Katz and Meyer 1990, Nekoei and Weber 2015), but exit unemployment faster through being recalled (Feldstein 1975, Katz 1986, Katz and Meyer 1990, Nekoei and Weber 2015). Recalled workers tend to receive their pre-unemployment wages, avoiding the usual post-layoff wage loss (Katz and Meyer 1990, Nekoei and Weber 2015, Fujita and Moscarini 2017).

Figure 1 The absence of a trend in the share of temporary layoffs among the unemployed

Temporary layoffs as a measure of firms’ forecast of the future 2

Note: Monthly shares from January 1967 to April 2020 in the US. The share is labelled for some historically high levels: June 1975, June 1980, March and April 2020. The last month’s share is attenuated to make the rest of the data visible. The horizontal line shows the average temporary layoff share of unemployment during the entire period. Seasonally adjusted. Source: Current Population Survey (CPS).

During the 2020 lockdown, temporary layoffs have spiked to an all-time high (Figure 1). What can we infer from the prevalence of temporary layoffs about the nature of the current economic crisis? We argue that the share of temporary layoffs, keeping the size of layoffs constant, reveals firms’ forecast of future recovery.

Temporary layoff share as an indicator of firms’ forecasts about their business

In recent research (Nekoei and Weber 2020) we test this hypothesis using unique administrative data from Austria where the type of layoff – temporary or permanent – is recorded for all job separations between 2004-2013. The novel aspect of our data is that it includes the expected rehiring date. To be considered as a temporary layoff, an unemployed worker in Austria is required to provide a written document specifying a projected date for rehire. This data is then matched to the social security and unemployment registers, which contain information on daily wages, exact employment spells, and unemployment benefits received.

Figure 2 Current TL shares and future recalls and separations (firm level)

Temporary layoffs as a measure of firms’ forecast of the future 3

Note: Current TL shares and future recalls likelihood of laid-off workers and separations likelihood of employed workers. 100-binned scatter plots of the relationship between (left-hand panel) the recall likelihood by layoff type, temporary layoffs (TL) or permanent layoffs (PL), (right-hand panel) the ratio of the number of layoffs in the next six months relative to the current month against the current TL share among laid off workers at the firm in the current month. Controlling flexibly for layoff size, the interactions of industry, region, and calendar month (seasonality). The revisualization stretches the range of x-axis beyond the [0,1] interval.

Figure 2 investigates the relationship between a worker’s future employment prospects and the current temporary layoff share of layoffs at her firm or industry (controlling both for the size of layoff as well as industry and region). Temporary layoffs have, in general, a higher chance of being recalled than permanent layoffs. Interestingly, the likelihood of a laid-off worker to be re-hired by her old employer increases with the share of temporary layoffs in her layoff cohort, both for those that were temporarily laid off, and for those that were permanently laid off. An increased temporary layoff share is thus a positive signal for future potential re-hiring even for those that are not promised to be recalled. 

Equally important, we find that a higher temporary layoff share among layoffs means that those workers who kept their jobs during a layoff are less likely to be let go during the following six months (Figure 2, right-hand panel). For similar results at industry level see Nekoei and Weber 2020.

These two pieces of evidence suggest that if the number of layoffs at a firm or in an industry indicates the magnitude of the shock currently faced, the temporary layoff share of those layoffs reveals employers’ forecast about the future of their businesses: a higher temporary layoff share implies that firms have more positive expectations about their recovery.

We hence believe that temporary layoffs can be used as a measure of employers’ forecast almost in real time. As employment offices usually record whether a layoff is temporary, the temporary layoff share is observable to policymakers, both at the sector, and down to the individual firm level. We use such data for Austria in Figure 2. Another potential data source to study the temporary layoff share are labour force surveys, like the CPS used in Figure 1. Contrary to administrative data from employment offices, surveys do, however, not cover the entire population of unemployed workers and thus cannot give a complete image of the economy.1

Prevalence of temporary layoffs

Theoretical models predict that temporary layoffs are common for workers who are difficult to be replaced at the same wage. This includes workers with high intrinsic or accumulated match-specific quality, or when a worker is especially productive at the current firm and has accumulated firm-specific human capital.

Empirically, we illustrate three key patterns for temporary layoff prevalence in our Austrian data (Nekoei and Weber 2020):

First, temporary layoffs are more prevalent in the upper-middle part of the wage distribution. This inverse U-shaped pattern might be due to two countervailing forces at play: on the one hand, higher-skilled workers are costlier to replace (e.g. they have higher firm-specific human capital); on the other hand, the opportunity cost of waiting for a recall is higher for higher-wage workers.

Second, temporary layoff shares and recall rates conditional on being temporary laid off are higher when more workers are let go simultaneously, such as in mass layoffs. Smaller layoffs are more likely due to worker-level inefficiency (high wage relative to productivity) rather than a firm-level productivity or demand shock.

Third and similarly, large layoffs at industry level (recessions) are also accompanied by higher recall shares.2 This is mainly due to higher recall rates of both temporary layoffs and permanent layoffs, rather than higher temporary layoff shares. Counter-cyclical recall rates of both temporary and permanent layoffs suggest that employers do not adjust their forecast adequately to economic cycles, if we assume that temporary layoffs partially reveal the employer’s forecast of the future of the business and the likelihood of recall. Employers are underestimating mean reversion, hence they are relatively pessimistic in downturns, and relatively optimistic in expansionary times.

Figure 3 New-job wage around a recall of a former colleague

Temporary layoffs as a measure of firms’ forecast of the future 4

Note: The log wage change between the pre-unemployment and post-unemployment jobs only for workers who find a new job (not recalled) around the reference date – the day when some of pre-unemployment colleagues are recalled.

Outcomes for temporary laid-off workers

A new aspect of the data we use in Nekoei and Weber 2020 is the information about the anticipated rehiring date communicated by the employers (recall date), which allows us to track outcomes for temporary layoffs relative to the original recall promise. Using the social security records, we show that most recalls happen around a communicated recall date. During an unemployment spell, the hazard of new-job acceptance jumps and the accepted new-job wage drops at the recall date. The same pattern is true after a former colleague has been recalled, which is perceived as lowering the chance of own recall (Figure 3).

Inspired by Katz and Meyer (1990b], we focus on non-recalled temporary layoffs who have not found a new job before the recall date. One half of them never find a job, implying an early sorting of temporary layoss relative to permanent layoffs. The further out the communicated recall date, the lower the accepted new-job wage, both unconditional and conditional on non-employment duration (Figure 4).

This suggests that waiting for a later recall date is costly. However, workers take this (at least partially) into account by reducing their target wage at each point of the unemployment spell when faced with a later recall date. The best evidence is that wages of new jobs found at the beginning of the unemployment spells are lower the longer the waiting period, whereas the pre-layoff wages are constant (Figure 4).

Figure 4 Temporary layoffs’ outcomes by length of waiting period between layoff and expected recall date

Temporary layoffs as a measure of firms’ forecast of the future 5

An interpretation

A framework emerges from these facts: temporary layoffs are used when firms face perceived temporary shocks for workers who are difficult to replace at the same wage. This includes workers with high intrinsic or accumulated match-specific quality. The temporary layoff share among layoffs indicates employers’ forecast about the future of their businesses almost in real time: a higher temporary layoff share implies a more positive forecast about firm recovery, and increases the likelihood of recall both for temporary and permanent layoffs, while reducing future layoff likelihood. Most recalls happen around the communicated recall date, and workers wait for it: they are more selective, search less, and are thus less likely to exit unemployment before the recall date. If not recalled, this wait is costly due to negative duration dependence. Knowing that, temporary layoffs wait less when the likelihood of being recalled is low and the communicated recall date is late. The target wage drops and the job finding rate increases once the recall date has passed and workers have not been recalled, suggesting that workers update their beliefs. However, recall is very selective, so that the employment prospects of those who are still unemployed after the recall date are dire: selection can lead to lower wages, if not a permanent exit from the labour market.


Feldstein, M S (1975), “The importance of temporary layoffs: an empirical analysis”, Brookings Papers on Economic Activity 3: 725–745.

Floud R, J Humphries, and P Johnson (2014), The Cambridge Economic History of Modern Britain: Volume 2, Growth and Decline, 1870 to the Present, Cambridge: Cambridge University Press.

Fujita, S and G Moscarini (2017), “Recall and unemployment”, American Economic Review 107(12): 3875–3916.

Katz, L F (1986), “Layoffs, recall and the duration of unemployment”, Technical report, National Bureau of Economic Research.

Katz, L F and B D Meyer (1990b), “Unemployment insurance, recall expectations, and unemployment outcomes”, The Quarterly Journal of Economics 105(4): 973–1002.

Nekoei, A and A Weber (2015), “Recall expectations and duration dependence”, American Economic Review 105(5): 142–46.

Nekoei, A and A Weber (2020), “Seven Facts about Temporary Layoffs”, CEPR Discussion Paper 14845.


1 Another difference, albeit much less important, is the nuance of the temporary layoff definition: In CPS, for example, a temporary layoff is defined as an unemployed worker who expects to be recalled by the pre-unemployment employer within six months at the time of survey, whereas in the Austrian data, it is defined as having an expected date of rehiring at the time of entering the

2 Relatedly, Fujita & Moscarini 2017 documents a negative relationship between TL and unemployment rate at national level.

Leave a Comment

Your email address will not be published. Required fields are marked *