• Recessions reduce employment and payroll, which impacts workers compensation premium
  • In addition, two distinct changes in workforce composition push injury frequency below trend during a recession and above trend during its recovery
    • The employment share of sectors with high injury frequency, like construction and manufacturing, falls during a recession
    • The employment share of short-tenured workers (who have higher frequency than longer-tenured workers) falls during a recession
  • In a future recession, payroll losses and the decline in injury frequency due to changes in workforce composition are likely to be milder than in the Great Recession
    • Because construction and manufacturing have lower employment shares today, employment losses in these sectors will have less impact on overall injury frequency
    • Since the share of short-tenured workers today is nearly the same as in 2008, the sensitivity of injury frequency to their employment losses is unchanged

The main effect of a recession on workers compensation is obvious. Employment and payroll fall, which reduces premium. But there is also evidence that injury frequency falls faster when the economy goes into recession and increases or falls more slowly during a recovery. This phenomenon is largely due to changes in the workforce that are linked to the recessionary cycle. More precisely, there are two complementary but distinct aspects of workforce change during a recession: changing sector mix and changing employee tenure.

In recessions, the percentage decline in construction employment significantly exceeds the overall percentage decline in employment. This drives down injury frequency because construction has the highest injury rate per worker of any sector. Likewise, manufacturing loses an above-average share of employment during recessions and has higher injury rates per worker than most other sectors. During recessions, firms also stop or slow down hiring, and recent hires are often the first to be laid off. This naturally shrinks the number of short-tenured workers, who have higher frequency than longer-tenured workers.

Our discussion will focus on injury frequency per worker (or per payroll, when employment is not available). NCCI typically reports frequency per premium, partly because this measure varies less by employment mix. However, here we are interested in measuring the impact of a recession on the total number of work injuries – including the effect it has on changing employment mix. See “NCCI Explains Its Top 3 Frequency Measures” for a more detailed explanation of different frequency measures. See also “Workers Compensation Claim Frequency – 2013 Update” for an expanded discussion of the construction sector’s contribution to frequency by payroll and by premium during the Great Recession.

Over more than two decades, NCCI research has examined the connection between injury frequency and fluctuations in employment, including the relationship between job flows and injury frequency during recessionary periods.

The Great Recession and the Long Expansion

What we know about the effect of recessions on workers compensation comes from research that mostly predates the Great Recession. Since then, workforce demographics have changed, and injury frequency has declined substantially. Here we describe changes in the labor market that are likely to make a future recession different from past recessions.

Employment and Payroll Change

Before the Great Recession, US nonfarm employment peaked at 138.4 million in December 2007 and January 2008. Employment fell by 6.3% to its lowest level of 129.7 million in February 2010. This was more than twice as large a percentage decline than any other recession in the last 50 years. Annual Quarterly Census of Employment and Wages (QCEW) growth in total wages and salaries, which had been above 5% each year from 2004-2007, increased by just 1.5% in 2008, fell 6.0% in 2009, and increased by 2.1% in 2010. It has since resumed steady growth, growing at an annualized 4.7% rate from 2011-2018.

The Great Recession’s most significant effect on workers compensation came directly from the hits to employment and wages, which together impacted payroll. But the Great Recession also impacted aggregate injury frequency via changes in the composition of the workforce, a topic which we discuss next.

Frequency Change

As shown in Figure 1, frequency has been in a nearly uninterrupted decline for over two decades. Injury frequency per 10,000 full-time equivalent workers fell faster than trend at the onset of the Great Recession, as in prior recessions. Frequency fell by 7.3% and 6.1% in 2008 and 2009. That combined 13.6% over two years is 3.1 percentage points more than it would have fallen if frequency had changed at the average rate from the previous 10 years. Frequency then rose in 2010, the only time in the last 20 years it has done so. The next sections discuss fluctuations in the mix of workers and how they help explain the observed frequency cycle.

Employment Changes by Sector

Every sector of the economy lost jobs during the Great Recession, although magnitudes varied widely. In Figure 2, we show the change in employment since December 2007 by NCCI’s categorization of industry groups. The shaded area indicates the NBER-defined recession period.

As shown in the first panel, Contracting (Construction) employment collapsed, losing over 25% of the total jobs it had at the start of the Great Recession. Contracting employment rebounded throughout the recovery, rising faster than overall employment each year, but it has only just reached its pre-Recession level of employment. Since overall employment has increased, the share of employment in Contracting has fallen by 10% since the end of 2007. Manufacturing’s share has fallen by 16% in the same time period. Manufacturing employment did not fall as sharply as Contracting during the recession, but fewer manufacturing jobs returned afterward.

The other industry groups were not as badly affected. Employment in Goods & Services was affected least, falling just 3% during the Great Recession − less than half the percentage drop in overall employment. Overall, its share of employment has increased by 4% since the end of 2007, driven primarily by increased health care jobs. Office & Clerical employment followed a very similar pattern of decline and recovery as overall employment, while Miscellaneous employment experienced a sharper decline and a sharper recovery. The employment share in each of the latter two industry groups increased by 1%.

Because Contracting and Manufacturing make up a smaller share of employment and payroll than in the past, a similar proportion of payroll decline to prior recessions in these sectors will have a smaller impact on overall frequency change in a future recession.

Employee Tenure and Workplace Injuries

In Figure 3, we show the proportion of employees with less than one year of tenure and the proportion of work injuries suffered by such employees.These estimates come from two BLS surveys. Employment share by worker tenure comes from a biannual supplement of the Current Population Survey, collected in January of even years, and work injury share comes from the annual Injuries, Illnesses and Fatalities survey. There are two key takeaways. First, short-tenured workers’ share of workplace injuries is always much higher than their share of employment. Second, the employment and injury shares for such workers both fell sharply during the Great Recession, from 2006 to 2010, and both series have inched up since. These observations support the finding that short-tenured workers have relatively high injury frequency and are hit hardest by recessions.

How much of the frequency change in the Great Recession can be explained by changing employee tenure? The data suggests short-tenured workers have about a 50% higher injury rate than other workers.We do not combine the BLS data to directly estimate injury rates for short-tenured workers because the timing and collection method for injuries and illnesses data collected is not a perfect match for the employment data. To acknowledge that uncertainty, we use the rough estimate of 50% higher injury rates to approximate the effects of changes in the distribution of employee tenure. If we assume exactly 50% higher frequency for short-tenured workers, then their decrease in employment share from 22.9% to 19% between January 2008 and January 2010 should have created a 1.7% total drop in injury frequency. By this approximation, the changing mix of employee tenure explains a little more than half of the difference in injury rates from trend during the Great Recession.

What Will the Next Recession Look Like?

To assess the impact of a future recession on workers compensation, we construct three recessionary scenarios for payroll and injury frequency across different industry groups, all deviating from the baseline forecast in 2020-2022. Our choice of this time horizon is suppositional – we are not predicting that a recession will begin in 2020.

We first establish a baseline forecast from which to create the scenarios. Payroll has been growing by an average of a little less than 5% a year since 2011. Moody’s Analytics forecasts annual payroll growth to continue at 5.9%, 3.1% and 4.2% in NCCI states in 2020, 2021 and 2022, which implies a 4.4% average annual rate over the three years.Moody’s also projects 6.1% payroll growth for the full year 2019. In this baseline forecast, changing industry mix contributes a cumulative 0.8% increase to workers compensation premium from 2019 through 2022. This is primarily because of above average payroll growth in Contracting. Table 1 shows payroll growth forecasts by industry group, as well as the cumulative three-year impacts of projected payroll growth and industry mix change.

Figure 4 shows the overall change in payroll in the baseline forecast and each of the three scenarios (Con indicates Contracting). Full assumed changes in payroll and cumulative effects on mix and payroll growth are shown in Table 2. Because Goods & Services is historically less sensitive to recessions, its payroll share increases in all three scenarios compared to the baseline forecast.

How much does the recession path affect premium solely through the channel of changing industry mix? We calculate that by holding loss costs by industry group constant and calculating the difference in the implied change in premium from the change in payroll.

In the Mild and Severe recession scenarios, changing mix contributes a 1.1% and 1.3% increase to premium, respectively. Each of these is a larger increase than in the baseline. The biggest reason for this is that during a recession, payroll growth in Goods & Services (driven by health care) will not fall as fast as other industries. Following prior recessions, the percentage decline in Goods & Services payroll from baseline in these scenarios is barely half as large as in other sectors. Goods & Services has above average losses, so this slower decline exerts upward pressure on premium.

In prior recessions, this upward pressure was more than counteracted by the downward pressure exerted by slower or negative payroll growth in other industry groups, especially Contracting. But Contracting’s current payroll share is barely one-third that of Goods & Services. Because of this declining share, payroll effects on Contracting would have to be especially strong in a future recession for changing industry mix to drive down frequency as it did in the past.

In the scenario designed to affect Contracting most, changing industry mix drives premium down by 0.4%. However, even the Contracting scenario generates a cumulative difference from the baseline of only 1.2 percentage points in premium, much less than the direct effect of slower payroll growth. The effects of industry mix in the scenarios are small, and which direction they will go depends on the specific nature of the recession.

During the Great Recession, the industry mix change had greater impact because Contracting and Manufacturing had higher employment shares than now and each suffered very large job losses. Using the same scenario calculations on the change in industry group payroll from 2007-2010 (but using current loss costs, for comparability to the scenarios), we find a cumulative effect of industry mix on premium in those three years would have been -4.9%. Changing industry mix affected the last recession much more than we expect it to affect the next.

While the distribution of workers across industries has changed since 2007, the distribution of employee tenure has not. Figure 5 shows that the proportion of workers by range of employee tenure in 2018, the latest available time of measurement, was very similar to that in 2008. We also know from the data in Figure 3 that short-tenured workers still have much higher injury rates than others.

Thus, any change in employment will likely have the same impact on the share of short-tenured workers and thus frequency in the next recession as in prior recessions. The total effect from changing tenure is likely to be smaller in the future than it was in the Great Recession only because the next recession should be milder, not because of the pre-recession composition of the workforce.


Recessions impact the workforce by shrinking employment and changing the mix of workers. Though exceptionally severe, the Great Recession exhibited both characteristics. The primary impact on workers compensation was the most obvious one: the sharp drop in aggregate employment during the Great Recession and the long recovery afterwards directly impacted payroll. But the primary payroll effect was amplified by secondary effects: disproportionate layoffs among short-tenured workers and in industries with high injury frequency, especially construction. Both dimensions of the change to the mix of workers tended to reduce overall injury frequency as employment fell during the Great Recession, and to increase frequency as employment recovered.

In a future recession, we think that changes in industry mix are likely to be less significant, and consequently that the induced frequency effect via this channel is likely to be more muted. The share of employment in construction and manufacturing has declined since 2007 while employment in service sectors, especially health care, has grown. Although health care has high injury frequency, health care employment is much less sensitive to recessions than construction and manufacturing. Consequently, employment losses during the next recession are likely to include a lower share of workers from industries with high injury frequency than during the Great Recession.

On the other hand, the distribution of employee tenure looks very similar today as it did just prior to the Great Recession. For this reason, we think that changes in worker tenure are likely to have a similar impact on frequency in the next recession as in the Great Recession, net of a difference in scale. In a future recession, short-tenured workers are likely to be laid off first and rehired last, but the overall magnitude of the resulting frequency impact will depend on the recession’s scale and duration.