As COVID-19 rapidly emerged through early 2020, Australia introduced a job retention scheme – JobKeeper. Amongst other aims, the program was intended to maintain the connection between workers and firms, and so prevent the breakdown of “good job matches” and job-specific human capital, which could take years to build back up. At the same time though, it had the potential prop up unprofitable and unproductive firms.
Understanding these issues, and more generally how labour flowed between less and more productive firms in Australia over the COVID period is extremely important for a couple of reasons. First, it can potentially help us to understand why productivity growth has been so weak over the past five years. Second, it can help us to design policies for future crises by teaching us what works, what doesn’t, and what unintended consequences various policies could cause.
In a recent paper, we looked at these issues using state-of-the-art high-frequency administrative tax data on firm-level employment, merged with business register data.
The pandemic and productivity-enhancing reallocation
In general, we expect more productive firms to grow, and less productive firms to shrink over time. This is often called productivity-enhancing reallocation, as it pushes up aggregate productivity by shifting resources to more productive uses. It is a key driver of aggregate productivity.
The literature on what happens to the link between firm employment growth and (prior) firm productivity during downturns is mixed. Much of the early literature found that downturns strengthened the link, as less productive firms were forced to exit during tough times. But the more recent literature has found the opposite, particularly when the economic shock is one that affects firms unevenly based on something other than their productivity – such as the credit supply shock during the global financial crisis.
So, what happened to the link between firm growth and productivity during the pandemic? Initially it strengthened.
During the acute phase of the pandemic in the June quarter of 2020, the gap in employment growth between more and less productive firms (within industries) got larger, relative to the years prior to the COVID. And this was particularly evident in harder hit industries, like Accommodation & Food Services, and Arts & Recreation. So, it seems that the most productive firms were best able to navigate the sharp downturn in economic activity when the pandemic hit.
Once this acute phase ended though, the relationship between firm growth and productivity weakened significantly, and fell well below pre-COVID levels. We saw this pattern repeated in Victoria during its lockdown in mid-2020, and in Victoria and NSW during the Delta lockdowns in 2021 (Figure 1).
Figure 1: Difference in performance between high and low productivity firms, by region.
Notes: Lines show estimated difference in employment growth between high productivity firm (LP one standard deviation above industry mean – 1.25 log points) and a low productivity firm (LP one standard deviation below industry mean). Source: Andrews, Bahar and Hambur 2025
The role of JobKeeper
Another interesting takeaway from Figure 1 is that the gap in outcomes between low and high productivity firms picks up substantially in January 2021, and from April 2021. These line up with the end of the second and third phases of the JobKeeper program. This provides some suggestion that JobKeeper may have been supporting lower productivity firms and weakening the link between growth and productivity.
To understand this better, we examine whether the relationship between employment growth and a firm’s productivity differed based on whether the firm was in an industry and state that was heavily using JobKeeper.
The first thing we found was that, at least early on, the relationship between productivity and growth was stronger where JobKeeper was used heavily, even after we accounted for the intensity of the downturn. So, it seems that early on JobKeeper actually disproportionately protected the more productive firms and jobs (within each industry).
This lines up with the fact that, on average, the firms that took up the first phase of JobKeeper were more productive than those that didn’t. This seems a little counterintuitive: why wouldn’t unprofitable and unproductive firms jump at this lifeline? Possibly because there was still some cost in doing so, in terms of running payrolls and administrative functions. Faced with extreme uncertainty, it was the more productive firms that were willing to take on these costs, as they saw the value in continuing to operate and exist during and after the pandemic.
These dynamics changed as the pandemic wore on. We see that the ‘boost’ that JobKeeper gave to productive firms waned over its first phase. Moreover, as firms rolled off the program at the end of the first phase and at the end of the program, we saw a burst of movement of workers from low productivity to higher productivity firms. This suggests that JobKeeper had been delaying this reallocation from low to high productivity firms.
Consistent with this, those firms using the later phases of the program tended to be slightly less productive (Figure 2). By this stage, there were two broad groups of firms who were likely to qualify for the program, based on their turnover decline compared to a year ago: (i) those in sectors still heavily affected by restrictions and (ii) those firms who were either already on a downward trajectory or who had not adapted. The latter set of firms could generally be expected to be less productive.
Figure 2: Distribution of labour productivity for JobKeeper recipient firms.
Notes: Plots residuals from regression of labour productivity on balance sheet controls from take-up regressions in Table 5. Residuals plotted separately for firms in JobKeeper 1.0 and 2.0. Source: Andrews, Bahar and Hambur 2025
Conclusions and implications
So, what do we take from these finding overall?
First, outside of the initial phase, the pandemic significantly weakened the link between firm-level employment growth and prior productivity, which, all else equal, will have weighed on productivity. Whether these effects are long-lasting, or are unwound over time, is an important question for future research.
Second, that JobKeeper initially supported more productive firms highlights the importance of such policies in preventing the closure of otherwise viable firms in the face of an economic shock, and therefore preventing medium-term economic scarring. However, our findings that the program reduced reallocation later in the pandemic highlights the potential for such policies to become distortive over time. This possibility should be weighed against other policy objectives and kept in mind in the design of such programs.
The opinions expressed are those of the authors and do not necessarily reflect the views of the Australian Treasury, the Reserve Bank of Australia, the Australian Government or the OECD.
Recent Comments