Can payroll tax cuts help firms during recessions?

Abstract

This paper estimates the effect of payroll tax cuts on firm-level employment and balance-sheet outcomes during economic downturns. We use two regional payroll tax cuts in Finland as well as the onset of the Great Recession to estimate the effect of the recession on firms treated by the payroll tax cuts compared to a similar control group. When implemented, prior to the Great Recession, we estimate that the payroll tax cuts had limited effects on employment and balance-sheet outcomes of firms located in the treated regions. However, when the recession starts, some of its negative effects were substantially hampered by the previously enacted payroll tax cuts in treated firms. These employment effects are exacerbated for men and low-skilled employees. We also find that sales and profits in treated firms respond differently in treated firms during the recession. We provide some evidence showing that firms that are liquidity constrained are the ones that exhibit the strongest response. This shows that payroll tax cuts can make firms more resilient during downturns, possibly by relaxing liquidity constraints.

Introduction

How do payroll taxes affect firms during economic downturns? The common wisdom in public finance is that the incidence of payroll taxes is mostly borne by workers, in which case, payroll tax changes should not affect firm-level outcomes. However, recent evidence has questioned this received wisdom, by showing that some, if not all, of the incidence of payroll taxes is borne by firms and can substantially affect firm level activity.¹ This paper empirically addresses this question by estimating the effect of payroll taxes on firm activity during the Great Recession.

Our empirical strategy relies on using two regional tax cuts in Finland that were enacted several years prior to the Great Recession and were still in place throughout the entire recessionary period. These two waves of payroll tax cuts were enacted in 2003 and 2005 and were repealed in 2012, which allows us to observe the behavior of firms in the treated regions before and after the onset of the Great Recession in 2008 in Finland.² The treated regions are all located in the Northern part of Finland, which is relatively poor compared to the rest of the country. However, not all Northern regions are selected into treatment, which allows us to use the remaining similar Northern regions as a control group. Using a difference-in-difference strategy, we compare firms located in regions treated with the payroll tax cuts to firms in similar control regions where payroll taxes were not changed.

First, we find that the payroll tax cuts, when enacted several years prior to the Great Recession, have very limited effects on earnings and on firm-level outcomes, implying that the savings from the lower payroll tax cuts are likely channeled into firms and likely saved since they do not appear to affect employment, earnings, sales or investment. The clearest effect of the cut in payroll taxes appears at the time of the Great Recession: here we estimate that firms located in the treatment regions exhibit a substantially different behavior than the ones in the control region. In particular, while the Great Recession caused a decrease in earnings in firms located in both control and treatment regions, these decreases are substantially smaller in the treated regions. And these effects seem to be particularly acute for low-skilled workers. Overall, the net of payroll tax wage bill is relatively higher for firms in the treated regions, consistent with them paying their employees relatively more. While the employment effects are somewhat noisier, we also find that treated firms employ relatively more workers, which could be due to either more hiring or fewer separations. In addition, treated firms also invest relatively more and realize more sales than control firms during the Great Recession. Importantly, some of these effects are long lived and tend to persist even after the payroll tax cuts are repealed in 2012. This is particularly true for the employment effects. One candidate explanation for the finding that payroll taxes help firms during recessions is that they relax liquidity constraints. We find evidence consistent with this interpretation: treated firms that have widely available liquid assets have a muted response to the Great Recession, whereas those firms that face liquidity constraints exhibit the strongest response. This suggests that payroll taxes help firms by relaxing their financial constraints.

Our main identification assumption is that firms in the treatment regions would have behaved similarly to firms in the control region had there been no change in payroll taxes. A common test of this assumption is to ensure that trends in the control and treatment groups are parallel prior to the change in payroll taxes. Our graphical evidence is consistent with this identification assumption. We also show evidence that the control and treatment regions are very similar along many observable characteristics. This is likely due to the fact that regions were selected into treatment not because of their economic conditions but rather because of political considerations: the payroll tax cuts were targeted at Northern regions only, which are typically poorer than the rest of Finland, but among these regions, it did not affect the poorest ones. While all the treatment regions are relatively poor compared to the rest of Finland, they are not the poorest regions in the area.

Our main contribution is to shed light on the effects of payroll taxes during recessions, which is a question that had not been addressed by previous literatures. In doing so, we contribute to three main strands of literatures. First, we contribute to the tax incidence literature and, in particular, to the few papers that address the question of who bears the incidence of payroll taxes. Our main contribution to this literature is to assess whether the incidence of payroll taxes depends on the business cycle. This literature has mostly focused on estimating the incidence of payroll taxes without paying much attention to business cycles. While older papers have mostly estimated that workers bear the incidence of payroll taxes (Gruber, 1997), recent results (Saez et al., 2012, Saez et al., 2019, Benzarti and Harju, 2020) show that firms are likely to bear a substantial burden of the incidence of payroll taxes. Relatedly, Bozio et al. (2019) show that this burden crucially depends on whether payroll taxes are linked to the benefits they fund. Notably, Korkeamäki and Uusitalo (2009) use the same variation from the first wave of the experiment as we do to study short-term responses and find no clear wage or employment responses.³ Finally, Huttunen et al. (2013) finds no employment or wage responses to a very targeted payroll tax cut for the employers of older and low-wage workers in Finland.

Second, we contribute to a public finance literature that assesses whether government intervention should vary during the business cycle. While we do not address this question directly – we do not derive optimal payroll tax rates over the business cycle – we provide evidence that the effect of payroll taxes varies over the business cycle, which can help inform future research on this question. In public finance, this literature has mostly focused on the optimal provision of unemployment insurance (see Landais et al. (2018a) and Landais et al. (2018b)) and on public expenditure (Michaillat and Saez, 2019).

Relatedly, our results have implications for the effectiveness of stimulus programs. Our paper is closely related to Ku et al. (2020), who show that place-based payroll tax cuts are partially shifted to employees and have large employment effects. We complement their analysis by focusing on the effect of place-based payroll tax cuts during periods of severe economic downturns. While there are several papers showing that Value-Added Tax cuts are ineffective at stimulating firm activity as they are mostly passed through to profits (see, e.g. Benzarti et al., 2020, Benzarti and Carloni, 2019, Kosonen, 2015, Harju et al., 2018), few papers have considered the stimulus effects of payroll taxes. Our findings suggest that they can have long-run effects on firm activity by increasing production, output and profits even after the cuts are repealed. They also lead to higher wages, which can also stimulate demand.

Finally, our results are also closely related to papers estimating the effect of place-based policies (reviewed for example in Kline and Moretti (2014) and Neumark and Simpson (2015)). In our empirical setting the payroll tax cuts were targeted to high unemployment areas and aimed at increasing employment. Our empirical findings support the view that reducing payroll tax rates in areas suffering from poor economic conditions can increase economic activity, especially during downturns.