COVID-19 business support illustration
VoxEU Column COVID-19 Productivity and Innovation

The impact of firm-level Covid rescue policies on productivity growth and reallocation

The Covid-19 pandemic caused a massive drop in economic activity and prompted governments to implement a variety of firm-level support schemes. This column uses administrative data on the universe of firm subsidies during Covid-19 for the Flanders region in Belgium to study the impact on productivity growth and reallocation. Subsidised firms have a 43% lower probability of exiting the market, and experience a 7% productivity growth, compared to firms that applied for but did not obtain subsidies. Exiting firms are on average less productive than incumbents. In the aggregate, both subsidised and non-subsidised firms contribute to productivity growth, with a larger average contribution per firm for subsidised firms.

On top of a massive health crisis with millions of lives lost, Covid-19 triggered the largest drop in world GDP since WWII. In most Western countries, GDP fell by between 5% and 10% in 2020. These steep numbers resulted from a variety of emergency policies to curb the spread of the virus, including lockdowns, social distancing, and industry closures. To flank the sanitary measures, many governments implemented firm-level support schemes, including direct subsidies, furlough schemes, and bank guarantees (see OECD 2021 for an overview).

While these measures have likely supported economic activity, they potentially hamper the process of creative destruction: firms that are not productive enough to survive otherwise might stay in the market, at the expense of resources that could have been allocated to more productive firms. As a result, these policies could unintentionally contribute to an aggregate productivity slowdown already present in many EU countries (Andrews et al. 2016), or an increased ‘zombification’ of the economy (Andrews et al. 2017). In a recent paper (Konings et al. 2022), we use administrative data on the universe of firm-level subsidies in the Flanders region of Belgium for 2020 to answer the following questions:

  1. What is the impact of these subsidies on within-firm productivity growth and survival?
  2. How do these subsidies contribute to aggregate productivity growth and allocative efficiency?

Firm-level subsidies during Covid-19

The Flemish government issued subsidies to support companies that were forced to shut down or saw a large reduction in their sales in 2020. The goal was to keep the economy afloat and to avoid firm failures, layoffs, and liquidity issues as a direct result of the imposed sanitary policies. Tandem measures were issued at the sector level. For example, if a lockdown on all non-essential stores was imposed, companies in these sectors would become eligible for the flanking support measures. The scale and speed of this program is truly impressive: the median pay-out rate was only two days after submitting a very light online application. Between 12 March and 31 December 2020, a total of €1.7 billion was paid out through this channel.

Table 1 describes the five support measures that have been paid throughout 2020, the requirements and subsidy amounts, and their eligibility period. As restrictive policies changed throughout the year, subsidies also evolved. The measures can be broadly categorised along two dimensions: (1) a premium for mandatory closure (premium 1) versus experiencing a drop of at least 60% in turnover relative to the same period in 2019 (premia 2-5); and (2) flat fee (premia 1-3) versus ad valorem subsidies (premia 4-5).

Table 1 Firm-level subsidies, 2020

Table 1 Firm-level subsidies, 2020

Figure 1 shows the top ten sectors in terms of support in 2020. Over €500 million was allocated to retail trade, followed by the food and beverage sector with almost €400 million. Other sectors include wholesale, construction, accommodation, and travel agencies. Some of these sectors were also hit disproportionally hard by the stringent lockdown policies, and often remained closed the longest. Over 100,000 firms were supported, and €1.5 billion of subsidies were allocated to micro firms (i.e. with at most ten full-time employees).

Figure 1 Allocation of subsidies across NACE two-digit sectors, 2020

Figure 1 Allocation of subsidies across NACE two-digit sectors, 2020

The impact of firm subsidies on firm exit and within-firm productivity growth

For each firm, we know for which subsidy the firm applied, the date of application, whether the subsidy was approved or not, and if granted, the payment date and amount of support. We combine this information with quarterly sales from their value-added tax (VAT) declarations, employment from the Social Security, and annual accounts from Bureau Van Dijk. Companies are classified as those that received a subsidy in 2020 (‘treated’) and those that applied for but did not get support (‘untreated’). We then estimate a difference-in-differences setup and compare within-firm outcomes before and after treatment (first difference) with companies that applied for but did not obtain subsidies (second difference).

Figure 2 shows the estimated impact of firm subsidies on quarterly within-firm productivity growth. Coefficients are normalised to zero in the quarter before the first subsidy. Point estimates for the pre-treatment quarters are not statistically significant different from zero, supporting the parallel trends assumption over the observed period. The estimated treatment effect is 5% in the first quarter post-treatment, and 4% in the following quarter. Estimating this model on yearly changes, we find an impact of 7-8%. We also estimate the model separately for each subsidy scheme and find that most of the effect can be contributed to the first – and largest – scheme. However, we do not find a differential effect across the other schemes, which varied in terms of flat versus ad valorem fees. These results are also robust to several alternative specifications, including nearest-neighbour matching, the Sun and Abraham (2021) estimator for staggered treatments, and federal furlough schemes.

We then turn to the impact of the support schemes on firm exit. Within firms over time, and controlling for a series of typical predictors of firm exit (debt-to-asset ratio, productivity levels, firm size, and age), we find that subsidised firms have a six percentage point (or 43%) lower probability of exiting the market in 2020, compared to non-subsidised firms. A counterfactual scenario of no subsidies to the treated group would have increased exit rates by 61% instead. Everything else constant, exiting firms are on average significantly less productive, across both treated and non-treated firms, suggesting that creative destruction still occurred across the board.

Figure 2 Impact of subsidies on productivity growth, quarter on quarter, 2020

Figure 2 Impact of subsidies on productivity growth, quarter on quarter, 2020

The contribution of firm subsidies to aggregate productivity growth

Is there any effect of these subsidies on aggregate outcomes? Figure 3 shows the evolution of labour productivity growth of the Flemish economy between 2006 and 2020. Aggregate growth (black line) was 6.1% in 2020, confirming similar official statistics for Belgium (National Council for Productivity 2021). Building on Olley and Pakes (1996) and Melitz and Polanec (2015), we decompose aggregate growth into several firm-level components. In 2020, the largest contribution to aggregate growth is the unweighted average productivity growth of surviving firms (blue), which is similar to other years pre-pandemic. The reallocation term (green) measures the covariance between firm-level growth in market shares and productivity growth. If recessions are times of cleansing and creative destruction, we expect a positive covariance term. However, the negative reallocation term indicates that less productive firms gained market shares, or that smaller firms became more productive. Yet, the figure shows that this reallocation term was already negative in 2018 and 2019 and is thus not particular to the Covid crisis in 2020. This might signal other structural issues in the Flemish economy. Finally, both entering (yellow) and exiting (red) firms are in general less productive than incumbents, so entry has a negative and exit a positive contribution to aggregate growth in 2020, which is again similar to other years in the data, and consistent with the firm-level analysis on exit.

Figure 3 Aggregate productivity growth and its decomposition, 2006-2020

Figure 3 Aggregate productivity growth and its decomposition, 2006-2020

Finally, we further decompose productivity growth into the contributions of subsidised firms and all other firms in the Flemish economy in Table 2. This generates the same four components for each subgroup separately, plus an additional term that measures the reallocation of market shares from treated to all other firms. Focusing on the contribution of surviving firms, we find that the within-firm evolutions of both treated (4%) and all other firms (4.3%) are similar across both groups. However, as the share of treated firms is much smaller (around 20% of all firms), this implies a larger contribution to aggregate productivity growth per firm. In both groups, we see the negative reallocation of market shares and productivity growth at -3.1% and -1.6%, respectively. The larger negative reallocation effect for treated firms might signal that a natural process of creative destruction was less prevalent for subsidised firms. Conversely, across groups, market shares of non-treated firms increase at the expense of treated firms (+1.1%). These aggregate results provide a subtle and more nuanced view on the impact of the Covid subsidies on the economy than the average within-firm effects above.

Table 2 Contribution of treated and all other firms to aggregate productivity growth, 2020

Table 2 Contribution of treated and all other firms to aggregate productivity growth, 2020

Our data end in 2020, and we can unfortunately not measure the persistence and decay in future quarters. We also have no information on how these subsidies have been used within the firm: were they used for fixed costs, the wage bill, or other costs? Some firms reported that they have been innovating by switching to, or additionally opening up, online sales platforms. Alternatively, it is possible that capacity utilisation increased, suggesting excess capacity before the crisis. In future work, with information on particular investments by the firm, it would be interesting to further explore these dimensions.

Authors’ note: The paper on which this column is based was initiated as an independent expert evaluation of Covid-19 support measures taken by the Flemish government. The report, written together with Technopolis Group, can be found here (in Dutch). We are grateful to Flanders Innovation and Entrepreneurship (VLAIO) for making the data on individual enterprises and support measures available for academic research.

References

Andrews, A, C Criscuolo and P N Gal (2016), “The best versus the rest: The global productivity slowdown, divergence across firms and the role of public policy”, OECD Productivity Working Papers 5, OECD Publishing, Paris.

Andrews, D, M Adalet McGowan and V Millot (2017), “Confronting the zombies: Policies for productivity revival”, OECD Economic Policy Papers 21, OECD Publishing, Paris.

Konings, J, G Magerman and D Van Esbroeck (2022), “The Impact of Firm-level Covid Rescue Policies on Productivity Growth and Reallocation”, CEPR Discussion Paper 17552.

Melitz, M J and S Polanec (2015), “Dynamic olley-pakes productivity decomposition with entry and exit”, The Rand Journal of Economics 46(2): 362–375.

National Council for Productivity (2021), Annual report 2021.

OECD (2021), “Country policy tracker for covid-19”.

Olley, S G and A Pakes (1996), “The dynamics of productivity in the telecommunications equipment industry”, Econometrica 64(6): 1263–1297.

Sun, L and S Abraham (2021), “Estimating dynamic treatment effects in event studies with heterogeneous treatment effects”, Journal of Econometrics 225(2): 175–199.

438 Reads