VoxEU Column Industrial organisation Taxation

Regional firm subsidies: Direct, spillover, and welfare effects

Increasing regional inequality has become a major concern for policymakers both in the US and Europe. This column investigates the effects of a large place-based investment subsidy targeted at manufacturing firms in East Germany. It shows that a decrease in the subsidy rate leads to a decrease in manufacturing employment, highlighting spillovers to untreated sectors in treated counties and untreated counties connected via trade and local taxes. It also finds that the place-based policy is at least as efficient as cash transfers for the unemployed but is more effective in curbing regional inequality overall.

In light of rising regional within-country inequality around the world (Moretti 2008, Ehrlich and Overman 2020, Rosés and Wolf 2018), many policymakers have discussed efficient and effective measures to counter the increasing spatial disparities. In this regard, place-based policies, which directly subsidise regions that are economically lagging behind, are a prominent instrument in policymakers’ toolkits (Glaeser and Gottlieb 2008). Between 2014 and 2020, the EU spent more than €350 billion – about a third of its budget – on regional policies (Ehrlich and Overman 2020). The US currently devotes about $60 billion to place-based policies – mostly through business tax incentives (Bartik 2020, Slattery and Zidar 2020). A recent wave of studies has demonstrated that place-based policies unfold positive economic effects on targeted regions (Kline and Moretti 2014, Neumark and Simpson 2015, Ku et al. 2020). But it is well-known that the overall welfare effects of place-based policies also depend on the indirect effects (also known as spillover effects).

These spillover effect may take various forms and signs. On the one hand, agglomeration effects in neighbouring regions, multiplier effects on other untreated industries, or human capital externalities due to the migration and the concentration of better educated individuals may exert positive spillovers. On the other hand, relocation effects due to shifts in economic activities and/or increasing house prices may have negative indirect effects. Last, place-based policies might affect other (local) policy instruments such as local tax rates or other tax bases, as well as government expenditures (such as income tax revenues or spending on unemployment insurance).

In a new paper, we take a fresh view on the direct and spillover effects of a prominent German place-based policy and assess their welfare and distributional implications (Siegloch et al. 2021). We study the case of a policy called the Joint Federal/Länder Task for the Improvement of Regional Economic Structures, or GRW.1  The GRW constitutes Germany's main regional policy scheme for underdeveloped regions. While not exclusively targeted at East Germany, the largest share of the subsidies went to the formerly socialist part of the country after reunification, and it was the main regional subsidy to revitalise the East German economy. The GRW's central instrument is investment subsidies for manufacturing firms in eligible regions. Depending on the region, the year, and the firm size, a company could receive a subsidy of up to 50% of the investment cost. These subsidies can be used for purchasing new machines or building new production sites. The explicit goal of the policy – and a criterion to qualify – is to boost investment, and thereby create new jobs and stimulate regional growth. 

We combine official data on the universe of subsidy cases with administrative social security data on firms and workers to estimate the reduce-form effects of the policy. We differentiate between the direct policy effects and various spillover effects across regions, sectors, and other local policies. Our identifying variation comes from multiple reforms of the maximum subsidy rate of investment cost between 1997 and 2014. These reforms mostly cut subsidy rates differentially across East German counties based on pre-determined economic performance. For each new policy regime, the measure of economic development is based on past performance measures, which are determined at a higher regional level. Explicitly, we compare counties in an event study setup that are below the threshold yielding a higher subsidy rate to counties that are above.2 Eligibility thresholds change across budgeting periods and these changes are partly triggered by EU legislation, which is exogenous to economic developments in East Germany.  

We derive three sets of results: (1) the direct effect of the policy, (2) the spillover effects of the policy, and (3) the resulting welfare and distributional effects of the policy. 

Direct effects

For the direct effects, Figure 1 shows our main result. A one percentage point decrease of the subsidy rate leads to a 1% long-run manufacturing employment decrease. This estimate is similar to a recent study investigating a similar regional industrial policy in the UK (Criscuolo et al. 2019). While most of the subsidy changes are decreases, we do not find asymmetric effects when looking at the few subsidy increases. We also show that a one percentage point decrease of the subsidy rate decreases subsidised investment by 14.6%, and total (subsidised plus unsubsidised) investment by 6.7%. We also show that wages are largely unaffected and regional unemployment tends to increase when subsidy rates are cut. 

Figure 1 Effect of one-percentage-point cut in the subsidy rate on manufacturing employment


Spillover effects

In terms of spillover effects, we derive the following results shown in Figure 2. First, a one percentage point decrease of the subsidy rate leads to a 0.26% and 0.47% employment reduction in the untreated retail and construction sector, respectively (see Figure 2a). Second, there is no evidence for positive or negative spillovers of a county-level shock within the local labour market (see Figure 2b). Third, we find evidence for negative manufacturing employment responses of counties that have a higher trade exposure to treated counties (see Figure 2c). Fourth, we demonstrate important negative policy spillovers. A decrease in the subsidy leads to a long-run increase in local business and property tax rates, which can be rationalised with fixed expenditure requirements of municipalities and a decreasing tax base (see Figure 2d). 

Figure 2 Spillover effects of one-percentage-point cut in the subsidy rate 


Notes: Panel (a) shows the effect of a one-percentage-point subsidy cut on manufacturing, commerce and construction employment. Panel (b) shows the effect of a one-percentage-point subsidy cut on manufacturing employment at the firm, county and local labour market level. Panel (c) shows the effect of a one-percentage-point subsidy cut on the business and property tax rate. Panel (d) shows the effect of a 1% trade exposure to a one-percentage-point subsidy cut on manufacturing employment counties.

Welfare and inequality effects

Last, we ask the question whether, given all these spillover effects, place-based policies are an efficient tool to use. To that end, we make use of the marginal value of public funds (or ‘MVPF’) framework proposed by Hendren and Sprung-Keyser (2020). This calculates the ‘bank for the buck’ of the policy, taking into account fiscal spillover on all tax bases. We calculate a marginal value of public funds of 0.96, which is in the same ballpark as estimates of the marginal value of public funds of unemployment insurance and cash transfers, which target a similar set of beneficiaries (Hendren and Sprung-Keyser 2020). 

As an alternative measure, a simple back-of-the-envelope calculation shows that the cost per job were about €24,000. Importantly, we show that both the cost per job and the marginal value of public funds are substantially downward-biased if one does not account for spillovers. Third, given the similarity in terms of the marginal value of public funds, we show in a simulation exercise that place-based policies are more effective in reducing regional inequality for given efficiency costs compared to cash transfers since place-based policies as more regionally targeted. This suggests that place-based policies can be an efficient measure to counteract the increase in regional inequality if one takes their spillovers into account.


Bartik, T J (2020), “Using Place-Based Jobs Policies to Help Distressed Communities”, Journal of Economic Perspectives 34(3): 99–127.

Criscuolo, C, R Martin, H G Overman and J van Reenen (2019), “Some Causal Effects of an Industrial Policy”, American Economic Review 109: 48–85

de Chaisemartin, C and X D’Haultfoeuille (2020), “Difference-in-Differences Estimators of Intertemporal Treatment Effects”, working paper.

Ehrlich, M V and H G Overman (2020), “Place-Based Policies and Spatial Disparities across European Cities”, Journal of Economic Perspectives 34: 128–149

Glaeser, E L and J D Gottlieb (2008), “The Economics of Place-Making Policies”, Brookings Papers on Economic Activity 39: 155–253.

Hendren, N and B Sprung-Keyser (2020), “A Unified Welfare Analysis of Government Policies”, The Quarterly Journal of Economics 135: 1209–1318.

Kline, P and E Moretti (2014), “Local Economic Development, Agglomeration Economies, and the Big Push: 100 Years of Evidence from the Tennessee Valley Authority”, The Quarterly Journal of Economics 129: 275–331.

Ku, H, U Schönberg and R C Schreiner (2020), "Do place-based tax incentives create jobs?",, 15 February.

Neumark, D and H Simpson (2015), “Place-Based Policies”, Handbook of Regional and Urban Economics 5: 1197–1287.

Rosés, J R and N Wolf (2018), “The return of regional inequality: Europe from 1900 to today”,, 14 March.

Siegloch, S, N Wehrhöfer and T Etzel (2021), “Direct, Spillover and Welfare Effects of Regional Firm Subsidies”, CEPR Discussion Paper 16129.

Slattery, C and O Zidar (2020), “Evaluating State and Local Business Incentives”, Journal of Economic Perspectives 34: 90–118.

Sun, L and S Abraham (2020), “Estimating Dynamic Treatment Effects in Event Studies with Heterogeneous Treatment Effects,” Journal of Econometrics.


1 The German name of the policy is Gemeinschaftsaufgabe Verbesserung der regionalen Wirtschaftsstruktur.

2 We show that our results are robust to heterogonous treatment effects using the recently proposed estimators by de Chaisemartin & D’Haultfoeuille (2020) and Sun & Abraham (2020).

735 Reads