Industrial policy at work: Evidence from an income tax break for IT workers
Industrial policy at work: Evidence from Romania’s income tax break for IT workers
Despite industrial policies making a comeback in policy and academic circles, the credible empirical evidence on both their effectiveness and efficiency remains scarce (for exceptions, see Criscuolo et al. 2019 and Barwick et al. 2020). Providing such evidence has become even more pressing in the context of new widespread adoption of industrial policies in response to the Covid-19 crisis (Motta and Peitz 2020, Hassler et al. 2020). In light of this, we present evidence on the effectiveness of an industrial policy with a unique design, targeting a sector seen as key to the transition to a knowledge economy – the IT sector.
Romania’s personal income tax break for programmers
In 2001, Romania introduced a full personal income tax break for programmers working for software firms (defined based on their sector of activity and a minimum of revenues from software creation). The programme also applied to those in an eligible programming-related occupation, as well as those holding an eligible IT-related bachelor’s degree. A 2013 amendment greatly expanded the policy’s scope by extending the list of eligible sector codes and bachelor’s degree specialisations.
This tax break (and its expansion) aimed to support the development of the IT sector in Romania. Despite its high perceived potential (owed to post-communist high-quality STEM education), Romania’s IT sector was underdeveloped. High labour taxes were considered the main reason why the sector did not take off, and programmers emigrated in large numbers.
In Manelici and Pantea (2020), we estimate the effects of this income tax break on incumbent firms in the IT sector in Romania, the overall IT sector in Romania relative to that in comparable countries, and the broader economy (in particular, high-intensity IT-using sectors).
Unique features of this industrial policy
Several features of this policy stand out. First, the policy focuses on the IT sector, which is regarded as systemically important. The later emergence and smaller size of the IT sector in the EU compared to the US is seen as a major cause of the EU’s lower productivity growth (Van Ark et al. 2008, Gordon and Sayed 2020). As a result, this policy is especially relevant for the EU and for countries whose ‘digitisation’ of the economy lags behind.
Second, despite the statutory incidence of the tax break on workers, its unique combination of rules (on the worker, the firm, and the activity performed by the worker in the firm) implies that the tax break rewards particular matches between workers and firms. The expectation was that lowering the tax burden on these matches would increase their prevalence. Our evidence of firm and sector growth is in line with a shared incidence of the tax break between workers and firms.
Lastly, this policy involves reductions in labour taxes, rather than the more commonly used reductions in corporate taxes or subsidies. In the EU, reductions in labour taxes are typically used to increase the employment of ‘hard-to-employ’ (Eurofound 2017) or research and development workers (European Commission 2014). However, labour tax reductions for research and development workers tend to affect researcher wages alone (due to the inelastic supply of researchers) (Goolsbee 1998, Lokshin and Mohnen 2013). In contrast, we find strong positive effects of the policy on firm and sector size.
Incumbent firms expanded after the introduction of the policy in 2001
We estimate the effects of introducing the policy in 2001 on firms in the eligible software sector. To that end, we use a difference-in-differences strategy on Amadeus data. The control group contains firms in other non-eligible high-tech knowledge-intensive sectors. For identification, we exploit the unexpected nature of the legislative initiative (which belonged to an independent member of the parliament and was adopted by emergency ordinance) and the similarity between firms in eligible and non-eligible high-tech knowledge-intensive service sectors.
Figure 1 shows that firms in eligible and non-eligible high-tech knowledge-intensive service sectors were following similar trends before the introduction of the tax exemption. However, after 2001, firms in the eligible sector embarked on an upward trend. By 2005, treated firms have, on average, 24% higher revenues and 13% more workers than firms in comparable non-eligible sectors (relative to 2000).
Figure 1 Firm-level effects of the 2001 income tax break on revenues and employment
Note: These figures use a difference-in-differences strategy to estimate the effects of the 2001 tax break on incumbent firms. Treated firms are those in the NACE Rev 1 sector 722 (software consultancy and supply). Firms in the control group are in non-eligible HTKI service sectors. The data comes from Amadeus, Bureau Van Dijk. Regressions include firm and calendar year fixed effects.
Incumbent firms also grew after the 2013 expansion in the scope of the policy
To estimate the effects of the 2013 expansion of the tax break, we leverage administrative data where we observe the share of exempted employees in each firm. The empirical strategy is still a difference-in-differences one, but treatment is now defined based both on the sector of the activity and a non-trivial increase in the share of exempted employees (to at least 20%) after 2013. The reference group contains firms in HTKI service sectors (eligible and non-eligible) with under 5% workforce exemption throughout the entire sample period. For identification, we rely on the unexpected nature of the reform, which was prompted by the transition from ‘NACE rev 1.1’ to ‘NACE rev 2’ and occurred during a period of political volatility.
Figure 2 displays a lack of differential trends (between treated and untreated firms in high-tech knowledge-intensive service sectors) in revenues and employment before 2013, and a gradual expansion of treated firms after 2013. By 2015, treated firms have 20% higher revenues and 10% higher employment than untreated yet comparable firms (relative to 2012).
Figure 2 Firm-level effects of the 2013 expansion of the income tax break on revenues and employment
Note: These figures use a difference-in-differences strategy to estimate the effects of the 2013 tax reform on incumbent firms. Treated firms are those whose share of income tax exempted workers jumps from under 5% to over 20% after 2013. Firms in the control group are in HTKI service sectors and have an under 5% share of income tax exempted workers throughout the entire sample period. The data comes from Amadeus, Bureau Van Dijk. Regressions include firm and sector-by-year fixed effects.
After 2001, the IT sector in Romania grew faster than the IT sector in comparable countries
We also estimate the effects of the tax break for the IT sector on the growth of this sector in Romania relative to growth rates in comparable countries. The IT sectors in these other countries are likely to be affected by similar sector-specific technology and demand shocks (such as the US dot-com crash and the subsequent ‘offshoring’ to Central and Eastern Europe). This analysis uses the synthetic control method and data from Eurostat and the World Bank. ‘Synthetic Romania’ combines those Central and Eastern European countries with pre-2001 levels of GDP per capita and IT sector performance that were similar to Romania’s.
Figure 3 indicates that in 2015 the revenues (employment) in the IT sector of Romania were 6.52 (1.83) times larger than the revenues (employment) in 2000. These values reflect the exceptional growth of the IT sector in Romania – plausibly owed to the 2001 policy –relative to the growth of revenues (employment) in all other sectors in Romania as well as the same difference in growth rates in ‘synthetic Romania’.
Figure 3 The growth of revenues and employment in the IT sector in Romania vs. “synthetic Romania”
Note: These figures use the synthetic control method to estimate the effect of the 2001 tax break on the growth of Romania’s IT sector (relative to the rest of the economy and the year 2000) compared to that of the IT sector in “synthetic Romania.” The data comes from Eurostat and the World Bank.
After 2001, high IT-usage sectors grew faster in Romania than in comparable countries
The improvements in the prices, quality, and variety of IT services (that plausibly ensued after the expansion of the IT sector) are expected to benefit the broader economy, especially in a country that had relatively low levels of IT adoption beforehand. To estimate these benefits, we compare the expansion of sectors that used IT services most intensively before 2001 to those that use IT less intensively. The intensity of the use of IT services is defined based on the shares of input expenditures from the IT sector in total input expenditures in the ‘2000 Input-Output table’.
As in the previous sector-level analysis, we rely on the synthetic control method and data from Eurostat and the World Bank. Figure 4 suggests that after 2001, high-intensity IT-using sectors grew more than low-intensity IT-using sectors in Romania relative to ‘synthetic Romania’ (0.75 times more for revenues and 0.61 times more for employment).
Figure 4 The growth of revenues and employment in the high-intensity IT-using sectors in Romania vs. “synthetic Romania”
Note: These figures use the synthetic control method to estimate the effect of the 2001 tax break on the growth of Romania’s high-intensity IT-using sectors (relative to Romania’s low-intensity IT-using sectors and the year 2000) compared to that of high-intensity IT-using sectors in “synthetic Romania.” The data comes from Eurostat and the World Bank.
Implications for policy
We find that the income tax break for workers in IT had positive effects on the size of eligible firms in IT sectors relative to non-eligible firms in other high-tech knowledge-intensive service sectors, the expansion of the IT sector in Romania relative to the IT sector in comparable countries, and the growth of higher-intensity IT-using sectors relative to the lower-intensity ones.
This evidence suggests that this industrial policy has been effective in supporting the development of the IT sector – a sector of systemic importance in the transition to a knowledge economy. Moreover, given the conditions of the tax exemption, the jobs created are high-skilled/high-wage (programming) jobs – a frequent policy priority. Finally, this study illustrates the potential of preferential labour tax rates, a less-frequently used form of industrial policy.
While establishing the efficiency of this industrial policy lies beyond the scope of this analysis, this policy is likely to have met the theoretical criteria required for it to be welfare-improving. Namely, the policy supported an activity that was ‘new’ for Romania in 2001 (Rodrik 2004), and skill/knowledge-intensive (Aghion et al. 2011, European Commission 2017, Cherif and Hasanov 2019). Moreover, Romania most likely had a latent comparative advantage in this activity and only lacked a policy signal to tilt resources towards it. Finally, the growth of the IT sector has supported the growth of IT-using sectors – a necessary condition for the sector to generate inter-sector positive externalities (Harrison and Rodriguez-Clare 2010). This is encouraging in terms of the ability of governments to design and implement industrial policies that are both effective and efficient.
Aghion, P, J Boulanger and E Cohen (2011), “Rethinking Industrial Policy”, Bruegel Policy Brief.
Barwick, P J, M Kalouptsidi and N Zahur (2019), “China’s Industrial Policy: An Empirical Investigation”, NBER Working Paper 26075.
Cherif, R and F Hasanov (2019), The Return of the Policy That Shall Not Be Named: Principles of Industrial Policy, Technical Report, International Monetary Fund.
Criscuolo, C, R Martin, H G Overman and J Van Reenen (2019), “Some Causal Effects of an Industrial Policy”, American Economic Review 109(1): 48–85.
Eurofound (2017), “Employment Effects of Reduced Non-wage Labour Costs”, Luxembourg.
European Commission (2014), “A Study on R&D Tax Incentives”, Taxation Papers, Working Paper 52.
European Commission (2017), Industrial Policy Strategy: Investing in a Smart, Innovative and Sustainable Industry, Policy Report.
Fumagalli, C, M Motta and M Peitz (2020), “Which Role for State Aid and Merger Control During and After the Covid Crisis?”, Journal of European Competition Law & Practice, forthcoming.
Goolsbee, A (1998), “Does Government R&D Policy Mainly Benefit Scientists and Engineers?”, American Economic Review 88: 298–302.
Gordon, R J and H Sayed (2020), “Transatlantic Technologies: The Role of ICT on the Evolution of U.S. and European Productivity Growth”, International Productivity Monitor 38: 50-80.
Harrison, A and A Rodriguez-Clare (2010), “Trade, Foreign Investment, and Industrial Policy”, Handbook of Development Economics 5.
Lokshin, B and P Mohnen (2013), “Do R&D Tax Incentives Lead to Higher Wages for R&D Workers? Evidence from The Netherlands”, Research Policy 42(3): 823 – 830.
Manelici, I and S Pantea (2020), “Industrial Policy at Work: Evidence from Romania’s Income Tax Break for Workers in IT”, working paper.
Rodrik, D (2004), “Industrial Policy for the Twenty-first Century”, Technical Report, UNIDO Working Paper.
Van Ark, B, M O’Mahony and M P Timmer (2008), “The Productivity Gap between Europe and the United States: Trends and Causes”, Journal of Economic Perspectives 22(1): 25–44.