Keywords

1 Introduction

As a part of the Europe 2020 strategy, fostering the conditions for innovation plays an important role among priorities of the policy-making in Europe. What concerns the tax rules conditions for companies as a factor influencing enterprise innovation process in Europe, a heterogeneity among the European Union (EU) member states persists. The existence of 28 tax systems means that enterprises need to adapt to a country-specific tax conditions when making all kind of decisions, including the innovation strategy decisions. In such circumstances, the development of two phenomena is specifically not desirable: (1) companies innovation strategies can become limited by tax system borders, and (2) multinational companies are motivated to waste their innovation capacity for tax planning strategies (so-called tax innovation) instead of using it for innovation activities in the areas with the potential of growing productivity and efficiency (e.g. the core-business activities). To avoid the occurrence of such phenomena, the projects in order to some forms of standardisation or harmonisation in this area are highly welcomed (see Uramová et al. 2016). The directive proposal of Common and Consolidated Company Tax Base (CCCTB) of the European Commission (EC) from 2011 was presented as an initiative with two principal objectives: to make the corporate tax framework in Europe to be more simple, and to reduce opportunities for multinational companies to avoid income tax payments. The main idea was that companies operating within the EU would have to comply with only one system for computing its taxable income, rather than different sets of rules in each member state in which they operate. It is important to underline that according to this project, each member state will keep its right to apply its own corporate tax rate. However, this proposal hasn’t met a necessary political support within the European Council yet.

As the priority is to harmonize the national corporate income tax base and establish a Common Company Tax Base (CCTB), the current debate of types of common rules which would best foster the innovation activity of companies is necessary. Our paper tries to contribute to this debate by identifying the potential causalities between tax rules and innovation activities in the EU member states (especially the business innovation activities) at three levels. Firstly, we focus on the links between innovation activity and nominal tax rates in order to confirm the relevance of the approach “tax base harmonisation only” for the innovation process in the EU. Afterwards, we directly concentrate on the links between innovation activity and tax base rules in the EU member states. Finally, we look to the potential causalities between country-specific tax incentives driving the innovation process in companies and the enterprise innovation performance in these countries. Our ambition in such testing was to identify the role of these incentives for innovation process in Europe from the perspective of an eventual impact of the CCTB proposal.

This research paper originated in partial fulfilment, and with support of, the project ITMS 26110230082 Mobility—Support of Science, Research and Education at Matej Bel University in Banská Bystrica (Mobility—podpora vedy, výskumu a vzdelávania na UMB) under the Operational Program Education co-financed by the European Social Fund within the bounds of financial subsidy contract No. 018/2012/1.2/OPV.

The paper has been also supported by the Scientific Grant Agency of the Ministry of Education of the Slovak Republic and Slovak Academy of Science (VEGA) under the contract No. 1/1009/16 Innovation potential of the regions of Slovakia, its measurement and innovation policy at the regional level.

2 Literature Review

Mulgan and Albury (2003) define innovation as the successful implementation of a new or significantly improved product, service, marketing strategy or new organization method that will bring substantial improvement to the economy, efficiency or quality of the outputs. According to this definition of innovation activity, firm’s investment in innovation comprises the R&D investments as well as the non-R&D investments, and both of these aspects should be taken into consideration. Concerning the R&D innovation only, Zemplinerová and Hromádková (2012) argue that private companies invest in this innovation less than would be socially desirable.

The literature is proposing several external factors which can drive entreprise innovation processes focusing mainly on enterprise R&D activities. In this context, the public financial support as a factor promoting R&D in business sector is usually tested. As reported by Hunady et al. (2014) the public financial support for R&D and innovation is one of the most important factors affecting the firm’s innovation activities. The authors also found many other determinants of innovation such as market competition, type of the industry as well as export focus of the firm. Based on the date from OECD countries, Falk (2005) found that there are two important political instruments supporting R&D in firms: special tax treatment for companies that invest in R&D and direct financial support.

What concerns the tax incentives efficiency, the evidence in literature is ambiguous. Based on the data from Canadian firms, Czarnitzki et al. (2011) conclude that R&D tax credit increase the R&D engagement at firm level. Similarly, Cappelen et al. (2012) found that use of tax credit often lead to successful developed of new production processes and products in the case of Norwegian firms. On the other hand, Tassey (2007) stated that R&D tax credits when applied in US, were ineffective. He proposes the changes that should be made to increase its effectiveness. He argues that a flat rate applied to all R&D is the most effective way to promote R&D. In the context of the European Union, Ientile and Mairesse (2009) also conclude that the impact of R&D tax credits on R&D investment is quite heterogeneous, likely sensitive to the country analysed and methodology used. The authors identified that while the R&D tax incentives appear to be efficient in Norway and France, evaluations for Spain and The Netherlands provide less convincing results.

Several characteristics of the existing literature appear to be relevant for our research: (1) the authors concentrate on R&D expenditures and the impact of tax conditions on innovation (e.g. the non-R&D innovation) is missing, (2) most of the studies focus on R&D tax credits and other forms of tax incentives (enhanced allowances and accelerated depreciation) are not analysed, (3) there is no one ‘perfect’ way how to assess the effectiveness of tax conditions and the results depend on data and methods applied. When choosing methodology of our research, we were trying to reflect this characteristics.

3 Data and Methodology

In order to reach the potential causalities between chosen aspects of the tax system and the innovation activities in the EU member states, diverse data sources were used (Table 1). Most of the data were retrieved from the following EC publications: European Innovation Scoreboards (2007–2009) and Innovation Union Scoreboards (2010, 2011, 2013–2015), Taxation trends in European Union (European Union 2014b), Tax reform in EU Member states 2015 (European Union 2015) and EC Study on R&D Tax Incentives (European Union 2014a).

Table 1 Description and data sources of variables used in the analysis
Full size table

We obtained panel data for the first six variables. In the case of SII, STR, EATR, TB and Firm’s R&D expenditures, we used the data for 28 EU countries in the period of 2007–2014. Thus, we gained 196 observations, but this number has been slightly decreased by the application of first difference in the models. Due to the several missing observations, the number of observation for non-R&D innovation expenditure was lower and included 162 observations.

As our approach took into consideration the potential impact of tax base composition, we needed to choose a quantitative indicator to capture this phenomena. For this purpose, we used the effective corporate tax rates (the third variable) which implicitly contains the effect of the tax base composition as well as the effect of a statutory tax rate level. Furthermore, we also calculated the difference between effective and statutory tax rate (the fourth variable) in order to approximate only the potential effect of tax base (without a rate dimension).

Concerning the tax incentive score (the seventh variable), data for a certain period (year 2014) were available for 26 EU member states. Data for Germany and Estonia are missing due to the fact that these two countries haven’t implemented a specific tax incentives to facilitate enterprise R&D activity in their tax systems.

Different types of analysis have been conducted in this dataset in order to test assumed correlations or causal relationships: correlation analysis, panel Granger causality tests and panel fixed-effect regression analysis. In the first two parts of our analysis, we applied the panel data analysis to search for potential dependencies between indicators of innovation performance and tax system specificities (corporate tax rates and corporate tax bases). In this case, we were able to capture the dynamic aspect as well as to test the lagged dependencies between variables. All the variables have been tested for the stationarity with various panel stationarity tests. Most of the tests indicated that all variables appeared to be non-stationary at level, but stationary at their first difference. In accordance with these results, we decided to use differenced data in order to avoid the potential problem of spurious regression, which seemed to be very high.

In the third part of our analysis, the correlation analysis based on the cross-section data were used. As this part of our analysis focused on examination of link between R&D tax incentives and innovation activity in EU countries, we put under the question the assumed correlation between R&D tax incentives (by country and by tax incentive type) and firm’s R&D and non-R&D innovation expenditures.

4 Results

In our analysis structure, three different approaches could be identified. Firstly, we were trying to focus on the relationships between the enterprise innovation activity (both R&D and non-R&D expenditures) on one side and the corporate statutory tax rates on the other side. To identify an eventual existence of innovation transfer between companies and other groups of economic subjects (like the spillovers effects of large companies), we proceeded to enlarge our analysis by taking the Summary innovation index into account. Secondly, we were trying to test the potential causality between the existing tax base rules (represented both by the effective tax rate and by the numerical difference between the statutory and effective tax rate) and the innovation activity (enterprise R&D expenditures, enterprise non-R&D expenditures and overall SII index) in the EU Member states. Finally, the links between chosen features of tax incentives and enterprise innovation activity as well as between tax incentives’ ranking and the innovation activity were tested.

As a first step of our analysis, we tested the Granger causalities between selected pairs of variables. The results of the tests are summarized in Table 2. In vast majority of cases, no significant Granger causalities between the observed variables can be identified. However, it seems that there is a significant Granger causality arising from statutory corporate tax rates to summary innovation index. This could represent a kind of causality in Granger sense between the level of corporate tax rates and the innovation performance of the whole economy. Surprisingly, no analogical significant evidence for statutory tax rates and enterprise innovation activity represented by R&D and non-R&D expenditures can be identified. Although it seems that level of corporate tax rate can have a positive impact on innovation activity in a specific country, there is no evidence that this impact passes through the innovation activity of the all companies sector.

Table 2 Results of Granger causality tests
Full size table

What concerns the effective corporate tax rates and the difference between statutory and effective tax rate (approximations of tax base), the Granger causality between these variables and firm’s R&D and non-R&D innovation expenditures appear to be insignificant. So it seems that different rules of tax base composition in the EU member states don’t influence the innovation activity in these countries, at least for the analysed period.

As a next step of our analysis, we decided to explore potential causalities using simple panel fixed-effects regression models. To keep it simple, each model contained one dependent and one independent variable. All variables have been used at their first differences, thus the number of observation have been redacted by one period for each country. Furthermore, the White diagonal robust standard errors have been applied in all the models. We alternated the cross-section and period fixed-effects in each model. The outcomes of the first models are shown in Table 3. In this case, the Summary innovation index is used as a dependent variable.

Table 3 Results of panel regressions with Summary innovation index as dependent variable
Full size table

In most cases, the outcomes of regression analysis are in line with the results of Granger causality tests. On one hand, there is no evident relationship between tax rates and innovation index, when using the variables from the same period. However, the negative effect of tax rates becomes significant at 10% level, once we lag the tax rates variables by one period. Moreover, the impact of statutory tax rates in period fixed-effect model seems to be significant even at 1% level of significance. Hence, there is some evidence that higher nominal corporate tax rates can have a negative effect on overall innovation performance of the country.

Furthermore, we continued in proceeding analogical regression analysis, but with the Firm’s R&D expenditure and non-R&D innovation expenditure as a dependent variable. The outcomes of the models are summarized in the Tables 4 and 5, respectively.

Table 4 Result of panel regressions with firm’s R&D expenditures as dependent variable
Full size table
Table 5 Result of panel regressions with firm’s non-R&D innovation expenditures as dependent variable
Full size table

Based on the results, we can say that there is no significant relationship between the firm’s R&D expenditures and effective or statutory tax rates. The same is true for the firm’s non-R&D innovation expenditure (Table 5). While performing 16 fixed-effect regressions with different specifications, we failed to find any statistically significant causality.

To sum it up, we can say that probably, there is an impact of corporate tax rate on innovation performance of the country as whole. However, this effect is delayed by at least 1 year. On the other hand, any comparable causality was not found in the case of firm’s R&D expenditures and non-R&D innovation activities.

In the context of innovation fostering, the existence of various tax incentives supporting the R&D activities in almost all EU member states can eventually represent an efficient channel. To test this assumption, we decided to study the impact of R&D tax incentives on enterprise innovation activity. As described in details by European Union (2014a), different types of R&D tax incentives as well as other tax rules and tax administrative features (eventually beneficial for the tax payer innovation activities) are applied by EU member states. From this point of view, Belgium and the United Kingdom are the leading member states with relatively more suitable tax rules for R&D and innovation. On the other hand, the tax system of Germany and Estonia do not use any specific initiative to focus on innovation activity. What concerns the form of the most widely used tax incentive, the tax credit for R&D expenditures are the most represented—this instrument which is not affecting the tax base rather decreasing the corporate tax rate, is applied in sixteen EU member states.

To find an evidence concerning eventual efficiency of different tax incentives, we decided to proceed the correlation analysis between selected features of R&D tax incentives and firm’s R&D and non-R&D innovation expenditures. Firstly, we calculated standard Pearson correlation coefficient for all selected variables in the sample and we found a positive, but weak correlation between most of the R&D tax incentives and firm’s R&D expenditure (Table 6). The total number of R&D tax incentives, calculated as the sum of tax incentives used in certain country, correlates positively with firm’s R&D expenditure, but this correlation is rather weak. The same evidence is true for tax credits and accelerated depreciation. On the other hand, there is rather significant negative correlation between total number of R&D tax incentives and firm’s non R&D innovation expenditure. Moreover, all tax incentives are negatively correlated with non R&D innovation expenditures.

Table 6 Pearson correlation coefficients for selected variable (cross-sections)
Full size table

According to these findings, the firms in the tax environment with more R&D tax incentives prefer to spend more on R&D. But this readiness to invest in R&D seems to have a negative impact on other forms of innovation activities (represented by non-R&D innovation). This could indicate that tax incentives could have more effect on the structure of innovation activities (the share between R&D and non-R&D innovation expenditures), rather than on the total volume of R&D and non-R&D innovation expenditure.

One can argue that the method we applied is not appropriate for the analysis of discrete binary variables, which are mostly used in the sample. Thus, we also decided to apply the tetra choric correlation, suitable only for binary variables. Therefore, the continuous variables had to be recoded to binary ones. The average value of each variable has been used as the threshold between zero and one. The results we obtained are to some extent similar to those concerning the Pearson correlation coefficients. However, the negative correlation between non-R&D innovation expenditure and tax credit, accelerated depreciation as well as patent box appears to be even stronger. This observation is especially true for the form of accelerated depreciation, where the correlation seems to be very strong, as indicated in Table 7.

Table 7 Tetrachoric correlations for binary variables (cross-sectional data)
Full size table

Since different types of tax incentives ensure different conditions for enterprise innovation activity, we also decided to apply the results of ranking of tax incentives in respect to R&D activities in European countries, calculated by EC Study on R&D tax incentives (European Union 2014a). The latter study takes into account three categories of features of the R&D tax incentives: (1) scope of the policy, including the type of R&D tax incentive and costs covered, (2) targeting of specific groups of firms, according to their size, age, region, etc. (3) organization, including administrative practices and evaluation (European Union 2014a, p. 73). According to this ranking, Denmark and Ireland seem to have the most suitable R&D tax incentives among the EU member states. On the other hand, the results of this ranking indicate that the least appropriate R&D tax incentive can be found in Malta, Cyprus and Greece.

The results of the correlation analysis between the tax incentives scores and other selected variables are presented in Table 8. These results are in compliance with the previous results gained by testing different forms of tax incentives independently. They indicate that there is a positive correlation between a country’s R&D tax incentives score and enterprise R&D expenditures. In addition, the positive correlation between better-scored country’s tax incentives and higher values of country’s summary innovation index can be found. In accordance to our previous results, the correlation between R&D tax incentives score and non-R&D innovation expenditure is again negative in our sample of the 28 EU member states.

Table 8 Pearson correlation coefficients (cross-sectional data)
Full size table

5 Discussion and Conclusions

The objective of our study was to find an evidence about the relationship between chosen features of national tax system related to tax base composition and the innovation activity in EU member states with a special emphasis on enterprise innovations. The empirical testing of such relationship is interesting by itself. However, as our research tries to contribute to the renewal debate concerning the CCTB proposal, we proceed in interpretation of our findings from this point of view.

The first characteristics of the CCTB directive proposal is related to the fact that harmonisation of tax base rules doesn’t need any harmonisation of corporate tax rates in the EU. As we found no evidence of the impact of statutory tax rate on enterprise innovation activity, we can support this approach of not bringing such politically difficult topic into consideration. However, one should notice, that a possibility of positive impact of the level of corporate statutory tax rate on overall innovation activity in a specific country may exist thanks to effects of spill overs between a certain groups of economic subjects (e.g. large companies) and other groups of economic subjects (e.g. innovation activity in the public sector). The probability of an existence of such effects seem to increase with the identification of the link between the statutory tax rates and Summary Innovation Index.

Further, an EC initiative towards harmonisation of tax base rules should have neither positive nor negative impact on enterprise innovation activity as the latter seems to be unaffected by the composition of tax base rules. At least in the analysed period, the approximation of tax base composition by two measurable variables—the effective tax rate and the difference between statutory and effective tax rates—seems not to be able to explain the differences in enterprise or overall innovation activity in EU member states. Eventually, other variables representing the tax base differences can be taken into account for further research, but in this case a firm-level data approach should be appropriate.

Although it seems that the enterprise innovation activity is not influenced by the tax base as a whole, our results indicate it might be affected by a certain part of tax base related to corresponding tax incentives effect. The use of R&D tax incentives to wider extent in some EU member state seems to lead to higher R&D innovation activity in companies in this country, as well as to higher overall innovation activity. On the other hand, our results indicate that the choice of a specific tax incentive might influence the structure of innovation schemes. For instance, while R&D tax incentives stimulate the enterprise R&D activity, they affect negatively the non-R&D activity in the companies.

From the perspective of the CCTB proposal, the effects of tax incentives having impact on tax base (enhanced allowances and accelerated depreciation) can be compared to those having impact on tax rate (tax credits and patent boxes). There is an evidence towards the preferable use of base-affecting tax incentives in the form of enhanced allowances which seem to be the only tool to affect positively both the R&D and non-R&D business activity. On the other hand, if the form of accelerated depreciation is applied in the new CCTB proposal in order to stimulate enterprise innovation activity in European companies, this can produce a strong negative effect in companies’ non-R&D activity. Thus, our results can lead to suggestion to implement the best practices of R&D tax incentives of the EU member states considered as having the best scores in tax incentives ranking (especially Denmark) into a new CCTB proposal. However, the results we obtained should be treated with attention because only a static approach was applied in this part of our research for the reason of a limited access to data about development of tax incentives in the EU member states. More detailed data in this field would lead to adoption of more appropriate methods (like panel data regression analysis) in empirical research of tax incentives efficiency. From this point of view, the further research in this area is needed.