R&D Tax Incentive Review

Transcription

1 SUBMISSION TO THE R&D Tax Incentive Review FEBRUARY 2016 AUSTRALIAN ACADEMY OF TECHNOLOGY AND ENGINEERING

2 29 February 2016 Mr Bill Ferris AC Chair, Innovation & Science Australia Dr Alan Finkel AO FTSE Chief Scientist of Australia Mr John Fraser Secretary to the Treasury Dear Sirs, R&D Tax Incentive Review The Australian Academy of Technology and Engineering (ATSE) welcomes the opportunity to provide input into the R&D Tax Incentive Review. ATSE advocates for a future in which technological sciences, engineering and innovation contribute significantly to Australia s social, economic and environmental wellbeing. The Academy is empowered in its mission by some 800 Fellows drawn from industry, academia, research institutes and government, who represent the brightest and the best in technological sciences and engineering in Australia. The Academy provides robust, independent and trusted evidence-based advice on technological issues of national importance. ATSE fosters national and international collaboration and encourages technology transfer for economic, social and environmental benefit. ATSE contends that R&D Tax Incentive is an effective measure when used in the spirit that the program is intended and as such it should be continued. However, the Academy suggests that the program would benefit greatly if the changes described in the attached submission are implemented. The R&D Tax Incentive is the most significant Government program for promoting innovation in the private sector, and as such it should be carefully targeted to achieve maximum benefit for Australia. ATSE strongly supports the objectives of the Incentive and endorses the Review of the program. The Academy would be pleased to assist the Review in any way, and the expertise of the Academy and its Fellows remain available to the Review. Should you require any further assistance, the contact at ATSE is Dr Matt Wenham, Executive Manager, Policy & Projects (matt.wenham@atse.org.au or (03) ). Yours sincerely, Professor Peter Gray FTSE President

3 SUBMISSION TO THE R&D TAX INCENTIVE REVIEW The Australian Academy of Technology and Engineering (ATSE) contends that the program is an effective measure when used in the spirit that the program is intended and should be continued. However, the Academy suggests that the program would benefit greatly if the changes described below were to be implemented. General Comments Confidence in the stability of a program such as the R&D Tax Incentive is incredibly important for its success. The Australian Council of Learned Academies (ACOLA) Securing Australia s Future1 Project 9 examined programs and measures to promote research translation in 14 countries2. The report findings highlighted the necessity of a stable suite of well-funded measures for Australia. Although ATSE argues that changes should be made to ensure that the Tax Incentive is able to more effectively and efficiently achieve its intended results, it is important to maintain confidence in the program for those that are using the program appropriately. Australia s fiscal support for industrial R&D is by no means excessive compared to other countries (see Figure 1). Any changes to the program should be about encouraging the right sort of activities, not reducing program costs. Any money saved should be redirected to other forms of support (e.g. higher concessional rates for collaborative activities, the CRC Programme, direct incentives such as the Business Research and Innovation Initiative (BRII) pilot announced as part of the National Innovation and Science Agenda). It should be noted that the impact of the Incentive is quite variable between different sectors. For example, the Incentive may be critical for SME-dominated sectors such as biotechnology, but be less so in sectors that are dominated by larger companies. Definition of R&D activities The definition of research and development should be tightened to ensure that the Incentive targets additional R&D activities with the potential for spillovers. As noted in the Issues Paper, there is evidence that business-as-usual activities were being claimed under the Tax Concession (i.e. activities lacking additionality). Anecdotal evidence suggests that this is still occurring to some extent despite the new definitions of core and supporting R&D activities introduced when the Tax Incentive replaced the Tax Concession. The rise of consultants that specialise in assisting companies to claim the R&D Tax Incentive is indicative that many companies are using the Incentive to reduce their tax bill after they have made investment decisions that are likely to have occurred under business as usual conditions i.e. firms are using the scheme to offset R&D expenditure rather than generate new R&D. It is acknowledged that there are consultants that do provide useful assistance in developing tax deductable R&D programs for businesses. 1 ACOLA combines the strengths of the four Australian Learned Academies (Science, Technology & Engineering, Humanities, and Social Sciences) to provide a nexus for interdisciplinary cooperation and develop integrated problem solving and cutting edge thinking on key issues for the benefit of Australia. Funded by the Australian Research Council and conducted by ACOLA for the Australian Chief Scientist and the Commonwealth Science Council, Securing Australia s Future delivers evidence-based research and interdisciplinary findings to support policy development in areas of importance to Australia s future. 2 Bell, J, Dogdson M, Field L, Gough P, Rowley S & Spurling T (2015) Research translation for economic and social benefits: country comparisons. Australian Council of Learned Academies.

4 The disproportionate increase in registrations relative to total R&D expenditure (as shown in Figure 1 in the Issues Paper) is likely to be evidence that the program has not been successful in driving significant additional R&D expenditure. ATSE contends that the R&D Tax Incentive is a highly effective measure in some circumstances. The Incentive is likely to be a significant enabler of R&D in SMEs and the refundable nature of the incentive can provide a critical funding stream for R&D focussed start-ups. For others, particularly larger companies (who claim the significant majority of expenditure), the difficulty in determining whether the Incentive drives any additional R&D, let alone whether there are significant spillovers makes the merit of the Incentive more difficult to determine. The existence of the Incentive may encourage multinational companies to do their R&D in Australia, but anecdotal evidence from ATSE Fellows based in industry suggests that the Incentive is unlikely to be the most significant driver of this decision. Ideally the Incentive should not apply to R&D that would have occurred anyway due to market forces. In theory the Incentive should drive more basic research as it is seen as optional, and is financially riskier, more likely to be additional, and has greater potential for spillover effects and societal benefit. Akcigit, Hanley & Serrano-Velarde (2013)3 argue that companies generally have sufficient incentives, if competition is high enough in the markets where they operate, to invest in applied research, i.e. research that mainly leads to further development of existing technology. This means that there is less market failure for applied research, while there is an under-investment (relative to what would be socially optimal) in basic research. In any review of the definition of eligible research, consideration needs to be given as to how this imbalance could be addressed. Reviewing international models (such as Canada s Scientific Research and Experimental Development Tax Incentive Program) may assist in identifying effective definitions. 0 Figure 1 - Business R&D Intensity and government support to business R&D, 2011 (as a percentage of GDP) 3 Akcigit, U., D. Hanley & N. Serrano-Velarde (2013). Back to Basics: Basic Research Spillovers, Innovation Policy and Growth, NBER Working Paper

5 Rates and thresholds ATSE supports the current approach of providing competitive and refundable rates to SMEs. These refunds frequently provide critical cash flow for small and start-up businesses in early stages of growth. ATSE supports the provision of competitive rates targeting activities that are likely to provide greater additionality or spillover benefits, as is common practice in other countries. Areas of potential focus include: o Cross-Sector Collaborative R&D (see below). o R&D that is likely to encourage growth especially employment growth and improved international competitiveness. o R&D aligned with Government priorities, such as the National Science and Research Priorities and Industry Growth Centres. However, any move in this direction would require careful definition to avoid unnecessary administrative burdens, and it may be more practical and effective to fund direct government programs such as BRII that reflect Government priorities. Driving Cross-Sector Collaboration R&D that involves collaboration between industry and publicly funded research organisations (PFROs) is recognised as being more likely to have significant spillover effects than R&D undertaken internally by firms. These channels are essential for allowing the benefits of R&D to flow to society as a whole rather than being contained within business. Australia s current rate of collaboration between the research and industry sectors is the lowest in the OECD. To drive greater collaboration, ATSE argues that the R&D Tax Incentive should be used to provide incentives to companies to engage in collaborative research expenditure. This reflects a key finding from the ACOLA Securing Australia s Future Project 9 study of overseas research translation schemes: o Australia s business R&D tax incentive could be adjusted to encourage collaboration with public sector researchers: A number of the countries reviewed are using R&D tax incentives to encourage collaboration with public sector research institutions. Countries that have adopted this approach have higher rates of business collaboration with public sector research institutions. Examples of such countries include Denmark and Chile. This suggests that a more favourable incentive for such collaboration is an effective incentive for business. Australia s R&D tax incentive could be adjusted to provide companies with a greater benefit for collaborative work with public sector researchers.4 Such collaborative arrangements would also be more transparent as they require an external financial transaction rather than an internal allocation of costs. ATSE is not suggesting that collaborative research should be a requirement for eligibility for the scheme, however the funding necessary to support a higher rate or threshold for collaborative R&D could be obtained from savings made by tightening the definition of eligible R&D. Threshold ATSE contends that the major focus of the R&D Tax Incentive should be to promote research, development and innovation in the small and medium enterprises that make up the majority of the Australian economy, especially those SMEs with high growth potential. To this end, the use of thresholds and differential rates for larger companies is seen as an appropriate structure for the Incentive. 4 Bell, J, Dogdson M, Field L, Gough P, Rowley S & Spurling T (2015) Research translation for economic and social benefits: country comparisons. Australian Council of Learned Academies. 3

6 Although it is recognised that the R&D Tax Incentive is a factor in large multinational corporations deciding to conduct R&D in Australia, anecdotal evidence suggests that it is only one of several factors. If the Incentive is to be made available above certain thresholds, ATSE suggests that applicants be required to provide further evidence of additionality or spillover benefits to be eligible. Administration and Compliance ATSE sees potential benefit in a pre-registration process. This will ensure that the Incentive plays a greater role in businesses decision making processes. Pre-registration should require that businesses demonstrate appropriate planning of their R&D programs. Although moving to a single agency delivery model for the Incentive offers potential benefits in reducing administrative burdens, it is important that specialist expertise in assessing R&D eligibility be retained. The Australian Tax Office will inevitably need to be involved in the delivery of the Incentive. A possible solution would be to relocate the specialist R&D assessment functions from Innovation Australia/AusIndustry into the ATO, although this may present different administrative and mission complexities. It should be noted that direct support incentives can more easily be provided through a single agency. Direct Support for R&D ACOLA s SAF Project 9 found o Australia can make greater use of direct support measures for business innovation to increase research translation: Firms that undertake R&D are more likely to become involved in the translation of public sector research. The project has found that Australia is overly reliant on indirect support for business R&D through the R&D Tax Incentive. Shifting the balance of government support for business innovation to greater use of direct measures such as grants, loans and procurement contracts would allow a more focused and targeted approach to support for research collaboration and translation.5 Further evidence of Australia s imbalance between direct and indirect support is provided in the literature review (attached below). This review was commissioned by the Academy to assist in preparing this submission and is made available to the Review to help provide a background on overseas evidence on supporting R&D. It should also be noted that the OECD has produced a large amount of literature on comparisons between countries support for business enterprise R&D (BERD). Several of these reports are cited in the ACOLA SAF Project 9 report, and the preceding Project 4 report, The role of science, research and technology in lifting Australian productivity.6 ATSE strongly supports the Government s Business Research and Innovation Initiative program and has long-recommended that the Government establish a program similar to the United States SBIR or United Kingdom SBRI programs. These overseas programs have been repeatedly demonstrated to work elsewhere, providing that they are stably and properly funded. Further evidence for this can be found in the ACOLA SAF Project 9 report. The Government should also consider establishing income contingent loans for SMEs conducting R&D, similar to the Higher Education Loan Program for university students. These loans would provide up front grants that would be repaid if and when a company becomes profitable. 5 Bell, J, Dogdson M, Field L, Gough P, Rowley S & Spurling T (2015) Research translation for economic and social benefits: country comparisons. Australian Council of Learned Academies. 6 Bell, J, Frater B, Butterfield L, Cunningham S, Dogdson M, Fox K, Spurling T & Webster E (2014) The role of science, research and technology in lifting Australian productivity. Australian Council of Learned Academies. 4

7 Funding for direct support (such as the CRC Programme, the proposed BRII program or future loan or voucher programs) should be sourced from savings produced by tightening the definition of eligible R&D, or by reducing the threshold. In other words, savings made by changes in the Incentive should be applied to other forms of support for R&D and not to reducing the overall amount of government spending on R&D. This is critical to achieving the aims set out in the NISA and the broader objectives of fostering a knowledgebased economy. 5

8 Attachment to Submission Background literature review R&D incentives Professor Goran Roos FTSE Tax incentives can be given in the form of expanded deductibility against corporate tax for certain, predefined, costs related to R&D activities (so-called tax deduction). Another form of tax incentives are tax credits which means that the company can, for a charge, postpone taxation until a later date. R&D expenses may also be taxed favourably by the tax code allowing an increased depreciation factor compared with other (long-term) assets (Bystedt & Fredrikson, 2015). As shown in Figure 1, several OECD countries have relatively extensive direct tax incentives for R&D. [The size of the "bubbles" indicates the absolute size (in purchasing power parity adjusted US dollars) in tax incentives for R&D. Black bubble indicates that the country has no tax incentives for R&D. For example, R&D intensity of the United States 1.89 percent, the total fiscal support 0.32 percent of GDP and the tax incentive is USD 8300, compared with South Korea, the corresponding figures are 3.09; 0.38 and USD Sweden is among the countries that have no tax incentives (black bubble).] (Bystedt & Fredrikson, 2015). Figure 1: Fiscal support (direct and indirect) to industrial R&D in 24 OECD countries (incl. China) and business R&D intensity in 2011, percent (OECD Science, Technology and Industry Scoreboard 2013) As can be seen from Figure 1 several countries have relatively extensive direct tax incentives for R&D. (A more detailed review of various OECD countries' system of tax incentives can be found in OECD 2011a, Chapter 2.). The OECD has developed a so called B-Index for a comparable way to measure the generosity of the tax incentives for R&D available in the member countries (Warda, 2002). B-index expresses the level of the profit before tax that a representative company must generate to go plusminus zero on an extra "R&D dollar" when taking into account any special fiscal conditions for R&D expenditure. The better fiscal terms are for R&D expenditure, the lower corporate profit requirements of an additional R&D expenditure. Countries with a higher value on B-index generally have a lower 6

9 rate of tax subsidies for R&D expenditure; 1-B index expresses thus tax subsidy rate for R&D expenditure. However, it should be noted that estimating the subsidy rate does not capture all aspects of how the tax incentives are designed or to what extent the deduction or credit facilities are actually used by businesses. Macroeconomic studies using econometric methods suggest that tax incentives have a positive effect on reported corporate R&D expenses (see, for example, Bloom, Griffith and Van Reenen, 2002 or summaries of studies by Konjunkturinsitutet, 2013). According to a comprehensive review of the empirical studies carried out by three European forecasting and research institutes (CPB, CASE, ETLA & IHS, 2015), the tax incentives in general have a positive effect on reported corporate R&D spending, but the studies differ in terms of the size of the effect. The studies that attempt to identify more precisely the effects show an elasticity with respect to a change in the generosity of tax incentives which is less than 1. This means that a country that wants to increase business R&D intensity in relation to the country GDP by 1 percent (on average) must spend at least 1 percent of GDP to make tax incentives more generous. Most studies come to a positive impact in the short term, but the impact appears to last - or even increase over longer the longer term. (Bloom et al. (2002) conclude that the long-term (price) elasticity of a change in the tax that hits R&D spending (as measured by the so-called user cost of capital) is close to 1 on average, but varies between industries). Westmore (2013) finds that a reduction of the B-index (i.e. more generous tax incentives) of 0.05 percentage points would lead to the R&D stock in the long term increasing by just under 6 percent for the average OECD country. The effect of Tax incentives on innovation are far less understood than the impact on R&D. According to the studies summarized by CPB, CASE, ETLA & IHS (2015), it appears that the positive effect on R&D of favourable fiscal conditions also apply to innovations, mainly incremental ones. However, it seems that tax incentives have a much greater effect on R&D spending than they have on patent intensity. (On the margin, according Wetmore s result, an increase in the B-Index with one standard deviation results in an increase in R&D by 5.5 percent, compared with an increase in patent intensity by 2.4 percent.) This suggests that the potential problems of reclassification of other expenses to R&D costs should not be neglected (ICS Ltd. (2010; Roos, 2012). In other words, the positive effects on corporate R&D spending could be solely an issue of reclassification of other expenses or that companies for tax reasons, choose to report their R&D costs in countries with generous fiscal conditions. This means that macro economically reported increases in R&D may in effect mean no increase in R&D activities on the micro-economic level (something the author have observed throughout many countries and firms) It is worth noting that the actual effectiveness of policies like R&D tax credits is very low what looks like a statistically reported increase in R&D spending (which explains why all econometric and research papers using reported R&D statistics end up arguing for its positive effects) is frequently just a reclassification of other expenses into R&D expenses and not an increase in the actual R&D executed (which becomes obvious when in-depth interviews are executed in firms) (Roos & O Connor, 2015, p.16) Though tax incentives appear to have a positive effect on both companies' R&D activities and innovation, the design of incentives plays a major role in how effective they are. A balance needs to be struck, for example, should tax incentives be given for all R&D related costs (given a certain definition of what costs should be the basis for a deduction or a credit), a so-called volume-based approach, or just for additional, R&D costs, a so-called incremental approach. The volume-based approach is usually easier to administer, but increases the risk of the companies being granted tax credit for research that would have happened anyway. The incremental approach reduces the deadweight costs, but is rather more complicated - and therefore more expensive - to administer (Bystedt & Fredrikson, 2015). Another aspect to consider when it comes to the impact of tax incentives for R&D is that the regulations must be stable over time. Westmore (2013) find that the positive effect of tax incentives for R&D is less - or even disappear - in countries that have had many changes in the regulatory system. Furthermore, the positive effects measured in econometric studies are partial effects, i.e., they do not take into account whether resources used for tax incentives could generate a greater effect on 7

10 innovation if they have been used in any other way (Jaumotte & Pain, 2005). Neither do the econometric estimates consider the societal costs of raising taxes or cutting spending in other areas in order to finance the tax loss that tax incentives generate. Too generous tax incentives may also lead to companies choosing to locate their R&D expenditure to countries solely - or at least to a large extent due to the tax treatment, rather than that the fundamental factors for R&D and innovation are better there. In some cases, the existence of tax incentives leads to a relocation of R&D from home to the country that has the more favourable tax environment for R&D expenditure. This can in turn lead to the total stock of R&D or innovation (patents portfolio) will be lower than it had been on fiscal conditions had been more equal between countries (Bystedt & Fredrikson, 2015). Direct (public) research and development incentives aims to achieve further R&D by compensating for companies only investing in innovative activities whilst these investments are economically viable but where there are projects that are societally economically viable that will not be achieved without the support since they may not be economically viable from the firm s point of view. (Examples of direct R&D support can be found in OECD (2011b, Ch. 2) and Cunningham et al. (2013)). The stimulating effect of these incentives and support can usually be called additionality (or complementarity). It can either refer to resources (inputs) or results (output). With resource-additionality we mean that the public support encourages companies to increase resource input in R&D in a way that would not have happened without this support. With result-additionality we mean that companies will increase the results - innovations - more than they had done in the absence of the incentive or support. Most evaluations of direct R&D support refer to resource-additionality (Bystedt & Fredrikson, 2015). Although R&D support can encourage businesses to engage in more R&D, it does not necessarily mean more innovation. It is likely that companies choose to, in the absence of support, fund R&D projects that have a higher probability of success in terms of commercialized innovations, but uses subsidies to fund riskier (and therefore expected unprofitable) projects. To ensure that the support results in more innovation (higher result-additionality) places high demands on the ability to target any incentives and support towards those projects that are not economically viable for the company but that are societally desirable. Akcigit, Hanley & Serrano-Velarde (2013) argues that companies generally have sufficient incentives, if competition is high enough in the markets where they operate, to invest in applied research, i.e. research that mainly leads to further development of existing technology. This means that there is hardly any (major) market failure for applied research, while on the other hand there is an under-investment (relative to what would be socially optimal) in basic research (Akcigit et al. (2013) reports data from French companies showing that also companies engage in what is classified as basic research. 11 percent of total business R&D expenditures consist of basic research and companies account for 15 percent of all basic research carried out in France.). Subsidising corporate R&D costs in a way that does not differ between applied and basic research could therefore lead to an over-investment in applied R&D. There are also reasons why direct R&D support can lead to private R&D resources being substituted rather than increased. A precondition for direct support generating additionality in private R&D spending, is that the supply of input factors, not least qualified workers, can respond to the increased demand for these factors as a result of the support. If supply of these input factors is sufficiently inelastic for the increased demand to result in increased cost for these input factors, which in turn increases the cost not only for the prospective new R&D, but also increases the costs of the existing input factors (also for those companies who may not receive any support), resulting in R&D support, if it is sufficiently large, substituting or crowding out existing R&D in companies. Crowding out can also occur through direct actions. For example, competition or product market regulations could be designed to benefit the larger R&D intensive companies resulting in these companies demanding more R&D labour hence increasing the cost for this labour and making it more expensive for small businesses to conduct R&D. (Bystedt & Fredrikson, 2015). The possibility that direct R&D support may "leak" in the form of higher wages for R&D personnel was highlighted in a study by Goolsbee (1998). Since then, several other studies have attempted to study the effects of R&D support on the wages of R&D personnel. Aerts (2008) reviews these studies and finds that although it is difficult to draw firm conclusions from these studies they indicate that in each 8

11 case there is some crowding out through increased wages. The fact that a less efficient matching of supply and demand for skilled labour can result in the effects of R&D support being reduced by the existence of "leaks" in the form of higher salaries for R&D personnel, makes it necessary to consider the labour market effects in both implementation and evaluations of direct R&D support. Another source of crowding out is if direct R&D support measures are designed selectively since selective support might give those companies (and industries) which can benefit from it, competitive advantages. The expansion of R&D activities in the eligible companies can then take place at the expense of non-eligible companies, which may even reduce their R&D. Cunningham, Gök & Laredo (2013) report the results of 43 studies that have looked at the effects on resource-additionality and 25 studies looking at the impact on result-additionality. They found, as stated above, that studies using macro data (regions or countries) generally find larger effects on additionality than those using micro date (company data or even lower level). In addition to the level of aggregation, industry- and company-specific factors play a major role in the observed effect of direct R&D support. Cunningham et al. finds it difficult to draw general conclusions from the studies of resource-additionality, which is in line with previous meta-studies. David, Hall & Toole (2000) review a large number of studies of the effects of direct R&D support on various levels (national, industry and firm). They reported that one-third of the studies show that direct R&D support crowds out private R&D spending. Garcia-Quevedo (2004) reports that about half of the studies in his meta-study find that public R&D support is a complement to the companies' own R&D resources (i.e. resource-additionality exists), while about a quarter of the studies demonstrated statistically insignificant effects, and the remaining quarter find that government R&D support crowds out the companies' own inputs. Correa, Andres & Borja-Vega (2013) applies so-called meta-study analysis to 37 studies of the relationship between direct R&D support and business R&D. Their conclusion is that there is support for direct R&D support leading to more innovation in enterprises. They note, however, that the included studies differ considerably in terms of estimated precision, choice of identification strategy and other relevant characteristics, which makes it difficult to discuss causal effects. Heshmati & Lööf (2005) investigated whether companies that receive public support for innovation in Sweden have a higher level of investment in R&D than other comparable companies. They also studied whether the public support has affected the firm s investment in R&D. They found that R&D support is primarily provided to R&D-intensive companies and to the extent that this support has any effect it is limited to small firms. It is also difficult to show clear causal relationships relating to results additionality. According to Cunningham et al (2013), a conclusion is that direct support in itself does not generate result additionality, but requires that companies have the capacity (in the broad sense) to convert the support into results. It is much more difficult than in studies of resource additionality, to ascertain the role industry and company characteristic play in making any support effective. Tillväxtanalys (2014) evaluated the short-term effects on employment, demand for highly skilled labour, labour productivity and turnover/growth of two Vinnova programs targeted at young research-oriented companies (VINN NU) and innovation-driven SMEs (Forska & Väx). They found no statistically significant effects on any of the outcome variables except for turnover. The impact from the two studied programs resulted on average in an increase in turnover of the supported enterprises by about 14 percent compared with the twin companies that were not provided any support. This positive effect was strongest for the smallest businesses with 1-5 employees and then declined with firm size. For companies with more than 20 employees, the impact on sales was not significant. The provided support showed no effect on turnover growth regardless of company size. A problem particular to studies of result additionality is that the performance measure that is often used e.g. sales, production, employment, exports and value added - is an insufficient and inadequate measure of innovation. As a side note, public support for R&D and innovation may sometimes have other goals than to lead to more innovation. If this is the case, these chosen variables may well be relevant. A particular problem is to evaluate the effectiveness of the support if multiple objectives exist and if it is unclear in which priority order these objectives are to be achieved. If the 9

12 purpose of direct R&D support is to stimulate R&D in those activities which are economically unviable on the firm level, but that has high societal benefit, measures like value added, production, etc., are less appropriate (Bystedt & Fredrikson, 2015). There are however, some studies looking at the effects on patent intensity or the introduction of new products to the market. Bérubé & Mohnen (2009) found that Canadian companies who, in addition to getting tax incentives also received direct R&D support, introduced more new products to the market and were more successful in commercializing their innovations. Schneider (2008) found in a study of patenting in Danish companies that patents that have been generated through being provided public support are of a higher social value than the patents that have not been supported. (Subsequent citations are used as a measure of social value. The study does not explicit take into account if the successful patents would have been developed even without public support). According to Westmore (2013) (a study not included in the compilation of previous studies mentioned above), direct R&D support to companies has on average a long-term positive effect on the R&D stock of industry for those OECD countries included in the study. This positive effect also seems to have increased over time. Some explanations for this may be that governments have become more careful with taxpayers' money, combined with tighter state aid rules, primarily within the EU, and often require matching contributions from the companies that receive support. This means that any support provided has a better opportunity of actually funding new, additional R&D. Another explanation may be that the level of direct R&D support has decreased over time, which may also have led to a more efficient allocation of the support provided. Similar to the evaluation effect of tax incentives, there are a number of methodological obstacles that any evaluator must overcome. OECD (2011a, Ch. 5) provides an overview of the methodological challenges to be addressed during impact evaluation of direct support. The programs also differ very significantly in terms of purpose and design, making it difficult to draw general conclusions. A particular problem in the evaluation of direct R&D support is that these are often decided by committees or similar, i.e. they involve some form of discretion in deciding which companies should receive support. According to David et al. (2000), there may be drivers that make the decision-making committees more likely to provide support to companies operating in industries with good technological opportunities. This results, as shown in empirical studies, in a relationship between R&D support and technological development, but this effect is not causal. Furthermore, the return on R&D is expected to be high in these companies, due to the good technological opportunities, even in the absence of public support, which increases the likelihood that the public support funds R&D that would have been executed without public support i.e. a crowding out effect. Einiö (2014) reports the effects on R&D expenditure, sales, employment and productivity, of Finnish support to innovative companies. This public support, which is administered by the Finnish Innovation Agency (Tekes), aims to help companies that have not yet commercialized their products. In accordance with EU state aid rules, this public support should endeavour to only subsidize such research and development that would not have taken place without the support. To identify the causal effect of R&D support, Einiö exploits the fact that the regulatory framework for the EU's regional development funds enable the support for certain pre-specified geographic areas to be potentially higher than the support for other areas. Since the division into geographical support regions is based on population density, it can be assumed that the higher amount of support (in areas with very low population density) is unrelated to factors affecting the firm s research productivity and hence the firm s likelihood of getting public support. The results indicate statistically and economically significant positive impact of R&D support (what is measured is the change in R&D expenses before and after the support which again raises the previously discussed issue around if increased reporting of R&D actually entails more R&D done). The effects persist and increase over time. Einiö also finds positive effects on both employment and sales in the short term, while the positive effects on productivity will take some years to show up. Although it is not without difficulty to draw general conclusions from these results - whether the quantitative effects or how the support should be designed - they confirm that properly designed support for firm R&D can have significant positive effects on both R&D resources in various forms of R&D outcomes (The study explicitly test for the existence of crowding out of corporate R&D resources and in all cases rejected this hypothesis). The study also highlights, 10

13 like David et al. (2000), that there is a significant risk that R&D support is granted to firms operating in industries that have a high level of technological opportunities. It is therefore difficult to know whether it really is the support that helps to stimulate firm R&D or if it is the increased return on R&D investments originating in the technological opportunities that are present. In a ground-breaking study Azoulay et al. (2015) finds that public funding of biomedical research, through the National Institutes of Health in the United States, generates additional patents in private firms. They find that approximately every other approved application leads to a new (additional) patent. The advantage of their study is both that they can take into account patent effects in several different research areas, not only those that happen to receive the funding, and that they exploit variations in the rules, to address the problem of research support having a higher likelihood of being granted to research areas with a proven potential to succeed. In conclusion, the evaluations that have been made of direct R&D support show that, under certain conditions, direct R&D support can lead to increased R&D expenses "net" in companies. Direct support can, if designed correctly, to some extent also give rise to more and better innovations and higher employment. The methodological problems that arise in the evaluation of R&D support are many and hence, the results must be interpreted with caution. It is also worth bearing in mind that most of the evaluations are measuring the impact in the short term and capture the effects of those companies that already have R&D when the support is granted (the so-called intensive margin ). In the longer term, the presence of R&D support can also encourage companies to start engaging in R&D activities (the extensive margin ). If this change in behaviour is also included, the long-term (positive) effects of direct R&D support - and also tax incentives - may be larger than those shown by studies that only measure the effects along the intensive margin. Conversely, support designed to primarily benefit existing companies will rather result in less R&D because it is more profitable for the existing companies to hire more R&D staff, which drives up the R&D personnel wages also for those companies considering starting investing in R&D. In the very long term, however, the higher salaries would attract more people to gain a qualification making them more attractive for R&D work, making the long-term labour supply for R&D work more elastic than it is in the short term. (Bystedt & Fredrikson, 2015). As indicated above there is no general, positive correlation between the extent of tax incentives and R&D intensity. Extensive fiscal incentives, especially in the form of tax incentives, and high R&D intensity exist in South Korea, the US and, to some extent, Japan. Sweden, along with Denmark, Finland, Israel, Switzerland, and to some extent Germany, have relatively weak fiscal incentives, but exhibit high R&D intensity. This suggests that the causality between fiscal incentives and R&D intensity also can go the other way: extensive fiscal incentives could be an indication of deficiencies in other parts of the innovation system or weak framework conditions that make the return on R&D lower, which governments of these countries choose to compensate for by offering generous subsidies and/or tax incentives, rather than improving the general conditions (for example, because it would challenge powerful interests). (Bystedt & Fredrikson, 2015). The absence of a (simple) connection between fiscal incentives and R&D does not necessarily mean that direct R&D support or tax incentives do not fulfil a role in countries' efforts to increase business R&D intensity and innovation. The fiscal incentives in general seem to result in the desired effect (again with the warning around macroeconomic data causality not necessarily indicating a microeconomic causality). However, several studies discussed show that the cost-effectiveness of these measures is questionable. To achieve a significant increase - so that the effects will be visible in the macro data - of corporate R&D or innovation requires fairly extensive income reductions or spending increases. These must then be weighed against other pressing purposes that require public funding. (Bystedt & Fredrikson, 2015). However, if policy makers consider it justified to put a certain part of the budget in fiscal R&D incentives will then tax incentives or direct support provide the most value for money? Unfortunately, there are few studies of the relative effectiveness of the two types of fiscal incentives (where the variation in the design of each measure can be great). Westmore s (2013) results at the country level suggests that direct support has a greater impact on R&D than tax incentives (as measured by B- 11

14 Index) while a Norwegian study on corporate data by Hægeland & Moen (2007) finds that tax credits provide greater impact than direct support. IAs can be seen, even in this respect, therefore, the empirical support for the design of fiscal incentives is not particularly robust. References Aerts, K. (2008). Who Writes the Pay Slip? Do R&D Subsidies merely Increase Researcher Wages?, K.U. Leuven, Dept. of Managerial Economics, Strategy and Innovation. Oktober Akcigit, U., D. Hanley & N. Serrano-Velarde (2013). Back to Basics: Basic Research Spillovers, Innovation Policy and Growth, NBER Working Paper Azoulay, P, J.S. Graff Zivin, D. Li & B. N. Sampat, (2015). Public R&D Investments and Private Sector Patenting: Evidence from NIH Funding Rules, National Bureau of Economic Research Working paper 20889, National Bureau of Economic Research, Cambridge, Ma. Bérubé, C. & P. Mohnen (2009). Are Firms that Receive R&D Subsidies More Innovative?, Canadian Journal of Economics, vol. 42(1), pp Bloom, N., R. Griffiths & J.V. Reenen (2002). Do R&D Tax Credits Work? Evidence from a Panel of Countries , Journal of Public Economics, vol. 85, pp Bystedt, F. & A. Fredrikson (2015). Forskning, innovationer och ekonomisk tillväxt, Bilaga 8 till Långtidsutredningen 2015, SOU 2015:107. Stockholm. Correa, P., L. Andrés & C. Borja-Vega (2013). The Impact of Government Support on Firm R&D Investments: a Meta-Analysis, Policy Research Working Paper WSP6532, Washington DC, World Bank. CPB, CASE, ETLA & IHS (2015). A Study on R&D Tax Incentives: Final Report, DG TAXUD Taxation Paper 52, Europeiska kommissionen. Cunningham, P., A. Gök & P. Laredo (2013). The Impact of Direct Support to R&D and Innovation in Firms, Nesta Working Paper 13/03. David, P.A., B.H. Hall & A.A. Toole (2000). Is Public R&D a Complement or Substitute for Private R&D? A Review of the Econometric Evidence, Research Policy, vol. 29(4-5), pp Einiö, E. (2014). R&D Subsidies and Company Performance: Evidence from Geographic Variation in Government Funding Based on the ERDF Population-Density Rule, Review of Economics and Statistics, vol. 96(4), pp Garcia-Quevedo, J. (2004). Do Public Subsidies Complement Business R&D? A Meta-Analysis of the Econometric Evidence, Kyklos, vol. 57(1), pp Goolsbee, A. (1998). Does Government R&D Policy mainly Benefit Scientists and Engineers?, American Economic Review, American Economic Association, vol. 88(2), pp Hægeland, T. & J. Moen (2007). Additionality of R&D Subsidies: A Comparison between Tax Credits and Direct Grants, Statistics Norway Report 2007/45. Heshmati, A. & H. Lööf (2005). The Impact of Public Funds on Private R&D Investment: New Evidence from a Firm Level Innovation Study, MTT Discussion Papers, vol. 3, pp ICS Ltd. (2010) Review of the Innovation Readiness of SMEs. A Short Study Undertaken for the Danish Agency for Science, Technology and Innovation, Copenhagen, Denmark. Jaumotte, F. & Pain (2005). Innovation in the Business Sector, OECD Economics Department Working Papers, No. 459, OECD Publishing. 12

15 Konjunkturinsitutet (2013). Tillväxt- och sysselsättningseffekter av infrastrukturinvesteringar, FoU och utbildning En litteraturöversikt, Specialstudier nr. 37, dec OECD (2011a). Tax Incentives for Business R&D, OECD Science, Technology and Industry Scoreboard 2011, OECD Publishing. OECD (2011b). Business Innovation Policies: Selected Country Comparisons, OECD Publishing. OECD (2013). OECD Science, Technology and Industry Scoreboard 2013, OECD Publishing. Roos, G. & O Connor, A. (2015). The idea of integrating innovation: Entrepreneurship and a systems perspective. Chapter 1 in Roos, G. & O Connor, A. (eds.). (2015). Integrating Innovation: South Australian Entrepreneurship Systems and Strategies. University of Adelaide Press. ISBN Adelaide. pp Roos, G. (2012). Manufacturing into the Future. Adelaide Thinker in Residence , Adelaide Thinkers in Residence, Government of South Australia, Adelaide, Australia. Schneider, C. (2008). Mixed R&D Incentives: the Effect of R&D Subsidies on Patented Inventions, Working Papers , Department of Economics, Copenhagen Business School. Tillväxtanalys (2014). Företagsstöd till innovativa små och medelstora företag: en kontrafaktisk effektutvärdering, PM 2014:16. Warda, J. (2002). Measuring the Value of R&D Tax Treatment in OECD Countries, STI Review 27, pp Westmore, B. (2013). R&D, Patenting and Productivity: The Role of Public Policy, OECD Economics Department Working Papers, No. 1047, OECD, Paris. 13