The process of innovation is rarely swift and problem-free. Progressing from an innovative idea in the course of a research and development (R&D) programme through the stages of prototype, production, marketing and ultimately a successful sales strategy is fraught with difficulty, even when the invention is truly novel and potentially a boon to the world. Investors may have to wait a long time before any return on their R&D project materialises, with eventual market success highly uncertain. Competitors may quickly find out about good and novel ideas, cutting out the possibility of any future monopoly profits that might otherwise be available. Many innovative firms drop out of the market altogether before being able to market their product, not because they don’t have a potentially profitable idea, but because they do not have sufficient funds to commercialise their invention on a timely and scalable basis. Recognising these difficulties, and the value of R&D to the economy, governments provide a substantial amount of financial support (often through the tax system) to private R&D and innovation.
Today, more than half of the UK Government’s support to private sector’s R&D efforts flows via tax incentive schemes. The UK is not alone in the large weight that it places on the tax system for promoting innovation: France, Canada, China and Korea are only a few examples of countries with tax incentives constituting more than 50% of the government funds for business R&D. That being said, tax incentives are not a necessity for a country that takes pride in nurturing and attracting innovative companies. Countries like Germany and Finland maintain high levels of business R&D spending (as a share of GDP), and they achieve this without major tax breaks for R&D.
From the policymaker’s perspective, tax incentives for R&D have a few key advantages over the alternative of direct subsidies. This is because tax incentives are usually much easier to administer. In order to decide on the recipients of direct subsidies, a government body has to gather a set of experts who judge the potential value of each project, and then it has to keep periodically monitoring the recipients for compliance with the specific rules of each subsidy policy. Tax incentives on the other hand allow firms to choose the most profitable projects, and as agents that know the market better than policymakers, those firms should be better placed to pick the projects with better prospects. The downside is that the government ends up subsidising many projects that would have gone ahead irrespective of the government support.
In my recent publication with Li Liu, entitled “Effectiveness of Fiscal Incentives for R&D: Quasi-experimental Evidence”  we set out to answer this additionality question: holding all else constant, how would two groups of very similar firms compare in terms of their R&D spending if the generosity of the scheme were increased for only one group – but not for the other? The findings are that, on average, medium-sized companies with the more generous R&D tax treatment increased their spending by around 33 percent relative to the group without the additional tax subsidy. The reform had induced a reduction in the user cost of capital by around 21 percent, so the response is sizeable (implying an elasticity of -1.6). Another finding is that the Exchequer recovers more than its nominal foregone tax revenue in terms of tax revenues arising from the newly-generated R&D under the policy. In other words, the policy is cost-effective.
Based on these results it is tempting to jump to the conclusion that R&D tax credits work.
But such a conclusion needs to be qualified. As with most topics in public policy, the relative effectiveness of R&D tax incentives is, for a number of reasons, a nuanced issue. First of all, our study establishes a causal link between the R&D policy and its positive effects for medium-sized companies. The results may differ for larger companies. Second, we find some evidence that younger firms respond more strongly than their longer-established peers, possibly suggesting that the policy helps the R&D efforts of firms with financial constraints, such as those in start-up mode. This point requires additional investigation, (and is a topic I am currently working on at the Centre for Business Taxation). Third, policymakers should be interested in further dimensions of heterogeneity across firms, since a positive effect for the average firm might not necessarily be what the policymakers should be targeting. Companies like Apple, Google or Tesla are not average by any measure. They also did not start out as average in their start-up years. Finally, my comments here focus on front-end tax breaks for R&D investments, i.e. those that reduce the cost of an investment. The case for (or against!) back-end regimes, such as patent box policies, which reduce the tax applied to the return of successful projects, will be a subject for a future blog post.
Current research shows that tax incentives for R&D have been effective in stimulating innovative activity. There is clearly much more work to be done to disentangle the mechanisms through which tax breaks for R&D operate and to identify what are the worthwhile “investments” the tax system should be undertaking. Are tax incentives, with the help of a cash refund, relieving a financing constraint? Or is a cash refund a waste of precious government resources? Given the scale of the financial commitments and the significance or R&D and innovation to the economy, these are important questions. Research on these issues is an important workstream within the Centre for Business Tax and we will report back with further findings in due course.
OECD R&D Tax Incentive Database, http://oe.cd/rdtax, November 2018.
Guceri and Liu, 2019. American Economic Journal: Economic Policy, Vol. 11 No. 1, February 2019. pp. 266-91
There are other studies which find similar average effects for larger companies in the United States For example: Rao, 2016. “Do Tax Credits Stimulate R&D Spending? The Effect of the R&D Tax Credit in its First Decade.” Journal of Public Economics, 140: 1-12.