The Thorny Problem of Tax Breaks for R&D

Innovation is a complex process and I’m always sceptical about simple solutions to encourage companies to be more innovative. Recently I have written about some evidence against the widely-held assumption that creating industry clusters will make firms in these clusters more innovative. It turns out the the relationship between industry clusters and innovation depends a lot more on the type of industry and where you are located. In other words, the belief that clusters = innovation breaks down when you look at why successful industry clusters form and what distinguishes them from unsuccessful clusters.

Another simplistic assumption is that tax refunds for R&D investments will make firms do more R&D and therefore be more innovative. At face value this seems like a reasonable assumption but someone with a more cynical mind might say that this will just result in business trying to classify more activities as R&D, and this is where a simple idea gets very tricky to implement. Many countries offer tax incentives for R&D investments but classifiying what R&D is compared to other types of business improvement is a serious difficulty.

In a provocative report for the Australian Business Foundation, Dr. Nicholas Gruen highlights the problems associated with legislators and courts deciding on definitions for innovation and R&D. These definitions are a problem and Gruen discusses the merits of a simpler definition of R&D. Under the new Australian legislation the focus is on R&D as experimental work that is different from production activities. As a case in point, his report uses the example of a mining business that is using new structural supports to support tunneling. Although there are a range of experiments to test and develop the supports, they do not count as tax deductable R&D because the tunneling would have happened anyway. On the other hand, if the experiments were conducted in a disused mine then they would be eligible for the tax concession.

Gruen points out that this definition of R&D as scientific experiment has a significant bias against process innovations – an extremely important form of innovation (think of the Toyota production methods).

…if production is part of an R&D project, one can argue that there is no purely logical reason for clawing back revenue from production… For process innovation is as important as product innovation, and process innovation must often be done on ‘live’ production which therefore produces the output that can be sold.

I’d extend that argument to two more categories of innovation in services and business models. In advanced economies, services consitute the majority of business activity and yet services experiments are almost impossible to disentangle from business activities (although maybe we can think about service experiments in Second Life). A 2005 report from the Australian Bureau of Statistics also found that nearly 1 in 4 businesses that reported innovation also engaged in a combination of product, service and process innovation.

This makes sense to me because it implies the importance of building a business model around an innovation to make it successful. In a recent report, the Boston Consulting Group has recorded the significance of business model innovation in highly performing firms. As we have said before on this blog, intellectual property – business model = 0 and yet the R&D tax concesssion gives some sort of sacred status to scientific product innovation and pretends that it can be isolated from other forms of innovation that are equally risky but can also create significant value for the business and create beneficial spillover effects in the economy.

I know there are firms that do undertake risky technical projects to create new products and these firms benefit from the tax concession. However, as Gruen points out, reductions in the Australian company tax rate since the 1980s has caused the effective value of the rebate to fall from 24.5 to 7.5 percent. While this reduction has made the scheme affordabe for the government, there is very little evidence that this rate of support does anything to stimulate R&D investments. Most of the R&D that is done and claimed under the tax rebate would have happened anyway.

So there we have it…. R&D tax concessions seem like a good idea at first glance but deeper examination struggles to find a rational case for having them. Should we scrap the program or is there another alternative?

Gruen’s report strongly favours an amendment which would allow loss-making firms to claim a refund rather than writing the R&D tax expense off on tax. Small firms are very dependent on cash flow and the Australian Bureau of Statistics report also shows that these small startups are more likely to engage in ‘new to the world’ innovation.

Changing the rules to help these businesses would be a sensible idea but it still leaves the problem of an R&D definition that is embedded in simple linear models of innovation where scientific discovery gets commercialised to a new product. Real world innovation processes in the 21st century are rarely that simple.