The problem of how to stimulate innovation is high on the agenda of many public policy-makers. It is generally understood that the key to long-term growth is improving productivity: high productivity is what makes national economies globally competitive and what enables firms to pay high wages. And although it is always possible to raise measured productivity by downsizing and making everyone else work a little harder, this kind of policy is unlikely to result in sustained growth. In the end, productivity will only increase if firms develop new products and processes – if they innovate.

To work out what kind of policy might be appropriate, we need to know what the problem is. Discussions of corporate innovative activity usually identify three types of risk or uncertainty that are likely to deter firms: technological uncertainty, competitive uncertainty and market uncertainty. These roughly speaking correspond to the following questions: ‘is it technically possible to make this new product or process?’, ‘will I be the first on the market?’ and ‘does a market actually exist?’

All of these risks are real – and all are hard to insure against, not least because moral hazard makes it difficult to provide insurance that does not undermine the insured party’s incentives to perform well. As a consequence, anyone contemplating the introduction of a new innovation must also be willing to shoulder the brunt of these risks.

No one knows for sure which of these sources of risk is most onerous. And it seems likely that different types of technology and different market settings create different types of risks. Nevertheless, many believe that the third type of risk – market uncertainty – is the most important source of uncertainty faced by firms.

The simple fact is that many of the new products and processes that firms introduce do not push back the frontiers of technological knowledge very far. Innovation for most firms is an engineering problem: implementing insights that have been gleaned from more fundamental or basic research. Hence technical risks are typically slight.

Similarly, although there are situations where a first mover can completely shut out all followers (or at least put them permanently at a disadvantage), this usually only happens in certain very special types of circumstances – for example, when a standard needs to be established or when an advance can be completely protected by a patent. While innovating firms can face very real difficulties in appropriating anything like the full benefits their new products or processes create, competition is more likely to stimulate than to inhibit innovation for most firms.

Market uncertainty, however, is always a problem. Innovative activities require firms to make what are often substantial investments. These are sunk costs: they are not recoverable if the innovation fails. Such fixed costs mean that the market for a particular innovation must reach at least some minimum size if the innovation is to pay. And users must want the innovation enough to enable the firm to set margins that enable it to recover its costs fairly quickly.

For smart firms, this often means that innovative activities are conducted jointly with users. (Indeed, in some sectors, they are actually user-led.) Among other things,

this enables the innovating firm to produce something that is genuinely valued by enough of its customers to pay. In fact, for some firms, the first step in any innovation project is to find one or several smart buyers or users to work with.

Government policies on innovation generally have to satisfy two criteria: they need to be targeted at the real problem; and they need to be ‘do-able’ by a set of civil servants whose administrative or management skills may be limited. In practice, this second criterion is often the more important. Consequently, many government policies designed to stimulate innovative activity take the form of R&D subsidies, tax cuts on certain types of expenditure, or the diffusion of information on best practice in one form or another.

Although these policies are do-able, it is not always clear just what it is that they do. Most studies of R&D subsidies suggest that they are used to finance second rate projects or that they actually displace private funding that might otherwise have been forthcoming. Hence the rate of return to such spending is typically very low.

A recent study of mine and some colleagues (CEPR Discussion Paper No. 1432) developed a large-scale dynamic econometric model of UK corporate patenting and innovation activity. We used it to simulate the likely effects of £500 million worth of public money being spent on either R&D subsidies, demand stimulation or corporate tax breaks. All three policy initiatives had only extremely modest effects on innovative activity – and in no case did £500 million spent in this way appear to generate anything like £500 million worth of increased innovation or patent production.

Are there any other types of public policies that might be used to stimulate innovation? To answer this question, it is worth returning to the three forms of uncertainty that are likely to be behind the problems innovating firms must overcome to succeed.

To deal with technological uncertainty, knowledge about basic science and technology must diffuse, firms must have access to specialized expertise that they would not ordinarily maintain in-house, and an appropriately skilled workforce must be available. The right kinds of public policy needed to solve these problems include investment in basic research, training and human capital formation, and flexible approaches to the problem of intellectual property rights.

To deal with competitive uncertainty, firms must have enough protection from imitators to be able to recover their outlays. But probably even more importantly, firms must face stiff competitive challenges in their markets. This means that an active and vigorous anti-trust policy is likely to help stimulate competitiveness.

Finally, governments themselves are active purchasers of new technology. Indeed, it is often the case that government procurement contracts are the initial market for a new technology. This, of course, means that government purchasing can be a potent policy instrument for stimulating innovation. If governments act like smart buyers and manage their suppliers carefully, then they ought to be able to have a substantive effect on the development of new products and processes. After all, this is exactly what smart private sector buyers do.

Paul A Geroski, Professor of Economics at London Business School and Research Fellow in CEPR’s Industrial Organization programme