they are, shy away from risky projects and discount for statistical variance of the ERR when ranking
R&D projects. Therefore, the risk-averse version of the ranked distribution of R&D projects can be
thought of as positioned lower on the ERR scale than the risk-neutral version. This creates a
divergence between the ex ante and ex post rate-of-return distributions, as the latter will more or less
coincide with the risk-neutral version.
Arrow and Lindner (1970, 1972) argue that in a typical public-investment situation,
governments can safely ignore risk as long as the investment is small relative to national income.
Given that this is true for most public agricultural R&D investments, risk aversion should not play
much of a role in the selection of public agricultural R&D projects or programs (Anderson 1991). In
other words, public agricultural R&D projects or programs with the same ERR, but with one being
riskier than the other (reflected by a higher statistical variance), should be treated the same. The
chance of a lower outcome is compensated by the chance of a higher outcome. Despite this theoretical
argument, public-research administrators most likely act moderately risk-averse, so that demands for
short-term accountability can be answered by at least some positive results (Greig 1981).
Risk and uncertainty are not static and may decline over time. R&D proposals that are
initially turned down as being too risky may be selected at a later stage when critical variables can be
predicted more accurately. For example, experience in a certain research field may increase the
confidence in research effectiveness over time.
The six underlying factors presented here are not necessarily exhaustive. Other factors may
play a role as well. Moreover, the relative importance of each of the six factors differs across research
fields. Market structure, for example, does not play much of a role when considering public
(agricultural) R&D. For other fields of research, however, this may constitute a highly relevant factor
that affects the ERR of R&D projects and hence shape up the available R&D opportunities.
Understanding which factors are the most critical is important when considering policies that could
shift the portfolio of possible R&D projects higher up on the ERR scale.
Table 1 summarizes some of the government policies that could affect each of the six factors
positively. Several of these policies are far broader than just R&D policy. These policies condition the
extent to which R&D can contribute to the overall economy. In developing countries in particular, the
10