(2) The size of the R&D underinvestment gap depends on three variables: (a) the relevant social
cutoff rate, (b) the actual cutoff rate, and (c) the slope of the ranked distribution curve.
To answer the questions whether there is underinvestment in R&D or not, and, if so, how
much, variables b and c need to be estimated and a set. In the literature one can find ample discussion
on the appropriate definition of the social cutoff rate and its corresponding value (see for example
Zerbe and Devily [1994]). Here we have settled on using a social cutoff rate of 7% real (that is, net of
inflation) for developed countries and 12% real for developing countries. The more challenging task is
to estimate variables b and c.
In the real world, selection committees usually do not rank R&D projects on the basis of
ERRs, nor do they have a concrete idea about the cutoff rate that they implicitly apply. However, the
selection criteria actually used tend to underpin at least some economic rationale. Since the selection
is less than economically optimal, a number of R&D projects with ERRs less than the “optimal”
cutoff rate will be selected at the expense of R&D projects with an ERR to or above the “optimal”
cutoff rate. This results in an ex ante ranked distribution of R&D projects that takes the form of a bell-
shaped distribution, although lopsided to the left (figure 2).
■> Expected rate of return (ERR)
Figure 2: The optimal versus suboptimal selection of R&D projects