As a simple benchmark a proportional (wage) income tax may be used. Raising the tax to
finance incremental government outlays reduces the after-tax wage rate and hence induces
households to replace consumption of market goods by leisure. The associated reduction in the
income tax base, in turn, generates a negative effect on tax revenue (essentially a “fiscal
externality”), which should be counted as part of the overall cost of government spending.
Augmenting the direct resource cost of (marginal) government spending by the distortionary
cost of taxation produces the “marginal cost of public funds” (or MPCF), which may be written
as:
MPCF =1/(1-(t/(1-t)*e))Where e is the (uncompensated) wage elasticity, t is the marginal effective tax rate on wage
income (including income and indirect taxes as well as the contribution from means tested
government transfers) and it is assumed that a proportional income tax is used.
If t=50 pct. and e=0.1, then MCPF=1.11. I.e., the total benefits of the marginal government
projects should exceed its direct resource costs by at least 11 pct.
Using an endogenous growth model empirical evidence suggests that if the government invests
in productive categories such as for example R&D, education, health etc. it enhances growth but
not if it is financed by distortionary taxes.^2
The overall conclusion therefore is that there is a good case for the government to subside R&D
- as long as the benefits exceed the marginal costs. The problem, however, is that there is
extreme uncertainty about what the exact net benefits may be. The empirical literature in the
field is comprehensive but unfortunately the estimates are very vulnerable and insecure. There
are a lot of methodological problems and collecting proper data can also be a problem,
especially when comparing different countries over time (panel data).
Going through the literature the overall picture is that the gain from private investments is
bigger than for public investments. Some studies even find that the effects from public
investments in R&D are negative or insignificant.^3
A conclusion like that needs some qualification. Taken at face value it means that publicly
performed R&D crowds out resources that alternatively could be used by the private sector,
hereby private R&D. However, regression analyses do not capture all aspects of the relationship.
For instance, while business-performed R&D is likely to be more directly targeted towards
innovation and implementation of new innovative processes in production, other forms of R&D
(e.g. health, energy, and university research) may not raise technology levels significantly in the
short run. They may, however, generate basic knowledge with possible “technology spillovers”.
These spillovers are not to be neglected as the business sector has the opportunity to take an
advantage of the research done in the government sector.
Comparing different studies can be difficult because the estimated effect depends on the
econometric method used, variables controlled for and time series looked at. Moreover, some
studies use growth in GDP as dependent variable while others use TFP or labour productivity.
(^2) Kneller et al. (1998).
(^3) See for instance Bassanini and Scarpetta (2001).