Public Finance, 2nd Edition

previous section. Also, the social marginal utilities may change with discrete

changes in individual gains and losses. In truth, a fully speciWed cost±beneWt

analysis is a hopeless task when the distribution is nonoptimal.

7

Policy ana-

lysts are forced to make many simplifying assumptions, yet they cannot really

know whether their assumptions are entirely appropriate or reasonable.

CONCLUSION

No matter how one views the issue, a nonoptimal distribution of income

poses grave diYculties for objective cost±beneWt analysis. As indicated in the

very Wrst chapter of the text, normative theory has never been able to resolve

the distribution question beyond the point of saying to the decision maker:

``Tell us your distributional preferences and we will tell you what to do.'' To

the extent society is unable to develop a consistent set of distributional

preferences, cost±beneWt analysis is Wrmly trapped in a dilemma. On the

one hand, there is no obvious simple and practical way of incorporating

distributional preferences into the analysis. On the other hand, there is no

way of judging the value of an analysis which chooses to ignore distributional

considerations. Indeed, if the distribution is nonoptimal, it may not even be

possible to compute the eYciency implications for any given project because

of various aggregation biases. Estimated market demand curves might yield

highly inaccurate compensation measures of aggregate gains and losses.

Finally, even if society's distributional preferences were well established,

the data required to incorporate them consistently into the shadow prices

for government projects are formidable, to say the least.

There is one apparent avenue for escaping the dilemma. The analyst can

adopt the Jorgenson±Slesnick framework for estimating social welfare and

social expenditure functions that we presented in Chapter 4. It is not a clean

escape, however, because this framework avoids aggregation problems by

assuming as a maintained hypothesis the so-called exact aggregation restric-

tions, which are restrictions on the parameters of the indirect utility func-

tions. These are in addition to the many other restrictions taken from

neoclassical consumer theory that are required for the estimation of the

underlying system of expenditure share equations. Unfortunately, these re-

strictions are very likely to fail standard statistical tests in any application,

which calls into question the reliability of their estimated indirect utility

functions. There is no sure escape from the dilemma.

7

Peter Hammond provides an excellent discussion of the diYculties of applied cost±beneWt

analysis that expands upon the discussion in the text in P. Hammond, ``Theoretical Progress in

Public Economics: A Provocative Assessment,'' Oxford Economic Papers, January 1990, sections

8±13. He includes some troubling econometric issues that we have ignored.

816 CONCLUSION