Public Finance, 2nd Edition

constant returns to scale. Even the government is assumed to transact at the

competitive producer prices to the extent it buys and sells inputs and outputs.

A second-best model might posit constraints in the form of noncompetitive

behavior in a small subset of markets, but the underlying market economy is

almost always competitive. Second-best results may not have competitive

interpretations, but the majority of models used to date have been competi-

tive through and through.

The newer literature on private information is somewhat of an exception

because it applies the techniques of game theory, and the games being played

may not occur in a market setting. Even so, the decentralized nature of the

competitive market has a correspondence in the public sector allocation

mechanisms that honor the revelation principle. A standard requirement of

truth-revealing mechanisms is that individuals have no control over their

opportunity sets. The public sector mechanisms must be decentralizable in

this sense.

In summary, although second-best theory has severely challenged all

Wrst-best policy rules, it has taken only the smallest, most hesitant steps

away from the highly stylized Wrst-best policy environment. Second-best

analysis is more realistic, but only slightly so.

PREVIEW OF PART III

With these reXections in mind, we will begin our second-best analysis with the

allocational theory of taxation, thereby reversing the order of presentation in

Part II. This happens to coincide with the historical development of second-

best theory, but that is really beside the point. There are two good analytical

reasons for considering tax theory Wrst.

One is that second-best tax theory is inherently simpler than second-best

expenditure theory, in this sense. Public expenditure theory requires the

speciWcation of a distinct problem (e.g., an externality) and one or more

distinct constraints (e.g., distorting taxation), whereas tax theory requires

only the speciWcation of a constraint. Saying that all taxes must be distorting

is at once an additional constraint on the system and the source of the

problem being analyzed in tax theory. Consequently, problems in tax theory

can be analyzed with much simpler general equilibrium models. This is an

important advantage. Second-best models speciWed in terms of prices are

quite diVerent from the Wrst-best quantity model of Part II, enough so that it

pays to begin the analysis as simply as possible. Thus, the initial chapters on

tax theory have two goals. Their main purpose is to demonstrate some

important theorems in the allocational theory of taxation, but they also

serve as an introduction to second-best methodology.

Second-best tax theory also logically precedes public expenditure theory,

so long as distorting taxes are one of the policy constraints. Having studied

12. INTRODUCTION TO SECOND-BEST ANALYSIS 397