Policy 1 Policy 2 Policy 3
A 1
> 2
> 3
B 2
> 3
> 1
C 3
> 1
> 2
Suppose now we do a "pairwise comparison" of policies
1 and 2--which will be preferred by majority voting? Clearly, Aaron
prefers 1 to 2, as does Cathy; hence, Policy 1 will be selected over Policy
2. But, suppose the comparison were between Policy 2 and Policy 3--here
Policy 2 has a majority preferring it to Policy 3. Thus, Policy 1
is preferred to Policy 2 and Policy 2 is preferred to Policy 3. So,
Policy 1 should be "best," right? Well, let's now compare Policy
1 with Policy 3 in a pairwise comparison. Aaron likes Policy 1 better
than Policy 3, but both Bob and Cathy prefer Policy 3 to Policy
1!! Social preferences over policies can be "intransitive" in this
way (transitivity is a property of the real number system, but also of
rational preferences--discuss). We have a real problem here: whoever
controls the "agenda-setting process" can control the outcome in settings
like this. While one might suspect that such cases are rare, one
must realize that "all the easy stuff" has already been done (everything
that we all agree on--we should have a national defense, a set of laws,
etc. has been done, leaving things about which there is more disagreement).
The agenda-setter, in short, can exert far more control on the outcome
than we would really like in a voting democracy under some circumstances.
A related, but more general, problem: inability
to make interpersonal utility comparisons means that we can never
know with any degree of confidence that a policy is "making society better
off." Arrow Impossibility Theorem. Discuss the role of income
transfers from the rich to the poor, assuming first identical preferences
(marginal utilities) for income; then, when the rich (who may be rich precisely
because they have such high preferences for the goods income can buy) have
a higher marginal utility of income than the poor. The beauty of
the market is that (in the absence of market imperfections) exchanges
are "Pareto Efficient" (Discuss--one or more individuals are made better
off, while nobody is made worse off). But, even this is not entirely
beyond criticism (equity may be viewed by many as more important than efficiency,
regardless of how efficient a policy is). Moreover, all the social
policies that have B>C for everyone have already been done!
This leads to the concept of "Kaldor Efficiency" (Discuss--those
who are made better off could compensate those who are made worse
off--i.e. B>C in dollars). The Kaldor efficiency notion is what underlies
benefit-cost analysis: if no group is systematically discriminated
against, and, if we do large numbers of policies, always doing things
with benefits greater than or equal to costs in dollars (despite
the non-comparability of their meaning among individuals) society as a
whole is made better off on average.