--- Eric Crampton <[EMAIL PROTECTED]> wrote:
> The important point is that there's a disjoint between the willingness to
> pay and the actual payment:

Since the relevant comparitive system is majority voting, there is a disjoint
in yes-no voting as well.  The disjoint is even greater, since with demand
revelation, each person pays the average cost, whereas with voting, the cost
is paid from arbitrary taxation.  We need to fold in the revenue aspect.
If the majority voting is tied to a head tax, then there is still a disjoint,
because given cost C/N, the a person votes yes if his value is > C/N, but
this does not reflect the intensity of his desire.

> faces a 1/million chance of being decisive

But it can be questionned whether the probability is known.
In the pure case, we don't know at all what values other will state.
If probabilities are known, how do we know them?  The method of knowing has
to be included.  If it is known by fiat, this injects the outcome in advance.

> saying "I like poor children so much, I just voted $1 million in favour of
> program X that will help the poor", the voter will "reveal a
> preference" of one million dollars for program X.  Of course, risk
> aversion will mitigate this somewhat, but if inefficient programs give
> larger expressive benefits, then instituting the Tideman-Tullock procedure
> will lead to worse outcomes.

This presumes (1) the ballots are not secret or else that they are secret and
people express the truth, (2) probabilities of paying the social cost (Clarke
tax) are known, (3) the utility derived from expression and wining does not
count.

Fred Foldvary

=====
[EMAIL PROTECTED]

Reply via email to