Re: Median Voter vs. The Sub-optimal Equilibria
On Thu, 5 Sep 2002, Fred Foldvary wrote: First of all, the outcome is not necessarily known in advance. The outcome could be to not obtain the good. We don't know whether one will have to pay the $100. It doesn't matter whether he knows he has to pay or knows he doesn't or knows he has to pay with probability x. All that matters is that he views the payment as exogenous. As soon as it's exogenous, expressive preferences rule. Even when the outcome has a high probability, one does not necessarily get expressive benefit from agreeing with it. Suppose the issue is to spend an extra $10 billion in enforcing the drug laws. Those who disagree with this goal will express a value of zero. Certainly true. People can get expressive benefits from opposing a proposition rather than supporting it. Anti-abortion people get as strong of expressive benefits from registering their opposition to abortion as do pro-choice folks from supporting it. The stronger the beliefs, the higher the expressive preferences. Same for the drug laws. If the net values are fairly close to zero, a high stated value such as $5 million could well change the outcome. It seems to me your are setting up a case with a predetermined outcome, whereas in fact, social choices generally do not have such outcomes. In cases where net values will be close to zero, you're definitely right. That would cause people to scale back their bids. But, for many issues, expressive benefits are effectively unidirectional. In any case, I don't need to set up a predetermined outcome to get the result. Say, after a dozen DRP elections at the national level, the mean net value is half a billion dollars and the variance is a hundred million. Surely people could feel fairly safe in bidding several million dollars -- quite unlikely that they'd ever be decisive with that size bid. I would then predict that total amount bid increases with the number of elections held as people realize the amount they can inflate without being decisive. Yes, but in social choice generally, is it not the case that the outcome can be unknown? Certainly. But, so long as each voter can be reasonably sure that his vote won't be decisive, my argument holds. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
On Thu, 5 Sep 2002, Fred Foldvary wrote: Each voter writes down $5 million, the project is undertaken, and social loss of $85 per voter follows. Eric If a voter writes down $5 million, he is declaring that the project is worth far more to him than the $100 cost he will have to pay. But he's lying!! He is willing to pay the $100, so what basis is there in concluding that there is a loss per voter? The demand has been revealed by the voters' statements. If they just feel good about having stated that, then it was worth $100 to feel good about it. He takes the payment of the $100 as totally exogenous. No matter what he bids, he'll either have to pay the hundred dollars or not pay it. So, ignore the $100 and get the expressive benefits. NOthing the voter can do will affect whether or not he pays the $100. The expressive benefits specified in the example were $5. But, it's still rational to state the $5 million preference since the vote doesn't change the outcome and since the bid gets him the expressive benefit. I don't see why this would be less efficient than simply taking a majority vote, and the majority voting yes because they feel good about it. Demand revelation is not perfect, but it is more efficient than simple majority voting. If there is any inefficiency, it is collective democracy itself that is at fault, not the method of social choice. Because the expressive votes are accorded more aggregate weight under DRP than under regular voting. Suppose that the degree of expressive benefits accorded to an individual for expressing a preference is distributed across a continuum of voters. Under regular voting, the people with the highest expressive benefits get as much say as people with low expressive benefits (whose stated preferences will be closer to their instrumental preferences). Under DRP, the preferences of those with the strongest expressive preferences are the ones accorded the most weight. While no voter is decisive individually, the collective failure under DRP can then be massive. If anybody voting instrumentally would never bid more than a few thousand dollars for one outcome rather than another, but expressive voters throw in bids hitting the millions of dollars, the preferences of the more rational voters are swamped far more quickly than under regular voting. To the extent that expressive preferences diverge from instrumental preferences, outcomes will be worse. Eric Crampton Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
--- Eric Crampton [EMAIL PROTECTED] wrote: I don't see why the total social cost would differ from the total private costs. Specify that I get $5 worth of expressive benefits from voting $5,000,000 for project A rather than project B. Specify also that I will not be decisive, and that there are many people like me, each of whom would not be decisive. Specify also that if B is actually chosen, I accrue net benefits of, say, $100,000. If I am not decisive, I will not bid $100,000 for B, because it's quite unlikely that I'll change the outcome. Instead, I'll vote for the sure bet of expressive benefits coming with a vote for A. The demand revealing process does not take a vote, by which I presume you mean (yes or no) or (A or B). The process involves having each participant state the maximum amount he would be willing to pay for either A or B. So your example should indicate how much the individuals are stating. Nobody compensates society under the demand revealing process. Check Tideman/Tullock 1976. All revenues collected must be wasted. Tullock has confirmed this in conversation. I don't regard that as authoritative. The revenues can be given to charity without disrupting the outcome. Hal Varian describes demand revelation in his textbook * Intermediate Microeconomics * (chapter on public goods). He states: The tax is not paid to the other agents - it is paid to the state. It doesn't matter where the money goes, as long as it doesn't influence anybody else's decision; all that matters is that it be paid by the pivotal people so that they face the proper incentives. The money must be turned into real resources which would be destroyed. So long as the benfit is not specific to those who participate, I don't see why the resources must be destroyed. It may seem that a person who sustains a large loss when his preference is not followed deserves compensation, but this cannot be given without motivating an excessive statement of differential value. ... In regard to the uncompensated losses that are produced, the demand-revealing process is similar to majority rule. Right. And this does not require the destuction of resources. If I get $5 in benefits if I bid $5,000,000 for project A over project B, and if I won't be decisive, then I will bid $5,000,000 for A. You are leaving out the cost. The demand revealing process posits a good with a cost C for N persons. The average cost is C/N. Each agent is assigned a specific cost, such as C/N, which he must pay if the outcome is to obtain the good. The participants are asked: How much is most you would be willing to pay for the good if the cost to you will be C/N if the decision is to obtain it? If indeed the decision is to do it or buy it, then all participants pay their assigned cost. I'm bidding a number IN EXCESS of the benefits I actually accrue. In the aggregate, many voters behaving this way impose social costs. If you state a value of $5 million knowing you will pay $100, then your revealed preference is to pay that cost to obtain that good, and the benefit is revealed to be greater than the cost for the group as a whole if the sum of their stated value exceeds the cost. Please tell me which version of the demand revealing process you're referring to. I'm looking at Tideman/Tullock, 1976. I read it, but don't have it in front of me. I am looking now at Varian's explanation. He cites T/T 1976. Nowhere does it specify that any cost/benefit analysis is undertaken. I'll look at T/T again, but the method does not make sense if there is no preassigned cost that the agent is comparing the benefit to. We're simply adding up the stated dollar valuation for project A (building the statue) versus the stated dollar valuation for project B (not building the statue). If neither has a stated cost, and there is no pre-assigned cost per participant, then of course you will get an inefficient outcome. Any rational economic decision must account for costs. Fred = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
--- Eric Crampton [EMAIL PROTECTED] wrote: Tideman/Tullock 1976. They begin the exposition of the demand revealing process with individuals being assigned arbitrary cost shares. These could be average program costs or could be proportionate to tax shares in the current system -- the demand revealing process works regardless of the taxation mechanism chosen to fund the public good. Yes, but an essential element is that each individual be assigned a specific cost share, in dollars, so that he knows ahead of stating a value what it will cost. That is Hal Varian's explanation in * Intermediate Microeconomics * . The tax payment extracted in the running of the demand revealing process is intended only to induce truthful preference revelation, not to fund the chosen program. The Clarke tax is thusly used. But the cost share assigned to a participant should fund the project. So, nothing specifies that tax payments approximate social costs, whether instrumental or expressive voting is assumed. The cost of the item should be paid by the assigned tax shares of the participants. Otherwise indeed it makes no sense. The method as stated by Varian is: 1) Assigen a cost c(i) to each agent i. He will pay this if the item is obtained. 2) Have each agent state a value s(i). The net value n(i) is the stated value minus the cost c(i). 3) If the sum of the stated values exceeds the total cost C, the item is obtained. 4) The decisive (pivotal) agents pay the social cost (Clarke tax) equal to the sum of the net values (all n(i)) other than that of the pivotal agent. Note that it is essential for each agent to be assigned a cost c(i). Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
On Thursday, August 29, 2002, at 02:55 PM, Fred Foldvary wrote: duals being assigned arbitrary cost shares. These could be average program costs or could be proportionate to tax shares in the current system -- the demand revealing process works regardless of the taxation mechanism chosen to fund the public good. Yes, but an essential element is that each individual be assigned a specific cost share, in dollars, so that he knows ahead of stating a value what it will cost. That is Hal Varian's explanation in * Intermediate Microeconomics * . The tax payment extracted in the running of the demand revealing process is intended only to induce truthful preference revelation, not to fund the chosen program. The Clarke tax is thusly used. But the cost share assigned to a participant should fund the project. So, nothing specifies that tax payments approximate social costs, whether instrumental or expressive voting is assumed. The cost of the item should be paid by the assigned tax shares of the participants. Otherwise indeed it makes no sense. The method as stated by Varian is: 1) Assigen a cost c(i) to each agent i. He will pay this if the item is obtained. 2) Have each agent state a value s(i). The net value n(i) is the stated value minus the cost c(i). 3) If the sum of the stated values exceeds the total cost C, the item is obtained. 4) The decisive (pivotal) agents pay the social cost (Clarke tax) equal to the sum of the
Re: Median Voter vs. The Sub-optimal Equilibria
On Wed, 28 Aug 2002, Fred Foldvary wrote: Is that not an integral part of the demand revealing process? To get one's net value, we need to know the cost he will pay, and that cost is the average cost. Tideman/Tullock 1976. They begin the exposition of the demand revealing process with individuals being assigned arbitrary cost shares. These could be average program costs or could be proportionate to tax shares in the current system -- the demand revealing process works regardless of the taxation mechanism chosen to fund the public good. Later, they play with estimating Lindahl taxes, but clearly those wouldn't be average costs either. But if the tax payment includes all the social cost, and value is revealed in the willingness to pay the opportunity cost (sacrifice of resources), why would the benefits be lower than the social cost? The value of the item would include that due to expression. The tax payment extracted in the running of the demand revealing process is intended only to induce truthful preference revelation, not to fund the chosen program. Indeed, extracted funds are to be wasted rather than used to fund the project. Tideman/Tullock also note in their section on Lindahl taxes that information gathered in the demand revealing process cannot be used for setting Lindahl tax shares as this would induce people to lie about their valuations in order to reduce their tax shares. Lindahl taxation is as difficult under the demand revealing process as under any other social choice mechanism. So, nothing specifies that tax payments approximate social costs, whether instrumental or expressive voting is assumed. Eric Crampton Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
On Wed, 28 Aug 2002, Fred Foldvary wrote: There will be individual costs that excede benefits, and benefits that excede costs, but I don't see why the total social cost would differ from the total private costs. The only voters that impose an externality are those who change the outcome, and they do compensate society. It's quite simple. Specify that I get $5 worth of expressive benefits from voting $5,000,000 for project A rather than project B. Specify also that I will not be decisive, and that there are many people like me, each of whom would not be decisive. Specify also that if B is actually chosen, I accrue net benefits of, say, $100,000. If I am not decisive, I will not bid $100,000 for B, because it's quite unlikely that I'll change the outcome. Instead, I'll vote for the sure bet of expressive benefits coming with a vote for A. The demand revealing process will then provide a strongly biased picture of net benefits. Inefficient projects will be chosen whenever expressive preferences diverge from instrumental preferences. Nobody compensates society under the demand revealing process. Check Tideman/Tullock 1976. All revenues collected must be wasted. Tullock has confirmed this in conversation. The money collected can't just be burned, which would confer a benefit upon all other holders of cash. The money must be turned into real resources which would be destroyed. It may seem that a person who sustains a large loss when his preference is not followed deserves compensation, but this cannot be given without motivating an excessive statement of differential value. ... In regard to the uncompensated losses that are produced, the demand-revealing process is similar to majority rule. Why is this regarded as a social waste, if it provides utility to the voters, even if that utility is from the expression? The total value stated by the voters excedes the total cost; where is the waste? We're talking past each other. The amount of expressive benefit enjoyed by the expressive voter does count as utility. But, we can expect that the expressive benefits will be less than $1 for each dollar bid for one project rather than another. If I get $5 in benefits if I bid $5,000,000 for project A over project B, and if I won't be decisive, then I will bid $5,000,000 for A. I'm bidding a number IN EXCESS of the benefits I actually accrue. In the aggregate, many voters behaving this way impose social costs. If someone states a value of $100 for a public good, why would the utility be less than than $100 spent for a private good? Because it's very very cheap for me to state a value of $100 for a public good if I'll not be decisive and if I won't pay the full cost of the $100. If I get even an infinitessimally small expressive benefit from saying that I value the public good at $100, then I'll do it regardless of how much the public good is worth to me. Let's say that each person gets $5 worth of expressive utility from voting for statue construction. The average cost to each voter if the statue is constructed is $50. And, each voter gets $3 worth of direct utility from looking at the constructed statue. Then the total utility per voter is $8, while the cost is $50, and the statue is not obtained. With demand revelation, the statue is only bought if the total stated value is greater than the total cost. Please tell me which version of the demand revealing process you're referring to. I'm looking at Tideman/Tullock, 1976. Nowhere does it specify that any cost/benefit analysis is undertaken. We're simply adding up the stated dollar valuation for project A (building the statue) versus the stated dollar valuation for project B (not building the statue). If no voter is decisive, each will state a positive value for project A because that's the only way of getting any utility, and the project IS undertaken. Since no voter is decisive, each voter takes the tax share as given and exogenous (the statue either will or won't be built, regardless of his vote). Optimization then requires voting for the project. He does not vote for the statue. With demand revelation, he only states the maximum he would be willing to pay. In this case it is $8. What makes demand revelation superior than ordinary voting is that one states a value for the good rather than voting yes or no. Ok. Same example as before, each individual voter obtains expressive utility of $5 from stating a value of $50. Project is built. But I still argue that nothing in the Tideman/Tullock system requires that the sum of the stated benefits has to exceed the known project costs; the system simply picks the option with the highest expressed valuation. In the example from before, option A is to build the statue, option B is to not build the statue. If there were zero expressive benefits anywhere, all voters would state a willingness to pay of $47 for option B ($50 average cost less $3 utility
Re: Median Voter vs. The Sub-optimal Equilibria
--- Eric Crampton [EMAIL PROTECTED] wrote: Why are we assuming that instituting the demand revealing process would get us to uniform average-cost taxation? Is that not an integral part of the demand revealing process? To get one's net value, we need to know the cost he will pay, and that cost is the average cost. For any such issues, it would be a rather safe bet to assume that one's dollar vote wouldn't be decisive. I concede that. [presuming] (3) the utility derived from expression and wining does not count. 3) Not true. Those benefits just have to be lower than social cost to get an inefficient outcome. But if the tax payment includes all the social cost, and value is revealed in the willingness to pay the opportunity cost (sacrifice of resources), why would the benefits be lower than the social cost? The value of the item would include that due to expression. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
--- Eric Crampton [EMAIL PROTECTED] wrote: But even if his expression reflects his stated value, he is still deriving utility from the good. Why does it matter the reason for the utility? Because of the divergence of private and social costs inherent in the voting act. There will be individual costs that excede benefits, and benefits that excede costs, but I don't see why the total social cost would differ from the total private costs. The only voters that impose an externality are those who change the outcome, and they do compensate society. So, massive social waste can be approved by majority vote because of the relatively small expressive benefits attached to voting for the inefficient program. Why is this regarded as a social waste, if it provides utility to the voters, even if that utility is from the expression? The total value stated by the voters excedes the total cost; where is the waste? the argument regarding expressive voting is that that utility is lower than the utility that could have been derived from alternate use of the funds that were used to construct the statue. If someone states a value of $100 for a public good, why would the utility be less than than $100 spent for a private good? Let's say that each person gets $5 worth of expressive utility from voting for statue construction. The average cost to each voter if the statue is constructed is $50. And, each voter gets $3 worth of direct utility from looking at the constructed statue. Then the total utility per voter is $8, while the cost is $50, and the statue is not obtained. With demand revelation, the statue is only bought if the total stated value is greater than the total cost. Since no voter is decisive, each voter votes for the statue constructionm in order to get the $5 in expressive benefits. He does not vote for the statue. With demand revelation, he only states the maximum he would be willing to pay. In this case it is $8. What makes demand revelation superior than ordinary voting is that one states a value for the good rather than voting yes or no. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
fabio guillermo rojas wrote: screwed-up institution. The outcome of democracy depends on the overall rationality of public opinion, but whatever outcome you get can be equally enjoyed by the rational and irrational alike. My question isn't about the quality of policy, but the difference between what the institution produces and what the median voter wants. It's easy for me to believe that voters are irrational, but it's harder for me to believe that every policy closely matches the MV. Why? What are the biggest unpopular policies that persist? Let me elaborate my question: isn't is possible than when voters put faith in some set of rules for generating policy that the outputs may be far from the MV? Fabio If I understand you, my paper talked about this too. If voters put an irrationally high level of trust in their leaders, this might give their leaders the slack to pursue their own agenda. The doctrine of papal infallibility is the extreme case - if your clientele considers you infallible, you obviously have a lot of slack! But it is plausible to see this as a special case of the MV. The only twist is that the median voter implicitly says My preference is whatever his preference is, within some limits. -- Prof. Bryan Caplan Department of Economics George Mason University http://www.bcaplan.com [EMAIL PROTECTED] He wrote a letter, but did not post it because he felt that no one would have understood what he wanted to say, and besides it was not necessary that anyone but himself should understand it. Leo Tolstoy, *The Cossacks*
Re: Median Voter vs. The Sub-optimal Equilibria
Fred Foldvary wrote: --- Bryan D Caplan [EMAIL PROTECTED] wrote: The outcome of democracy depends on the overall rationality of public opinion, but whatever outcome you get can be equally enjoyed by the rational and irrational alike. Does this not depend on the structure and method of voting? For example, would not the demand revelation method remedy the problem? It makes those who change the outcome pay the social cost. If I remember correctly, demand revelation mechanisms are useless if the probability of decisiveness is low and voters get some direct utility from expressive voting or holding irrational beliefs. Thus, suppose I get a $10 direct expressive benefit from voting for tons of useless health care spending, and the probability of decisiveness is 1-in-a-million. I don't see how any demand revelation mechanism is going to help. -- Prof. Bryan Caplan Department of Economics George Mason University http://www.bcaplan.com [EMAIL PROTECTED] He wrote a letter, but did not post it because he felt that no one would have understood what he wanted to say, and besides it was not necessary that anyone but himself should understand it. Leo Tolstoy, *The Cossacks*
Re: Median Voter vs. The Sub-optimal Equilibria
Thus, suppose I get a $10 direct expressive benefit from voting for tons of useless health care spending, and the probability of decisiveness is 1-in-a-million. I don't see how any demand revelation mechanism is going to help. Prof. Bryan Caplan If someone states a value of $10 not because he wants the health care spending but because he enjoys stating that value, then it seems to me that the person is willing to pay that amount even if for that odd reason, and so his statement of value would be just as significant as one who actually values the health care. If the average cost, say for mosquito abatement, would be $5, and that person states he would pay up to $10 for it, then he is knowingly contributing to the total stated value, and if the total value exceeds the cost, is willing to pay the $10, regardless of his reason. All we really should care about is how much he is willing to pay. There can be all sorts of reasons why one would favor a program. One could, for example, favor a freeway because he thinks it helps birds navigate, even if in fact it does not. No matter, he is willing to put up bucks, and that is what counts. The whole point of demand revelation is that one's true subjective value is unknown, and so is the reason for it. All we can know is the stated value, and that is sufficient, and makes it superior to majority voting. After all, someone could vote for a Democrat candidate because he likes the sound of the candidate's name, so there are odd reasons why people vote in any system. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
Actually, I'm in the process of writing up a short article for Public Choice on this topic (as Bryan may recall). If expressive voting theory holds, and if expressive benefits are increasing in the amount of money voted under the Tideman-Tullock procedure, then the demand revelation process should induce worse outcomes than the current system. Specifically, people who have the strongest preferences are accorded only one vote each in the current system. Under the demand revelation process, their influence will be magnified relative to more instrumentally-oriented voters. Under the demand-revealing process, if my vote is decisive in changing the outcome, I have to pay an amount equal to the amount of my dollar vote that was necessary to swing the outcome: say that $1 million votes for option A and my $20,000 vote was enough to generate $1,010,000 in votes for B; then I owe a tax of $10,000. Say that I get expressive benefits of $1000 by telling everyone I've put in a bid of $10,000 for policy option A under the demand revealing process. So long as my chances of being decisive are less than 1/10, it's rational for me to bid $10,000 and gain the $1000 in expressive benefits. In equilibrium, the people with the strongest expressive preferences bid the most and none of them are likely to be decisive, and outcomes are worse than under a one-man one-vote system, so long as expressive preferences diverge from instrumental preferences (which Bryan's work strongly suggests). Eric Crampton On Mon, 26 Aug 2002, Bryan Caplan wrote: If I remember correctly, demand revelation mechanisms are useless if the probability of decisiveness is low and voters get some direct utility from expressive voting or holding irrational beliefs. Thus, suppose I get a $10 direct expressive benefit from voting for tons of useless health care spending, and the probability of decisiveness is 1-in-a-million. I don't see how any demand revelation mechanism is going to help. -- Prof. Bryan Caplan Department of Economics George Mason University http://www.bcaplan.com [EMAIL PROTECTED] He wrote a letter, but did not post it because he felt that no one would have understood what he wanted to say, and besides it was not necessary that anyone but himself should understand it. Leo Tolstoy, *The Cossacks*
Re: Median Voter vs. The Sub-optimal Equilibria
Fred Foldvary wrote: Thus, suppose I get a $10 direct expressive benefit from voting for tons of useless health care spending, and the probability of decisiveness is 1-in-a-million. I don't see how any demand revelation mechanism is going to help. Prof. Bryan Caplan If someone states a value of $10 not because he wants the health care spending but because he enjoys stating that value, then it seems to me that the person is willing to pay that amount even if for that odd reason, and so his statement of value would be just as significant as one who actually values the health care. No, the point is that might *really* get a $10 benefit from SAYING you get a $1 M benefit. If your probability of decisiveness is under 1-in-100,000, it would pay to do so. But the social cost of this behavior could drastically exceed the private benefit. -- Prof. Bryan Caplan Department of Economics George Mason University http://www.bcaplan.com [EMAIL PROTECTED] He wrote a letter, but did not post it because he felt that no one would have understood what he wanted to say, and besides it was not necessary that anyone but himself should understand it. Leo Tolstoy, *The Cossacks*
Re: Median Voter vs. The Sub-optimal Equilibria
On Mon, 26 Aug 2002, Fred Foldvary wrote: If the average cost, say for mosquito abatement, would be $5, and that person states he would pay up to $10 for it, then he is knowingly contributing to the total stated value, and if the total value exceeds the cost, is willing to pay the $10, regardless of his reason. All we really should care about is how much he is willing to pay. There can be all sorts of reasons why one would favor a program. One could, for example, favor a freeway because he thinks it helps birds navigate, even if in fact it does not. No matter, he is willing to put up bucks, and that is what counts. The important point is that there's a disjoint between the willingness to pay and the actual payment: an expressed vote preference, even in the dollar terms of the Tideman-Tullock procedure, leads only probabilistically to actual requirements to pay the amount. If the expressive voter is risk neutral, faces a 1/million chance of being decisive and having to pay, and derives expressive benefits of $1 from 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. Eric
Re: Median Voter vs. The Sub-optimal Equilibria
--- Eric Crampton [EMAIL PROTECTED] wrote: If expressive voting theory holds, and if expressive benefits are increasing in the amount of money voted under the Tideman-Tullock procedure, then the demand revelation process should induce worse outcomes than the current system. Specifically, people who have the strongest preferences are accorded only one vote each in the current system. Under the demand revelation process, their influence will be magnified relative to more instrumentally-oriented voters. First of all, the demand-revealing method does not require that the identities of the persons stating a value be public. Each voter can be given a password, and he enters a stated value on a web site. The administrator of the system knows his identity, but this is not public knowledge. The voter can then express anything he likes, just as in a secret ballot. But even if his expression reflects his stated value, he is still deriving utility from the good. Why does it matter the reason for the utility? In equilibrium, the people with the strongest expressive preferences bid the most and none of them are likely to be decisive, and outcomes are worse than under a one-man one-vote system, What matters is that the voter is willing to pay the average cost of the good plus the expected social cost of being pivotal. Suppose a community is voting for a public sculpture. One may not really want to have a sculpture, but one gets esteem from the approval one gets from expressing support for the arts. Suppose further that one gets disutility from not voting in accord with one's public expression. The statue is still providing utility, although in an indirect way, as those who get utility from approval of expression still obtained that utility from the sculpture. It is like admitting an Albanian into your club. The members don't really like Albanians, but they are proud of being regarded as appreciative of ethnic diversity, so they all vote to let in the Albanian. They all feel good about being diverse, so admitting the Albanian was rational after all. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
--- 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]
Re: Median Voter vs. The Sub-optimal Equilibria
No, the point is that might *really* get a $10 benefit from SAYING you get a $1 M benefit. If your probability of decisiveness is under 1-in-100,000, it would pay to do so. But the social cost of this behavior could drastically exceed the private benefit. Prof. Bryan Caplan Right. But we need to do a complete comparative systems: 1) how prevalent is expressive voting, empirically? 2) does expressive voting still apply in secret ballots? 3) how do we get the probabilities of decisiveness? If the total values are close to the cost, the likelyhood of being decisive rises. It seems to me that in pure demand revelation, the probability is unknown and unknowable. When we add the probabilities, we mix in that with demand revelation, and it becomes a different system. When we inject a fiat probability such as 1/100K, then we have rigged the outcome. A premise of pure demand revelation is that the subjective values of others are unknown, and stated values could be lies. The person stating a value of $1 million has no way of knowing what the stated values of the others will be. He just knows the size of the voting pool and the cost of the public good. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
On Mon, 26 Aug 2002, Fred Foldvary wrote: First of all, the demand-revealing method does not require that the identities of the persons stating a value be public. Each voter can be given a password, and he enters a stated value on a web site. The administrator of the system knows his identity, but this is not public knowledge. The voter can then express anything he likes, just as in a secret ballot. I haven't argued that the inefficiency would arise from a public ballot, though that could make things worse, certainly. We currently have a secret ballot, but people still feel the urge to go and make their mark rather than just pretend as though they had. Expressive benefits seem to key into a few things, not least of which is a person's own self-image. Voting for a candidate that expresses values with which a voter wishes to identify gives But even if his expression reflects his stated value, he is still deriving utility from the good. Why does it matter the reason for the utility? Because of the divergence of private and social costs inherent in the voting act. What matters is that the voter is willing to pay the average cost of the good plus the expected social cost of being pivotal. Even under the demand revealing process the voter has an exceedingly small chance of being pivotal. So, massive social waste can be approved by majority vote because of the relatively small expressive benefits attached to voting for the inefficient program. If the voter knew ex ante that he would be pivotal, then your statement above would be correct. But, voters each are not likely to be pivotal, and each one knows it at some level. Suppose a community is voting for a public sculpture. One may not really want to have a sculpture, but one gets esteem from the approval one gets from expressing support for the arts. Suppose further that one gets disutility from not voting in accord with one's public expression. The statue is still providing utility, although in an indirect way, as those who get utility from approval of expression still obtained that utility from the sculpture. Sure, but the argument regarding expressive voting is that that utility is lower than the utility that could have been derived from alternate use of the funds that were used to construct the statue. Let's say that each person gets $5 worth of expressive utility from voting for statue construction. The average cost to each voter if the statue is constructed is $50. And, each voter gets $3 worth of direct utility from looking at the constructed statue. Since no voter is decisive, each voter votes for the statue constructionm in order to get the $5 in expressive benefits. Net social waste per voter is $42. I'm not denying that people get utility from expressing their preferences; I am arguing that they're in a massive PD game that leads to an inefficient outcome. ERic It is like admitting an Albanian into your club. The members don't really like Albanians, but they are proud of being regarded as appreciative of ethnic diversity, so they all vote to let in the Albanian. They all feel good about being diverse, so admitting the Albanian was rational after all. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
On Mon, 26 Aug 2002, Fred Foldvary wrote: Since the relevant comparitive system is majority voting, there is a disjoint in yes-no voting as well. Agree with you so far 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. Why are we assuming that instituting the demand revealing process would get us to uniform average-cost taxation? 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. Tullock speculates in the article that the total amount taxed through the demand revealing process would be quite low, which implies that the probability of being decisive will also be quite low. In any case, after a few such elections, it seems likely that the total amounts bid and the total amounts collected would be public knowledge, and if the total amount collected were indeed a very small fraction of the total amount bid, people would rightly conclude that the probability of decisiveness is low. Additionally, many issues enjoy reasonably broad public support. For any such issues, it would be a rather safe bet to assume that one's dollar vote wouldn't be decisive. 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. 1) neither is necessarily assumed. All expressive benefit could simply be internal to the voter: the voter expressing to himself what kind of person he is. Expressive voting in that case helps build self-image. But, we currently do have a secret ballot, and expressive preferences still reign. 2) They need not be known with certainty. They just have to be known to be relatively low. 3) Not true. Those benefits just have to be lower than social cost to get an inefficient outcome. Fred Foldvary = [EMAIL PROTECTED]
Re: Median Voter vs. The Sub-optimal Equilibria
--- Bryan D Caplan [EMAIL PROTECTED] wrote: The outcome of democracy depends on the overall rationality of public opinion, but whatever outcome you get can be equally enjoyed by the rational and irrational alike. Does this not depend on the structure and method of voting? For example, would not the demand revelation method remedy the problem? It makes those who change the outcome pay the social cost. (See Tideman and Tullock, 1976, A New and Superior Process for Making Social Choices, Journal of Political Economy 84 (6) December.) Fred Foldvary = [EMAIL PROTECTED]
