If their representatives don’t bring home the bacon, Americans are free to fire them on Election Day. But what if members of Congress didn’t run for reelection in their home districts but were randomly assigned to run somewhere else? A 2011 paper, “Randomizing Districts for Reelections: A Thought Experiment,” tried to find benefits in a legislature divorced from geography. Under such a system, “legislators cannot focus their attention upon pleasing a geographically-concentrated special interest while neglecting the broader national interest.” (
So why the anti-tax bias ? Because of ill-informed voter opinions ? How about no taxes at all, we could get rid of legislators and bureaucrats altogether. Let's all be road warriors, could be fun.
Today they put far more effort into increasing local spending than in reducing total taxes. The whole point is to create incentives to put similar efforts into spending and taxes. So yes, they would reduce their effort into more spending when they have a similarly attractive option to reduce taxes.
Doesn't this provide just another way people can focus on parochial interests at the expense of general interests? It's not valuable if taxes are cut for one region - the tax bill that must be raised is set by total spending. This would only work if representatives reduced the effort they put into getting spending increases.
Also, marginal deadweight losses from taxes increase more than linearly with the rate. So having different rates in different regions, while raising the same revenue, should result in greater inefficiency.
If you want that politician focus on the national interest that's easy. You use party-list proportional representation instead of majority voting for districts. Germany doesn't have a real pork issue.
If you don't want that politicians represent the people from their district, why do you have districts in the first place?
It is hard to find hard-to-manipulate ways to divide policy by issues, to assign weights to issues, to pick which citizens vote on which issues, and then to say how any particular policy issue is decided.
Logrolling is the main driver of spending. You support my unneeded fighter planes and I'll support your unneeded tanks.
We need ways for the general public to understand the true cost of things. I'm calling for a national sales tax to pay for defense, intelligence and homeland security. Congress would have to say "Here;s what we think is necessary to spend and here's the sales tax rate needed to pay for it."I'd let Congress take three years for a true national emergency or war, but otherwise spending would have to be balance with revenue.
Having those few candidates repeat and focus on local issues aids info specialization.
But the whole question is whether "aggregating information" (to use your phrase) by geography produces better results than "aggregating" it under other rubrics, such as pertain to issues construed more abstractly.
[Your proposal to tie taxes to local benefits is pure ultrarightistism, that would produce a race to the bottom, as districts are forced to lower taxes to attract "job creators." (Funny how that obnoxious phrase has vanished from M. Romney's rhetoric.)]
We could just have an at large election of representatives across the nation. Relative to that, the reason to have local representatives is to let local focus focus their attention on a few people, and gain from info specialization. But if those few people change every few years, and appeal to national issues, it is harder to specialize. Having those few candidates repeat and focus on local issues aids info specialization. Yes, this also has the potential for problems, if the political power of different districts differs from their importance. But I'm much less worried about that than about common incentives toward inefficiency, such as happens when many benefits are local but taxes are national.
I'm glad to see our idea getting some attention. A working version of our paper on this can be found on SSRN here: http://papers.ssrn.com/sol3...
Robin, we have received comments along those lines and intend to work it into a later version of the paper. Our randomization idea has some tension with the traditional view of representative democracy. Voters are used to their respective Congressman doing their bidding for them in Washington, not just with respect to pork but across a range of issues. Voters reward Congressmen who "fight for them," which works well for the Congressman trying to get reelected, but ultimately this leads to suboptimal national policies. Representatives "not learning their district's issues" and not having an incentive to fight for district-specific provisions could be a feature, not a bug necessarily. We don't put it exactly in those terms in the paper, but there is substantial discussion of why national policy would be more efficient if Congressmen did not have to cater to district specfic interests to the extent that they do now.
In our most recent revision of our paper (not yet available online), we get more specific with respect to modeling our randomization rule formally. We base our model off of a JPE paper by Weingast, Shepsle, and Johnson (1981). In that paper, Weingast et al propose a rule similar to what you are proposing. I think there are a lot of pro's for this kind of proposal.
With either your proposal or the current system, legislators still have an incentive to maximize benefits and minimize costs for their respective districts.It might be a step forward if we could vary taxes by district in proportion to their benefits, but public choice considerations arise, as the self-interested representatives who are trying to get reelected would still try to make the case for the lowest taxes possible for their district, while ALSO making the case for higher benefits (in other words, these may not offset in practice). Moreover, there are local projects that have national benefits, in addition to the local benefits. Military bases or interstate highways come to mind. The practical side of implementing such a tax system gets fuzzy. Enter self-interested politicians representing their districts. I think there are legitimate objections to my narrower points here, which we could go back and forth on, but the wider point is the important one: in either your proposal or the current political system, legislators still know whom they are accountable to, and their public choice incentive to cater to those interests remains.There is more discussion of this incentive issue in the paper.
Randomizing districts for reelection addresses the incentive problem in an important way. If a legislator does not know who they will be accountable to (because the district in which they would be up for a vote in the next election will be selected at random), then they do not have an incentive to fight for a particular district more than another one. We refer to this as a legislator standing behind a 'veil of randomness.' As a result, self interested legislators would then have an incentive to craft or vote for more efficient legislation with broad appeal to increase their chances of getting reelected.
A big problem is that spending seems almost completely disconnected from revenue these days. "Deficits don't matter" since Reagan. When Bush Sr. tried to raise taxes in exchange for lower spending, he didn't get the lower spending. When Clinton and the glory that was the 1990s economy obtained a surplus, GWB set about finding ways to spend it (and then some). Being fiscally prudent seems a politically losing move. Giving politicians a new way to lower taxes might have seemed a halfway good idea 17 years ago, but today we know that "starving the beast" is backwards and makes spending seem costless.
This is pie in the sky but the geographical representation seems somewhat out dated. With Government doing so much today I think it would be good to replace geographical representation, at least partially, with a system that breaks elections up by function. So you would vote for a head of social security who would set the amount and method of taxation and the amount and method of payout for social security. You would elect a head of defense and he would set the defense tax level and control defense spending. The same for medicare, medicaid and medical licensing. And so on for transportation, education etc. You would still need a President to declare war and run the courts.
So why the anti-tax bias ? Because of ill-informed voter opinions ? How about no taxes at all, we could get rid of legislators and bureaucrats altogether. Let's all be road warriors, could be fun.
Today they put far more effort into increasing local spending than in reducing total taxes. The whole point is to create incentives to put similar efforts into spending and taxes. So yes, they would reduce their effort into more spending when they have a similarly attractive option to reduce taxes.
Doesn't this provide just another way people can focus on parochial interests at the expense of general interests? It's not valuable if taxes are cut for one region - the tax bill that must be raised is set by total spending. This would only work if representatives reduced the effort they put into getting spending increases.
Also, marginal deadweight losses from taxes increase more than linearly with the rate. So having different rates in different regions, while raising the same revenue, should result in greater inefficiency.
If you want that politician focus on the national interest that's easy. You use party-list proportional representation instead of majority voting for districts. Germany doesn't have a real pork issue.
If you don't want that politicians represent the people from their district, why do you have districts in the first place?
It is hard to find hard-to-manipulate ways to divide policy by issues, to assign weights to issues, to pick which citizens vote on which issues, and then to say how any particular policy issue is decided.
Logrolling is the main driver of spending. You support my unneeded fighter planes and I'll support your unneeded tanks.
We need ways for the general public to understand the true cost of things. I'm calling for a national sales tax to pay for defense, intelligence and homeland security. Congress would have to say "Here;s what we think is necessary to spend and here's the sales tax rate needed to pay for it."I'd let Congress take three years for a true national emergency or war, but otherwise spending would have to be balance with revenue.
Having those few candidates repeat and focus on local issues aids info specialization.
But the whole question is whether "aggregating information" (to use your phrase) by geography produces better results than "aggregating" it under other rubrics, such as pertain to issues construed more abstractly.
[Your proposal to tie taxes to local benefits is pure ultrarightistism, that would produce a race to the bottom, as districts are forced to lower taxes to attract "job creators." (Funny how that obnoxious phrase has vanished from M. Romney's rhetoric.)]
The problems arise with the analogies that are drawn from the intellectualization of metaphors.
We could just have an at large election of representatives across the nation. Relative to that, the reason to have local representatives is to let local focus focus their attention on a few people, and gain from info specialization. But if those few people change every few years, and appeal to national issues, it is harder to specialize. Having those few candidates repeat and focus on local issues aids info specialization. Yes, this also has the potential for problems, if the political power of different districts differs from their importance. But I'm much less worried about that than about common incentives toward inefficiency, such as happens when many benefits are local but taxes are national.
I'm glad to see our idea getting some attention. A working version of our paper on this can be found on SSRN here: http://papers.ssrn.com/sol3...
Robin, we have received comments along those lines and intend to work it into a later version of the paper. Our randomization idea has some tension with the traditional view of representative democracy. Voters are used to their respective Congressman doing their bidding for them in Washington, not just with respect to pork but across a range of issues. Voters reward Congressmen who "fight for them," which works well for the Congressman trying to get reelected, but ultimately this leads to suboptimal national policies. Representatives "not learning their district's issues" and not having an incentive to fight for district-specific provisions could be a feature, not a bug necessarily. We don't put it exactly in those terms in the paper, but there is substantial discussion of why national policy would be more efficient if Congressmen did not have to cater to district specfic interests to the extent that they do now.
In our most recent revision of our paper (not yet available online), we get more specific with respect to modeling our randomization rule formally. We base our model off of a JPE paper by Weingast, Shepsle, and Johnson (1981). In that paper, Weingast et al propose a rule similar to what you are proposing. I think there are a lot of pro's for this kind of proposal.
With either your proposal or the current system, legislators still have an incentive to maximize benefits and minimize costs for their respective districts.It might be a step forward if we could vary taxes by district in proportion to their benefits, but public choice considerations arise, as the self-interested representatives who are trying to get reelected would still try to make the case for the lowest taxes possible for their district, while ALSO making the case for higher benefits (in other words, these may not offset in practice). Moreover, there are local projects that have national benefits, in addition to the local benefits. Military bases or interstate highways come to mind. The practical side of implementing such a tax system gets fuzzy. Enter self-interested politicians representing their districts. I think there are legitimate objections to my narrower points here, which we could go back and forth on, but the wider point is the important one: in either your proposal or the current political system, legislators still know whom they are accountable to, and their public choice incentive to cater to those interests remains.There is more discussion of this incentive issue in the paper.
Randomizing districts for reelection addresses the incentive problem in an important way. If a legislator does not know who they will be accountable to (because the district in which they would be up for a vote in the next election will be selected at random), then they do not have an incentive to fight for a particular district more than another one. We refer to this as a legislator standing behind a 'veil of randomness.' As a result, self interested legislators would then have an incentive to craft or vote for more efficient legislation with broad appeal to increase their chances of getting reelected.
A big problem is that spending seems almost completely disconnected from revenue these days. "Deficits don't matter" since Reagan. When Bush Sr. tried to raise taxes in exchange for lower spending, he didn't get the lower spending. When Clinton and the glory that was the 1990s economy obtained a surplus, GWB set about finding ways to spend it (and then some). Being fiscally prudent seems a politically losing move. Giving politicians a new way to lower taxes might have seemed a halfway good idea 17 years ago, but today we know that "starving the beast" is backwards and makes spending seem costless.
This is pie in the sky but the geographical representation seems somewhat out dated. With Government doing so much today I think it would be good to replace geographical representation, at least partially, with a system that breaks elections up by function. So you would vote for a head of social security who would set the amount and method of taxation and the amount and method of payout for social security. You would elect a head of defense and he would set the defense tax level and control defense spending. The same for medicare, medicaid and medical licensing. And so on for transportation, education etc. You would still need a President to declare war and run the courts.