CHAPTER 1
THE CASE FOR THIRD PARTIES AND ELECTION REFORM
The theme of the 2008 election was change with an emphasis on a new face and a promise of better policies. But in the campaign there was virtually no attention given to the issues of what changes we could make in our political system so that it would serve us better. This is surprising given that prior to the 2008 election many felt that the system had broken down into a cycle of excessive partisanship and endless costly campaigning. If we want changes we can really believe in, we must open up our political processes to third parties. There are three reasons for this.
Why We Need Viable Third Parties
There is a need for candid, informative discussions of policy issues which under our two party duopoly we have not had since 1992. Given, for example, the realities of our twin deficits and peak oil, the American people needed to be told in the 2008 election that they were going to have to downsize and could not be offered middle class tax cuts. But both candidates unabashedly pandered to the electorate on this. Yet in the 1992 campaign, Ross Perot as a third viable candidate was willing and able to tackle the issue of our deficit head on.
Polls have shown consistent support for a more vigorous policy on illegal immigration than either of the major candidates in the 2008 election supported. Yet this issue was not discussed candidly in the election. In Europe, one sees clear examples where third parties compelled major parties to give this issue more serious attention.
Because of political correctness in one party and cowardice in the other, as also occurred in the 2006 Michigan mid-term election, candidates of neither major party in 2008 honestly discussed the misuse of affirmative action in public policy. In the Michigan case in 2006 you had a third party intervention in the form of the Michigan Civil Rights Initiative which outlawed discriminatory policies based on race, gender and ethnicity by a margin of close to 60%. That was a good correction to the refusal by far too many liberals and even Republicans to dialog honestly on the issue of what affirmative action had morphed into.
The point is that viable third party presence can be a useful device for promoting in our campaign the open discussions needed on serious policy issues.
The second reason to open up our political processes to third parties involves accountability. During the initial debates on the TARP (Troubled Asset Relief Program) Speaker Pelosi blamed the 2008 financial meltdown on Bush. But in fact the Democrats were also complicit in contributing to the meltdown. In 2005 when some Republican senators and the White House tried to raise capital requirements on Fannie Mae and Freddie Mack, they were thwarted by Nancy Pelosi and Barnie Frank among others. Surely it is clear now that such a move would have been prudent.
It certainly is true that the Bush Administration had a very laid back attitude towards financial regulation and bears its share of responsibility for the 2008 crisis. But it was Allen Greenspan of the Federal Reserve Bank and Robert Rubin and Lawrence Summers of the Clinton Treasury Department who amongst others supported, or at least didn't oppose, the 1999 bipartisan repeal of the Glass-Steagall Act during the Clinton Administration.
AS WIKIPEDIA STATES IT, "The repeal enabled commercial lenders such as CITIGROUP, which was in 1999 then the largest U.S. bank by assets, to underwrite and trade instruments such as mortgaged-backed securities and collateralized debt obligations and established so-called structural investment vehicles, or SIVs, that bought these securities." These of course, represent the toxic assets and arrangements which greatly contributed to the 2008 financial meltdown.
This in turn led to the 10s of billions of capital infusion and the 100s of billions of asset guarantees by the government to prevent CITI's meltdown.
Under the circumstances of our two-party system, in responding to the
2008 economic crisis voters had to choose between punishing only one party or none. Had there been a viable third party presence in the
2008 election, voters would have had the alternative of holding both parties to account for the bipartisan economic mismanagement encompassing at least the last four presidential terms.
Voters face a similar dilemma in terms of holding candidates accountable for campaign promises that should not have been made or not broken. Surely the most dramatic example of this during the 2008 election campaign was Senator Obama's November 2007 promise to negotiate a level campaign finance field with his Republican opponent and his, at best hypocritical, repudiation of the promise in April
2008 after securing his party's nomination. Again, had there been a viable third party alternative, voters could have taken Senator Obama to task on this without having to support McCain to do so. As it was, we wound up with by far the most expensive presidential campaign in our history that has surely set a bad precedent for the future in terms of the ability of money to buy a political office.
The third reason why we need to open up our political processes to third parties is the dysfunction of our two major parties, especially the Republican Party. The contemporary Republican Party has totally eschewed traditional Republican values, as I argue in the second chapter, and really has nothing to offer the country in terms of 21st Century approaches to 21st Century problems.
I included the text of my essay on Bryan Caplan's book on irrational voters in this book for a reason. That is in part to highlight something important that I feel his book neglected and that I know contemporary Republicans ignore: That different times require different economic, and other, institutional arrangements. Opening up our political processes to new parties is surely needed to respond to this fact and to assure that we have a credible check on what the other major party will do with its power.
The question that arises from the above three points--the need for better discussion of policy issues, for better accountability, and for better alternatives--is: Why do we have a two party state?
Why We Have a Two Party System
A two party system is not mandated by our constitution, neither is the specific voting system we use. But since the foundation of the Republic, and even in colonial times, our voting system has been predominately based on plurality voting. Plurality voting is a voting system in which each voter can only vote for one candidate with the candidate having the most votes, not necessarily a majority, winning.
Even in those places like Georgia and Louisiana with run-offs in the event no candidate wins a majority, the voting system is largely based upon the principles of plurality voting.
Plurality voting is the underlying basis of the political duopoly enjoyed by the Republicans and Democrats. There is a strong tendency in countries that use this system for one of two patters to emerge: 1) the two party state as in the UK and the US, the prevailing pattern, or 2) the dominate party state which existed in Mexico for around 70 years under the PRI and in the American South between the end of Reconstruction and the beginning of the civil rights movement.
It is clear that the appearance of the two party state, while reinforced by limits on ballot access, is not dependent upon such limitations. Great Britain is a good example of this because of its quite low barriers to third party access. Anyone in Britain can run for parliament by making a 500 pound, less than a $1,000 dollar, deposit which is fully refundable if a candidate receives at least 5% of the votes. Yet third parties in Britain are quite marginalized much like they are in US elections. In the 2005 British parliamentary election, for example, Labour received 36.9% of the popular vote, the Conservatives 33.86% and the Liberal Democrats 23.09%. But in parliamentary seats the parties received 55.11% for Labour, 30.65% for the Conservatives and only 9.6% for the Liberal Democrats.
Changes in the methods of election will inevitably bring about changes in election outcomes. Until relatively recently New Zealand, for example, used plurality voting resulting in a two party system as in the UK and the US. However, after it introduced in 1996 elements of proportional representation as in the German model, a more complex system of party representation has emerged in which the disproportional representation of parties in the New Zealand parliament has steadily declined. In Mexico, movement away from the dominant party state was accompanied by the introduction of some proportional representation in Mexican elections. While I do not advocate such representation in our system, such examples do illustrate that changes in voting systems do change political systems.
This is an important factor to consider when asking the question of what is the purpose of a voting system and how should one judge its performance.
Purpose of a Voting System.
A voting system is a mechanism for combining the wishes of individuals into a decision for society or a group of people. In judging a voting system's performance two questions need to be asked:
1) How well does the outcome represent voters' desires?
2) How conducive is the system to giving the voters a reasonably
wide, as opposed to a narrow, range of choices?
In answering the first question, we generally accept that the majority's view best represents voters desires in terms of election outcomes. Hence in a yes or no vote on a single proposition a majority vote in favor or against is accepted as legitimate. The same would hold in a two candidate race. But when one has an election with three or more viable candidates, the issue, particularly in a plurality election, becomes messy, unless there is clearly one candidate with majority support.
The 2000 Florida presidential vote is a good example of the problem here. Did the outcome that gave the election to George W. Bush really represent the wishes of those voters whose votes were counted? Most of us would surely answer no because clearly most of those who voted for Nader would have supported Gore had the election been confined to a two man race between Bush and Gore.
The spoiler role is one of the fatal flaws that infests our single vote plurality system in that elections with three or more viable choices can easily elect a candidate that the majority does not want.
It is to get around this problem that voters in a plurality voting system tend to coalesce into a two party system which takes advantage of the wasted vote syndrome and assures a political duopoly. Once a two party system becomes entrenched voters face the wasted vote syndrome because in terms of influencing the outcome, it pays to vote strategically, not sincerely as Nader voters did in Florida when they inadvertently threw the election to Bush. The basic problem with plurality voting is that the one-person-one-vote principle which underlies this system limits the expression of the voter's will, or preference, and hence its expression in the election's outcome.
In his book Gaming the Vote, which is reviewed in this collection of essays, William Poundstone makes the obvious point that plurality is probably the worst voting system of those systems purportedly designed to reflect voters wishes. The most notable attempt to grapple with this problem occurred in the late 18th Century when two Frenchmen, Jean-Charles de Borda and the Marquis de Condorcet, took up the issue in the context of electing members to the French Academy.
Both realized that the system used, plurality voting, was not giving in many cases the correct answer. Each gave different answers as to what to do about this. Borda's solution was what has become known as the Borda count. This system involves asking voters to rank all the choices according to their preferences. For each ranking the top candidate is assigned a numerical score of one less than the number of candidates, the second ranked two less and so on to the last candidate who is assigned no score. The scores are added up and the candidate with the highest score wins. The advantage of the Borda count is that it allows voters a reasonable broader range for expressing their preferences since under the system they can express their judgement on every candidate rather than only one candidate as under plurality voting. This undoubtedly has been the inspiration for other ranking voting systems.
The disadvantage of the system is its flagrant manipulability or vulnerability to strategic voting which obscures the true preferences of voters in the outcome of a Borda election. That is, it pays Borda voters to rank the candidate who most threatens their top choice last even though otherwise they might prefer that candidate over all the choices besides their favorite. When this was pointed our to Borda he is reported to have said, "My scheme is for honest men." To this day the Borda count has been used in voting but only on a limited basis because of the problem of manipulability.
Condorcet's solution to the problem was both more sensible and impractical. In a muticandidate election Condorcet argued that the proper choice to represent voters' wishes was that candidate who could beat all other candidates in all of the pair-wise (one-to-one) elections possible from the field in such elections. There are, of course, two problems with this. First, in a multi-candidate contest, it may take an inordinate number of elections to determine such an outcome. Thus, for example, if you had five candidates for an office it could take up to ten one-to-one elections to determine a winner.
In the case of ten candidates, finding the Condorcet candidate could take up to 45 elections! A second problem with the criteria is that there are situations in which you get a Condorcet paradox or cycle in which no such winner emerges. An example would be the situation in pair-wise elections where A beats B in one, B beats C in another and C beats A.
Despite this the idea of the Condorcet candidate is used as a paradigm in judging how well alternative voting systems behave in determining the winner of an election. From this it would follow that a voting system that has a greater tendency to elect a Condorcet candidate in a multi-candidate election is preferable to one that does not.
In a situation in which there does not appear to be a Condorcet candidate, one would have to use some other criteria in judging how well a voting system performed in terms of representing voters'
preferences in the outcome. A candidate for this, which I would label a second best solution, is the election of a candidate who is acceptable to the largest number of voters in an election.
It is my contention that the idea of a Condorcet candidate, where one would exist, is a reasonable paradigm to use in judging how well alternative voting systems perform in determining a winner in an election. Where, as can be the case, there is not a Condorcet candidate in a multi-candidate election, then I would argue that we should revert to the second best mentioned above. Based on this and some other considerations, I maintain that the system of approval voting is the best system currently available to us.
Approval Voting
Approval voting is a very simple voting system which can be used in multi-candidate elections in place of plurality voting. In an approval voting election voters can give one vote each to the candidate or candidates they approve of with the candidate having the most votes winning. It is that simple. Approval voting has a number of attributes which make it a compelling system.
The first is that it totally gets rid of the wasted vote syndrome because a vote for one candidate does nothing to limit the possibility of voting for another as it would under plurality voting. It also negates the spoiler role since in an election like the 2000 presidential election in Florida a vote for Nader would not preclude voting for one of the other major candidates which is a necessary condition, besides not supporting them, of lessoning their chances.
Because of this, approval voting, in the absence of unreasonably restrictive ballot access, will open up our political processes to third parties.
A second desirable aspect of approval voting is that it encourages and allows voters to vote both sincerely and strategically at the same time. Because there is no wasted vote syndrome under approval voting, every voter always has an incentive to vote for his or her favorite candidate among those whom they approve. If a voter's favorite candidate has little chance of winning or a voter is indifferent between a number of candidates he or she can respond to this as well by casting additional approval votes. In short, approval voting significantly expands the expression of a voter's will, or preference, and hence its expression in an election's outcome.
A third desirable attribute of approval voting, as the technical literature has well established, is that it has a strong tendency to elect the Condorcet candidate where there is one. In the event that it does not elect one, it still comes up with a choice which I would qualify as at least a second best choice, something a number of other voting systems cannot necessarily do. The final desirable attribute of approval voting is that as a voting system it is both quite transparent in its results and is an easy system to administer and adopt in current conditions.
There are no perfect voting systems as work by Kenneth Arrow, the economics Nobel prize winner, and others have established. And there certainly are other voting systems which people might consider better than approval voting. In current circumstances there are two voting systems which need to be considered in addition to approval voting as an alternative to plurality voting. The first is instant run-off voting, or IRV. It deserves attention because it seems to be considered by many in this country to be the best replacement for plurality voting and is also the widest know voting alternative. The second is range voting which needs to be considered because it has been championed by Poundstone in his book Gaming the Vote and by Professor Scott Stevens in a very good video lecture on voting systems.
Instant Run-off Voting
Like the Borda count, IRV uses ranking of candidates to get its results. But instead of assigning scores to candidates, it simply tries to see if any candidate has a majority of the first place votes.
If no one does, the first place votes of the candidate who came in last are dropped and his or her supporters' second place votes are reallocated to the remaining candidates to see if a majority of votes can be reached for any one. If not, the next candidate with the lowest votes is dropped and his or her supporters votes for second place reallocated. The process goes on until a candidate gets a majority of the votes counted.
Supporters of IRV do not appear to understand that if you look at the technical literature on voting systems, there seems to be general agreement that the Hare system on which IRV is based "is really a very bad system." IRV has a number of problems which do not infest approval voting or infest it to the same degree.
Since it is a run-off system, IRV suffers from the problem which can occur in run-offs in that it may eliminate the candidate most acceptable to voters. Were, for instance, the 2002 French election for president confined to the top three candidates with the same relative proportion of votes between them as in the actual election, Jospin would have been eliminated, just as he was in the actual run-off, in an IRV election. Yet it was quite likely at the time that had Jospin faced Chirac in a one-to-one election, he would have won just as he would have in a one-to-one election against Le Pen. The
1970 New York US senate race was another example of a similar situation. Had the race among the three credible candidates; James Buckley, Richard Ottinger and Charles Goodell; been run under IRV between the three, Goodell would have been eliminated in the first round although based on polls at the time he was probably the Condorcet candidate and would have won under approval voting. The point of all this is that IRV is probably not as likely to choose a Condorcet candidate as approval voting.
But there is a far more serious problem with IRV as a voting system which many of its supporters do not realize--the monotonicity problem.
It is well established in the technical literature that it is possible for a candidate who would win an IRV election to wind up losing if he were to garner in the course of a campaign more first place votes.
This is a rather perverse voting result which does not even infest plurality voting. The reason that this can occur is that such changes in preferences can alter the order in which candidates are eliminated under IRV. In a November 2005 lecture at the University of Michigan Mathematics Department, Professor Steven Brams of New York University suggested that this can occur in about 10% of IRV elections.
Another disadvantage of IRV is that the system is flagrantly manipulable while approval voting is not. That is, under IRV a candidate can manipulate the outcome in his favor by how he encourages his supporters to rank lesser choices. Approval voting by its very nature contains no such possibility because it entails no ranking and thereby is an inherently less manipulable system.
There are a number of other problems with IRV which surely call into question its suitability as an alternative voting system. Given the difficulties of explaining how the system works and how it determines specific election outcomes, no one can credibly argue that it is a more transparent system than approval voting. In addition, it is inherently more difficult to integrate into our current structures for voting which Ferndale, Michigan voters learned after passing an initiative to institute IRV into its elections. Approval voting by contrast is easily integrated into our existing structures for voting since it involves nothing more than additional counting not unlike what occurs when you have significantly higher voter turnout as in the
2008 presidential election.
Despite these problems, the system of IRV is used in a significant number of elections, the most notable case being elections to the Australian House of Representatives. Proponents of IRV such as many members of the Green Party in the US frequently argue that the advantage of the system is that it will open up the political system to third parties. Supporters of Range Voting are quite vociferous in arguing that this is not the case and frequently cite the Australian experience. Indeed looking at the results of the last three Australian elections at the time of this writing, it does not appear that IRV has done much to move Australia away from two party dominance in its parliament.
Proponents of IRV tend to be quite avid in their support of the system. What they generally like about the system is that it permits them to express their preferences more completely by ranking the candidates. What they do not seem to realize is that in arriving at a result, IRV tends to deal with such rankings in a rather chaotic and arbitrary way such that its outcome does not really reflect voters'
preferences very well.
For those who rate ranking very high in voting systems, the system that Steven Brams labels fallback voting, a variant of approval voting, might be an alternative to consider. In this system voters are asked to rank the candidates to whom they would give an approval vote. If no candidate has a majority of first place votes then the approved candidates ranked second are added in. The process proceeds to the point when a majority is reached or the lists of approved candidates are exhausted. Since the process of adding approved candidates in proceeds according to ordinal ranking; first, second, third, etc.; one does not run into the monotonicity problem which infests IRV. While such a system might be an improvement over regular approval voting, it would not seem to me to make much sense to try to introduce such a more complicated system into our existing system of voting machines and the like. It might make sense to experiment with such a system at the local and state level.
Range Voting
Range voting is an alternative voting system which clearly has gotten more attention recently in part because of the attention given to it in Poundstone's 2008 book on Gaming the Vote. The system rests on an entirely different principle for judging voting systems than outlined in this essay. It assumes that the criteria for assessing the effectiveness of a voting system should be based on how much satisfaction results overall for voters from the results of a voting system. That is, it approaches voting from a perspective of trying to get the greatest satisfaction from a given result for the greatest number.
What the system involves is each voter assigning a score to each candidate on a scale like one to five or one to a hundred. Such scores are added for each candidate and then averaged to get a score to determine the winner. Proponents of the system have used computer simulations of assigned utility, or satisfaction, to try to show that this system gives the best results compared to systems like approval voting, the Borda count or IRV. As I explain in latter chapters, I have a number of reservations about this system.
As an economist, I object to their approach as I regard it as illogical because of the well stated view of most economists that the type of interpersonal comparisons underlying such a system are not valid. One concrete example of the difficulty of interpersonal comparisons is the problem of what is known as scale norming. Scale norming occurs in an election in a situation in which people assign on the scale used, such as a one to five range, different values for what are in fact the same subjective evaluations of the candidates. Under such circumstances, which are probably not uncommon, the resulting scores would be misleading guides to voters' collective evaluations of candidates.
An additional problem with range voting that Professor Hillinger pointed out is that it is quite subject to preference misrepresentation in terms of gaming the system to your favorite candidate's advantage rather than reveal your true preferences. This is not unlike what can take place in the Borda system. Thus, to promote my favorite candidate, I assign a higher score than my real judgement of him and assign to the others a lesser score than my real evaluation of them.
Given these considerations, I have serious reservations about the robustness and validity of the range voting approach to election systems as outlined later in this book.
Conclusion
Under current circumstances, it is my contention that a compelling case can be made for introducing approval voting into federal elections now. As I explain later, this can be done by a simple congressional statutory act. Given the Democrats' control of Congress and the White House after the 2008 election such an act could easily be enacted if the Democrats were truly committed to change we can believe in and really concerned, as they should be, about making changes in our political system to improve its functioning.
The focus of this book on opening up our political processes to third parties and voting reform does not mean that other political reforms are unimportant. On the contrary, such reforms also urgently need public discussion. But it is my contention that we need to start with voting reform because of my belief that changes in the methods of our elections will have the profoundest, and most beneficial, impact on how our system of governance operates. In addition to trying to promote discussion of this issue, I would like to see discussions of the following as listed in the order of importance I would assign to them.
1. Abolishing the presidential primary system and returning
presidential nominations to real party conventions.
2. Reducing the size and content of congressional acts with
more reliance in terms of carrying out their intent on
professionals within our executive branches.
3. Replacing many political appointees in the executive
branch below top cabinet levels with professional civil
servants as you have in Britain and Australia.
4. Restructuring our campaign finance system to serve
us better.
5. Amending the Constitution to make House terms four
years.
The materials which make up the rest of this book are op-ed pieces, memos and manifestos I have written on third parties and voting reform over the years and items from exchanges I have had on these subjects.
Some of the points made in the materials are date specific and may no longer be relevant to our current situation. Nevertheless, they are useful in understanding the evolution of the ideas which underly the themes of this book. In addition, there is in these texts some repetition which I have deliberately left in so that individual chapters can stand on their own which is meant to be useful for readers.