Normal
Some Thoughts on the Problems of Democracy
found at http://www.mercatus.org/
Gordon Tullock
March 15, 2002
Democracy is frequently referred to as a system of majority voting. Granted, the
last election in the United States the opposing candidate received more popular
votes than the winner, and neither received a majority of popular votes because of
the existence of other candidates, it is surprising that our system is called that." Of
course the winner did get a majority both in the Supreme Court and in the electoral
college. The loser got a majority in the Florida Supreme Court. 1
Why do we call it "majority voting" when . . .
The failure to get a majority in the popular vote is not particularly uncommon.
Lincoln, for example, got only 35 percent of the popular vote and if one of the
three of his opponents in the election, Douglas, had met him in a 2 candidate
election, he would have won. The 1912 election, once again, had a winner who
received less than half the popular votes. In Wilson's case he had two opponents,
either one of which could have beaten him if the other had not been present. The
current government in Canada received less than half the popular vote and, except
in wartime, no British government has been elected by more half the voters since
1920. In both of these cases, of course, the winner had more than half of the
representatives in Parliament. During its long reign in India, the Congress party
never received a majority of the popular vote. Normally it received less than 45
percent.
Another and interesting case is the election of 1960. Nixon had more popular votes
than Kennedy but lost in the electoral college. The fact that Nixon had more
popular votes, although not a majority of all votes, is almost a secret. 2
The election of 1960 is interesting and here I foreshadow the major part of this
article in that the two strongest candidates for the final election were eliminated in
the selection processes of the two parties. It is reasonably certain that Johnson
could have beaten Nixon in both popular and electoral votes and that Rockefeller
could have beaten Kennedy . All of this does not indicate the system is inferior, but
it does suggest that we stop calling it majority voting.
The problem of more than two alternatives
As on my readers will know, there are two general types of democracy, one
originating in England long ago in which the voters directly select individual
candidates. The other, which originated on the continent in the 19th century, is
called proportional representation and I will to a large extent leave it out of the
discussion below. Actually, I rather prefer proportional representation, but it is a
different subject and requires different analysis.
But to return to my main subject, the reader will have noticed that in each of the
cases with a winner who did not have a majority of the votes, more than two
alternatives were presented to the voters. With only two candidates or proposals
before the voters these problems do not occur. Unfortunately, there are many
people who would like to be the president and many ways which the government
income could be spent.
In the real world, it is likely that the system is confronted with more than two
alternatives and it must either cut them down to two by some means or be willing to
accept a candidate or proposition chosen by less than a majority. Of course it may
happen that although there are three candidates or proposals one of him gets more
than half of the votes. But we should not have a system which depends on that
chance.
The paradox: from Condorcet to Black and Arrow
This problem has been known for a very long time and procedures to either simply
take the one that has the most votes or reduce the number of alternatives to two
and then vote on the those two are orthodox. Unfortunately, none of these
procedures really overcome the problem. Shortly before the French Revolution
two French aristocrats, Condorcet and Borda, looked into the problem and
Condorcet found what is now known as Condorcet Paradox while Borda
produced system which avoided that Paradox but introduced another.
This was of course the period of the great expansion of democracy and, perhaps as
a result of the enthusiasm for democracy, the problem was largely forgotten
although some mathematicians seem to have known about it. In the mid-19th
century, Lewis Carroll rediscovered the paradox and did considerable work on it
without finding a solution. He too was largely forgotten and the problem was
rediscovered by Black who also discovered that he had predecessors lost in the
obscurity of minor mathematical publications.
Black succeeded in reviving interest in the issue which is a little paradoxical since it
is obvious he did not like the idea of democracy having paradoxes. Working with
Newing he produced a proof that this paradox could not the avoided if there were
more than two alternatives. 3 Notably he did not put out his work as a criticism of
democracy because he was a firm believer in democracy. Nevertheless careful
reading of his book with Newing shows the impossibility of avoiding the Paradox
except in special cases.
While Black and Newing were working, Arrow produced a general proof that the
paradox cannot be avoided in the general case. 4 This was his famous general
impossibility theorem. Notably it was only a remarkably long delay in the refereeing
process which prevented the Black and Newing paper from being published before
Arrow.
In the early days of research in public choice, papers dealing with the paradox
were a major component of papers submitted to me as editor of the journal. I think
that most of these papers were inspired by desire to avoid the paradox or at least
to minimize it. If this was the inspiration, it failed. The paradox continues and indeed
I have invented a much simpler although less elegant proof. 5
After considerable delay I discovered a proof that the problem was not
nonexistent, but of less importance than had been thought. If the voting preceded in
the usual way of legislatures with anyone free to introduce an amendment is, it
would precede to an outcome very close to the center of the cloud of individual
optima. A dictatorial chairman, however, could lead the votes to almost anywhere
he wished. Since the usual procedure is not involving a dictatorial chairman, this
meant that the voting normally would lead to a more or less satisfactory outcome.
Professor Arrow, in a very kind letter, accepted in the bulk of my reasoning but
pointed out that the latter part was not really mathematically strict. He was right,
but the reasoning was very strong even if not perfect, mathematicians standpoint. 6
What are fans of democracy to do?
The end result of all of this work is most disappointing for proponents of
democracy. Since the author and all of the readers of this paper are such
proponents of democracy we should all the unhappy about it. As far as I can see,
however, the usual response is not to be sad, but to sweep the problem under the
rug. Psychologically this is no doubt an optimal response, but it seems to me that
scholars should busily search for some better way of dealing with the paradox. I
frankly admit I have not found one, but the point of this article is to attempt to
interest other scholars in a revival of voting problem, so important in the early days
of Public Choice, but normally today not even mentioned in the average issue or,
indeed, in the average 10 issues.
This is particularly surprising granted that intellectuals are currently vigorously
producing arguments for democratic governments, and hence would be one would
think particularly interested in eliminating paradoxes in democracy. The new
arguments for majority voting, and the authors normally refer to majority voting
rather than simply democracy fall into two categories. The first is the allegation that
democracy's tend to produce capitalistic economies and hence are prosperous.
Dictatorships are allegedly less efficient in the economic field.
Democracy and economic progress
In the first place, economists have long known for (or at least many of them have)
that a capitalistic system is a better method and of producing prosperity than a
centrally directed one. In the last few years most intellectuals have adopted this
idea. Note however, that except for a certain number of "reactionary" economists
this idea was not popular among intellectuals 50 years ago. Warren Nutter
undertook a major research project in communist statistics and decided that the
rate of growth of the Soviet Union was not in any way remarkable. Indeed it was
about that of United States and markedly lower than that of Japan.
As a result he was practically drummed out of the economic profession. Note that
efforts to duplicate his research turned out to be surprisingly similar although the
CIA which funded them never overtly admitted that. It was however not just the
CIA which disagreed. To repeat, Nutter's career was more or less terminated in
the economic profession. The University of Virginia did not give him significant pay
raises and moved him to an inconvenient office. He could not go anywhere else
because other universities also thought that he was following his ideology and not
his science.
This was a personal tragedy for Nutter and a policy tragedy for many governments.
The view that economic progress required central control by someone not
hampered by democratic mechanisms was widely held. Many of the people who
felt this way remained in favor of democracy for reasons other than its economic
effects. Nevertheless, the recent view sweeping intellectuals that democracy works
better in economics than a communist dictatorship is recent. Further, as far as I can
see, like the earlier enthusiasm for communist dictatorships in the economic field, it
was based on nothing more than one of the waves of enthusiasm that tend to sweep
the intellectual classes, although the collapse of the old order in Russia undoubtedly
contributed
If we look at the real world, Japan, Germany, France, and India are all
democracies and doing very badly in their economies while Singapore and Hong
Kong continue to be prosperous with dictatorships and China, a clear-cut
dictatorship and a rather impressive one, currently claims the highest rate of growth
of any significant country. Speaking for myself, I tend to distrust communist
statistics, but observation on a recent trip to China indicates it is very much better
off than it was not too long ago, at the end of the great proletarian cultural
revolution.
Democracy and peace
Another recent argument for democracy is that democracies are relatively peaceful.
This began with view that democracies did not engage in aggressive wars.
Apparently, after a while, some intellectuals read the history of 19th-century in
which democracy's conquered much of the world, and the claim was reduced to
one in which democracy's do not fight other democracies. The original claim was
particularly surprising since many of the intellectuals who made it were American
citizens and therefore should have been aware that United States had seized is
present geographic area by a series of minor but certainly aggressive wars.
Consider the reduced claim that democracies rarely fight other democracies.
During a rather long period in which democracies were rare and hence had little
opportunity to fight each other this was undoubtedly true. Still it should be kept in
mind that in 1914 it could easily have been argued that both England and Imperial
Germany were democracies, indeed the major undemocratic country in the war
was Russia. Of course at that time all the monarchies were moving toward
democracy and England and Germany were much farther along than Russia. The
end product of the war was that what progress Russia had made towards
democracy was canceled and one of the world's worst autocracies put in its place.
The dissolution of Austria-Hungary was disastrous by any standard. The history of
the fragments was complicated, but on the whole a setback for democracy. Since
shortly after the war Italy became a dictatorship, albeit a rather mild one as
compared to Russia, it is not very obvious that the allied victory expanded
democracy.
World War II had a very nasty dictatorship, the Soviet Union, on the same side
with United States and England. Indeed it did most of the fighting. Japan had an
elected legislature and the Emperor's powers were very modest. It seems likely that
they should be regarded as a constitutional monarchy rather than as a dictatorship.
The end product of the war was of course the great expansion of the Soviet
dictatorship so that Europe in 1945 was less democratic than it had been in 1930.
Once again to refer to the war as a war for democracy is misleading
Better alternatives?
I have been presenting all of this material which could be regarded as an attack on
democracy, not in order to run down democracy but indicate that the enthusiasm
for democracy in the last 10 years and for communist dictatorship in the early '50s
were simply examples of the instability of intellectual opinion. Personally I prefer
democracies, but I must admit that the arguments for a democratic state are much
weaker than those for a capitalistic economy. Further, when you look around the
world you realize that Prince Bismarck's invention, the welfare state is largely
dominant in democracies. Since the long run prospects of that system are poor, this
could be an argument against democracy. It is however quite possible to feel as I
do that in spite of its defects, democracy is better than the currently known
alternatives.
So far this paper has not had any dominant theme except that it shows little
enthusiasm for democratic methods of government. I have to admit that I am not
enthusiastic about democracy. Like Churchill, I favor it over its current
competitors, but it seems to me that we should be looking for something better.
The purpose of this note is first to deal with certain popular arguments for
democracy, which I think are false, as a preliminary to encouraging my listeners to
look for new forms of government.
I find when talking to people about improvements in government the first response
with respect to any idea is to ask, "Is it more democratic than our present
methods?" If it were it would have the same defects. It could be, however, that
there is another form of government which has all of the advantages of our
democracy but adds something on to it. I have no suggestions at the moment but I
believe that seeking a better form of government then democracy is sensible policy.
It may be of course that the better form is a modification of our present democracy.
If we look around the world we observe some democratic governments that are
more successful than others. In my opinion the Swiss government is not only the
best democratic government in the world but also the best government. Thus
copying it would be an improvement even though I am great admirer of our
Constitution. Melding the two might be an excellent idea. But I hope that the public
choice scholars can do better.
But let me return to my main theme which is problems with our present democracy.
I will begin by once again turning to either the Arrow theorem, or the Black
Newing proof that the voting process has paradoxes.
Back to the voting paradox
Roughly speaking, simple majority voting works very well if they're only two
alternatives. When there are more the outcome may be close to random. In
general, in democracies there are more than two people who would like given job
and they're more than two policy suggestions for any given problem. In most
functioning democracies these multiple choices are winnowed down to two which
are then voted on. It is not obvious that one of the alternatives eliminated in the
winnowing down process could not get a majority over whoever wins in the
election limited to the two survivors.
Of course the winnowing down process does not always get the final choice
pattern down to two alternatives. I above mentioned cases in the American system
in which no one got a majority because there were more than 2 alternatives which
attracted significant votes. If one looks at policy choices we once again find a
number of alternatives offered which are then winnowed down to two by
successive votes in accord with the procedural rules in use in that particular voting
body. Granted the prospect of paradox, it is by no means obvious that one of the
other alternatives which lost earlier could not be the winner selected by the
operation of the rules of order.
To take a simple and artificial set of choices, suppose that the various alternatives
actually offered as amendments in the Senate are: A, B, C, D, E, and F. Note that
these are only a set of amendments offered on the floor of the Senate. If we
considered those offered in committee, the set of amendments to the House version
and what was done in the conference committee, probably at least 25 or 30
possible variants are proposed at one or another stage in the process. One of these
numerous alternatives could be capable getting majority against the winner in the
formal vote. Suppose for example that the votes are taken F against E, the winner
against D. and so forth. This leads to B winning, but that B as never been offered
against E, so it's perfectly possible that E could beat B but not beat D.
This is simply an example of the standard possibility of circular majorities. We have
no real measure of how often it happens. The first effort to find out how often was
a joint article by Colin Campbell and myself which generated random preferences
for a large body of synthetic voters in the computer memory and then tested them
for cycles.
The method is obviously crude but the only improvements on our design have been
based on the same method but with larger collections of random numbers assigned
to voters. Obviously, we need improvements but it is clear that the cycles are
reasonably common. In this paper, I will make no effort to find out how common
they are. I simply assume that they sometimes happen, and when they do, the
outcome is not one we should respect. Of course, if they were rare we could
regard the possibility as a minor defect, and go ahead. If they were common they
would not be a minor but a major defect. Unfortunately we do not know.
The effect of the paradox on Congressional elections
But let us go on to the candidates selected by election. In addition to the paradoxes
mentioned above, in the United States there is another very serious problem. In the
2000 election year, 31 members of the House of Representatives chose not to run
-- presumably mainly in order to retire. Of the remaining to 394, all but 11 were
re-elected. It is sometimes said that the security of tenure held by members of the
House of Representatives is greater than that held by members of the House of
Lords.
The basic reason for this security is that the members of the House are able to get
the federal government to expend great resources for their re-election. They have
large staffs, considerable free travel, access to a free television studio in the
basement of the capital, and many other advantages. There doesn't seem to be any
exact accounting for this money, but I asked a professional lobbyist how much he
thought it was worth and he gave a figure of two and half million per congressman
per election. It's obvious why they are so secure and obvious why they want to
restrict campaign expenditures by potential opponents.
Senators do not have as firm a grasp on their offices as the members of the House
and a President probably even less. In the case of the President resources spent by
the federal government to get him re-elected are immense. It is probably true that
resource expenditures have the highest payoff when there is little other information.
Thus, the value of these large government expenditures would be highest for the
House. It should be noted that in the 19th century when these expenditures were
not as large congressman normally served only one term.
This leads us perform a mental experiment. There must be several thousand people
who would like to be President and who think at least a little bit of running. Most of
them drop the idea almost immediately, but some will give it further consideration
and talk to people about the possibilities. At this stage we're down to something
like the 50 possible candidates. They "test the water." There was a senator
unknown to me who recently visited Washington to a give a speech to an
organization which he thought might support him. The Washington Post reported
him as " testing the water" and said he received an enthusiastic response. This put
him a bit above many of the above 50, perhaps in the top 10 or 20.
At this point he would begin trying harder to get support from one of the parties,
from special-interest groups, etc. he would also begin trying to raise money.
Eventually he would either be nominated or not. Whether he became President
would depend on who else was nominated and, in particular, how many were
nominated. The prospect that somebody who could beat the eventual winning
candidate was eliminated early in the game is certainly good. In other words the
paradox exists here as well as in the choice of issues. Indeed it appears to even
stronger here.
Some suggestions for improvement
This rather lengthy essay has raised a number of difficulties of a fundamental nature
with democracy. They do not prove that democracy is inferior to any other given
system, but they do indicate that we should substitute hard research for enthusiasm.
In recent years there's been very little effort to solve these problems. I would like to
encourage my audience to go back to the fundamentals.
As a sort of study aid for people looking for improvements in the structure of
government I should like to list some suggestions that already been made. This is
not intended to be a complete list of possible changes, but only a set which have
been made but have received little or no attention. I think they should be given
more attention by students in this field, but I hope that someone will invent even
better ones.
I begin with Clarke's demand revealing process. This has had more attention than
the others including a complete issue of public choice devoted to it. As the reader
may know, I have done some work in this area and am an enthusiast for the
system. It permits the voter to express not only which alternative he prefers, but
also how much he prefers it. The outcome might lead to a minority of intense voters
defeating a majority of near indifferent voter.
My second proposal is Earl Thompson's suggestion to put individual policy changes
up, in essence, to a bid. Among other things this permits compensation of the
losers. The third proposal is Hanson's betting procedure in which the voter may be
rewarded for favoring a decision for certain proposals if ex post an impartial
commission finds that it adds to the national welfare.
All three of these methods are radically different from our current voting
procedures. Indeed, it could be argued that they're not really voting at all.
Nevertheless they are a start. Note that all three permits a minority to win over a
majority. Also, all three of these suggestions are subject to the Arrow, Newing and
Black problem. If there are more than two alternatives, the order in which they are
taken up may lead to different outcomes. Voting on all alternatives at the same time
may also lead to different outcomes. As in other cases where these problems arise,
voting on the order of taking them up to also leads to paradoxes. Still these are a
start in investigating radically different ways of making decisions, and they're not
monarchical.
Alternative voting procedures
These are variants on the actual voting procedure. Other possible changes involve
the problem of who can vote. Dennis Mueller suggested that voters to given a short
examination before being permitted to vote. The intent, of course, is to restrict the
voting to people who at least know some elementary facts about the government. I,
myself, perhaps influenced by my knowledge of China, have suggested an
examination for the candidates. Following the Chinese precedent the examination
for different offices would be of different degrees of difficulty. A man who would
fail the president's exam might well pass that for alderman. Again following the
Chinese precedent the examination would leave several candidates for each office
to be selected by the voters. They would remain in control but the number of
alternatives would be reduced. If the examination were well-designed the
alternatives would also be improved.
It is not obvious that one man one vote is ideal. Corporations give different people
different votes depending on how many shares they own. As an aside, this system
was originated by Lord Clive. He not only started the British Empire in India by his
victory at Plassey, but he also, in an unsuccessful effort to control the Honorable
Company, caused changes from one man one vote to one share one vote. It's not
at all obvious which is the best system. I find, however, that most people are
strongly opposed to, let us say, the one vote for every dollar in taxes paid. Whether
this is merely opposition to a new idea or a rational position, I do not know. When
I talk to people about it their objections normally are not well reasoned. Indeed
they normally offer no reason at, all, merely opposition.
There are other proposals for giving different people to different numbers of votes.
Nevil Schute wrote a whole novel, In the Wet, devoted to this idea, and one of the
richest men in United States - Hunt -- also produced a book on it. The additional
votes could be distributed in terms of payment for services. For a modest example,
suppose any war veteran who actually got shot at gets an extra vote.
Another radical idea is to permit people to hold their votes in abeyance. I could, for
example, decide not to vote for president in 2004 and then cast two votes in 2008.
Or perhaps I could have credit, testing two votes in 2004 and none in 2008.
Perhaps to charge interest, I might be prohibited from casting votes for Senators in
2008.
Undemocratic governments and conclusion
But let us consider some undemocratic ideas. Gibbon thought the period of the
adoptive Emperor's in Rome was the happiest in human history. Mexico used a
somewhat similar system from 1931 to 1987. It also had a reasonably good
government by the rather moderate standards of Mexican government's. For a list
of other governments which are not democratic, I suggest my forthcoming
"Undemocratic Governments".
The reader should keep in mind that it is not necessary for the government form to
be suitable for a nation state. It may be suitable only for use in individual
governments which form only a part of the nation. If these were undemocratic, they
would nevertheless be subject to popular control because of the possibility of
moving. Note that some government functions cannot conveniently be broken up.
In military matters economies of scale require large government units. There are
other areas where breaking up government is unlikely to be successful. As an
obvious example consider the regulation of the electromagnetic spectrum. No
doubt the reader can think of many others. Nevertheless, pre-1870 Germany
seems to have been pretty well governed although none of the constituent
monarchies were democratic.
I do not claim that any of the alternative governments I have listed is ideal.
Nevertheless, I think they should receive careful study. Further, I think we should
look for other forms of government. There is no natural law which says we cannot
invent new forms of government which are better than the current sample. It is to
encourage the reader to engage in such radical thought that this essay is dedicated.
This Web Page Created with PageBreeze Free HTML Editor / Web Hosting