Futarchy: vote on values, bet on beliefs

[archerships]

http://hanson.gmu.edu/futarchy.pdf

Page 1
Shall We Vote on Values, But Bet on Beliefs?
Robin Hanson
Department of Economics
George Mason University
September 2000
Abstract
A key question to ask about any social institution is how well it generates, aggregates, and
distributes information. Speculative markets seem to do well at this, while familiar democratic
institutions, relying in part on academic institutions, seem to fail in many ways. So perhaps we
should consider "futarchy," a form of government where betting markets become our primary
common source on matters of fact. Democracy would say what we want, while speculators
would say how to get it. That is, elected representatives would formally define and manage
an after-the-fact measurement of national welfare, while market speculators would say which
policies they expect to raise national welfare. If we are willing to recommend policies that
macroeconomic data suggest are causally related to GDP, it seems we should be willing to
consider futarchy. Using an qualitative engineering-style approach, this paper considers thirty
one design issues with futarchy, and then presents a relatively specific proposal which responds
to those concerns.


For their comments, I thank Kathryn Aegis, Tom Bell, Peter Boettke, Nick Bostrom, Tom Breton, Damein
Broderick, James Buchanan, Bryan Caplan, Roger Congleton, Tyler Cowen, Hal Finney, David Friedman, Amihai
Glazer, Karl Hallowell, Bernardo Huberman, Craig Hubley, Peggy Jackson, Ken Koford, Hassan Masum, Peter
McCluskey, Jim McKinney, Eli Lehrer, Nicolaus Tideman, Karen Vaughn, Eliezer Yudkowsky, Richard Zeckhauser,
and participants of the Monomedia Berlin: Value conference and the 2000 GMU Public Choice Outreach Seminar. I
thank the Mercatus Center for financial support, and Edward Stringham for research assistance.
rhanson@gmu.edu http://hanson.gmu.edu 704-993-2326 FAX: 704-993-2323 MSN 1D3, CarowHall, Fairfax VA
22030
1
Page 2
Introduction
A fundamental social problem seems to be this: which social institutions are better at generating,
aggregating, and distributing knowledge? This is a key question at three levels: economics, politics,
and abstract information.
1
That is, this question seems to be a key issue for understanding which
economic institutions do better, which political institutions do better at regulating them, and
which abstract information institutions, like academia, are better at informing ordinary citizens
about economic and political institutions.
If we call this topic "social epistemology," then we should want our position on social episte-
mology to be consistent across these three levels. For example, a standard position is that current
economic institutions do best, that current political institutions best regulate them, and that current
academic and media institutions reliably tell people these things. A common alternative position
is that academia, while usually reliable, has misestimated the harms that democratic interventions
inflict on economies, a mistake that advocates within academia will eventually overturn.
This paper will explore another position, that betting markets are in many ways superior to
familiar academic-style institutions for aggregating and distributing knowledge on many important
topics, both specific and abstract. If so, perhaps those who disagree with the current academic
consensus on political institutions should work to have betting markets displace such academic
institutions, at least in part, as sources that ordinary people rely on for evaluations of political
institutions.
This third position also suggests that we give betting markets a more direct role in the way
political institutions evaluate the promise of economic interventions. While there are some more
limited ways in which this could be done, this paper will explore an extreme proposal, to see just
how far we can take this idea. After reviewing the centrality of information institutions and how
democracies may have failed while betting markets have succeeded as such institutions, this paper
will present an "engineering-style" proposal for a new form of government called "futarchy," and
will discuss thirty one issues and objections to the concept.
In futarchy, the public would vote on values, but bet on beliefs. The democratic process would be
limited to managing an after the fact measurement of "GDP+", a measure of national welfare that
they would define. The basic rule of government would be that when a betting market estimates
that some proposed policy would increase expected GDP+, that proposal becomes law. Democracy
would still tell us what we want, but betting markets would tell us how to get it. If we are willing
to recommend policies that macro data suggest are causally related to GDP, it seems we should be
willing to consider futarchy.
1
In this paper, "knowledge" and "information" are treated as synonymous.
2
Page 3
Information Institutions
The peculiar character of the problem of a rational economic order is determined pre-
cisely by the fact that the knowledge of the circumstances of which we must make use
... exists ... solely as the dispersed bits of incomplete and frequently contradictory
knowledge which all the separate individuals possess. ... Which of [central planning or
competition] is likely to be more efficient depends mainly on whether we are more likely
to succeed in putting at the disposal of a single central authority all the knowledge
which ought to be used but which is initially dispersed among many individuals, or in
conveying to the individuals such additional knowledge as they need in order to enable
them to fit their plans in with those of others. (Hayek, 1945).
Economic theorists today wholeheartedly accept Hayek's once-revolutionary claim that an un-
even distribution of knowledge, now called asymmetric information, is the key problem in social
systems. They also accept that policy interventions are mainly constrained by a very wide dis-
persion of relevant knowledge, and by the fact that those who know may not want to tell. Most
even accept Hayek's claim that unregulated competition is better than pure central planning. Most
economists today do not, however, accept Hayek's stark choice being these polar extremes. Hayek's
argument that our choice is effectively binary was based on human cognitive biases toward the seen
over the unseen:
When we decide each issue solely on what appear to be its individual merits, we always
over-estimate the advantages of central direction ... [Thus] freedom can be preserved
only if it is treated as a supreme principle which must never be sacrificed (Hayek, 1973).
Most economists, however, are not persuaded that they suffer from such severe cognitive biases.
2
Instead, most economists now study what can go wrong in economies with asymmetric infor-
mation, and how they might best intervene given their knowledge constraints. That is, they study
how to pursue policy goals when one understands the overall structure of some policy area, but is
ignorant of many details. And few economists believe that the exact best policy is always exactly
zero intervention.
3
An institutional critique of this perspective is that our choices are much more limited. Ordi-
nary citizens can not choose from all possible intervention policies; they can at best only choose
from feasible stable institutions, which will then choose more specific policies.
4
The question then
2
Many economists instead argue that cognitive biases are against government intervention.
3
One might deny that we can know much about the overall structure of an area of life. But at best this argues
for using more abstract economic theory, which makes fewer specific assumptions.
4
A "constitutional" critique appears to go further, arguing that we can at best choose among constitutionally-
defined forms of government, which will then choose more specific government institutions.
3
Page 4
becomes which institutions, including a "no intervention" institution if that is feasible, produce
better policies on average.
Public choice critiques of government intervention argue further that familiar high-intervention
democratic political institutions suffer greatly from problems like rent-seeking, coordination failures,
commitment failures, adverse selection, and especially low incentives for voters to become informed
(Holcombe, 1985). Defenders of democratic interventions, in contrast, suggest that such problems
seem to be manageable in practice, as rational agent theory suggests that they should be (Wittman,
1995).
Whichever side has the better arguments, such arguments seem too complex for ordinary citizens
to evaluate personally. So most citizens must again choose institutions, instead of specific policies.
That is, ordinary citizens with little time to directly study political institutions must instead defer
to some indirect information sources for answers to such abstract questions. And since citizens also
have little time to evaluate the reliability of information sources, a handful of such sources must
serve them for all such topics. Thus at the abstract information level, citizens must again typically
choose between a few broad long-lived institutions, perhaps as specific as particular academic
disciplines or news media, but more likely academia and mainstream news media taken as wholes.
Academic opinion on the effectiveness of political institutions has for many decades seemed to
lean more toward democracy's defenders than its detractors. Mainstream media has on the whole
leaned the same way. Critics who think democracy results in too much intervention, as well as
those who think it results in too little, often explain this mistaken judgment as due to various
institutional failures in academia and news media, failures which reduce the informativeness of any
current consensus (Yeager, 1997; Rosen, 1997).
Many academic studies do in fact suggest real failures in academia-style institutions (Redner,
1987), such as universities, research labs, and expert advisory committees. In statistical studies,
these failures include overconfidence in variance estimates (Henrion & Fischhoff, 1986), strong
unacknowledged selection biases (Long & Lang, 1992), and very high error rates (Dewald, Thursby,
& Anderson, 1986). In peer review, failures include very low levels of agreement (Chubin & Hackett,
1990), strong biases against obscure institutions (Peters & Ceci, 1982), and biased objections to
methods that give disliked results (Mahoney, 1977).
Critics, however, mostly continue to work to change academic opinions, apparently accepting
academia as the appropriate forum for such debate. (Even Hayek granted that "a body of suitably
chosen experts may be in the best position to command all the best [scientific] knowledge available"
(Hayek, 1945).) The alternative information institutions that critics offer, such as specific factions
of academic disciplines (e.g., Austrian economics) or specific "ideological" news media, seem far
too small to gain a reputation among ordinary citizens as reliable information sources.
If the currently dominant information institutions suffer from serious failures, however, critics of
4
Page 5
standard views on political institutions might do better to promote better information institutions.
After all, even if they succeed in changing academic consensus, their criticisms of academia could
then come back to haunt them. In contrast, they should expect widespread acceptance of better
information institutions to make their political views more likely to be widely accepted. They
should also find common cause in creating better information institutions with those who hold
other non-standard views. Finally, since economic and political institutions are in part information
institutions, better information institutions may also directly allow better economic and political
institutions. This possibility should be particularly interesting to those who think current political
institutions do poorly as information institutions.
To put it another way: a fundamental social problem seems to be social epistemology. Episte-
mology has traditionally been concerned mostly with the conscious strategies that a rational person
should use when deciding what to believe given primary sources of evidence. Most real problems
in deciding what to believe, however, concern how the social institutions people are embedded in
influence who we should believe. Social epistemology was first defined as,
How should the pursuit of knowledge be organized, given that ... knowledge is pursued
by many human beings ... each equipped with roughly the same imperfect cognitive
capacities ...? (Fuller, 1988)
For our purposes, we can take social epistemology to be the study of how effective various social
institutions are at generating, aggregating, and distributing knowledge and information. This seems
to be the central subject of economics, politics, and of abstract persuasion regarding such topics. We
should prefer a reflexive position on social epistemology, i.e., a position consistent across economics,
politics, and the crucial subject of why ordinary people should believe this abstract position.
For example, a standard roughly-consistent position is that decentralized markets do well as
information institutions, that they do even better with limited regulation, that familiar democracy
does sufficiently well as an information institution to provide such regulation, and that academia
is reliable enough as an information institution for ordinary people to rely on it to answer abstract
questions about economics and politics. This position considers academia reliable enough to be
given important political roles in deciding policy.
Another position, with perhaps more internal tensions, is that while academia is the most
reliable known information institution which ordinary citizens can turn to for answers to abstract
questions about economics and politics, it is not reliable enough to be given a large role in deciding
policy, and it has made a serious error on the subject of democratic interventions into the economy.
This position suggests that the academic minority that knows of this error should continue to push
academia to correct it, and that until they succeed little else can be done to change the opinions
of ordinary people.
5
Page 6
Several other common positions could be described, but this paper will instead present a new
position, that another general information institution is superior to current academic and news
media institutions at many aspects of delivering policy-relevant answers to abstract questions, that
the relevant academic research endorses this conclusion
5
, and that the reason ordinary citizens have
not yet been persuaded of its robust reliability is both that technology made it very cheap only
recently, and that regulatory error has kept it effectively illegal. This regulatory error is of a sort
one expects to be common, a broad activity ban intended to reduce a "seen" problem, which then
prevents an "unseen" new industry from developing.
This alternative institution is betting markets, which are decentralized markets that avoid many
of the perceived disadvantages of markets in general. Betting markets can not only be relied on to
tell us what sorts of economic interventions will get us the things we want, but can also be formally
embedded in our political institutions in order to directly and officially fulfill that role.
One refinement of this position says that economic interventions are almost never a good idea,
and that a better abstract information institution will reveal this fact. Another refinement, however,
says that a better information institution will reveal that we want even more intervention than
we see now. Both of these positions have a common interest in promoting a better information
institution.
Information Failures of Democracy
If familiar political institutions were effective at producing informed policy choices, there would
be relatively little reason to seek better information institutions, either to imbed within political
institutions, or to inform ordinary people about the effectiveness of their political institutions. So
how well-informed is current democratic policy?
Half a century ago, empirical research on individual U.S. voters seemed to confirm the worst
fears of democratic skeptics.
Most people made up their minds long before the election ... few citizens paid much
attention to politics ... Even on important issues such as government help with jobs,
aid to education, or the stationing of American troops abroad, large proportions of the
public did not know what the government was currently doing, where the opposing
parties stood, or even what they themselves wanted to government to do ... less than
20% ... had `real and stable' attitudes on ... electric power and housing ... (Page &
Shapiro, 1992).
5
This position thus accepts academia as a reasonable forum in which to make the argument that another institution
is better than academia.
6
Page 7
Such high levels of ignorance continue today (Delli-Carpini & Keeter, 1989). For example, only
29 percent of U.S. adults can name their congressman, and only 24 percent can identify the first
amendment.
Formal analysis found many problems with democracy as well, including instability, rent-
seeking, coordination failures, and commitment failures (Holcombe, 1985; Besley & Coate, 1998).
Formal models did not, however, as easily validate what many considered the most serious problem:
ignorant policies due to low incentives for voters to become informed (Wittman, 1995).
In theory, it is enough to have citizens just vote selfishly and retrospectively, and to have policies
centrally determined. That is, if it is clear who is responsible for good policy, and if citizens voted
for incumbents only when their personal lives seemed better than expected, then citizens would
not need to understand much about abstract policies. Incumbents would then have the power and
incentive to make voters feel good about their lives. Real voters, however, do not in fact seem to
vote very selfishly (Sears & Funk, 1990). Thus in order to get informed voter-driven policy, we
seem to need most voters to obtain information about broader policy consequences.
In this case ignorant voters, such as those who do not understand the consequences of import
quotas, can indeed induce bad policies, such as by voting for politicians because they support
import quotas. Democracy seems reasonably capable of overcoming such ignorance, however, at
least in theory given rational voters. After all, voters learn many things as a side effect of just
living, large elections average out random errors by individual voters, and political entrepreneurs
can take the initiative to inform voters via advertising. Also, a small fraction of informed citizens
can determine elections if uninformed citizens either abstain, infer what the informed know from
opinion polls, or trust political parties and other sources regarding how to cast their votes.
More recent empirical work has similarly found less to complain about in voters.
Collective public opinion is rational [meaning] ... real, stable, differentiated, consistent,
coherent; reflective of basic values and beliefs; and responsive (in predictable and rea-
sonable ways) to new information and changing circumstances ... On most domestic
matters, about which elites often compete and provide multiple sources of information,
the public can ... form ... opinions that approximate fully and correctly informed
preferences (Page & Shapiro, 1992).
This is not to say there are not information problems with democracy, however. The above
quote goes on to say,
In foreign affairs, on the other hand, government monopolies of information (and con-
sensus among elites) may sometimes lead the public astray from preferences it would
hold if fully informed (Page & Shapiro, 1992).
7
Page 8
% In U.S. Agree With This Opinion
Cite
47 God created humans in basically present form in last 10,000 years (Gallup, 1999)
52 `astrology has some scientific truth'
(Davis et al., 1996)
72 believe in angels
(Gallup, 1998)
80 `US government is hiding that it knows of the existence of Aliens' (CNN, 1997)
85 `Jesus Christ was born to a virgin'
(Group, 1994)
Table 1: Contrarian Public Opinions
And in fact these authors goes on to claim that the U.S. public was lead far astray regarding World
War II and the Cold War.
Democracy can also suffer from information failures due to long delays in information getting
to many people. Such delays, or irrationality, can plausibly explain the fact that time series of
public opinions do not tend to look like random walks. Even on questions of fact such as the risk
of a nuclear power accident, it seems that one can predict future average opinions from trends in
past average opinions (Page & Shapiro, 1992).
More relevant evidence comes from many specific contrarian public opinions, such as those listed
in Table 1. These opinions are not just cases where the public is ignorant of expert opinion, such
as they might be on foreign trade. Instead, these are cases where the public seems largely aware
that expert opinion disagrees with them.
Contrarian public opinions suggest not only ignorance or a failure to share information, but
also irrationality more directly. In theory, rational agents with common priors, who would have the
same estimates were it not for having different information and computation, should use the fact
that someone disagrees with them in a certain direction to update their beliefs. As two rational
agents alternate telling each other their new opinion, each one should not be able to predict in
which direction the other will disagree with them next time, and the only stable endpoint of this
process is to eliminate such disagreements (Aumann, 1976; Hanson, 1997). Yet people seem to
persistently disagree on factual claims, especially on political topics, even when these people seem
well aware of the directions that others disagree. This lends some support to theories that try
to explain otherwise puzzling policies and political behavior in terms of weak but positive voter
preferences for irrationality (Caplan, ).
Of course misinformed and irrational voters need not prevent informed policy, if voters allow
policy to be determined by informed elites such as academic advisors. In many areas, however,
such as tariffs or immigration, policy often seems closer to what public opinion would suggest than
to what relevant experts advise (Dixit, 1997; Stiglitz, 1998). This suggests that the public does not
defer to experts in many areas. Furthermore, many case studies suggest that when governments do
use academic experts, they do so frequently for legitimation of predetermined policies, rather than
8
Page 9
for information to help determine policy (Barker & Peters, 1993).
Even when democratic policy does defer to academic experts, the failings of academic-style
institutions may lead to poorly-informed policy. In academic-style policy information institutions,
such as government research labs, advisory agencies, or advisory committees, experts simply declare
their policy advice, without much in the way of clear or direct incentives to be right. Even when
such experts are selected on the basis of their "academic reputation," this basically means they
are selected by other experts, and those other experts again have little clear or direct incentive to
choose those who would offer accurate policy advice. (Data suggesting real failures in academia
was mentioned in the last section.)
If familiar political institutions suffer from information failures, how large are those failures?
The huge variation in economic growth rates across nations, depicted in Figure 1, suggests that the
effects of information failures may be very large (Maddison, 1995). There are surely some random
and uncontrollable reasons why some nations are rich and others poor. An important fraction of
the variation, however, seems attributable to some nations adopting policies which relevant experts
knew to be bad, and thereby becoming poorer than nations which adopted better policies (Olson,
1996; Sachs & Warner, 1995; Ascher, 1999). Since the subgroup of democratic nations also have
large variations in growth rates, democracies also seem to have serious failures to adopt good
policies.
Many factors contribute to bad policy, including commitment and coordination failures. But
it seems hard to imagine that nations would adopt bad policies nearly as often as they do if it
were common knowledge that such policies are bad. Thus at some level bad policy seems to be
fundamentally due to a failure to aggregate and distribute relevant information.
Information Successes of Speculative Markets
While democratic policy seems to suffer from serious information failures, speculative markets have
shown some dramatic information successes.
Most markets for stocks, bonds, currency, and commodities futures are called "speculative
markets" because they allow people to bet on future prices by buying or selling today in the
hope of later reversing such trades for a profit. Such opportunities for speculation occur when
identical durable items are frequently traded in a market with low transaction costs. Given such
opportunities, everyone is in essence invited to be paid to correct the current market price, by
pushing that price closer to the future price. Such invitations are accepted by those sure enough
of their beliefs to "put their money where their mouth is," and wise enough not to have lost their
money in previous bets.
"Betting markets" are speculative markets which trade assets that are specifically designed to
9
Page 10
Figure 1: Log GDP Per Capita vs. Year For 56 Countries
allow people to bet on particular matters of fact, such as which horse will win a race. The final
values of such assets are defined in terms of some official final judgment about the fact in question.
By construction, such assets are durable, identical, and can be created in unlimited supply.
Betting and other speculative markets have been around for many centuries, and for many
decades economists have studied the ability of such markets to aggregate information. The main
finding of this research is that such markets tend to be relatively "efficient" in the sense that it is
hard to find information that has not been incorporated into market prices (Lo, 1997; Hausch, Lo,
& Ziemba, 1994). The main apparent exceptions seem to be long-term aggregate price movements
in real asset markets, and a long-shot bias in high-transaction-cost betting markets.
Compared to what seems rational, asset markets seem to have too much long-term aggregate
price variation, such as stock market "bubbles" (Shiller, 2000). These price movements are where
risk and delay most reduce the extra returns to speculators for correcting pricing errors, and where
theory says that "noise" traders can actually gain superior returns (though not utility) from their
irrationally-large risk-taking (Long, Shleifer, Summers, & Waldmann, 1990). Long-term aggregate
prices, however, are also where it is hardest to empirically distinguish irrationality from rational
shared information about changes in the nature of the economy (Barsky & Long, 1993), and where
selection effects most pollute the available data (Jorion & Goetzmann, 2000). So it remains unclear
10
Page 11
just how irrational long-term aggregate price movements are.
Speculators have a much stronger incentive to eliminate price biases that vary over shorter terms
and that are independent of the few aggregate price dimensions that command a "risk-premium."
High transaction costs, however, can deter speculators from eliminating such biases. Parimutuel
betting markets, for example, typically charge about an 18% transaction fee, and also typically
display a "long shot bias," over-estimating the chances of unlikely outcomes (Hausch et al., 1994).
This bias occurs because those who favor a low probability outcome need to spend proportionally
less money to have the market price reflect their views, and transaction costs deter speculators
from correcting this bias.
Over the last few decades economists have also studied speculative markets in laboratory experi-
ments, where they have more control over trader information and preferences. The name "Hayekian
hypothesis" has even been given to the roughly confirmed thesis that speculative markets aggre-
gate information well even when traders know little about their environment and other traders
(Sunder, 1995). Experimenters have found, for example, that traders can aggregate information
well when they are experienced in their role and abstractly know the payoffs of players in other
roles (Forsythe & Lundholm, 1990). If the structure of traders' information is complex enough
relative to the number of assets available to trade, however, information "traps" can occur where
individual traders have no direct incentive to reveal their information (Noeth, Camerer, Plott, &
Webber, 1999). Such problems are typically, though not always, reduced by allowing trading of
more kinds of related assets, since that allows more kinds of arbitrage, i.e., risk-less profits from
correcting market prices.
The key policy question about any institution is how it performs relative to alternative institu-
tions in the same situation or environment, or averaged over a given range of environments. A few
studies have looked at this empirically, directly comparing speculative markets with other institu-
tions for aggregating information. For example, racetrack market odds improve on the prediction
of racetrack experts (Figlewski, 1979). Florida orange juice commodity futures improve on govern-
ment weather forecasts (Roll, 1984). Betting markets beat opinion polls at predicting U.S. election
results (Forsythe, Nelson, Neumann, & Wright, 1992). Finally, betting markets consistently beat
Hewlett Packard official forecasts at predicting Hewlett Packard printer sales (Chen & Plott, 1998;
Plott, 2000).
Unfortunately, no studies have directly compared estimates from speculative markets to esti-
mates from academic-style institutions (though one has been proposed (Hanson, 1995a)). We do
know, however, that those who do best at betting on horse races are smart in ways they can not
articulate, and in ways unrelated to I.Q. (Ceci & Liker, 1986). Academic-style institutions, in
contrast, seem largely limited to aggregating articulated knowledge from those with high I.Q.
Academic institutions place a great deal of weight, perhaps too much, on the opinions of experts
11
Page 12
relative to more ordinary people. One might worry that betting markets where anyone could join
would place too little weight on experts relative to ordinary people. We know, however, that
speculative markets seem to, if anything, put too much weight on advice from experts, both public
and private (Figlewski, 1979; Metzger, 1985; Lichtenstein, Kaufmann, & Bhagat, 1999). Thus one
cannot reasonably claim that while academic institutions may place too much weight on experts
relative to ordinary people, betting markets place too little weight. Speculative markets thus seem
to do better at this weighting choice.
How can betting markets beat opinion polls when they use the same fallible human sources?
A study of election betting markets found that traders overall suffered from standard biases such
as expecting their favored candidate to win, and seeing that candidate as having won debates.
"Market makers," however, were found to be unbiased on average. These are traders who make
offers that others accept, rather than accepting offers made by others, or making offers that others
do not accept. Compared to other traders, market makers invest twice as much, trade more, earn
higher returns, and make one sixth as many errors. They also tend to be more highly educated, and
experienced at trading (Forsythe et al., 1992; Forsythe, Rietz, & Ross, 1999). Betting markets seem
to beat opinion polls and other competing institutions because of the disproportionate influence
such markets give to more rational and informed traders.
The number of topics on which markets can create estimates is limited by the number of
markets one can create. There are fixed costs to create and run markets, and dividing attention
among more markets also raises trader liquidity and volatility costs, i.e., costs due to difficulties
in finding trading partners and fluctuations in prices (Pagano, 1989). However, while it was once
thought that speculative markets could only be viable if they annually traded millions of dollars,
say 10,000 trades of $100 each (Carlton, 1984), it is now clear that much smaller markets are viable.
For example, laboratory experiments consistently show that markets with a couple of traders are
viable. Very low internet mechanical transaction costs are also now spurring a burst of innovation
exploring a great many new market forms, many of which are small (Varian, 1998; Shiller, 1993).
Play money web markets are now available where anyone can create new betting topics, and where a
handful of traders betting play pennies once every few weeks are typically successful at aggregating
information into prices (see, for example, hsx.com, ideosphere.com (Kittlitz, 1999)).
It remains widely illegal, however, to create real money markets like these play money markets,
and so most speculative markets still trade millions of dollars a year. This regulatory block on finan-
cial innovation should not be surprising, however, because all of our familiar financial institutions
were once prohibited by laws against gambling and usury. For example, a thirteenth century decree
by Pope Gregory IX prohibited maritime insurance as usury. The 1570 Code of the Low Coun-
tries outlawed life insurance as gambling (Brenner & Brenner, 1990). In response to speculation
in the South Sea Bubble, in 1720 Britain basically banned the formation of joint-stock companies
12
Page 13
(Kindleberger, 1984). And futures markets were banned as gambling in the late nineteenth century
U.S. (Brenner & Brenner, 1990).
The history of financial regulation can thus be roughly summarized as everything being banned
as gambling (or usury) until an exception was granted for some newly legitimized higher purpose.
For each purpose, such as capitalizing firms, insuring idiosyncratic risk, or insuring common risk,
laws and regulations were created to ensure that generic gambling could not slip in. We may thus
reasonably hope to someday legitimate, and thereby legalize, markets whose main function is to
aggregate information on questions that matter (Bell, 1997).
6
The Engineering of Institutions
It is tempting to use the success of betting markets as information institutions to solve the problems
of democracies as information institutions. But do we know anywhere near enough about either
type of institution to be proposing radical new forms of governments based on this idea?
It depends on whether one thinks like a scientist or like an engineer. A scientist (or at least a
caricature of one) insists on saying "I do not know" about a theory until it has robust empirical
support, or has clear theoretical support from some other empirically-supported theory. A scien-
tist bases policy recommendations only on relatively direct data, or on well-supported theory. A
scientist who studies systems tends to assume that existing systems are functional, and uses that
as data to refine theory. A scientist therefore stays quiet about radical new forms of government,
which can not possibly have direct empirical support, and which are too complex for our theories
to make direct predictions about.
An engineer, on the other hand, is more interested in improving systems than in improving
theory. An engineer is thus willing to make cruder judgments, farther removed from theory. An
engineer is happy to work on a concept with a five percent chance of success, if the payoff from
success would be thirty times the cost of trying. An engineer uses theory explicitly as far as it
will go, but also uses theory-informed intuitions to more informally think about a wide a range of
design issues. An engineer then typically moves on to a series of increasingly realistic and expensive
"proof of concept" prototypes, from computer simulations, to laboratory "wind tunnel" models, to
field tests. While scientists have little use for prototypes and their tests, being neither basic theory
nor data that tests theory, prototypes are what make the engineers' world go round.
Social science is now mostly dominated by a scientific, rather than an engineering, style. The
most respected proposals for new economic and political institutions are thus based firmly on
6
General betting has been legal, if highly regulated, in England for several decades now. The number of betting
topics, however, is limited because bookies are expected to set stable odds, rather than allow odds to fluctuate and
so be set by bettors. Bookies thus mainly allowquestions they feel they can afford to estimate well (Sharpe, 1994).
13
Page 14
established theory. For example, authorities were persuaded to consider new institutions such as
pollution emissions trading and new forms of auctions largely on the basis of theoretical arguments
and endorsement by economic theorists. Efforts to test prototypes of such institutions, in contrast,
have gained relatively little academic prestige.
Economic theorists of "mechanism design" and "implementation" have considered, often in
quite some generality, what complex mechanisms would let you achieve your arbitrary policy goals
by inducing people to tell you what they know. Assuming that human behavior is exactly described
by standard game theories, such theorists have found that it often seems possible to get people to
tell everything they know relevant to achieving almost arbitrary policy goals.
For example, regarding a simple "public goods" decision, i.e., should the government implement
a project and if so who should pay how much, one should in theory be able to get people to tell
everything relevant they know as long as their information is correlated in some way, and no one
runs out of money (D'Aspremont & Gerard-Varet, 1979). More generally, it should be enough
that policy makers are just more patient than ordinary people (Kalai & Ledyard, 1998), or that
anything any one person knows is also at least implied by a combination of things other people
know (Postlewaite & Schmeidler, 1986; Duggan, 1997), or that the probability of some observable
event varies depending on each relevant thing people know (Bergin & Sen, 1998). Few economists,
however, believe that these complex mechanisms would work with real people, and at least one
experiment has supported this expectation.
Regarding political institutions, formal voting models have supported alternative voting rules
such as "approval voting," where voters can vote for as many or few candidates as they like (Brams
& Fishburn, 1983). Formal models have also supported several simple institutions for making public
goods decisions (Thompson, 1966; Tideman & Tullock, 1976; Groves & Ledyard, 1977), a few of
which have even done well in some laboratory experiments (Chen & Tang, 1998; Falkinger, Fehr,
Gachter, & Winter-Ebmer, 2000).
A few social scientists have informally proposed more radical institutional changes, such as
random selection of legislators (Carson & Martin, 1999; Callenbach & Phillips, 1985; Burnheim,
1985), or allowing fine grain private choice of legal systems (Friedman, 1989). Prototypes of these
proposals have even been explored to varying degrees. But none of this has gained much academic
prestige.
The radical new form of government proposed in this paper, "futarchy," is intended to be taken
in the engineering spirit. While motivated in part by theory, it seems difficult to use established
theory to in non-trivial detail compare futarchy to existing institutions, and in any case no such
models are presented here. The purpose of this paper is not to induce high confidence in readers that
futarchy would work well, but merely to raise readers' confidence up to a level that would justify
further exploration via the next level of prototype. This paper thus, from this point forward, mainly
14
Page 15
takes on an engineering tone, qualitatively identifying and addressing a wide range of design issues.
Proposal Concept
Directly or indirectly, academic-style institutions are today important information institutions that
ordinary people defer to, when they defer to anyone, for answers to abstract questions about political
institutions and policy.
7
The resulting democratic policies often seem less well-informed than they
could be. Yet betting markets seem to do a remarkably good job of aggregating information.
So perhaps we should consider substituting betting markets for academic-style and other familiar
institutions in answering abstract questions about policy. There are four levels at which we might
consider such substitution: parameter advice, decision advice, agency decision control, and entire
government control.
Parameter Advice
At the simplest level, policy makers could get into the habit of deferring to betting markets, instead
of commissioning studies or expert committees, to predict parameters such as budget surpluses,
lifespan changes, trust fund depletion rates, global warming rates, AIDS infection rates, or prob-
abilities of asteroid strikes. The elites who would have otherwise been on the elite committee, or
influenced its conclusions, would instead have to gain influence by persuading market speculators.
This general idea of using speculative markets more widely for information aggregation has been
discussed before, both in journals (Hirshleifer, 1971; Leamer, 1986) and in science fiction (Brunner,
1975), as has the more specific application of aggregating policy-relevant information (Zeckhauser
& Viscusi, 1990; Hanson, 1995b).
The thicker a market is, the more offers and trades it has, and thus the harder it is for one
person to change the price. In order to ensure that such speculative markets are thick enough
to have good estimates, one could either directly subsidize a market maker, or credibly commit
to having the market price influence policy. One can subsidize a market maker in a way that
eventually gives money to whomever turns out to be right (Hanson, 1992). This can allow even a
single other trader to be paid to create informed estimates on some topic. We will later show that
an influential market can not be very thin.
Decision Advice
Since one can ask betting markets to estimate just about any parameter that one might reasonably
measure after the fact, one can ask betting markets to make conditional estimates. So, for example,
7
Most of this deference is indirect, via intermediating institutions such as news media.
15
Page 16
not only could a betting market estimate future sea levels, it could also estimate sea levels condi-
tional on future CO2 levels. (One way to do this is with called-off trades, i.e., undoing trades if
the condition is not met.) And since the choices we make can be treated as conditions, we can ask
betting markets to estimate the observable consequences of decisions. We could, for example, ask
betting markets to estimate murder rates conditional on passing gun control laws, and conditional
on not passing gun control laws. If market estimates of murder rates were clearly higher given gun
control than given not, that would argue against passing gun control laws.
Thus, at the decision advice level, policy makers could get into the habit of deferring to betting
markets, instead of commissioning studies or expert committees, to estimate the consequences of
policies. Markets might estimate student test scores given school vouchers, lifespans given universal
health care, attendance if a new stadium is built, or the chance of war given a treaty. In general,
one would ask a market to estimate an outcome of interest, conditional on choices one might make.
This idea has been described before, both in journals (Hanson, 1999) and in science fiction (Stiegler,
1999).
Agency Decision Control
The above approach could help if government agency decisions are uninformed due to agencies
lacking access to relevant information. Agency heads, however, might instead lack incentives to
make their decisions reflect available information. They might get away with this behavior if
legislators have limited abilities to oversee the agency, or if voters have limited abilities to oversee
legislators. To deal with this problem, we might put betting markets more directly in control of
policy.
8
For example, monetary policy seems to be a policy area where people largely agree on a few
statistics by which one can tell, after the fact, how effective a policy was, and on a few control
variables one might use to get good outcomes. Policymakers try to achieve high and steady GDP
and employment, and they control short term interest rates and reserve requirements. The dispute
seems to be mostly about what control variable choices, in what contexts, lead to good outcomes.
Thus monetary policy seems a good candidate for more direct "futarchy," or government by betting
markets.
9
In monetary futarchy, the government would choose some explicit function describing good
outcomes. This function would presumably be increasing in future GDP and employment, and be
concave in these in order to reward stability. The rule of monetary policy would then be that we
8
We later discuss hownon-in-control betting markets can also be biased due to fears that decision makers have
private information.
9
The name "futarchy," comes from government by futures markets, since one can think of bets as claim-judgment
futures.
16
Page 17
would raise or lower interest rates whenever a betting market clearly estimated a higher value for
the good outcome function conditional on such a change. Information that is now given privately
to monetary officials would instead be made public, and hence available to market speculators.
Other possibilities for agency futarchy would be to approve drugs based on estimates of costs
and lived saved versus lives lost, to set crime policy based on estimated costs and crime rates, or
to set organ transplant policy based on estimates of costs and life years gained. In general for each
agency one would officially choose some value function that describes the goal of that agency, and
then officially let market speculators determine which policies best achieve that goal.
Entire Government Control
A more ambitious level at which we might substitute betting markets for other information institu-
tions is that of controlling an entire government. We would vote on values, but bet on beliefs. That
is, elected representatives would mainly define and oversee the ex post (after the fact) measurement
of some "GDP+," a measure of national welfare. The basic rule of government would be:
When an approved betting market clearly estimates that a proposed policy would in-
crease expected GDP+, that proposal becomes law.
Democracy would still say what we want, but betting markets would now say how to get it.
Futarchy seems to be a promising form of government, if we accept three assumptions:
1. It is not that hard to tell rich happy nations from poor miserable ones.
2. Existing democracies fail primarily due to not aggregating available information.
3. Betting markets are the best known institutions for aggregating information.
Many have argued that we typically know too little about individual preferences to help people
via detailed interventions into their daily lives (Buchanan, 1988). As we aggregate more and more
across people, time, and contexts, however, we seem able to make more accurate judgements about
preferences. It seems reasonably clear, for example, that on average people in Ethiopia today are
less satisfied with their lives than people in the United States today. This can be true even if we
would find it difficult to advise any random person on what they should eat for lunch, at least from
among the common luch menu items around them.
Most empirical research in growth economics seems willing to presume that policies that induce
high GDP are good policies. Of course we know that GDP leaves out many things we value, and
is often imperfectly measured. But it seems hard to believe the errors are very large; frequent
travelers find it hard to escape the impression that, comparing nations with large differences in
measured GDP, most people who live in the high GDP nations are richer and better off than most
17
Page 18
people who live in the low GDP nations. Thus researchers have typically recommended policies
which their research suggest causally raise GDP.
But this same attitude also suggests that GDP-futarchy, i.e., a government which does whatever
betting markets say would raise measured GDP, would do well for most citizens. More specifically,
imagine that national welfare were defined as a few-percent-annually-discounted average of future
log GDP, with GDP defined and measured as it is now in the U.S. or similar countries.
It seems plausible that most citizens of most nations would do better under this GDP-futarchy
than under their current forms of government. If nations are typically poor because of adopting
known-to-be-bad policies, then GDP-futarchies which grow their GDP as fast as the best-informed
policies allow should become rich. And futarchy policy would be well-informed, both because in
betting markets those who know they are ignorant have strong incentives to stay away, and because
those who do not know they are ignorant will decline in influence as they lose their money.
If a GDP-futarchy would do well, then futarchy based on GDP+ should do even better. That
is, if elected representatives, freed from thinking about how to get what they want, can develop
substantially better ex post measures of what citizens want, then most citizens should be even
better off. And since economists spending modest budgets have already found many promising
extensions to current GDP measures (Boskin, 2000; Nordhaus, 2000), we might do even better
with both increased funding and the full attention of elected representatives.
Futarchy is intended to appeal to a wide range of ideologies and political factions, remaining
neutral on most such disputes. Futarchy could in principle result in either extreme socialism or
extreme minarchy, depending on what elected representatives say they want, and on which policies
speculators believe will achieve those goals. Futarchy does require that there be some community
of speculators who have assets they are free to bet, and that some extra benefits consistently accrue
to those who win their bets. These benefits need not be large, however, nor need the community
need be very large. Futarchy should appeal most to those who can accept the values democracy
would favor, and who think that speculators could be persuaded to agree with their assessment of
how to best promote such values.
Design Issues
Many ideas seem promising at first glance, but then seem less so after one considers some details.
Let us therefore, in engineering fashion, identify and briefly discuss thirty one design issues, each
expressed as an objection to this overall approach. After this we will present a more detailed
proposal for "futarchy" as a form of government.
18
Page 19
The Rich Would Get More Influence
A basic fact of life is that in general the rich get more influence. After all, the point of being rich
is to be able to get more of whatever it is you want. If we do not like that, we can tax the rich
more, and then they will not be as rich.
Perhaps the concern here is that the rich should not have proportionally as much influence in
betting markets as we accept them having elsewhere. Betting markets are not opinion polls where
the rich get more votes, however. The rich have more potential votes, but they should lose money
if they use those extra votes without having proportionally extra information. And if there's one
thing we know about the rich it is that they don't easily throw their money away; those who do,
do not stay rich very long.
It Would Be Better To Appeal To Higher Motives Than Money
If one could reliably pay off bets in units of "higher" forms of value, then we might want to do so.
We want to induce people to tell us what they know, and we need to offer them something that
we can reliably produce, can distribute in controlled amounts, and know that most people value.
Money satisfies these requirements, and no other forms of value yet seem to. We do not seem to
know how to design productive information institutions based on controlled appeals to other values.
People Would Volunteer Less Information And Lie More
Many of who think democracy works well expect policy-relevant information to be revealed by
people who have little other way to benefit from their information other than to honestly reveal
it because they feel it is their civic duty. Such optimists may fear that if a betting market were
available, such people might be instead tempted to keep quiet about their information in order to
benefit by betting on it. They might even publicly lie in order to deceive other traders.
Of course betting reveals information directly to markets, and once people with secrets run
out of ways to bet on their secrets, they should want to reveal those secrets and persuade other
speculators to believe them, so that they could reverse their earlier trades for a profit. Secret-holders
might, however, never think they have run out of ways to bet on their information.
This is indeed a potential concern, at least in theory. The studies mentioned earlier showing
that speculative markets beat other institutions at aggregating information suggest that institutions
which now try to rely on weak incentives and civic duty must have other serious drawbacks. But
perhaps a new institutions will one day be discovered to better tap this potential.
19
Page 20
People Prefer Their Comfortable Illusions
The beliefs that speculators would use to estimate policy consequences would probably not be the
same beliefs as the comfortable illusions that voters prefer. So if futarchy forced voters to accept
specific speculator beliefs, it might thus force them to forgo their comfortable illusions.
Fortunately it seems that voters need not be directly confronted with speculator beliefs, and
therefore need not forgo their illusions. Voters need to keep abreast of news about government
policy, and stay informed about choices made by their elected representatives, but this can all be
done without becoming much aware of speculator beliefs.
Speculator estimates of welfare that determine policy would not directly explain speculators'
reasoning, and specific approved policies could typically be rationalized in many ways. Other mar-
kets and forums might more concretely explain speculator reasoning, but news media that catered
to viewers with cherished illusions would likely not confront their viewers with uncomfortable be-
liefs. Even today, media avoid telling viewers how legislative sausage is made, when viewers find
that distasteful.
Maybe Low Information Is Not Democracy's Major Failing
Futarchy has been designed primarily to fix information problems with democracy. But perhaps,
contrary to the impression one might get from the evidence presented earlier, democracy's biggest
problem is some other problem, such as a coordination or commitment failure, or preventing the
government from being overthrown.
In this case the big questions are whether futarchy makes that bigger problem any worse, and
whether the institutions which might fix that problem are incompatible with futarchy. Futarchy
does still "vote on values," and leaves open the question of how exactly the democratic component
is to work. This would seem to leave a lot of room for adjustment to deal with other problems.
Thus in the absence of particular reasons to think that futarchy makes some bigger problem much
worse, or that futarchy precludes some essential fix, it seems reasonable to pursue futarchy as a
way to fix information problems with democracy.
For example, a bigger problem governments face might be to find a way to pay off powerful folks
who could otherwise overthrow the government. However it is that current democracies do this, it
would seem that the "vote on values" democracy part of futarchy could also do this. Instead of
paying them off via direct financial payments or pork barrel programs in their favor, they could be
paid off via raising the weight on outcome measures that they favor.
20
Page 21
Democracy Might Become Unstable
In theory, democracies seem to have problems of policy instability, or "cycling," since for any policy
there is always another policy that some majority likes more. In practice, however, this does not
seem to be a large problem (Tullock, 1981). If voter ignorance is part of the reason for such stability,
as some theories suggest (Hanson, 1997), perhaps futarchy would make instability worse, by better
informing citizens about the consequences of their choices.
This might be a reason to lean toward more stable democratic institutions within futarchy.
So we might prefer one or more large decentralized legislatures whose members change gradually,
and we might shy away from powerful presidents or proportional representation systems in which
coalitions rule as a group.
If values change more slowly than beliefs about the consequences of policy, and if less expertize is
required to make value judgments than to make full policy choices, then we might reasonably accept
a democratic part of futarchy that is slower and less expert than today's democratic processes. Thus
futarchy might allow more experimentation with non-other democratic forms, such as demarchy or
more direct democracy.
Time-Consistency Might Be A Problem
The inability to commit has been identified as a major reason, in theory and in at least some
practice, for government failures (Levy & Spiller, 1994; Besley & Coate, 1998). The basic rule of
futarchy described above does not seem to overcome this problem, but it also does not seem to
make this problem any worse.
While a constitution might declare that the government could commit itself to future choices,
it is not clear that courts could be relied on to enforce such commitments. It is also possible that
allowing earlier generations to take advantage of later generations creates worse problems than
commitment failures. Perhaps we should keep the standard approach, which is to typically allow
governments to commit only if they can arrange to do so themselves, such as via transaction costs,
posting bonds, external reputation, or other external institutions.
Expressive Voting Could Still Cause Problems
Voters often seem to vote expressively, i.e., to care about other things when voting besides influ-
encing what policies the winners will promote (Brennan & Lomasky, 1993). For example, voters
want to take sides, to show themselves and their friends that they are knowledgeable and that they
care, and to have people they like and respect represent them. And while these may be worthwhile
goals, policy choices can suffer as a result.
Futarchy, by limiting democracy primarily to values, would presumably turn voters more toward
21
Page 22
showing that they care, and away from showing that they are knowledgeable. It is not clear that
futarchy makes expressive voting any more of a problem overall, but if this is a concern, we might
consider forms of democracy that are harmed less by expressive voting, if such can be identified.
Markets Might Be Too Thin To Make Good Estimates
In general, there are an infinity of possible markets, and so finite costs of creating markets ensure
than most possible markets do not actually exist. And even if one created some arbitrary market,
there is no guarantee that people would trade much there; the market might end up being too
"thin."
A market that influences important policies, however, cannot be very thin. After all, if a very
thin market were to influence policy, interested parties could then by definition pay very little to
move the price and favorably influence policy. If interested parties on both sides were similarly
funded, their combined trades would thicken the market. Alternatively, if one side was willing to
spend much more, then speculators who knew that fact could make easy profits by trading against
that better funded side, again thickening the market. So either way, the market would end up being
thick.
People Could Buy Policy Via Betting Markets
Imagine that speculators knew that a certain interested party had a strong interest in a price
moving in a certain direction, but had no special information about that topic. If this interested
party then began trading large amounts in order to move the price, speculators would see easy
profits in moving the price back, which they would do given low transaction costs.
Thus the only way for such an interested party to substantially influence the price is to possibly
have substantial special information, or to trade via parties that might have such information. And
this means that all the interested party can really do is choose whether to hide or reveal some
private information they hold.
For example, assume some proposed city stadiums would be profitable, while others would be
unprofitable, and that many speculators know how profitable proposed stadiums are on average.
For each stadium, assume only few interested insiders know how profitable that stadium would be,
and that these insiders would benefit from the stadium being built even if it were not profitable.
Imagine first that speculators initially drove market profitability estimates for all proposed
stadiums to the average value. Then imagine that many insiders tried to buy a favorable impression
of their stadium by bidding up its market estimate of profitability. Speculators could then profit
by driving some prices back down so that the average price returned the known average stadium
profitability.
22
Page 23
If it was known that all insiders bid up their estimate equally, speculators would push all
estimates back to the same average value. Speculators would only let estimate differences stand
if they expected a correlation between insider trading and stadium profitability. If insiders tend
to push more when they know their stadium is more profitable, they would thereby reveal some
information through their trades, and market estimates would correlate with stadium profitability.
But if there were no correlation between insider trading and profitability, all estimates would end
up being the same.
Therefore, as claimed, in this case all that insiders can really do is either reveal their inside
information, or try to hide it. And it turns out that all cases are basically of the same form as in
this stadium example. When insiders face speculators with deep pockets and low transaction costs,
they can only influence market-estimate-determined policy by revealing or hiding information.
One Rich Fool Could Do Great Damage
Even Bill Gates, the richest person alive today, has only a small amount of wealth compared to that
available to all market speculators. If most all those speculators were therefore confident that Bill
Gates were wrong about some estimate, even Bill Gates could not substantially change a betting
market estimate. If Mr. Gates pushed the price one way, all those speculators would push it right
back.
More likely, however, speculators would allow Mr. Gates to move the price some because they
reasonably suspected that he had access to hidden information. And given his previous successes,
this would not be an unreasonable assumption.
You Need A Way To Tell If A Proposal Was Implemented
It seems reasonable to treat proposals like contracts, in the sense of being responsible for indicating
how ambiguities should be settled. A contract typically specifies a jurisdiction whose law is to
interpret it, and can also specify other rules of interpretation or the arbitrator who is to do the
interpretation. Contract law remains in the background to fill in remaining ambiguities. Similarly,
futarchy proposals could specify their interpretation regime as well as who would do the interpreting.
If it were not clear to speculators how a proposal would be interpreted and implemented, then
speculators would have to average over the kinds of interpretations they think likely. An unclear
proposal could be a sloppy but sincere attempt to improve national welfare, or it could be an
insincere attempt to slip in policies that would not be approved if proposed clearly. Insincere
proposals probably reduce national welfare, while sloppy sincere proposals might increase it.
Speculators must thus estimate the relative contribution of sincere and insincere proposals to the
class of unclear proposals in any given policy area. When insincere proposals dominate, speculators
should be wary of endorsing unclear proposals, just as anyone is wary of signing a contract he or she
23
Page 24
does not understand. But when sloppy but sincere proposals dominate, speculators may accept the
occasiona