How to Fix Democracy Robin Hanson Dept. of Economics George Mason University
Transcription
How to Fix Democracy Robin Hanson Dept. of Economics George Mason University
How to Fix Democracy Robin Hanson Dept. of Economics George Mason University Outline • • • • • • Science vs. engineering of government Big democratic failure: not aggregate info But speculative markets do this well They can estimate policy consequences Futarchy: vote on values, but bet on facts Discuss rationale and objections Science vs. Design/Engineering • E.g., astronomy • Learn what is • Path: puzzle, theory, implications, data test, inference, cite, in texts • Focus on few high probability theories • E.g., rocketry • Create what can be • Path: problem, idea, prototype, lab test, field test, common use • Try out many low probability ideas Can know what works, without know why it works. Can know why things go wrong, yet be unable to fix. Observed Forms of Government • Dictator – roving? inherited? – Limits: constitution, regional exit, emigration • Democracy – – – – Franchise: all, land, money, inherited, guns Representation: direct, random, elected Power: united, divided, legislature, federalism Voting Rule: first past post, proportional rule • Many alternatives studied, e.g., approval voting Win/Win vs. Doing Battle What You Want Win/Win Doing Battle What I want Issues In Designing Government “Do battle, Find win/win deals.” • • • • • • • Promote invest, limit rent-seek (time-consistent) Stable against revolution, policy cycles Allow frequent, timely coordination Feel belong, represented, listened to Induce citizens to reveal preferences Induce reveal info on policy consequences Big peaceful changes must not be ideological Public Goods Mechanisms • Each says value of each choice, pick choice & tax – Ignore agenda setting, incentives to learn, secrets • Clarke Tax [71, Groves73, Tideman/Tullock76] – Fails in lab experiments, coalitions can beat • Walker [81] - fails in lab experiments. • Groves/Ledyard Mechanism [77, Chen/Tang98] – Has worked in lab, but not win/win for all w.r.t. no p.g. • Lots theory on complex mechanisms for secrets – Says can do great if signals correlated or acts repeated – But no experiments! Project Approval Example Number of People 10 10 14 or 16 6 or 4 Payoff if Project Payoff if Project Succeeds Fails 1 1 -1 -1 1 -1 -1 1 If project will succeed, 4 people get + clues and 2 get – clues. If project will fail, 4 people get – clues and 2 get + clues. Probability of project success = 1/3. Massive Info Aggregate Failure! Evidence • • • • Agreeing to disagree on facts is irrational Opinion polls are not random walks Many stupid public opinions Huge variation in national growth rates Theory • Small outcome push for informed voters • Bigger other pushes: beliefs are like clothes Stupid Public Opinions • • • • • • 85% “Jesus was born of a virgin” (’94 Barna) 80% “U.S. Govt. is hiding the fact that it knows of the existence of aliens” (’97 CNN/Time) 72% believe in angels (’96 Gallup) 52% “astrology has some scientific truth”, “theory of evolution probably not true” (’96 GSS) 66% “foreign aid spending is too high is major reason economy is not doing better” (’96 SAEE) 74% “was an official cover-up …[of] the truth about the Kennedy assassination?” (’98 CBS) 4.4 4.2 Log GDP Per Capita vs. Year 4 Japan, Argentina, Venezuela highlighted Source: Maddison 1995 3.8 3.6 3.4 3.2 3 2.8 2.6 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 An Old Idea 1651 British chemical docs dare schools: Oh ye Schooles. ... Let us take out of the hospitals, out of the Camps, or from elsewhere, 200, or 500 poor People, that have Fevers, Pleurisies, etc. Let us divide them into halfes, let us cast lots, that one halfe of them may fall to my share, and the other to yours; ... we shall see how many Funerals both of us shall have: But let the reward of the contention or wager, be 300 Florens, deposited on both sides: Here your business is decided. Continental Drift Scenario • 1912, Alfred Wegener proposed • Was: experts said nuts, believed in 1960s • Alternative: – – – – – – Create claim that experts will agree in 2012 Offer to bet at high odds Experts put up or shut up; can’t laugh it off. If they keep odds down, his grandkids make $. Speculators tempted to check on experts Eventually experts agree, bettors paid off Speculative Markets Work Info Aggregate Wonders • It’s hard to find info not reflected in prices • O.J. futures beat govt. forecasts [R. Roll AER ‘84] • Election markets beat polls [Foresythe et.al. AER ‘92] – most traders biased, but not active traders • Markets at horse races tap non-IQ smarts [Ceci & Liker, J. Exper. Psy. ‘86] • In lab, reveal info if traders experienced & know net payoffs [Forsythe & Lundholm, Econometrica ‘90] Recent Market Prototypes • Iowa Electronic Markets - on elections • Play-$ web games show small markets work – FX: 2K players, 200+ claims, 80+ sci/tech – HSX: 100K players, ~500 movies, stars – More .coms: newsbet, myhand, funbets, … • Internal company markets on sales, delivery – at HP, Siemens, Xanadu – Limit participation & stakes Why Speculative Markets Work • Instead of the rest of us deciding who the experts are, people self-select as experts. • People have incentive to be honest with themselves, and to stay out if don’t know. • Those who aren’t honest with themselves eventually lose money, and so influence. • Can specialize in correcting any bias see. “Put Where Is” Vision • Markets on most topics of debate – Can ask any question we may eventually agree on • Interpret betting odds as answer estimate – Can subsidize a market to get better estimate • Public, media defer to market estimates – To influence world, experts convince speculators – Likely: labs bet on advice from hired experts • Ignorant stay out – are self-aware or go broke – Yet non-credentialed masses have voice • Arbitrageurs kill biases, inconsistencies Decision Markets Give E(O|C) Choices Outcomes • • • • • • • • FED money policy Who elected president Health care regulation School vouchers • Who is CEO • Which ad agency GDP per capita War deaths Lifespan School test scores • Stock price • Product sales Predicting Gun Control P(G) $1 P(not G) $1 if Gun Control 1 $1 if No Gun Control Predicting Gun Control Outcomes $1 if Gun Control & Less Crime $1 if Gun Control $1 if Gun Control & More Crime $1 $1 if No Gun Control $1 if No Gun Control & Less Crime $1 if No Gun Control & More Crime Gun Control Decision Markets P(G) $1 if Gun Control & Less Crime $1 if Gun Control P(L | G) $1 Compare! P(L | not G) P(not G) $1 if No Gun Control $1 if No Gun Control & Less Crime Regulation of Speculation Racetrack Casino Capitalize Firms “Entertain” Lottery Sweepstake Sports Bet Gamble Aggregate Information Hedge Individual Risk Hedge Common Risk Does Ignorance Stabilize Politics? • • • • Of role – like behind Rawls’ “veil” Of candidate positions – mix -> pure str. Of project success – see voting example Expressive voting seems more altruistic Rationale for “Futarchy” • The main reason why nations are richer, happier is that they follow better policies. • Betting markets are the best institution we know of to aggregate information on specific later-verifiable topics. • It’s not that hard to tell the rich, happy nations from the poor, miserable ones. – Personal hidden info not that big a deal – Measurable aggregates get close enough Vote On Values But Bet On Beliefs E[GDP+|Alt.] >? E[GDP+|S.Q.] Futarchy’s One Rule When a market estimates currently-defined GDP+ to be higher given some proposed alternative policy, that policy becomes law. • Like contract, proposal says how handle conflicts. • Refine GDP+ to fix unintended policies. • Recurse to settle details to clarify this rule, and to pick local proxy outcomes for smaller decisions. • Treaties via include other nation’s GDP+ in own. Issues In Defining GDP+ Claim: Plain GDP futarchy beats our democracy. (Few-% time-discounted average of future log) • • • • Risk aversion, time discounting Leisure, environment, art, lifespan Inequality, population, foreigners Counting moral outrages Issues In Defining Official Prices • • • • • Valid “banks” issue contingent assets Valid “cash” contingent assets pay off in Valid “marketplace” guarantees offer “Enough” difference: size, duration, breadth Valid judges of whether alternative tried Ideological Neutrality • If speculators think minarchy best gets citizens what they say they want, futarchy should deliver minarchy. • But leftists have also endorsed, thinking speculators will agree with them about the value of regulation, social insurance, etc. – Expect savvy selfish capitalists to bet beliefs Objections to Futarchy • • • • • • • Is election where rich get more votes Markets bubbles, markets too thin Interested parties buy policies via market Insiders get policies want by redefine GDP+ GDP+ measurements will be corrupted GDP+ only measurable now when not matter Max GDP+ will max plus errors in measure