10 · James Buchanan · Gordon Tullock · George Stigler · Mancur Olson

Public choice

Economists long treated market actors as self-interested and government as a benevolent fixer of market failures. Public choice, built at Virginia and Chicago from the 1950s on, dropped that double standard: voters, politicians, lobbyists and bureaucrats respond to incentives too. James Buchanan and Gordon Tullock's The Calculus of Consent (1962) analysed voting rules, Mancur Olson's The Logic of Collective Action (1965) explained which groups organise, Tullock (1967) found the hidden cost of chasing privileges, and George Stigler (1971) asked whom regulation really serves. Buchanan called the result "politics without romance", and won the 1986 Nobel for it. Each model below runs the actual incentive logic.
An import quota raises the domestic price of sugar. A few thousand growers gain hundreds of thousands of dollars each; hundreds of millions of consumers lose a few dollars each, more in total than the growers gain. Each person acts only if their stake beats the cost of paying attention, and in big groups each member's contribution matters less, so free-riding grows.

An import quota on sugar: who notices, who pays to lobby

Growers4.5k people
+$560k
gain per person per year
100% care enough to act
Consumers331M people
−$9.81
loss per person per year (higher prices)
4% care enough to act
money raised for lobbying
growers
$60.3M
consumers
$0
growers gain
$2,500.0M
consumers lose
$3,250.0M
incl. deadweight
net to society
-$750.0M
stake ratio
57,022×
per grower vs per consumer
The quota survives. The losers lose more in total, but no single consumer finds it worth an afternoon to fight a few dollars a year.

Try: raise the number of producers toward a million. The total transfer is unchanged, but each grower's share of it shrinks, free-riding grows, and their war chest falls several-fold. Then shrink consumers to a few thousand (think industrial candy makers instead of households) and the losing side finally outspends the winners.

Rent-seeking: the cost of a monopoly is more than the triangle

pc-rent-seekingpc-rent-dissipation
The government will grant one firm an exclusive licence. Classic welfare economics counted only the red deadweight triangle as the cost; the yellow rectangle was “just a transfer” to the monopolist. Tullock asked what firms will spend to win that rectangle. Here firms bid in a lottery contest, each adjusting toward its best reply to the others.

A monopoly created by government licence

002020404060608080100100quantitydemandcost (competitive price)monopoly pricerent 1,60011% burneddeadweight 800

Round 0: what each firm spends chasing the prize

firm 1
20 · 12%
firm 2
35 · 21%
firm 3
50 · 29%
firm 4
65 · 38%

Each firm's chance of winning = its spending ÷ total spending (a Tullock lottery). Every round each firm moves toward its best reply to the others.

total spent
170
equilibrium
1,200
(n−1)/n of rent
rent dissipated
11%
social cost
970
triangle + lobbying

Try: press Play with 1 firm (it spends almost nothing), then 2, 4 and 10. With more contenders, total lobbying approaches the whole rent, so the true cost of the monopoly is the red triangle plus most of the rectangle. Switch to "sold at auction": the bids become a transfer to taxpayers, not resources burned on lawyers and lobbyists.

Rational ignorance: why studying the ballot rarely pays

pc-rational-ignorancepc-paradox-of-voting
Anthony Downs pointed out that information is costly and a single vote almost never changes the result. Below, the chance of being decisive comes from the forecast distribution of the vote share (the method used by Gelman, Silver and Edlin), multiplied by what the better outcome is worth to you.

Expected private payoff of becoming an informed voter

0.0010¢$1$100k10010k1.0M100Mvoters (log scale)dollars (log scale)cost of informing yourself $600100M
P(your vote decides)
1 in 9.4M
expected benefit
0.1¢
cost of study
$600
20h × $30
stops paying at
200
voters
Staying uninformed is the rational choice here. That is not stupidity; it is a correct reading of your odds.

Try: slide voters from a village (100) to a nation (100M). The chance of being decisive falls roughly as 1/N, and faster still if the race isn't close. Raise your stake to $1M: even then, for a national election, the expected value of studying is cents.

People do vote, and many do read up, mostly for civic or expressive reasons rather than private gain. The public choice point is narrower: the incentive to be well-informed is weak for the many and strong for the few with concentrated stakes, which is why the previous two models work.

Logrolling: trading votes to pass bills nobody would pass alone

pc-logrollingpc-pork-barrelpc-calculus-of-consent
Nine legislators each want a project for their own district. Each project's benefit goes to one district; its cost is paid by all nine through taxes. Voting sincerely, each bill gets one vote. Then legislators start trading.

Nine pork-barrel bills, nine districts (cells = net gain to that district if the bill passes)

bill forD1D2D3D4D5D6D7D8D9votes
D1+8.3✓−1.7−1.7−1.7−1.7−1.7−1.7−1.7−1.71/9 fail
D2−1.7+8.3✓−1.7−1.7−1.7−1.7−1.7−1.7−1.71/9 fail
D3−1.7−1.7+8.3✓−1.7−1.7−1.7−1.7−1.7−1.71/9 fail
D4−1.7−1.7−1.7+8.3✓−1.7−1.7−1.7−1.7−1.71/9 fail
D5−1.7−1.7−1.7−1.7+8.3✓−1.7−1.7−1.7−1.71/9 fail
D6−1.7−1.7−1.7−1.7−1.7+8.3✓−1.7−1.7−1.71/9 fail
D7−1.7−1.7−1.7−1.7−1.7−1.7+8.3✓−1.7−1.71/9 fail
D8−1.7−1.7−1.7−1.7−1.7−1.7−1.7+8.3✓−1.71/9 fail
D9−1.7−1.7−1.7−1.7−1.7−1.7−1.7−1.7+8.3✓1/9 fail
net0.00.00.00.00.00.00.00.00.0
bills passed
0/9
net welfare
0.0
trader's net
0.0
outsider's net
0.0
Every bill destroys value (benefit 10 < cost 15), so none can pass on its merits. Trading votes lets a majority push the costs onto the minority.

Try: start with no trading (every bill fails 1–8). Switch to the 5-member logroll: five wasteful bills pass, traders gain, outsiders pay. Then let everyone trade: all nine pass and every district ends up worse off than if nothing had passed, yet no legislator wants to be the one left out.

Regulatory capture: the agency drifts toward the industry

pc-regulatory-capturepc-revolving-doorpc-government-failure
An agency is founded after a public outcry to police an industry. The industry has a permanent, concentrated interest in its rules; the public pays attention only in bursts. Each year three forces act on the agency's position: lobbying, the revolving door, and public attention, which decays.
revolving-door ban

An agency created to police an industry · year 0

rules: 0 = public interest, 100 = industry's wish listpublic attentionstaff via revolving door
0501000204060years since the agency was founded
agency position
8
serves the public
public attention
90%
revolving door
5%
of senior staff
scandals
0
none yet

Forces each year

Industry pulls every year: it has the stakes, the expertise, and shows up at every hearing.

Revolving door: staff who came from, or hope to join, the industry share its view.

The public pushes back only after a scandal, and then forgets.

Try: press Play and watch the agency drift as attention fades, until a scandal (random, likelier the more captured it is) snaps it back for a few years. Set "forgets" low, or ban the revolving door, and the drift slows; it never fully stops while lobbying is the only steady force.

Majority rule: the median voter, and when there is no majority will

pc-median-voterpc-condorcet-paradoxpc-arrow-theorempc-agenda-control
Two candidates seeking a majority move toward wherever votes are, and on one dimension that is the median voter (Hotelling, Black, Downs). But with several options and no single dimension, majority preferences can go round in a circle, as Condorcet noticed in 1785 and Arrow generalised.

Voters on a left–right line, each voting for the nearer candidate

median voterABleftright
A's vote share
53.7%
B's vote share
46.3%
median voter
48.3
gap A↔B
70
round 0

Try: press Campaign. Each candidate in turn shifts one step if it wins votes, and both end up at the median voter, however the electorate is spread. On the polarised electorate they still meet in the empty middle, where few voters actually sit: a prediction that real primaries, turnout and activists often break.

Preset:

Each bloc's ranking (click to change)

  • A · build a school
  • B · fix the roads
  • C · cut taxes

Head-to-head majorities

A 67–33B 67–33C 66–34ABC

A beats B, B beats C, C beats A: majority preference is a cycle

Whoever sets the agenda picks the winner

first A vs B, winner vs C
final choice: C
first B vs C, winner vs A
final choice: A
first A vs C, winner vs B
final choice: B

3 different agendas, 3 different "wills of the people".

Try: with the cycle preset, each option wins under some voting order, so the chair, not the voters, decides. Switch to single-peaked preferences (everyone ranks options along one left–right line) and the median option B wins every agenda. Then click a bloc to scramble its ranking and break it again.

The rise and decline of nations: coalitions pile up

pc-distributional-coalitions
Olson's later book applied the logic of collective action to whole economies. Small groups slowly overcome their organising costs and form cartels, guilds and lobbies that fight over shares of output instead of growing it. In a long-stable country they accumulate; a war or revolution can sweep them away. Two identical countries start here; only B can be shaken.

Output per head (log scale) · year 0

Country A: stable for a centuryCountry B
1003161,000020406080100year

Entrenched distributional coalitions

lobbies in Alobbies in B
0204060020406080100year
growth in A
0.00%
growth in B
0.00%
lobbies A / B
2 / 2
output A / B
100 / 100

Try: press Play and let both countries accumulate lobbies (cartels, licensing boards, protected trades) for 50 years, then hit "Upheaval in B": most of its coalitions are swept away and its growth jumps back toward the frontier while A keeps slowing. Olson's reading of post-war West Germany and Japan versus Britain. (Only the coalition channel is modelled here; real post-war booms also had catch-up and rebuilding.)

Niskanen's bureau: twice the efficient size

pc-niskanen-bureaucracy
William Niskanen modelled the head of a government bureau as wanting a bigger budget. The bureau knows its true costs; its legislative sponsor knows mostly the value of the service. The bureau offers 'this much output for this budget', take it or leave it, and the sponsor accepts anything that is worth at least what it costs.

A bureau sells its sponsor a whole output level for a whole budget

total value to the publictotal cost (the budget)
02k4k6k050100output of the bureaubest for publicbudget-maximisingchosen
output
120
best: 60
budget
4,800
best: 2,400
net value to public
0
surplus absorbed
100%

Try: at 0% oversight the bureau grows until total cost equals total value, twice the efficient size, and every bit of public surplus is used up. The vertical gap between the two curves is what the public gains; raise oversight to watch the budget shrink and that gap reopen.

Later work (including Niskanen's own) softened the model: bureaucrats may maximise discretionary slack rather than output, and sponsors do learn. The lasting point is the information asymmetry, not the exact factor of two.
  • takeawayPolitical outcomes follow who has the incentive to organise and pay attention. Small groups with big stakes beat large groups with small stakes, even when the policy makes society poorer.
  • takeawayThe cost of privileges, monopolies, tariffs and licences includes the resources spent chasing them. Agencies and bureaus have their own interests, and drift toward whoever watches them most closely.
  • takeawayMajority rule has no single "will of the people" to discover: results depend on rules, agendas and vote-trading. Hence Buchanan's focus on constitutional rules that limit what any majority can impose.

Key concepts · 20

James Buchanan · Gordon Tullock · George Stigler · Mancur Olson
  1. Politics without romance

    Analysing government with the same assumption of self-interested, incentive-driven people used for markets.

    Comparing a real market with an imagined, perfectly benevolent state is a rigged contest. Public choice compares real markets with real governments, staffed by ordinary people facing their own incentives and limited information.

    Buchanan, “Politics without Romance” (1979)

    ↑ see it in the visualization
  2. Concentrated benefits, diffuse costs

    A policy that gives a small group large gains while spreading larger total costs thinly over many people.

    The few have a strong reason to organise and lobby; the many lose too little each to notice. Such policies survive even when they make society poorer overall.

    Olson, The Logic of Collective Action (1965)

    US sugar program (import quotas)Agricultural subsidies in the EU and USOccupational licensing
    ↑ see it in the visualization
  3. Logic of collective action

    Large groups with a common interest tend not to act on it, because each member can free-ride on the efforts of others.

    Olson overturned the assumption that shared interests produce political pressure. Small groups organise easily; large, latent groups like consumers or taxpayers rarely do.

    Olson, The Logic of Collective Action (1965)

    ↑ see it in the visualization
  4. Selective incentives

    Private benefits (insurance, journals, discounts, or penalties) offered only to members, so large groups can still organise.

    They explain why big lobbies such as farm bureaus, unions or retiree associations exist despite the free-rider problem: the lobbying is a by-product of something members buy for themselves.

    Olson, The Logic of Collective Action (1965)

    ↑ see it in the visualization
  5. Rent-seeking

    Spending resources to obtain a transfer or privilege from government rather than to create new wealth.

    Tullock saw that the cost of a monopoly, tariff or theft is not only the deadweight triangle: the resources spent competing for the privilege (and defending against it) are also lost. That made the true cost of such policies much larger than earlier estimates.

    Tullock, “The Welfare Costs of Tariffs, Monopolies, and Theft” (1967); Krueger (1974)

    Taxi medallionsImport licences in pre-1991 India (the “licence raj”)Lobbying for tariff protection
    ↑ see it in the visualization
  6. Rent dissipation (Tullock contest)

    When many contenders compete for a prize by spending, total spending approaches the value of the prize.

    With n equal contenders in a lottery-style contest, together they spend (n−1)/n of the rent. Whether bids are burned on lobbying or paid to the treasury as an auction makes the difference between waste and a transfer.

    Tullock, “Efficient Rent Seeking” (1980)

    ↑ see it in the visualization
  7. Rational ignorance

    Voters choosing not to learn about policy because the chance their vote decides anything is tiny.

    Becoming informed costs hours; the expected payoff is the stake times a one-in-millions chance of being decisive. Ignorance is then the rational choice, which leaves the field to those with concentrated stakes.

    Downs, An Economic Theory of Democracy (1957)

    ↑ see it in the visualization
  8. Paradox of voting

    If the expected private benefit of voting is near zero, why does anyone vote?

    Turnout is explained by civic duty, expressive motives and social pressure rather than instrumental calculation. Brennan, Lomasky and Caplan add that cheap, expressive votes let people indulge biases they would not act on with their own money.

    Downs (1957); Brennan & Lomasky, Democracy and Decision (1993)

    ↑ see it in the visualization
  9. Logrolling

    Legislators trading votes: I support your project if you support mine.

    It lets intense minorities be heard, but it also lets a coalition pass bundles of projects that each cost more than they are worth, with the costs pushed onto everyone else.

    Buchanan & Tullock, The Calculus of Consent (1962)

    ↑ see it in the visualization
  10. Pork barrel and universalism

    Spending targeted at particular districts, often agreed in a norm where every legislator gets a share.

    When everyone trades, every bill passes and every district can end up worse off than if none had, yet no single legislator gains by refusing their share.

    Weingast, Shepsle & Johnsen (1981)

    “Bridge to Nowhere” in Alaska (2005)US highway bill earmarks
    ↑ see it in the visualization
  11. Regulatory capture

    A regulatory agency coming to serve the industry it regulates instead of the public.

    Stigler argued regulation is often acquired by the industry and designed for its benefit: entry barriers, price floors, restrictions on substitutes. Industry has steady, concentrated interest; public attention is sporadic.

    Stigler, “The Theory of Economic Regulation” (1971)

    Interstate Commerce Commission and railroads/truckingCivil Aeronautics Board fares before 1978Pre-2008 financial regulators
    ↑ see it in the visualization
  12. Revolving door

    The movement of staff between regulatory agencies and the firms they oversee.

    Expertise often exists only inside the industry, so agencies hire from it, and officials who hope for industry jobs later have a reason to be accommodating. Cooling-off rules slow but do not remove this.

    Stigler (1971); Peltzman (1976)

    ↑ see it in the visualization
  13. Median voter theorem

    With voters on one dimension and single-peaked preferences, two-candidate majority competition converges to the median voter’s position.

    It explains why platforms often look alike and why policy tracks the middle. It also shows the limits of democracy as a preference-aggregator: the median decides, and minorities on either side get nothing from the vote itself.

    Hotelling (1929); Black (1948); Downs, An Economic Theory of Democracy (1957)

    ↑ see it in the visualization
  14. Condorcet paradox

    Majority preferences can be cyclic: A beats B, B beats C, and C beats A.

    Then there is no coherent “will of the majority”, and whoever controls the order of votes can produce any outcome.

    Condorcet (1785)

    ↑ see it in the visualization
  15. Arrow’s impossibility theorem

    No ranked voting rule over three or more options satisfies a short list of reasonable fairness conditions at once.

    Every collective decision rule has some flaw: it can cycle, ignore some preferences, or be manipulated. Buchanan’s answer was to stop treating “society” as an agent with preferences and focus on rules individuals can agree to.

    Arrow, Social Choice and Individual Values (1951)

    ↑ see it in the visualization
  16. Agenda control

    The power to choose which options are voted on and in what order.

    When preferences cycle, the agenda setter, a committee chair or a speaker, effectively picks the winner. Institutions that limit agenda power are a large part of constitutional design.

    McKelvey (1976); Riker, Liberalism against Populism (1982)

    ↑ see it in the visualization
  17. Distributional coalitions

    Interest groups that seek a bigger share of output for their members rather than a bigger total.

    Olson argued that stable societies accumulate such coalitions over time, which slow adoption of new technology, reallocation of labour and entry, and so slow growth. Upheavals that destroy them can be followed by fast growth.

    Olson, The Rise and Decline of Nations (1982)

    Post-war West Germany and Japan vs BritainGuilds in early-modern Europe
    ↑ see it in the visualization
  18. Budget-maximising bureaucracy

    Niskanen’s model of bureau heads seeking larger budgets, which bring salary, staff, power and prestige.

    Because the bureau knows its costs and the legislature does not, it can offer a whole output for a whole budget, expanding until total cost equals total value. Output ends up too large, and the public surplus is used up.

    Niskanen, Bureaucracy and Representative Government (1971)

    ↑ see it in the visualization
  19. Government failure

    Collective action producing worse outcomes than the alternative, because of the incentives and information of political actors.

    Market failure is a reason to look at government action, not proof that it will help. The fair comparison is between two imperfect institutions.

    Buchanan & Tullock; Wolf, Markets or Governments (1988)

    ↑ see it in the visualization