The Politics of Corruption

Lecture 4: Corruption as an Equilibrium: Norms and Collective Action

Bogdan G. Popescu

Tecnológico de Monterrey

The ask

Twenty dollars now, or a week she cannot afford to lose.
The fee is posted on the wall. The twenty dollars is for skipping the wait.
Nothing here is hidden. She knows exactly what is being sold.

And she is not alone in this queue. Behind her stand a shopkeeper, a cook and a driver, all with the same paperwork and the same officer.

Three answers, one at a time. Listen for the one who disagrees.

Everyone agrees

Sofia, the cook, and the driver all say the same thing.
Not one of them is defending bribery.
Put them in a survey and you would record three clean answers.

The norm is not missing, then. Is having it enough?

An hour later, at the window

The line has not gone away. Neither has the officer.

What Sofia is watching

She has not moved. The others have.
Nothing she believes about bribery has changed.
What changed is what she now expects the queue to do.

If everyone else pays, what exactly does her refusal buy?

What just happened?

Four people who all call bribery wrong. Four people who pay.
No one was argued out of their principles.
Each acted on what they expected the others to do.

Everyone kept the rule in their head. Nobody kept it at the window.

Is this hypocrisy — or is it a trap?

The Puzzle — and Today’s Route

Four people who all said bribery was wrong. Four people who paid. Not one of them changed their mind.

So the puzzle is not why people approve of corruption. Almost nobody does. It is why disapproval so often fails to stop it.

Four questions:

What makes a rule bind at all — and what went missing here?
Sofia is invented. Does any of this happen for real?
One dishonest official, or a trap everyone is caught in?
If it is a trap, what can a reform actually do?

I. Why Do People Join In?

What a norm actually is, why it can stop binding, and whether it travels with people.

Two questions you ask without noticing

Almost everyone agrees bribery is wrong. So the rule is not what is missing.

Then what makes a rule actually bind? Before following one, you quietly check two things:

Does anyone actually do this?
Would people mind if I didn’t?

Empirical expectation — what I think others really do.

Normative expectation — what I think others expect of me.

A rule grips people only when both answers are yes. Bicchieri calls that a conditional preference: I want to follow the rule provided others are.

Bicchieri (2006), The Grammar of Society, ch. 1.

Four borders Sofia could have walked into

Sofia is in the top-left box. One yes out of two — the rule is still in everybody’s head, it just no longer describes the queue. Both answers have to be yes.

Why three people were enough

Each person who pays makes paying more sensible for the next one in line.
So small changes bounce off: for years, the norm does not budge.
Then enough of them accumulate, the tipping point is passed, and the old norm goes all at once.

Tipping point — the moment when enough people have switched that switching becomes the sensible thing for everyone else too.

Burke & Young (2011), “Social Norms,” Handbook of Social Economics, ch. 8.

Flat, then a cliff, then flat again

Sofia did not need everyone to pay. She needed enough.

So why don’t people just copy the locals?

Move somewhere honest and you would expect to pick up honest habits. Often people don’t.
Simpser studies people born and raised in one country, under the same laws and the same police, whose parents came from different ones.
Their attitudes to bribery still track the parents’ country, a full generation later.
Staying close to family and community is what makes fitting in costly.

An attitude can outlive the place that produced it.

Simpser (2020), The Culture of Corruption across Generations. The cost-of-fitting-in argument is Lazear (1999).

Same country, different parents

The actual question (European Social Survey, 2004):

“How wrong, if at all, do you consider the following ways of behaving to be — a public official asking someone for a favour or bribe in return for their services?”

Answers: not wrong at all · a bit wrong · wrong · seriously wrong

What this does and does not settle

It does not make culture destiny. Simpser finds attitudes shifting across generations too.
They shift slowly where bribery is tolerated, faster where it is punished.
And it cannot explain Sofia: nothing about her upbringing changed between morning and afternoon.

Two forces, then: what you bring with you, and what the queue in front of you is doing today.

II. Does It Show Up in Behaviour?

How Can We Separate Rules from Norms?

People from corrupt countries also face weaker enforcement at home
So comparing countries cannot tell us which one matters
We would need people from many countries under identical rules
And a moment when the rules change but the people do not

Fisman and Miguel (2007) found a place where both were true.

Diplomatic Immunity as a Natural Test

A black sedan with a diplomatic CD licence plate parked beside a red NO STANDING ANYTIME sign outside United Nations headquarters, four parking tickets tucked under its windscreen wiper, the row of member-state flags and the UN emblem behind it.

Around 1,700 UN diplomats from 149 countries
Nearly all missions sit within a mile of UN headquarters
Same city, same streets, same parking signs
Before Nov 2002, an unpaid ticket cost them nothing
So how they parked revealed what they had brought with them

Design from Fisman & Miguel (2007).

Same city, same streets, same signs

Same streets, same signs, same wardens. Whatever differed, it was not the rules.

Three blocks by the East River

Where you may not stand

Germany’s diplomats: one ticket all year

Egypt’s diplomats: 141 each

The signs, the wardens and the penalty were the same for everyone here. Only the drivers differed.

150,000 tickets. $18 million. Nothing collected.

Unpaid parking violations by UN diplomats in New York City, November 1997 – December 2002.

Not one of them carried a penalty. Whatever governed how these diplomats parked, it was not the law.

Fisman & Miguel (2007). Around 1,700 diplomats from 149 countries.

When Nobody Was Watching

What the Comparison Can and Cannot Show

The pattern is real, but it is not perfect
Kuwait, the very worst violator, scored as relatively clean
Small missions own few cars, so they park badly less often
The researchers therefore compared countries adjusting for mission size

On that comparison, moving from Nigeria’s score to Norway’s implies roughly an 80% drop in violations.

Then Enforcement Arrived

What the Diplomats Teach Us

For five years there was no penalty at all. All that was left was what each diplomat brought with them.
And that differed hugely: 141 unpaid tickets a year, against 1.
Then the plates started coming off. Violations fell 98% in months.

Both mattered, at different moments: what they brought with them while nothing was enforced, and the plates once something was.

Still open: something travels with people. But what? A belief about others, a habit, or a conviction? That is the next study.

Does Believing It Is Spreading Make People Bribe?

The question: does hearing that corruption is rising make people join in?
The test: 4,200 people interviewed across Costa Rica’s main urban area
Some, chosen at random, were shown a flyer reporting rising corruption
Everyone was then asked whether they would bribe a traffic officer

Because the flyer was assigned by chance, any difference is caused by the information.

Corbacho, Gingerich, Oliveros & Ruiz-Vega (2016).

What the flyer actually said

A third of people saw this. A third saw nothing.
The last third saw a placebo: same design, equally bad news, but about slow courts.
The placebo changed nothing. So it is the message, not the bad mood, that moved people.

Believing Corruption Is Rising Makes People Willing to Bribe

What the Two Studies Add Up To

Diplomats acted on norms brought from home, where nothing was enforced
Costa Ricans became more willing to bribe once told corruption was rising
The two studies do different jobs — one tracks behaviour, the other isolates belief
And expectations can make themselves come true

Mind the gap: Corbacho measures stated willingness, not observed bribery. Fisman & Miguel measures real behaviour but cannot tell us which learned disposition is doing the work.

Next: if expectations are the problem, what should reform actually target?

III. Principal–agent vs. Collective Action Problem

The container clears. Now she needs a permit.

Sofia paid the $20. The barrier lifted; the cargo is through.
To move it out of the yard she needs a permit from the Northland Permit Office.
A different building, a different queue, the same question waiting for her.

Two versions of that office. Watch what changes.

Story A — one window asks

Four windows. Three of them process her neighbours without comment.
At window 2, the clerk names a figure: twenty dollars — the same price as the barrier.
Behind the glass at the back sits a supervisor who wants permits issued properly.
He simply cannot see what happens at the counter.

Ask yourself: what exactly is the problem here — and who could fix it?

Story B — every window asks

Same office, same queue, same Sofia. This time every clerk names a figure.
The supervisor has left his desk. He is at the counter, taking his share.
Reporting it would mean reporting to the man collecting the cut.
Refusing does not get her a clean permit. It gets her no permit.

Two Names for Two Problems

The same corrupt payment, two very different situations. Author’s summary of Persson, Rothstein & Teorell (2013).
	Story A	Story B
What goes wrong	One person hides bad behaviour	Everyone expects everyone else to cheat
Who can fix it	The boss, by watching more closely	Nobody acting alone
Honest people are	The normal case	Punished as fools
Political scientists call it	A principal–agent problem	A collective-action problem

Principal–agent problem

Someone hands a job to someone else — and cannot watch them do it.

The principal — the supervisor. Wants permits issued properly.
The agent — the clerk. Wants the twenty dollars.
They want different things, and only one of them is at the counter.
The clerk knows what happened there. The supervisor does not.

That gap has a name: information asymmetry. It is what makes the betrayal possible.

So the fix is obvious — watch more closely, audit, raise the penalty. It works because somebody at the top wants it fixed.

Rose-Ackerman (1978); Klitgaard (1988), who summarises the diagnosis as monopoly plus discretion minus accountability. Framing follows Persson, Rothstein & Teorell (2013).

Collective-action problem

Everyone would be better off if nobody paid. Each person alone is still better off paying.

Nobody is exempt — clerks, applicants and the supervisor are all following their own interest.
What is sensible for me depends on what I expect from you.
The more people expected to pay, the more sense paying makes.
Past a point it flips: enough others pay, and joining in is simply what a sensible person does.

And now there is no principal. Principal–agent theory assumes somebody wants the rules enforced. Collective-action theory asks whether anybody does — including the supervisor, who is taking a share.

Persson, Rothstein & Teorell (2013), drawing on Ostrom (1998).

Equilibrium

A situation nobody will step out of on their own — even when everybody would rather it were different.

Sofia would prefer an office where nobody pays.
So would the applicant next to her. So, most likely, would the clerks.
But given what everyone else is doing, paying is still her best move.
Refusing costs her the permit, and nobody else notices.

Add up everyone’s best move and you get a situation nobody chose and nobody can leave on their own.

So “just stop paying” is not advice. It asks one person to carry the whole cost of a change that needs everybody.

The collective-action reading of corruption: Persson, Rothstein & Teorell (2013).

IV. What Does That Mean for Reform?

If corruption is an equilibrium rather than hidden misconduct, what follows for policy?

Why Standard Reforms Can Fail

You report to someone who has already been bribed. And then it ends there.

The usual diagnosis: corrupt agents evade monitoring.
The usual remedy: more monitoring, stronger agencies, higher penalties.
But audits need honest auditors; reporting needs someone willing to act; punishment needs prosecutors willing to prosecute.
The problem: if the supervisor is taking a share too, who enforces the reform?

Interviews with 60+ Kenyan and Ugandan officials, journalists and NGO staff (Persson, Rothstein & Teorell 2013).

Kenya and Uganda Did Everything the Playbook Says

Kenya: an Anti-Corruption and Economic Crimes Act, a Public Officers Ethics Act, and the UN Convention against Corruption.
Uganda: an Inspectorate General of Government as lead anticorruption agency, plus an Auditor General and a Directorate of Ethics and Integrity.
Global Integrity rated both legal frameworks “very strong” — 100 out of 100.

Then it measured how much of it was actually being applied.

The gap between law and practice: “large” in Kenya, “huge” in Uganda. Since 1998, corruption in neither country has been anything but rampant.

Watching more closely is not wrong. It quietly assumes the very thing that is missing — somebody who wants to watch.

Global Integrity (2008), reported in Persson, Rothstein & Teorell (2013).

What Would Have to Change?

Behaviour — people must actually meet different rules and different consequences.
Expectations — people must have reason to think everybody else has met them too.

Suppose the police really have been cleaned up.
You believe it.
But you think everyone else still expects to pay.
So you pay anyway, and so does everyone reasoning the same way.

I have to believe the rules changed, and I have to think you believe it too. The second one is the hard part.

What Might Make a Reform Credible?

Visible — people can see that conduct has actually changed, not just that a statute passed.
Credible — enforcement really happens, including against insiders.
Coordinated — reporting, prosecution and staffing move together, rather than one rule changing on paper.

A prosecuted minister works mostly as information: it tells everyone that everyone else has now seen the same thing.

Careful. Collective-action theory predicts these three should matter. The studies in this lecture do not establish the best reform package.

What Do We Actually Know?

Where the findings stop and the prescription begins.
The evidence supports	The theory infers from it
Norms travel with people, even where nothing is enforced	Corruption can settle into a self-holding equilibrium
Telling people corruption is rising raises their willingness to bribe	Reform must shift what people expect of one another
Credible enforcement changed diplomats’ behaviour within months	Enforcement works through belief, not penalties alone
(no direct evidence here)	Coordinated overhaul beats isolated fixes

That last row is what Session 2 would call plausible but not established — and it is the row the whole reform argument rests on. Press it.

Exercise — You Take Over the Permit Office

Every clerk at the Northland Permit Office takes twenty dollars. Your predecessor took a share. Three months, no new budget.

Choose two:

Fire the two worst clerks
Put a camera at every window
Raise the penalty for taking a payment
Publish each week how many permits were issued without payment
Publicly prosecute your predecessor

For each one, answer:

What happens the next morning? Does a clerk actually stop asking for the twenty dollars — and why?
What would make it fail? Who could ignore it, sabotage it, or fake it?

Answers for the world they are in: every clerk takes, and the boss took a share.

1. Fire the two worst. The other two keep asking. Replacements learn the price within a week — and the supervisor picks them.

2. Cameras. The payment moves off camera: outside, by phone, “come back Thursday.” And whoever reviews the footage is taking a cut.

3. Higher penalty. Nothing happens. The old penalty was never imposed either. Cost = penalty × chance of enforcement, and the second number is near zero.

4. Publish the weekly count. The only one aimed at beliefs rather than behaviour. Two catches: the numbers can be faked, and if week one is near zero it advertises that paying is normal — Corbacho’s flyer in reverse. If the share climbs, it works.

5. Prosecute the predecessor. The only one that touches somebody above the clerks. Visible, credible, an insider — all three at once. Fails if it reads as a one-off.

Push on 2 and 3. Both sound serious; both die on the same question — who acts on it?

Takeaways

Everyone already agrees bribery is wrong. That was never the missing piece. What failed at Sofia’s window was expecting anyone else to refuse.
Both the person and the room matter. Diplomats kept their home habits once nothing was enforced; Costa Ricans told corruption was rising became readier to pay.
One corrupt clerk and a captured office look identical from outside. The first needs a watchful boss. The second has nobody left to watch.
So the diagnosis comes before the cure. Cameras and penalties fix the first office and bounce off the second.
And we know the trap is real better than we know the way out.

Corruption survives not because everyone wants it, but because everyone may be waiting for everyone else to stop first.

References

Corbacho, A., Gingerich, D. W., Oliveros, V., & Ruiz-Vega, M. (2016). Corruption as a self-fulfilling prophecy: Evidence from a survey experiment in Costa Rica. American Journal of Political Science, 60(4), 1077–1092. https://doi.org/10.1111/ajps.12244

Fisman, R., & Miguel, E. (2007). Corruption, norms, and legal enforcement: Evidence from diplomatic parking tickets. Journal of Political Economy, 115(6), 1020–1048. https://doi.org/10.1086/527495

Global Integrity. (2008). Global Integrity report 2008. Global Integrity.

Kaufmann, D., Kraay, A., & Mastruzzi, M. (2005). Governance matters IV: Governance indicators for 1996–2004 (Policy Research Working Paper No. 3630). World Bank.

Klitgaard, R. (1988). Controlling corruption. University of California Press.

References (continued)

Persson, A., Rothstein, B., & Teorell, J. (2013). Why anticorruption reforms fail—Systemic corruption as a collective action problem. Governance, 26(3), 449–471. https://doi.org/10.1111/j.1468-0491.2012.01604.x

Rose-Ackerman, S. (1978). Corruption: A study in political economy. Academic Press.

Rothstein, B. (2011). The quality of government: Corruption, social trust, and inequality in international perspective. University of Chicago Press.