The Politics of Corruption

Lecture 3: Incentives, Rents, and the Principal–Agent Problem

Bogdan G. Popescu

Tecnológico de Monterrey

Four questions for today

%%{init: {"fontSize":20,"theme":"base","themeVariables":{"fontSize":"20px","fontFamily":"Arial, Helvetica, sans-serif","primaryColor":"#eef1f4","primaryTextColor":"#1e293b","primaryBorderColor":"#334155","lineColor":"#334155","textColor":"#334155","edgeLabelBackground":"#f9fafb"},"flowchart":{"htmlLabels":false,"useMaxWidth":true,"nodeSpacing":50,"rankSpacing":70,"padding":34,"width":1150,"height":620}} }%%
flowchart LR
  A["1. Why can the<br/>officer ask?"] --> B["2. Why do both<br/>sides go along?"]
  B --> C["3. What are the<br/>warning signs of<br/>corruption risk?"]
  C --> D["4. One bribe<br/>or many?"]
  classDef step fill:#eef1f4,stroke:#334155,stroke-width:2px,color:#1e293b;
  classDef last fill:#e3ece9,stroke:#4a7c6f,stroke-width:3px,color:#1e293b;
  class A,B,C step;
  class D last;

By the end of today

Explain how a monopoly over a permit turns into a rent.
Read a bribe as each side’s best response, and weigh the gain against the expected cost of being caught.
Name the principal–agent–client problem behind an official’s discretion.
Score an office on monopoly, discretion and accountability — and say when discretion is safe.
Tell apart corruption with theft and without theft, and say which one spreads.
Say why gatekeepers who must all be paid take more, and leave less trade, than a single monopolist.

I. Why can the officer ask?

Sofia cannot go anywhere else

She cannot ask a different officer.
She cannot bring the goods in somewhere else.
She cannot wait: her customers will not.
He controls something she badly needs.

Whenever one person controls the only door, that person has power.

Why the officer has power

%%{init: {"fontSize":20,"theme":"base","themeVariables":{"fontSize":"20px","fontFamily":"Arial, Helvetica, sans-serif","primaryColor":"#eef1f4","primaryTextColor":"#1e293b","primaryBorderColor":"#334155","lineColor":"#334155","textColor":"#334155","edgeLabelBackground":"#f9fafb"},"flowchart":{"htmlLabels":false,"useMaxWidth":true,"nodeSpacing":40,"rankSpacing":60,"padding":32,"width":1180,"height":600}} }%%
flowchart LR
  A["Sofia needs her<br/>shipment released"] --> B["Only this officer<br/>can release it"]
  B --> C["She cannot go<br/>anywhere else"]
  C --> D["The officer<br/>has power"]
  classDef base fill:#eef1f4,stroke:#334155,stroke-width:2px,color:#1e293b;
  classDef prize fill:#f0e6d6,stroke:#b7943a,stroke-width:3px,color:#1e293b;
  class A,B,C base;
  class D prize;

Where does the bribe come from?

Sofia’s shipment at the border (dollars)
Sofia’s situation	Amount
The legal customs fee	$100
What a week’s delay would cost her	$500
What the officer asks for	$20

She will happily pay $20 to avoid losing $500.
That $500 is what the officer is really selling.

Two words for what we just saw

Monopoly — one gatekeeper, and no way around him.

A rent — money you earn by controlling a door, not by doing anything useful.

Rules create doors: permits, licences, inspections.
Many of those rules are necessary.
Every one of them still creates a door.

Both terms follow Shleifer & Vishny (1993, pp. 600–601) and Rose-Ackerman (1978, ch. 1).

So what?

We now know why the officer can ask.
That is an opportunity, not yet a bribe.
Somebody still has to say yes.
So why does Sofia pay, and why does he risk it?

II. Why do both sides go along?

Sofia’s choice

If she refuses

The shipment waits a week.
Her customers buy from someone else.
She loses about $500.

If she pays

The bribe costs her $20.
The shipment leaves today.

Paying is wrong. For Sofia on her own, it is also much cheaper.

Sofia and the officer, side by side

What each side gains or loses, in dollars
	Sofia pays $20	Sofia refuses
Officer asks for a bribe	Sofia −$20 · Officer +$20	Sofia −$500 · Officer $0
Officer just does his job	—	Sofia $0 · Officer $0

Sofia’s best outcome is the bottom row: nobody asks.
But once he has asked, paying beats refusing.
They land top-left — the cell neither of them wanted.

Source: Klitgaard (1988, pp. 69–70).

Why does he ask in the first place?

Figure 1

Best response — the choice that pays you most, taking the other side’s choice as given. He asks because he can already see what her best response will be.

Three reasons firms pay

“Pay $20 and I will stamp it today.”
“Pay me and I will make sure you win the contract.”
“Pay me and I will forget the tax you owe.”

Speed money · Buying an advantage · Buying non-enforcement

Categories after Rose-Ackerman (1978, ch. 1) and Shleifer & Vishny (1993, pp. 601, 604).

The boss cannot see everything

A minister cannot inspect every shipment herself.
So she hires officers to do it for her.
The officer is at the border. She is not.
He knows what happened. She reads a summary.

The person you trust to do the job may use the job for themselves.

Boss, official, firm

%%{init: {"fontSize":19,"theme":"base","themeVariables":{"fontSize":"19px","fontFamily":"Arial, Helvetica, sans-serif","primaryColor":"#eef1f4","primaryTextColor":"#1e293b","primaryBorderColor":"#334155","lineColor":"#334155","textColor":"#334155","edgeLabelBackground":"#f9fafb"},"flowchart":{"htmlLabels":false,"useMaxWidth":true,"nodeSpacing":85,"rankSpacing":165,"padding":32,"width":1150,"height":600}} }%%
flowchart LR
  P["THE BOSS<br/>the minister<br/>(principal)"]
  A["THE OFFICIAL<br/>the customs officer<br/>(agent)"]
  C["THE FIRM<br/>Sofia<br/>(client)"]
  P -->|"gives the job"| A
  A -->|"serves"| C
  C -.->|"bribe"| A
  A -.->|"reports"| P
  classDef principal fill:#e3ece9,stroke:#4a7c6f,stroke-width:3px,color:#1e293b;
  classDef agent fill:#f0e6d6,stroke:#b7943a,stroke-width:3px,color:#1e293b;
  classDef client fill:#eef1f4,stroke:#334155,stroke-width:2px,color:#1e293b;
  class P principal;
  class A agent;
  class C client;

Source: Klitgaard’s principal–agent–client model (1988, pp. 69–73). Dotted arrows are what the boss cannot easily see.

What he gains, what he might lose

What the officer gains

The bribe: $20.

What the officer might lose

Chance of being caught: 1 time in 10.
Fine if caught: $100.
So on average the risk costs him $10.

$20 gained against $10 expected to be lost. He takes the bribe.

When is the bribe too risky?

Figure 2: He refuses once the expected punishment — the chance of being caught times the fine — passes the bribe.

So what?

Sofia pays because refusing costs her more.
The officer asks because the risk looks small.
We need not assume either is unusually wicked.
Change those numbers and the behaviour changes.

III. Three warning signs of corruption risk

Three warning signs of corruption risk

No alternative. Can the citizen go somewhere else?
Unclear rules. Can the official decide almost anything?
Nobody checks. Will anyone find out, and act?

Sofia’s border post scores badly on all three.

A way to remember them

more monopoly + more discretion − less accountability = more corruption risk

Robert Klitgaard writes this as C = M + D − A.
It is not real mathematics. It is a memory device.
The terms have no units; you cannot add them up.
It tells you where to look, not how much there is.

Klitgaard (1988, box 12, p. 75) calls it “a stylized equation” — his own wording is the warning.

The three signs at the border

What each warning sign directs you to look for
Warning sign	The question	At Sofia’s border post
Monopoly	Can she go elsewhere?	One lane, one officer on duty
Discretion	Who decides, and on what basis?	He picks which trucks to open
Accountability	Who checks, and what follows?	No records, no rotation, no appeal

Source: Klitgaard (1988, pp. 74–75, and box 15, p. 88).

Exercise 1 · Read the office

A city office issues licences for food stalls. One inspector covers the whole market. He decides whether a stall is “clean enough”. He writes the result on a paper form he keeps in his desk.

Three warning signs are hiding in that description. Name them.
Which of the three is the most dangerous here — and why that one?

5 minutes · one pair reports, the others say where they disagree

Question 1 — one sign per sentence.

One inspector covers the whole market → monopoly. A stall-holder who is refused has nowhere else to go. That is what makes any threat credible.
He decides whether a stall is “clean enough” → discretion. There is no published standard, so “clean enough” is whatever he says it is. Two identical stalls can get opposite answers, and neither owner can show he was wrong.
A paper form he keeps in his desk → accountability. Nobody else sees the decisions, so nobody can compare them, count them, or notice a pattern.

Question 2 — no single right answer. Accept any of these if it is defended:

Monopoly, because she cannot walk away. He can simply wait her out.
Discretion, because “clean enough” is where the money is. He can invent a problem out of nothing, or overlook a real one.
Accountability, because it is the cheapest to fix here, and because it is what makes the other two survivable.

The strongest answers notice that the signs multiply rather than add: discretion is manageable when somebody checks, and a monopoly is manageable when the rule is clear. Here neither holds.

Two things to push back on:

Anyone who makes it about the inspector’s character. Nothing in the description says he is dishonest — the point is that this situation would tempt an honest person.
Anyone who concludes “just write a rulebook”. Hold that thought: the next slides show that taking discretion away is not automatically the fix.

Is letting officials choose always bad?

“Unclear rules” says an official should not decide freely.
But you want the bridge built by someone who can build it.
The cheapest bid is not always the best builder.
So: does letting officials choose let dirtier firms in?

Decarolis, Fisman, Pinotti & Vannutelli (2025, pp. 214–216). Over 200,000 Italian public works contracts, 2000–2016.

Discretion is much less dangerous when competition stays strong

Figure 3: Choice alone: almost no change. Choice with few bidders: two more contracts in every hundred.

In these contracts, letting an official pick the winner mattered little — as long as plenty of firms were still allowed to bid.

Checking everyone is expensive

Auditors, inspectors, cameras and courts all cost money.
Chasing the last small bribe costs more than it saves.
Klitgaard’s blunt version: the best amount is not zero.
So aim the effort where the harm is greatest.

Source: Klitgaard (1988, ch. 2, pp. 24–27). He means this as a budgeting point, not as an excuse.

So what?

Three warning signs let you inspect any office.
Letting officials choose is not the danger by itself.
Choice with no competition is the danger.
Now: what if Sofia faces more than one officer?

IV. One bribe or many?

Bribery when the state still gets paid

He records the shipment and charges the $100, exactly as the rules say. He cannot waive that fee — but he does decide when the goods move. The $20 buys today instead of next week.

Sofia pays the legal fee, and a bribe on top (dollars)
Who	Amount
Sofia pays in total	$120
The government receives	$100
The officer keeps	$20

Sofia is $20 worse off than if the rules were followed.
She has a reason to complain.

Bribery when the state gets nothing

This time he never writes the shipment down. No record, so no $100 fee is ever owed. He waves the truck through and keeps $60 for himself.

The officer writes nothing down and keeps the lot (dollars)
Who	Amount
Sofia pays in total	$60
The government receives	$0
The officer keeps	$60

Sofia pays $60 instead of $100. She is better off.
Now nobody in the room wants to complain.

Where the money goes

Figure 4: When the fee is skipped, the loser is the treasury — and it is absent.

Why almost nobody reports it

When the fee is skipped, Sofia saves money too.
An honest importer who pays $100 is undercut.
To survive, she must find her own friendly officer.
Both sides now want the same thing: silence.

When the state is robbed, both sides come out ahead — so both sides want silence.

Source: Shleifer & Vishny (1993, p. 604). Their fix: first make theft from the state visible in the accounts.

Sixty tolls on the Rhine

A merchant sails down the river. Sixty different rulers each stop him and demand a toll.

What happens to the final price?
Who stops making the journey at all?

Pfalzgrafenstein Castle, a white toll fortress with a pointed tower, standing on a small island in the middle of the Rhine River at Kaub, Germany, forcing all passing river traffic past it.

Pfalzgrafenstein, a toll station built in the middle of the Rhine at Kaub (from 1326). Photo © Gary Bembridge, CC BY 2.0, via Wikimedia Commons.

Around 1400 the Rhine carried some 60 independent tolls; English rivers carried none. Reported in Shleifer & Vishny (1993, p. 608), citing Heilbroner (1962).

Three ways to arrange the offices

%%{init: {"fontSize":20,"theme":"base","themeVariables":{"fontSize":"20px","fontFamily":"Arial, Helvetica, sans-serif","primaryColor":"#eef1f4","primaryTextColor":"#1e293b","primaryBorderColor":"#334155","lineColor":"#334155","textColor":"#334155","edgeLabelBackground":"#f9fafb"},"flowchart":{"htmlLabels":false,"useMaxWidth":true,"nodeSpacing":45,"rankSpacing":150,"padding":30,"width":1180,"height":640}} }%%
flowchart LR
  A1["EITHER WILL DO<br/>two offices,<br/>either can approve"] --> A2["Bribe near zero<br/>Most trade happens"]
  B1["ONE OFFICE<br/>one office approves<br/>everything at once"] --> B2["One medium bribe<br/>Some trade happens"]
  C1["ALL MUST AGREE<br/>five offices,<br/>each can stop her"] --> C2["Bribe after bribe<br/>Least trade happens"]
  classDef good fill:#e3ece9,stroke:#4a7c6f,stroke-width:3px,color:#1e293b;
  classDef mid fill:#f0e6d6,stroke:#b7943a,stroke-width:3px,color:#1e293b;
  classDef bad fill:#f4e0da,stroke:#b44527,stroke-width:3px,color:#1e293b;
  class A1,A2 good;
  class B1,B2 mid;
  class C1,C2 bad;

Source: Shleifer & Vishny (1993, sec. III, pp. 604–608). The Rhine tolls are the third case.

When every office can say no

Figure 5: Each office adds its own demand, so the merchant pays more and fewer shipments are worth making.

Exercise 2 · One river

Figure 6

One merchant, one river. Her cargo is worth a number between $1 and $100 — it is on my card, and you do not get to see it.

She finds out what the tolls add up to before she sets out. So it is the whole trip or none of it.

Tolls under the card → she sails → every toll gets paid
Tolls over the card → she turns back → nobody gets paid

What you need. A card with a number from 1 to 100 for each round (pick them before class, keep them face down). One scrap of paper per team. Two columns on the board.

How many teams. Five is what the figures assume — with about 25 students that is five to a team. Four to six behaves the same way. The equilibrium toll is $100/(n+1) each, so five teams is about $17 each and $83 in total.

The rule that matters most. If she turns back, nobody gets paid. Say it twice. The next-but-one round does not work unless every team has that clear before they write anything down.

The question a sharp student will ask. “Doesn’t she pay the first station and turn back at the third?” No — and answer it, or they will debate it instead of the point. She learns the going rates before she loads the boat, so she either makes the whole trip or does not start, and a half-trip pays nobody. Sequential tolls collected along the way are a hold-up problem — real, but a different model from the one on the previous slides.

First: just one toll station

Figure 7

What should it charge?

Too much and she turns back with nothing paid. Too little and you left money on the table. The best answer is about $50 — she sails roughly half the time.

Now a toll for every team

Figure 8

Same river. Same merchant. One thing has changed: now she has to get past all of you.

Your team is one station. Write your toll. Do not show it. Do not talk to the other teams.

We reveal together and add them up. We play this twice.

Scenario 1 · The ship passes

Figure 9

The card said $72. The tolls came to $62.

She sails — and every team is paid.

Scenario 2 · The ship turns back

Figure 10

The card said $72. The tolls came to $110.

She turns back — and nobody is paid. Not one team.

Same arithmetic, out loud: $20 plus $25 plus $15 plus $30 plus $20 is $110, which is over the $72 on the card. The voyage never starts, and every team earns zero — including the team that only asked for $15.

This is the slide to come back to after the round. When their own numbers add up past the card, put this back on screen and ask:

“Which team made an irrational decision?” On its own, none of them. $20 is a modest toll.
“Then why did every one of you earn nothing?” Because each team set its toll without counting the other tolls stacked on top of it. Every team made the voyage less likely for all of them, and no team paid that cost by itself.

Then play it again. They moderate without being told to, the voyage survives, and everybody collects — which is Scenario 1. That second play is the learning; do not skip it to save two minutes. Point out what they have just done: the tolls went down and the teams earned more.

When somebody asks what the solution is — and somebody will — do not go down the equilibrium road with a first-year class. There is no tidy answer: if the others are charging $17, your best reply is to charge a bit more, not less, which is exactly why the total blows up. Say this and move on: the point is not how to solve this game. It is that when everyone sets their own piece separately, the total blows up.

Two questions to finish on, then straight into the next slide.

Which arrangement asked for more in total — the single station, or a toll for every team? Every team, by a long way: about $83 against $50.
Which one actually paid the toll-keepers more? The single station.

That is the whole result. Spread across the teams the demands were higher and the takings were lower, because they destroyed the voyage that would have paid them. More power to block is not more money.

And the line to leave them with: the same teams sat in the same seats the whole time. Nobody got greedier between the two arrangements. Not the people — the structure.

So what?

Splitting corruption between many hands makes it worse.
The many offices even collect less in total.
A second office issuing the same permit cuts monopoly power.
Elections and a free press work on accountability instead.

Source: Shleifer & Vishny (1993, pp. 605–611). Note their uncomfortable warning about decentralisation reforms.

Five questions to ask about any office

Monopoly — what does the official control, and can the citizen go elsewhere?
Discretion — who decides, and against what written rule?
Accountability — who can find out, and who can punish?
Theft — does the state still get paid, or does the fee disappear?
Many hands — how many officials can block the same transaction?

Corruption is not only about bad people. It is about institutional design.

References

Decarolis, F., Fisman, R., Pinotti, P., & Vannutelli, S. (2025). Rules, discretion, and corruption in procurement: Evidence from Italian government contracting. Journal of Political Economy Microeconomics, 3(2), 213–254. https://doi.org/10.1086/732654

Heilbroner, R. L. (1962). The making of economic society. Prentice-Hall.

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

References

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

Shleifer, A., & Vishny, R. W. (1993). Corruption. The Quarterly Journal of Economics, 108(3), 599–617. https://doi.org/10.2307/2118402