Thursday’s exam is Sessions 1–6 only.

Closed book. No notes, no books, no browser tabs except Canvas.

On Thursday, each list has one option per question: you use each option exactly once. The exception is Part F, which has one extra option.

Today, by contrast, every exercise has one extra option. That way you cannot get the last answer by elimination: you have to know it.

Bribery and extortion can look identical: the same money, the same office, and in both the official may speak first. What tells them apart is whether the problem was already there. Sofia’s queue was real. The health inspector’s regulation was not.

The delay at the border is genuine, and the officer really does control it. Sofia is paying to move up a queue that already existed.

Compare this with Sofia. Her delay was real. This regulation was not.

The contractor never had the city’s money. He obtained it by making the invoice say something untrue.

Where this stops being ordinary lobbying is contested, and countries draw the line in different places.

That is why the previous picture is clientelism, and not because of the flour. The clues are the word again, the years on the banner, and the list. Ask what happens after the election: if the deal ends there, it is vote buying; if it comes back at the next one, it is clientelism.

Diego is paid with a job, and he is paid for the campaigning he did rather than for his vote.

Nobody hands over money here. The industry gets what it wants when the rule is written, and once it passes, what it protects becomes legal.

Compared with what? Usually one of two things: other countries at the same moment, or the same country before and after a change.

Comparing countries leaves in everything else that makes them different. Comparing before and after keeps the country itself fixed.

And when an author says he has taken income into account, he has removed what that one number explains, and nothing else. Whatever he never measured is still in there.

Both halves matter. Publishing a decision lets people find out about it, but if nobody can act on what they find, publishing is not enough.

She has paid $20 more than the law asks. She is worse off than if everyone had followed the rules, so she has a reason to complain about it.

She has paid $40 less than the law asks. She is better off than if everyone had followed the rules.

So she will never report it. That is why this one is harder to stop.

Ask who loses. Without theft it is Sofia, and she can be a witness. With theft it is the state, and the only two people in the room are both better off.

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

The fix looks obvious. Watch more closely, audit, raise the penalty. It works because somebody at the top wants it fixed.

Refusing costs you something and changes nothing, so you stop refusing. What makes sense for you depends on what you expect everyone else to do.

Almost every anticorruption reform is built for the top row. It assumes somebody in charge wants this stopped, and only needs better tools to catch the people below.

Persson, Rothstein and Teorell’s point is that in the bottom row that person does not exist.

What a clerk actually risks is not the fine on paper:

\[\text{what it costs me} = \text{the fine} \times \text{the chance anyone applies it}\]

The second group is the placebo. It shows that bad news in general was not what moved people.

The groups were formed at random, so nothing about the people explains the gap. Only the corruption figures do.

One caution. This is what people said they would do. Nobody watched anyone pay a bribe.

But countries differ in many ways at once, so you cannot tell which difference is doing the work.

Comparing countries cannot tell you what corruption costs. You need something that separates the cause from everything travelling with it.

The lottery decided which villages were audited, so nothing about a village explains the gap. Only the audit does.

Compare exam results against sixty things about students and something will line up: shoe size, birth month, backpack colour. The clue is how many comparisons were made.

“Systematic” does not mean “widespread”. It describes how the corruption keeps a party in power.

This is a different question from Session 3’s. With or without theft asks what the treasury ends up with. Venal or systematic asks what the exchange buys. One case can be both.

Ask what the exchange buys.

Venal corruption needs bad people. Systematic corruption needs only the arrangement.

That is why reform worked by taking away what there was to sell, rather than by replacing the people.

Items 2 and 3 are the ones to get right. Ask who made the problem. Item 3’s queue was there before anyone spoke; item 2’s offence never existed at all.

Left over: vote buying — a single purchase at election time. Item 7 happens after every election, and it is that repetition that makes it clientelism.

Jail everyone in case 4 and the arrangement still works. Do the same in case 3 and it stops.

Left over: state capture — that is about who writes the rules, not about which way a single exchange runs.

The label is worth one mark. The sentence in the right-hand column is worth the other two.

Left over: replace the official — Session 4 showed why. Fire the worst two and the replacements learn the price within a week.

In item 4 you have to say what the something else is. Answering “something else causes both” and stopping earns 1 mark of the 3. Naming “family income” earns the other two.

Left over: too few cases to tell — half a city’s bus routes, sixty explanations, a whole hospital system, a whole cohort of children. None of these is short of cases.

The problem: executives may be rating a country’s wealth, its press freedom, or its reputation rather than its conduct. If they partly read “free press” as “clean”, the finding is built into the way the score was made.

(b) Each country is compared with other countries, never with its own past. There is no before-and-after inside any country.

The rival: how strong the state is — how old the administration is, how well staffed, how closely it follows its own rules. Strong states print more newspapers and run cleaner offices. Allowing for income only removes the part of that history that one income number captures.

Two or three sentences for each answer.

1 a checklist for offices at risk · 2 why the payer keeps quiet · 3 why a tougher law changes nothing · 4 how economic favours kept a group in power · 5 bureaucracy and investment.

On the paper the options are worded differently and carry different letters. Match the idea, not the letter.