In my last video, I described corruption as a collective action or principal agent problem.  In this video, I use these models to discuss different ways to reduce corruption.  I describe different ways to resolve collective action and principal agent problems, use examples of anti-corruption reforms to illustrate them, and comment on how well each type of reform has worked.

As a reminder, we can think of corruption as a principal agent problem in the sense that someone is not fulfilling their expected responsibilities.  In the case of low-level (or petty) corruption, a bureaucrat is not doing their duty to provide public services without bribes (for example).  Or in the case of high-level (or grand) corruption, a leader is not fulfilling their responsibility to protect the public interest.

Corruption can also be a collective action problem in the sense that if it is prevalent throughout society (if it is a norm), then it becomes rational to engage in corrupt practices (like bribery).

These distinctions make a difference because they imply different solutions to the problem of corruption. If corruption is a principal-agent problem, then it can be changed by changing people’s incentives. But if corruption is a collective action problem, then you need to change overall norms.  Institutional change can often help with both.  Let’s get into more detail:

Principal agent problems are often resolved by changing individual cost-benefit calculation.  You can give people more incentives to do what you want and higher penalties if they don’t do what you want.  In the context of corruption, increasing bureaucrats’ basic pay scales reduces their incentive to ask for bribes, while prosecuting corruption cases increases the individual cost of engaging in corrupt practices.

Similarly, since principal agent problems are at core about contract violations, you can also resolve them by monitoring the agent more closely.  Just as you can (try to) make sure a contractor working on a home renovation does not run over their initial cost estimate by checking on their performance regularly, you can also allow more monitoring of the performance of government officials.  The most common way to do this is by requiring asset declarations – in many countries high level government officials need to submit asset declarations, forms (available to the public) where they have to state how many houses they own, which stocks they trade in, etc.  This makes it possible for people to see if officials have incomes much higher than their government salaries, or are profiting off of inside information.

Finally, one of the most effective ways to resolve a principal-agent problem is by changing the agent.  Corrupt officials benefit by staying in their jobs because they are able to build networks of allies, patrons, and clients of their own.  Rotating bureaucrats through different offices every few years can reduce corruption by breaking up those networks.  And when a leader is the source of corruption in a country, elections that allow for a change in office can similarly remove the problem.  The new officials or leaders may not be any more honest than the old ones, but the change at least means the new officials will need some time to build up new corrupt networks.

Which of these actually work?

There is strong evidence that raising salaries and rotation work very well at reducing corruption in a country.  The main problem with these policies is that they are difficult to implement.  Salary increases may be beyond the means of lower-income countries that collect little in government revenue.  Rotating mid-level staff can similarly be expensive, requiring additional training or relocation expenses. 

The benefits of rotating leaders show both why democratic institutions can reduce corruption and why it’s hard to make that work.  One reason why corruption is low in established democracies is because they regularly change leadership.  But corruption is often very high in new or weak democracies, and in these cases corruption undermines this aspect of democracy by allowing leaders to use their ill gotten gains to swing votes in their favor and never lose elections.

There is actually little evidence that monitoring alone reduces corruption.  Asset declarations can actually backfire by making everyone realize how common corruption is (creating a collective action problem). They tend to only work when paired with a free media (that can read and interpret the declaration, and an independent judicial system that can evaluate whether they were truthful and prosecute violations.  Similarly, prosecutions can be effective, but tend to require systemic change, which I’ll discuss in a few minutes.

Collective action problems can also be resolved by changing cost-benefit calculations (this is the main point of state institutions in the first place – to create a penalty for not contributing taxes to a public good).  But they can also be resolved through information sharing and creating focal points for action.

We’ll talk about collective action in its traditional sense (organizing protests and social movements) in a few weeks, but creating an information cascade – where you tip public opinion on an issue (like the possibility of a democratic transition) by widely sharing information – is one of the most common ways to resolve them.

Unfortunately, as with transparency and monitoring, information sharing alone rarely resolves collective action problems related to corruption.  Public awareness campaigns against corruption rarely have an impact – either because they make corruption look more prominent or because the public already knows that corruption is a problem and is wrong.  Even successful awareness raising campaigns like in India (which created an app for people to report on officials who asked for bribes) often have only a short-term effect (usage of the site has declined significantly since its peak in 2011).  As with all awareness raising campaigns, they tend to only result in change when accompanied by institutional reform.

And that’s where thinking of corruption as a collective action problem can be helpful.  When reformers are trying to generate collective action – to get an information cascade going – they don’t start just anywhere.  They will start at what are called focal points, building on a key idea, person, event, or organization to generate support more quickly.  If we are talking about political protest, organizers might choose a symbolic date to hold the protest on to get more attention.  In the context of anti-corruption reform, what this means is choosing key institutions to reform first.

Taking high-profile symbolic measures against corruption – like prosecuting a high level official for taking bribes – can be a great way generate a focal point in an anti-corruption campaign. But it will only lead to reform if it is connected to other systemic changes.  Prosecuting officials for corruption requires investigators to identify cases, trained prosecutors to bring legal cases against them, and independent judges willing to try those cases (amongst other things).  Since the regular legal system is often a source of corruption in countries, those seeking reform will often create a separate institution to bundle these functions together and prosecute not just one high profile corruption case, but many.

For example, in the early 2000s, Indonesia established the Corruption Eradication Commission (abbreviated KPK) that had investigative and prosecutorial powers to conduct sting operations against corrupt officials.  It succeeded in part because of support from civil society, international pressure, its expansive powers, and a smart strategy.  It started with prosecutions of mid-level officials.  This built public support (for effectiveness) without alienating the powers that be, putting it in a strong position to later successfully prosecute high-level officials, including relatives of the president.  That is the idea behind a focal point – you choose and empower an initial set of institutions to reform, then build upon that success to change not just public opinion but private practice.


This has been a quick introduction to anti-corruption policy.  The key things to remember are that the model you use to define corruption in a country – what you think the causes are – shapes what policies you will choose to implement.  Changing incentive structures can work for specific cases of corruption (particularly when you can remove leaders for corruption), but increasing monitoring alone rarely reduces endemic corruption.  Rather, it is often necessary to implement comprehensive institutional reform, starting with key functions (like prosecutions or contracting) and building support from there.