Lecture 25: Policies, Politics: Can Evidence Play a Role in the Fight Against Poverty?, cont.

• Download English-US transcript (PDF)

The following content is provided under a Creative Commons license. Your support will help MIT OpenCourseWare continue to offer high-quality educational resources for free. To make a donation or view additional materials from hundreds of MIT courses, visit MIT OpenCourseWare at ocw.mit.edu.

PROFESSOR: Might as well get started.

So this is all carrying on the conversation that we had last time. We talked about, possibly for very long-term historical reasons, some countries might have very different kinds of institutions from others. In particular, some might be much worse than others, at least using the standard measures. And then what do you do about it?

So the first reaction is, sometimes, well, tough. If the story is that you had to have some lucky break 400 years ago to get the institutions right, then it's not entirely clear that there's much you can do about it. That's the pessimistic view.

Now, for the people who really worry about these issues, there's an interesting debate going on. It's really between these five people listed above, are on the two sides of the debate. One side of it is a Stanford professor called Paul Romer, who has come up with this idea of charter cities, like charter schools.

So he wants, if you'd like, to import institutions. So his basic idea is, the US has good institutions, so a US firm, or a US NGO, would start-- if South Sudan wants to have a city, then they should just hire a US firm to set up the city for them, have the same guarantees that the US has. So it'll be a part of the US institutional frame.

So it's not clear exactly how that works. Because for example, suppose you set up a court in that country, and it has the US laws, but the judge in the court decides not to implement those laws. Does the US invade? Who's making the commitment? So this is a set of real issues about how you get that to work. Yeah.

AUDIENCE: Isn't there even more fundamental issues of what foreign country would be willing to sponsor a charter city? Because seemingly, all the advantages, even if you could find a country to host it, like a country that would be willing to give up the land, the outside sponsor, for lack of a better term-- at least the way you guys talk about it in the book-- they basically give it up as soon as the city's kind of successful.

He alludes to Hong Kong being an example. But Hong Kong had very different circumstances, because Britain originally colonized it in the hopes of having it as its own, having it as a commercial center. And then they got the benefits for 100 years before they handed it over.

PROFESSOR: And equally importantly, they handed it over under circumstances when they really didn't have any choice. So Britain is certainly in no position to, let's say, disagree with China about a city that's next doors to China. So there was no leverage is there whatsoever.

Yeah. So there's several issues there. You've just brought up one, which is, how do I make the, to use your word, the sponsoring country, give up the city. Then there's the opposite problem. How do I make sure the city actually runs, and who's guaranteeing it?

Suppose the judge who's supposed to enforce US laws in, I don't know, Benin, doesn't do it, who's liable to make it happen?

AUDIENCE: But there's no one preventing those countries [INAUDIBLE] extracting institutions, they have no incentive to be [INAUDIBLE] either.

PROFESSOR: Right.

AUDIENCE: So in the same was as colonies where Europeans did not settle permanently [INAUDIBLE]. They set up [INAUDIBLE] institutions that would be [INAUDIBLE].

PROFESSOR: So you're right. There's the question of the demand side for it. Maybe there's no country that wants it. I think Paul's presumption is that there are countries that would like to have better institutions. They cannot domestically manufacture it. But if they just somehow imported institutions, and there was a [? guarantor ?] behind it, then it would work. So what you're saying is completely right. There needs to be a demand for it. If the country doesn't want it, it's never going to happen.

But let's say the country even wants these institutions. Even in that case, there are several problems. One is, how do you actually run a charter city? I mean, who's in charge? Is the US prepared to back up its commitment to the charter city? If it says the US laws, does that mean the US is going to send in an army whenever the laws are violated?

AUDIENCE: Likewise, does it mean that if the country gets overrun by an outside invader and the charter city gets overrun, will the US extend its nuclear arms-- like, let's say--

PROFESSOR: Right. Exactly. So [? what is ?] commitment to it.

AUDIENCE: Foreign policy.

PROFESSOR: Absolutely. So all forms of commitment are an issue, obviously. It's not enough to say that we have US-style laws. In fact, if you look across the world, laws don't vary a lot. Legal systems vary relatively little. What varies is the enforcement of the law. So you have to make-- who's watching that the laws are getting enforced? Yeah. Melissa.

AUDIENCE: I guess as an example of this on a much smaller scale is you have these cruise lines set up these enclaves in Caribbean islands. And it'd be a town in Haiti, and it functions much better than the rest of the country. And part of that's because they have economic demand. But the cruise line has basically bought the town and runs things. So like I said, it's all on a much smaller scale. But it seems like that that's the only thing.

AUDIENCE: In those instances, generally speaking, the cruise line owns all of the land. And it's private property. And they setup the town. So it's really no--

PROFESSOR: It's like a hotel.

AUDIENCE: It's like a privatized government.

AUDIENCE: Yeah. It's like a large hotel.

AUDIENCE: But it's larger than a hotel.

PROFESSOR: But a large hotel. Some hotels have their internal currency. I was once trapped into this completely horrible hotel which had this internal currency with high inflation. This was a hotel. Because whenever I wanted to buy something-- so basically, there you go into the hotel. It's like a resort. And they give you a certain amount of currency, and you spend it anywhere you want in the resort. But it turns out that the prices are extraordinarily high. So that's sort of the same idea.

AUDIENCE: It's quite different. But I guess it's just that we don't really have anything on a smaller scale to this that we could even say, does it work on a smaller scale?

PROFESSOR: So why is it different from a resort where you can go from one shop to another? The shops are competing. Why is it different?

AUDIENCE: At least one thing that's different is there's a major industry there that creates demand. Whereas if you set some city up in the middle of Africa, where there's not already an economic industry, and everybody's--

PROFESSOR: So part of it is that there is an economic engine that's already there. Then there's a chicken-and-egg problem, which is that the city will only generate value if it comes with an economic model, and the economic model isn't there already. If you just guarantee the institutions, how good does the guarantee need to be before the economic engine arrives? That's a challenge.

As you might imagine, the problem might be that if I tell you that I'm in Cote d'Ivoire in the middle of the civil war, but we have this enclave that's protected, you really need very good assurances of protection before some industry's going to be willing to move to this enclave in Cote d'Ivoire in the middle of the civil war. So you're always this question, who guarantees?

But also, I think there is a third question, which is, what's to stop the country from withdrawing its concessions? So let's say the country gave you some land. You build a lot of buildings. You've got a few factories running there. What's to start a country with bad institutions with coming back the next year and saying, well, now that we have all of these factories, I want them. And who's going to stop that? Will the US send in an army to stop that? It seems like there are myriad issues of implementation. It would make this a rather wacky scheme.

AUDIENCE: [INAUDIBLE] also the fact that [INAUDIBLE] affect the rest of the country. [? They're ?] going to build a whole new city [INAUDIBLE]. And all the existing cities and things surrounding that will not necessarily be receptive to it. [INAUDIBLE].

PROFESSOR: Yeah. That too. Just how do you deal with the inequality and all that within the country? Yeah.

AUDIENCE: So when Paul Romer proposed this, it sounds very theoretical, and I don't think he actually proposed how to implement it. But did he mention the timeline for these cities? Because although you might have the framework for good institutions, you might have all the laws, you still need the people to implement it, the people to respect it. And that's somewhat a cultural issue, that if you grew up in a corrupt society, you have it ingrained in you how to go about things.

So it would take at least one generation, or several generations, to get people to change their ways and actually follow the law and help make these institutions work. For example, Russia transiting from the Communist society to democracy, they have all the laws and all the rules and all the institutions. But people still have this old mindset that's not working.

PROFESSOR: That goes back to the first point we were saying, which is that the evidence on institutions says that if you got the right accident which made your institutions the right ones 200 or 400 years ago, then that persists. That doesn't say anything about transplanting institutions.

But I think Romer is asking the right question, which is, imagine that the US has good institutions. Can a US company take advantage of the fact that it is a US company, and the US has good institutions, to set up a kind of a charter city in Cote d'Ivoire which will have good institutions?

So he's thought about this issue. He thinks that, well, the way you do it is, you import institutions. That's the idea. And so in some sense, you're totally right. It's not enough to just write down the rules of the institutions. And that's completely right. But he's not saying that. He's saying that plus implementation will come from US company. The question is whether a US company can implement anything like that. Yeah.

AUDIENCE: Couldn't to some degree-- I think it'll be interesting to see in 20 or 30 years, what Afghanistan and Iraq look like? I mean, obviously, they had a pretty distinct population. But to some degree, the closest thing to what Romer suggested we're seeing in action in Iraq.

I mean, the level of US monetary involvement has been huge [INAUDIBLE] military expenses. And they're trying to create these new institutions. So that could be an interesting example case to build off of in the future, when changed institutions on a wide scale he was talking about. I mean, obviously, it's not very sustainable, I imagine. And there will be a series of issues. But just as an example.

PROFESSOR: This is an aside, but it's worth reading the book, I think it's called Imperial Life in the Emerald City, or something. Wonderful book. The Washington Post bureau chief in Iraq, after a year and a half there, wrote this book about the implementation of US institutions in Iraq.

And what was wonderful about the story in that book was-- and this is relevant for what we're going to talk about in a bit. So it's not entirely an aside. So the book was wonderful in describing exactly how the US went about implementing good institutions in Iraq.

So basically, rule one was that the people who got hired to do that were either staffers of prominent Republican congressmen or senators, or prominent contributors to the Republican cause. Those were the only people who were selected.

Second, most of the staffers who were given the responsibility for, for example, designing the stock-market laws, were 23-year-olds who had an undergraduate degree in accounting were sent to the set up these laws. And there was usually one person. A 23-year-old with an undergraduate degree was assigned to be in charge of setting up the stock market or figuring out how to reform the legal system.

So these where people with strong background in being Republican operatives, and no background in actually running an institution. So the book is wonderful in that it goes through how the whole process was doomed from the start. Because the people who were sent to do these institutions had no competence whatsoever in doing the job, because they were all political appointees.

And the book is actually worth reading. And it comes back to point I'll come back to later. Which is that, imagine you wanted to make institutions better. You still need competence. And it's not just a matter of having some playbook from the US. It's a matter of actually having the competence to turn that playbook into reality. And I think that was a very major problem in Iraq.

So then there is Paul Collier, who has a different view. In some ways, this view I find less unrealistic-- maybe less defensible, but less unrealistic. Which is, if the country has really horrible institutions, just invade. This is, I think, internally coherent. Much more, I think, internally coherent than the Romer suggestion, which I don't see how it would be implemented. But this one, at least the policy prescription's simple. Invade, and then just set up colonial rule for a while.

This has been tried. It's been implemented in the past. We know how it's to be done, so there's no implementation problem. You might have various objections to it. But in many ways, this is a more difficult one to challenge on purely a priori grounds.

One of the sad facts is that we know very little about the effect of colonial rule from history. That is to say, if you look at places that didn't get colonized, so few of them didn't get colonized, and they're so unique, like China is one country that didn't get colonized. But basically, poor countries, most countries outside Europe and North America were colonized. And as a result, there's just very little. And when you compare the ones that got colonized with the ones that didn't, the ones that didn't get colonized were Japan and China. And those are not random draws from any population. So it's very hard to see what the effect of being colonized is.

The one exception to that that I know, which is not quite perfect, but it does a little bit of that, is, an MIT Ph.D. student some years ago, Lakshmi Iyer, did this paper where what she did is, she looked at, in India, there was what was called the Doctrine of Lapse. The Doctrine of Lapse was a beautiful idea.

So the way India was colonized was bit by bit. Some bits got left out. At any point, they would fight a battle, take over some land, and then some would be left out. And then they'd have a treaty with the local potentate at that time to have a kind of a dependent ruler.

Basically, this process of colonizing India ends in 1857 with about 2/3 of India under direct British rule, and one third of India under rule of a whole bunch of rulers-- 500-odd rulers who had a little small territory, and was the king of that territory, and had a treaty with the British. The British basically were overseeing this person. So that's how the colonization happened.

Now, one of the ways in which the British would take over one of these little kingdoms was under what was called the Doctrine of Lapse. The Doctrine of Lapse was, one might say, a cynical idea. But it was the idea that if the king didn't have a natural heir, then the kingdom would lapse to British rule.

So basically, what this paper does, is she compares places which, during-- Doctrine of Lapse was enforced very heavily for eight years, from 1848 to 1856. During this period, about 20 places had the king die. And I guess some of those places had a male natural heir. Natural heir means a son. And the other places didn't.

So what she does is, she compares the places which had a natural heir with the places which didn't have a natural heir. So if you look at these places, if you think that the birth of a male child is kind of a random accident, then you might say that these are comparable. And on most things, they look comparable.

If you look at these places 150 years later, the places that were under British rule do worse in almost every attribute that can be measured. Less good infrastructure, less good education, less good other things. So that's the only piece of evidence I've seen that has some bearing on this question. It's still a very special set of 20 places. So I would say, we know very little about even the economic effects of colonialism.

And then there are all the political effects, which are that colonists often create hierarchies, and those hierarchies create all kinds of resentment, and a whole bunch of other stuff that potentially can go wrong. But I don't think there's a great natural experiment that tells you what would have happened had Algeria not been colonized. I don't think anybody knows the answer. Yeah.

AUDIENCE: [INAUDIBLE] the fact that everywhere from the Congo to the US and Canada have been colonized suggests that it doesn't matter so much on the nature of it. Why do we want to make a general statement?

PROFESSOR: I guess the question is whether or not once you interact race with colony, you get much worse outcomes. You almost surely do. The question is whether that's interpretable or not. If I said, take colonies which are nonwhite colonies, they do a lot worse than white colonies.

Now, what that means I don't know. But I do think that you could make a firm hypothesis, which I don't think is testable given the data, that if you were nonwhite and you were colonized by white people, you ended up badly. You could make the hypothesis. It wouldn't be rejected.

AUDIENCE: [? Half the ?] people in the US who were not white who were colonized by white people, it still ended up badly, economically, relative to white Americans.

PROFESSOR: Exactly. So I'm not saying that people resolved this question. I'm saying the hypothesis would not be prima facie rejected by the data. But my point is only that we don't know. I wasn't saying more than that. I would just say that it's not implausible hypothesis, but one that I don't think we know enough to answer Collier's question.

I actually don't think that this is a realistic idea, because I don't think anybody's prepared to actually-- I think if you look at the US, the real problem with getting the US to be a colonist is that the US actually loves the idea, maybe, of being influential in the world, but it doesn't actually like the idea of having young US soldiers fighting to protect some foreign land from its own internal conflicts. That is not a politically popular idea.

So I don't actually think any country is volunteering to be a colonist. And that might be a good thing, but it is certainly true that, I think, after the Vietnam war, the US has had enough of long-term colonialism. I think that's the given. So I don't think there is any country that's volunteering.

So again, I don't think Collier's solution, whether it's right or wrong-- I'm inclined to believe it wouldn't work, even if you tried, but that's a [INAUDIBLE]. I don't think there is any evidence that anybody's willing to do it. So it's maybe moot.

And then there is Bill Easterly, who's been very critical of both Romer and Collier. And his position is very internally consistent. His view is that there's nothing you can do to help countries. Countries have to help themselves. There is no sense in which there is really any outside wisdom that's useful, with an important caveat I'll come back to in a minute. But therefore, colonialism won't help, because countries help themselves. They solve their own problems. You can't really help any country.

He doesn't quite stop there, which would be really internally consistent-- whether right or wrong is a different question. He says, except you should have free trade, democracy, and capitalism. Those are the three things that he's in favor of. And then he's not really in favor of anything else. So he does believe that the country should not have choice on those dimensions, but once you have implemented those, you can leave them alone.

Now, again, the problem is that there is not much evidence-- I mean, certainly some countries have figured it out. It's not clear that there's any evidence that the process of figuring out is a particularly efficient process. We've talked a lot through this semester about policies that were tried and didn't work, and they didn't work for very good reasons. So in other words, there is a lot of knowledge out there that's useful for countries to use. So it's not clear that you want to take the position that countries need to make every mistake and figure it out.

And it's not clear what a country means. Is it really the case that, if I think of Tunisia under Ben Ali, there was a collective experiment in dictatorship? It doesn't seem like that's the right description. So a country is just a set of people.

But it's not clear that there is any organizing body which is implementing a national experiment. I mean, maybe some countries are, but most countries, when they fail, they seem to be not doing anything particularly interesting. Some guy is stealing a lot of money, and the rest of the people are suffering. It's not clear that's a symptom of the process of figuring out where you're going, rather than just a symptom of failure.

And then, I think the most internally consistent are definitely our colleagues Acemoglu and Robinson, whose view is very much that there's nothing you can do to help, but also that mostly nothing good will happen. So their view is, I think, most solidly internally consistent. Basically, countries are probably screwed. Yeah.

AUDIENCE: I don't think that's a fair assessment. Jim Robinson's doing all this stuff in Sierra Leone now with the World Bank and [INAUDIBLE].

PROFESSOR: But I know Jim very well. I know Jim very well. I think he would say he's a pessimist. He might be trying to do something, but I think he's-- I would say that that's the most internally consistent of all of these positions that you see, is the one of saying that, look, if I really believe that institutional changes are very difficult, and they only happen because of major historical events, then it's internally consistent be pessimistic. Because why would we have much hope?

And I think the example they give of changes are the French Revolution and the Glorious Revolution. That's the two examples that they mostly spend a lot of time talking about. So their view is that change is possible, but change happens when internal forces within a country just happen to be aligned in the perfect way.

And they give these two examples of the Glorious Revolution in Britain in 1688 and the French Revolution in 1789 as the two examples of where you just had the right alignment of forces. And then the internal compulsions of the country just drove the country towards good outcomes. That in each case, these were just moments of liberation, and they did wonderful things for the country.

There are many, many other such revolutions, and they have been less successful. The Russian Revolution, the Mexican Revolution, the Chinese Revolution. There are many others you could go through, and they have been less successful, clearly.

So their view is that it's not enough to have revolutions. You need to change societies massively. You need to have revolutions. But revolutions are not easy. Mostly they go wrong. It's only when you have the lucky matching of different forces that you get the right outcome.

And therefore, mostly there's not much you can do. You can get lucky. If you get lucky, things will work. If they do not, there's not much you can do about it. And that's very consistent with the view that they have from their evidence that all we really know is that, if countries got it right several hundred years ago, then that seems to persist. So when good things happen, good things build on each other, but we have no recipe for creating good things.

So that's a very, very pessimistic and tough-minded, but, I think in some ways, admirably internally consistent point of view.

AUDIENCE: I think [INAUDIBLE] has even tried to see [INAUDIBLE] institutions, changing institutions, is that US conventions, which are in the treaties, even on things like child labor or [INAUDIBLE]. So in your assessment, or your generally economic [INAUDIBLE], is it part of the reason, in many case, not effective enough, or [INAUDIBLE] will not be effective?

PROFESSOR: So I'm going to actually argue that lots of institutional changes are possible. Just, they're possible at a different level from the ones that we are talking about. And so I would agree that many things change within countries.

Indeed, I would say that if you look at status of women, the standing fact about the world is how much has changed over the last 150 years, not that there's not a lot of change. And I think the change has happened in China as much as it's happened in the US. And China was an extraordinarily, I would say, anti-woman country. And you see massive changes going on.

So I'm going to argue mostly that there is actually lots and lots of possibilities for changing institutions, just that you don't-- this is asking a different question. So what we want to argue in the book is, lots of things change. They don't necessarily change wholesale, and you don't necessarily get better institutions in any macro sense. But that doesn't mean that things are not changing all the time, even within political systems which are bad.

So it isn't the case that-- for example, countries which have had very large improvements in education and health. Indonesia under Suharto-- which was extraordinarily corrupt, was often rated the most corrupt country in the world, and was under an extraordinarily repressive dictator who at some point, it's claimed, killed a million people by labeling them Communists in the mid-'60s.

Nevertheless, if you look at what's happening in Indonesia, between the period of 1972 and 1990, is one of the world's biggest, most successful investors in human capital in the world. You see big changes in education levels, you see big changes in health Suharto was very, very concerned about Indonesian health, nutrition, and he created a nationwide campaign of young people going back to villages and bringing message of better nutrition. And it's claimed to have improved nutrition massively in Indonesia.

So in other words, that's not to say that Indonesia became a better country immediately, or that corruption went away. In fact, during the same period that education is improving in Indonesia, corruption is also going up. Suharto is becoming more and more corrupt. As his children get older, they're becoming more and more richer and richer, and buying up larger and larger shares of the country.

So there was not that there was a one-way of movement in institutions. And indeed, that's my point. There is no sense in which there's one institution in place, one set of institutions in place. There are many, many things going on in any country. And many of those are getting better. Others are getting worse.

I think we're always looking for leverage points. What are the best places to make a small push and see where we can make a change? So once we start thinking about institutions in small letters, rather than institutions in capital letters-- so not capitalism, but how a specific market is run-- we do see lots of changes. And that's the level at which we should think of institutional change. Yeah.

AUDIENCE: I get the sense that, looking at it from one perspective, you should almost always prefer gradual reform. Because all the things that you showed us on the last slide are so out of sample that the chances that they'll actually work, just because they're so far beyond what we can observe or test and day-to-day things, that doing gradual reform would be the path of lesser resistance.

But then Jim and [? Daron ?] and [? Pavel ?] have this paper where they talk about this thing called the seesaw effect, where you make one dimension of institutions better, another gets worse to compensate. Or Jeff Sachs would say that we need to change everything at once. Do we actually have-- I mean, I guess that Jim and [? Daron and ?] [? Pavel ?] have a little bit of evidence in their paper. But do we really have much in the way of systematic evidence for that?

PROFESSOR: So I think the answer's no. It could be that they're right, that you can never improve anything without making something else worse. I find that implausible, let's say. But I'll leave that as a--

My main point I'm trying to make here is a simpler one, which is that there is no grand theorem which says that everything that you can change is determined by some overarching institutional frame. That even within bad institutions, there's lots of things that do change-- some for the better, some for the worse-- and that there is enough slack within the system that you can often change things even when things look pretty dire.

So here's an example. In Chinese democracy-- this is not well-known, because China's not supposed to be a democracy. But in fact, China introduced village elections quite early. And the elections were phased in, so you could compare places which had elections with places which had a centrally-appointed village head. So this was just a comparison of what happens to a place when it becomes elected.

So you been basically seem to see that when these institutional changes happen in a place like China, which is, after all, not a democracy. Huge control of the Communist party. Certainly, whenever there is any real attempt to change institutions at the high level, it's been stamped out with a great deal of clarity. There's never been any debate on how-- challenges is to the power of the Party have not been entertained. So you might imagine that when China introduces village elections, this would do nothing. Because power has not shifted. The same people have power.

Turns out that when you introduce elections, you get very large changes. And this is not now. This started, I think, in the '80s. So when you start introducing elections, you get very large changes. Basically, two things happen. One is that the central policies that are unpopular are less enforced. So the one-child policy, which is the policy of not allowing people to have more than one child, is relaxed much faster than places which have elections. So somehow, despite the fact that the central government is all-powerful, when the village has an election, it does enforce different things.

Second, one other very important part of Chinese policy is reallocation of farmland. Basically, what they were doing was, they would, for example, take farmland from farmers and build cities, or roads, or industrial estates, or special economic zones, and things like that. And basically, when the village had elected government, that stopped, or went down a lot. So you couldn't take land away from the farmers that easily.

And this is where the all-powerful state which was supposed to have impunity. It was still true that you could make many changes within that system. And it's so striking, even in a system like China, how much responsiveness there is to local demands once you start democracy.

So you might have thought that, well, the big institution is, there is no democracy in China. China is classified as a non-democratic state. Yet at some level, China is actually quite democratic at the local level. And that's a way to understand, also, why China has managed its conflicts relatively effectively. It's because there is actually pressure involved at the local level, which allow it to let off pressure.

So there's a sense in which this is an example of why we shouldn't look at institutions from a thousand miles up, but we should look at how they function on the ground a little bit.

And, I think, equally importantly-- so I just talked about why bad institutions in capitals, like lack of democracy, doesn't really necessarily always mean that is not democracy on the ground-- the opposite is also true. That you could have good institutions, meaning good electoral laws, nominally fair elections, but those elections may actually not be at all fair.

And here's a nice example. In Brazil, there was a complicated paper ballot system. Brazil has, for the last many years, had free and fair elections, since about the mid-'80s. And yet, in the Brazilian electoral system, for a long time, people just didn't think about it. So you would have to write in the name of the person who you were electing, or the number. Either name or number. You have to say, Mr. seven, or something.

And since many people were not very literate, when they write the names of somebody, it is illegible. So as a result, 11% of the names were rejected in any election. That's a lot, right? 11% of the votes were rejected. That's a huge amount.

So this was taking too much effort. Counting votes was too complicated. So they replaced it with electronic voting. So you to press a button. So that meant that the invalid votes went away. And as a result, the fraction of the elected representatives who were uneducated or poor went up. The poor were voting for people like themselves.

As a result, health expenditures went up, and a bunch of pro-poor policies were implemented. So this is an example of where Brazil had good institutions in capitals to start with, but somehow those institutions weren't doing what they were supposed to do.

So if you looked at it from a thousand feet up, you will see good institutions. But that doesn't mean that the institutions were actually doing what the were supposed to be doing. And so about 10% of the population, and a huge proportion of the poor, were disenfranchised. 11% percent of the votes were eliminated. That means that, if you take the poor, probably 30, 40% of the poor were disenfranchised. They couldn't vote. That changed the nature of voting and changed the nature of outcomes.

So it's a beautiful example of how something-- nobody thought about this. It's not that anybody had this as a conspiracy. This was not somebody's plan. It just was something that was just there. People thought paper ballots were fine. Nobody ever thought about it. Once you change the rules, and you mostly changed the rules not to fix it, but to just make it easier to count votes, it turned out that that had massive consequences.

This is the picture. I'll just skip this.

So here's another couple of nice experiments.

So one argument that Paul Collier makes a lot of is, he says the reason why you have to invade this country is that these countries, have too much internal conflict, that they can't govern themselves because they're fighting each other all the time.

Paul is very sympathetic to these people, you have to understand. He's a very, very soft-hearted colonist-- imperialist. I've known him for many, many years. And he's certainly not someone who's cynical about this. He certainly does really believe that these people are just caught in this eternal fight against each other.

They hate each other's guts. And so, therefore, they can never govern. Whenever one group takes over, it starts beating up on the others. And that's why you always have a permanent state of civil war and conflict. They're just not capable of governing each other because there's so much fundamental hatred of each other.

So to look at this, Leonard Wantchekon, who is a political scientist from Benin, did the following. Leonard was in the student movement in Benin in the late '70s, early '80s. And that's when, I think, Benin had a dictatorship. And so he went to jail with all the other people who were leading the student movement. And so he's very good friends with all the leading politicians in Benin now.

So Benin is now a democracy. And in the democracy, he's good friends with all the leaders. So he basically talked them into doing something quite remarkable. His idea was, every leader, he basically got the following deal. He said, let's go to the area where you are really dominant. Go to the area where you are really powerful. So you're never going to lose that area. You're always going to win that area. So it doesn't really matter if you get a little less votes.

So let's do an experiment where you send different messages in different areas. So within in that area, let's say there's this region like that, where this is where you're totally dominant. You're always going to win. So now let's chop this up into little pieces and randomly choose some of them. This is all within the area where you're sure to win. We're going to randomize the message you send.

So there were two messages they were sending. So this is the message that primes people on their narrow tribal interests, if you like. This says, the Bariba people will get jobs. I am the representative of the Bariba people. This is what Paul Collier is particularly concerned with.

So this is the same guy going to different in villages and giving different messages. The villages are randomly chosen, so you can compare what happens. And they were willing to do it because they were sure of winning. So they didn't have to worry about winning there. The question was many votes do they get. This was the Clientelist message, the message that I'll serve you guys, my people. That's what people are worried about.

Here's the public policy message. Same person, different village, goes and delivers this message. We will fight corruption and promote peace among all ethnic groups in all regions of Benin. This was completely rigged. They were given these messages. They agreed to deliver it because they were friends of his.

So this is what makes Paul collier worried. So this was the first experiment that Leonard did. So if you look at what happens to vote shares, if you go to the villages where you did the public policy message, you got only 59% of the vote. When you did Clientelist message, you got 79% of the vote. So this is what gets Paul Collier worried. He says it's not possible to have democracy in these places because people are so narrow-minded.

Now what happened-- let me just go back-- is-- OK. Let me see. Where did it go? OK.

Let me mention one more, and I'll come back to-- no, actually-- so let me show you one other experiment before I come to this. Because this is sort of interesting. I'll show you one more.

So here's an experiment that we did in India, which is kind of the opposite of this Wantchekon experiment, and has the opposite result. And I'll try and interpret why. This is probably the most corrupt place in India. About 40% of the people who are elected from there have criminal charges against them.

In this area, we did an experiment where we randomly chose villages. And in some of the villages, we basically showed up with a puppet show. Basically, we showed up with a puppet show. The point of the public show was to say, you should not vote based on your caste, which is the equivalent of ethnicity in India. You should vote based on who's going to do more development for you.

So that was the message. It was a completely neutral message. And the neutral message was delivered by an NGO. When you look at what happened, the people who voted for their ethnic party went down by a quarter. If you just went and told people, don't do it, they don't do it.

So this is sort of the opposite story, which is, in Benin, the experiment was that you went and told people, please vote for me based on my ethnicity. Then they seem to vote on ethnicity. If you tell people, don't vote on ethnicity, they do the opposite. Now, is Benin different? I'll come back to that later.

But what this made us think about was, maybe the reason why you observe this ethnicity-based voting is not because people are passionate about their ethnic roots. Sometimes they are. But maybe a lot of people are not. But because they have no other information, maybe they don't know anything about the candidate. And therefore, if I have no idea who to vote for, there are three people running, I just vote for the one who has the same name as me.

Why? Because I have no idea who they are. I've never met them. I have no idea what they believe in. I don't trust them to have told me the truth in any case. So do you actually get the right outcome?

So maybe the reason why people vote based on ethnicity, whereas we think that somehow people to just have some biological or sociological hatred of those other people and they can't resist voting, in fact, maybe the reason why they vote for their own ethnicity is only because they have no better choices. They don't know anything about any of the candidates, so they might as well vote for their ethnicity. So this is what we were investigating.

Keep going. Yeah.

AUDIENCE: In the Benin experiment as well, it seemed like it was a way to target things. Like, I'm this ethnicity, and I'm going to get things for people of this ethnicity, versus, I'm going to do good things for everybody. I don't know if that would be so different than if somebody was running for Senate in the US and was like, I'm going to do good things for the people of Massachusetts, versus, I'm going to do good things for everyone in the US.

PROFESSOR: So you may well be right. So that's another point. I'll come back to the Benin in a minute. I just wanted to show this experiment. And I'll come back to the Benin one in a minute.

So based on this idea that maybe voters don't know who they are voting for-- and therefore, they just vote at random, and maybe they just vote for the guy who has the right name or something-- we did another experiment where we gave information about the candidates to the voters. We told them, this candidate works harder than that candidate. We showed newspapers printed people's performance.

And when you do that, you see that people do change their voting patterns, suggesting that people don't really know a lot about the candidates. If you give them information, they react, which suggests that information is valuable to them. They don't really know what's going on.

OK. Let me now go back to Benin, and I'll come back to this. So here's what happened in Benin.

Wantchekon did a second experiment. Second experiment was, instead of people just saying, I'm going to do good for everybody, he actually had them run a conference where, in the conference, they discussed real policies.

So they had a long discussion of real, substantive policies. Experts were invited. And so parties took up the policies that came out of that conference and put it on the platform. These were well-worked-out policies, not claims about, I'll do good for everybody.

So the only difference-- then he does the same experiment as before in a different election, later election. He does the same experiment, which is telling some people, either send an ethnic message or a non-ethnic message.

And now, when you go and tell I'm going to do good for everybody, that actually goes the other way. So when you say, I'm going to do good for everybody, but it's backed with a real policy proposal, you get much more positive results.

So it looks like the voters were actually rightly cynical about these politicians. They were claiming they're going to do stuff for everybody, but nobody believed them, because they had not put any content into it. When you actually put content into what you're claiming, voters react very differently. They feel like there's something that's going on that's maybe worth supporting.

So you see, instead of this strong reaction for the ethnic point of view, now you get the opposite, which is a strong reaction for the non-ethnic point of view.

Another example. So the point I'm trying to make here is that in some sense, what looks like structural resistance to any change-- countries are just hopeless, they have such ethnic conflict that they can't do anything about it-- that might be true under some situations.

But, for example, if you want to name countries which have had some of the most serious ethnic violence in the last 20 years, they include many countries which are actually very economically successful countries. So it's not clear that that's something that the effects are that strong.

But even if you believe it, it's not clear whether ethnic conflicts are a result of other failures or the cause. So in other words, is it the case that people are basically cynical about the political system, and therefore they'll vote for the person whose name is the same as theirs, or is it really the case that they vote for people who have the same name as theirs, and therefore the political system fails? That's the challenge in figuring out.

And at least there seems to be some data which suggests that-- this evidence that I've been talking about-- suggests that it may well be that, to a certain extent, if people are cynical about the political system, makes them then vote more unreliably. And if you could actually get the political system to be more effective, then people would actually be less inclined to vote on the ethnic basis.

So in other words, I'm trying to make the case that what looks like-- in Collier's view, the fundamental problem is ethnic hatred. Whereas it may well be that the fundamental problem is just economic failure, and ethnic hatred or ethnic conflicts are just a result of that. And it's not clear which way that goes.

And in particular, it could be very innocent. People just are completely-- it's not that they hate anybody. They just think, OK, well if I don't have any information, I'm going to vote for the guy who has the same name as me. That might be a significant part of what is called ethnic voting.

Here's another instance that's interesting. So going down this same path of making the point that it is possible to--

So in other words, I made two points. One is that having good institutions is no guarantee that you get good outcomes. And the good institution at the 1,000-feet level is not a guarantee that you get good outcomes on the ground level.

And conversely, if you look at it from 1,000 feet, what looks like bad outcomes may not be necessarily some things that you should just take as, well, it's not possible to do anything in this country. There may be lots of slack in the country. Because, for example, even ethnic preferences, which is seen as a fundamental constraint, may not be that much of a constraint. They may just be an outcome rather than a constraint.

To pursue that point, that it's often possible to have substantial changes even when the economic incentives have not been changed very much-- so the structures of power have not changed, the institutions are the same-- here's an example. This is not someone saying whether the institutions are good or bad to start with, but to say that it's possible to change institutions at the margin, a little, and get big effects.

So here's an example from Brazil. Brazil is interesting. The political system in Brazil is among the most interesting in the world. One thing they do is every month, 60 municipalities are chosen at random, and their accounts are audited.

So you audit the accounts, and then the audits are basically-- there's a lottery. Literally. That's on TV. They're doing a lottery. The winner of the lottery doesn't get a lot of money. The winner of the lottery gets audited. So you don't want to do be the winner. It's on TV, so it's very, very transparent. Everybody can see that it is really a lottery. That's a big advantage of lottery. The audit is also given to the government and disclosed to the media.

So there's a nice paper studying the effect of these audits. And basically what they do is, they compare places that were audited just before the election and places that were audited just after the elections. And in particular, it turns out that if you look at the average effect of being audited before or after elections, there's not a big difference.

Why isn't there a big difference? Well, turns out most people don't find out the results. But if you look at the places which have a which have a radio station, and compare the effect of being audited on places which have a radio station with the effect of not being audited on places which have a radio station, you see a massive effect.

If you're corrupt, and you were a mayor from a place that had a radio station, and you were audited before the election, you are 25 percentage points less likely to get elected than if you were a non-corrupt mayor from a place which has a radio station and which was audited before the election. So the difference in election probability--

So what we're doing here is comparing places that were audited before the election and after the election. And we're comparing good mayors with bad mayors. So good mayors are much more likely to get elected in places where the audit was before the elections. Whereas there's little difference between good mayors and bad mayors if the audit was to happen after the elections.

So just this act of publishing the audits seems to have substantially changed the incentive for corruption in Brazil. So this is an example of something where the system has not changed. The whole system is the same. Brazilian democracy has not changed a lot. But just the fact that they instituted these audits means that being corrupt is much more costly.

This is a picture which makes that point. The blue is those who the audit was after the election. And being corrupt has no effect on your reelection rate.

Whereas, if you had zero corrupt violations and you had a pre-election audit, you were 35 percentage points-- I said 25-- 35 percentage points less likely to be elected then if you had three or more, if there was a pre-election audit. If you have a post-election audit, it doesn't matter, because the election is done.

So getting information has huge effects here. These effects are on a magnitude which-- clearly, you go from having 50% chance of being elected to being a 20% chance of being elected, just when you publish the results.

So again, making the point that small changes in the institutional frame, the little institutions, changing them a little might actually give you big effects.

So, so far, what I've been trying to say is a bit that, even within bad institutional frames, or without changing the institutional framework generally, you can make good things happen. That doesn't mean that good things are guaranteed to happen-- whenever you try things, good things happen. That's not at all what I'm saying.

So I'm saying that what's true about economy at the 1,000-feet level is entirely one thing. And then there is possibly what's irrelevant at the 10-feet level, and those things are only weakly correlated. So in other words, things fail at the ground level whether or not you have good institutional structures.

So let me give you an example of where the institutional structures were all good, and the system completely failed.

So we did an experiment where there was a district administration in India worked with an NGO to set up monitoring system to make sure that nurses come to work. Nurses often don't come to work. You saw that, I think, at the beginning of the semester with Esther. And this is a big concern for the health system. Because you show up, nobody's there, that's not so good.

So the way it was implemented was that there was a date and time stamp given to every nurse. This was just on Mondays when she was required to be in the center. She was required to stamp her presence on something that was stuck to the wall.

So she would have stamped it there so that people could check whether she was there or not. So she had to stamp it on the wall. So she had to be there. And this NGO was in charge of collecting the stamps and counting them up and giving them to the government.

So the district administration was very taking this very seriously. They announced that anybody who doesn't come to work at least 50% of the time will get fired. So there was a quite major top-down attempt to reform. So the institutional frame was doing everything it needed to do.

So when this was announced, immediately-- just compare the dashed lines to the dashed lines. I didn't tell you the story of the solid lines. So just compare the dashed line to the dashed line. You see there's a very major effect.

At the beginning, the presence of the nurse goes from 15%-- 15%. They don't show up at all-- to about 55%. It goes up massively. When you announce this program, it goes up massively. They start to come to work. And that's not surprising. You told them.

And then what happens, you see those curves getting closer and closer, and they finally cross. So after a while, it turns out that in the places where the nurses were threatened with losing their jobs, they're coming to work less than where they were not threatened. So what happened? Why did it get reversed?

So what happened was very interesting. So two things happened. The nurse has an immediate boss whose job is to decide whether the nurse needs to come to work or not. Why? Because the nurses have things like training. So some days, they're not required to come to work. Those are called exempt days.

And then the date and time stamp had to work in order to monitor the nurses. So the date and time stamp got broken, you couldn't monitor them. Now there was a rule which said that, if it's broken, they would have bring it in to the center and get it replaced. But still, once it gets broken, it's less easy to monitor them.

So what happened was very simple. Absence did not go up. It's not that the absence went up. The exemptions went up. Meaning that their immediate boss basically declared, fine, you don't have to come to work. From now on, you're not absent. You're exempt. Meaning the boss said, I don't care if you don't come to work. So I'll give you an official excuse to not come to work. That was part of what happened.

Part of what happened was that they started breaking the machines. The machines started breaking in increasing frequency. And the bosses who should have complained about that didn't complain. So the machines kept breaking and nobody was objecting, so the whole program collapsed.

Now, why did this happen?

So clearly, this was a result of the fact that the nurse and her immediate boss were colluding. That's not surprising. But the reason why it actually happened was not because the institutional frame was bad. In some sense, there is democracy. There was supervisor who wanted it to happen. There was an NGO that was implementing it.

The reason why didn't work was mostly because there was no demand for it to work. It wasn't so much that the institutional frame wasn't there. It was just that, basically, people figured out that if it didn't work, nobody would complain.

And why would nobody complain? Well, you saw this before. Nobody ever showed up in these health centers in any case. So there was no one to complain. So there's no political compulsion to fix it. And that meant that at some level, there was no pressure to make it work.

This is what sometimes we call triple-i. In other words, the reason why this intervention didn't work is not because the institutional frame was not working, but because there was no one on the ground who actually had any stake in making it work. Nobody went to the health center, so nobody complained.

The system was designed with the idea that these nurses have some internal desire to serve of the people. So the system has no effective culture of making sure that the nurses one want to come to work. They never invested in it.

And finally, there was no one had thought about the fact that there were these exempt days, and that these exempt days were not regulated. So that it was mostly just the fact that, when this intervention was introduced, nobody had actually put enough thought into how the system would react.

And so it was not so much that the institutional frame was particularly bad here. Other interventions do work. It's just that this was a system where there was no one who actually had thought about how to design an intervention that would work, and therefore it didn't work.

Indeed, when we went back, we figured out why you get this crossing. Why do you think we get this crossing? Why do you think the nurses came less when they were incentivized to come?

AUDIENCE: I think it's a situation where there were new measure put in place. Once it became clear that those measure weren't going to be enforced at all long-term, it was an even clearer indication that no one would check up on them.

It's one thing to have a policy that in theory exists, and no one tries to ever enforce it. But if then they try to institute this new policy and indeed, it was very [INAUDIBLE] that there was no enforcement, that sends an even stronger message that you can shirk out your responsibilities.

PROFESSOR: Right. So they realized that their bosses were very willing to collude with them to not let them work. So because they had had a policy which required their bosses to enforce, now they could learn that the bosses had no interest in enforcing. Before this, they didn't know that. So in the control villages, they don't yet know that the bosses don't care. In the treatment villages, they really learned that the bosses don't give a damn whether they come to work or not. So they stopped coming to work.

So whether these institutions at the 1,000-feet level work or don't work-- I mean, there's a democracy here. Lots of people vote. There is political competition. All of those things that are supposed to make these things work are all present. The reason why it doesn't work is simply, at the micro level, the incentive to make it work isn't there. Because people don't go to the health center, so nobody actually demands that the health centers work.

AUDIENCE: So actually, more people started to go to the health centers. And yet there were few people [INAUDIBLE]?

PROFESSOR: No. Basically, people have checked out of the system. So more people use the health center, but only by accident, when it's open. So the probability, per day, more people are not present. It's open more often, but it doesn't have any effect on who gets treated per day.

Yeah. Alyssa?

AUDIENCE: The fact that nobody is using these can't be that there wouldn't be a demand for health care if they functioned. It's itself an outcome of the fact that--

PROFESSOR: Absolutely. But this attempt to fix it came from a bureaucratic impulse which said, have health center, will make it work. And they just did not think about what was required to make it work, which is to get a demand for it. Otherwise, it's never going to work.

The political system only provides incentives for things people want. If people don't want it, the political system doesn't provide incentives. It was a classic example of trying to fix institutions without really having understood how the institutions work.

So I'll continue with this next time.