Melissa Dell on the Significance of Persistence (Ep. 100)

Subscribe: Apple Podcasts / Spotify / RSS

Mercatus Center
Conversations with Tyler
39 min readJul 15, 2020


Explaining 10 percent of something is not usually cause for celebration. And yet when it comes to economic development, where so many factors are in play — institutions, culture, geography, to name a few — it’s impressive indeed. And that’s just what Melissa Dell has accomplished in her pathbreaking work. From the impact of the Mexican Revolution to the different development paths of northern and southern Vietnam, her work exploits what are often accidents of history — whether a Peruvian village was just inside or outside a mine’s catchment area, for example — to explain persistent differences in outcomes. Her work has earned numerous plaudits, including the John Bates Clark Medal earlier this year.

On the 100th episode of Conversations with Tyler, Melissa joined Tyler to discuss what’s behind Vietnam’s economic performance, why persistence isn’t predictive, the benefits and drawbacks of state capacity, the differing economic legacies of forced labor in Indonesia and Peru, whether people like her should still be called a Rhodes scholar, if SATs are useful, the joys of long-distance running, why higher temps are bad for economic growth, how her grandmother cultivated her curiosity, her next project looking to unlock huge historical datasets, and more.

Listen to the full conversation

You can also watch a video of the conversation here.

Read the full transcript

TYLER COWEN: I am chatting today with Melissa Dell, who is one of the world’s leading economists. She is a professor of economics at Harvard and recently has won the John Bates Clark Medal, which, in the eyes of many, is actually harder to win than a Nobel Prize. Melissa, welcome.

MELISSA DELL: Thank you.

On persistence

COWEN: Now, as you well know, a lot of your research concerns the idea of persistence of historical effects through time. If we look at the economic history of Vietnam, do we, on net, see persistence effects on the large variables? In particular, I have in mind the relative performance of North and South Vietnam. North Vietnam is, initially, more Chinese-influenced, yet today it has ended up poorer. What does that tell us about persistence?

DELL: Yeah, I’ve done some work looking at the persistence of economic development in Vietnam. The work I did, actually, was limited to what was South Vietnam because there’s also been huge events, obviously, that have happened in the past hundred years in North Vietnam, with a war that destroyed much of the country and was fought over an extended period of time.

But when you look, in general, at places in Vietnam that have a similar recent history, but going back in time, one of them was part of a much stronger, more centralized state. The other one was part of what is today Cambodia, a much weaker state, generally ruled by local lords instead of by a strong centralized state.

You see the towns that were part of the stronger, centralized state going back before colonialism, so several hundred years ago. More recently, they have better-functioning local governments. They’re richer. They’re better off, which shows that places that have a long history of governance seem better able to do that more recently.

So places going back a long time ago — they were part of the central state. They had to collect taxes locally to send up to the central state. They had to organize military conscripts. The central state mandated that they had certain laws.

More recently, those places also have more functional local governments and are also better off economically, whereas the places that never had that structure that comes from a state — it was, essentially, if you were living in that area, there’d be a warlord that you sent tribute to. But there was never any regular taxation, never any organized local government under a central state. Those places much more recently — when there were constitutional reforms in Vietnam that gave them a degree of self-government — they weren’t able to do that very effectively.

They weren’t able to keep the positions on their local city council filled. They weren’t able to provide, as effectively, local public goods, like education or health services. So really, having this long history of governance makes places more able to do that today. That’s relevant because there’s been a big push to have local governments provide an important role in providing public goods, et cetera.

If places don’t have a history of doing that, perhaps not surprisingly, they tend to have a much harder time. When the World Bank says, “Well, we want to give local autonomy to let local governments decide how to provide schools in the way that works best for them,” that’s going to work in places that have a long history of providing education. In places that don’t, they’re more likely to have a hard time.

COWEN: But if you select your cases on the basis of having similar histories, aren’t you selecting for persistence because the locales that have reversals of fortune — a big war in North Vietnam, Communism coming to North Vietnam — that’s a kind of mean reversion. It deliberately gives them a not-similar history. Do you then not overrate the degree of persistence in the dataset by just taking the sliver that is indeed continuous with its own history?

DELL: I think that you could imagine writing papers about different things. Our motivation was, we wanted to think about if the historical state could have a role at all. In order to do that, you don’t really want to compare South Korea to the Philippines — which is what most of the historical literature on this does — because they’re different in so many ways. We know that South Korea looks really different from the Philippines, but there’s so many ways that they can be different.

By looking within South Vietnam, we wanted to say, “Okay, these are places that had a much more recent modern history. Can their past history still matter?” But we’re not saying that that begins to explain everything. There’s other forces that happened more recently that we think are also important.

Certainly, there can be mean reversions, and the argument is not that things are always persistent. I think part of the literature is about understanding why sometimes things persist and sometimes they don’t. Certainly, more recent events can matter, and we’re not claiming that there’s an R-squared of one that a place’s history is its destiny, but that there are these forces.

I think that there’s ways in which looking beyond just household consumption today, I think you do very much see the legacy of the historical state in both places that were part of North and South Vietnam. If you look at coronavirus, Vietnam has been remarkably effective at dealing with it. They share a border with China, and yet they have almost no cases. They’ve been super successful at isolating their outbreaks.

Life is much more normal there than it would be in the United States. And I think that that, in large part, goes back to the fact that much of Vietnam does have this history of having a strong, centralized state that coordinates very well with local governments, and having local governments that have been effective at providing public goods like healthcare. That has come in really useful in this moment when you really need a strong, centralized response. That’s an example where I don’t have a way to prove that econometrically, but if you look throughout Vietnam, the parts that were part of North Vietnam as well as South Vietnam — they’ve been able to deal with this really effectively, and I don’t think that’s an accident.

COWEN: Let’s say we take your example of South Korea and the Philippines. If we go back to 1960, as you know, many people in development economics were somewhat optimistic about the Philippines. They have the English language. They had a burgeoning education sector, ties with the United States. And not many people were optimistic about South Korea. It had a per capita income roughly at a central African level.

If persistence is so important at the aggregate level, shouldn’t development economics be more predictive successfully than it in fact is? Because few people really predicted the relative fates of South Korea and the Philippines, but we could have read off their entire past histories in 1960.

DELL: It comes down to the fact that the world is fundamentally a complex place. If you were sitting there in 1960, what is it that you’re emphasizing? I think, actually, the people that were saying South Korea is a basket case were emphasizing the current circumstances because this was the midst of the Korean War, which, by the way, was more about external influences in Korea than about Korea itself.

And it was about the Cold War, and it was a really bad situation. Much of the country had been destroyed. The politics were a bit of a mess. But if you had looked back at the longer history of Korea, you would have seen that they have this long history of having functioning institutions that are able to provide conditions that promote economic development, and that’s important. So in some sense, the 1950s and the ’60s were an aberration in Korea’s history.

Whereas, if you go back further in time in the Philippines, they were organized very differently. They did not have a history of a strong, centralized state, and that matters potentially for what they were able to do going forward.

I think, also, the argument is not necessarily that a strong state is always good for economic development. There’s a huge literature looking at examples like Nazi Germany that shows in places where the Prussian state was stronger, the Nazis were better at exterminating Jews. So depending on what the state wants to do, having that state capacity can be a good thing or a bad thing. It doesn’t necessarily promote economic development at all points in time, but it does, in some sense, seem to be largely a prerequisite.

If you can’t have a state that’s capable of monopolizing violence and promoting the conditions that we need for economic development, like providing public goods, then most of those places that have chronically had a weak state are very poor. If you have a strong state, it can be good or bad, depending on exactly what that state is doing. And it can change over time, and there can be reversals. But it provides potential opportunities — if that state does pursue policies that are helpful instead of harmful for economic development — for that to happen.

Whereas in the Philippines, there’s some arguments that Marcos actually aspired to be a developmental dictator, and he just had no capacity to do that because nobody at the local level would really do what the central government ordered them to do. Instead, it was much easier just to steal things than to try to direct the state towards development because the central state in the Philippines didn’t have that capacity.

COWEN: But say predictively, how much weight do you give persistence in your own understanding of the world? If we compare Pakistan and India throughout much of history, what we now call Pakistan was somewhat richer than much of India. It may well have had more state capacity. That’s harder to measure. Right now, Pakistan is considerably poorer, and people are more pessimistic about it. Do you look to the past and think we should upgrade our expectations about Pakistan because of persistence?

DELL: I’ll give an example here from a little bit of a different context. I was presenting some work that I’d done on Mexico to a group of historians. And I think that historians have a very different approach than economists. They tend to focus in on a very narrow context. They might look at a specific village, and they want to explain a hundred percent of what was going on in that village in that time period.

Whereas in this paper, I was looking at the impacts of the Mexican Revolution, which is a historical conflict in economic development. And this historian, who had studied it extensively and knows a ton, was saying, “Well, I kind of see what you’re saying, and that holds in this case, but what about this exception? And what about that exception?”

And my response was to say my partial R-squared, which is the percent of the variation that this regression explains, is 0.1, which means it’s explaining 10 percent of the variation in the data. And I think, you know, that’s pretty good because the world’s a complex place, so something that explains 10 percent of the variation is potentially a pretty big deal.

But that means there’s still 90 percent of the variation that’s explained by other things. And obviously, if you go down to the individual level, there’s even more variation there in the data to explain. So I think that in these cases where we see even 10 percent of the variation being explained by a historical variable, that’s actually really strong persistence. But there’s a huge scope for so many things to matter.

I’ll say the same thing when I teach an undergrad class about economic growth in history. We talk about the various explanations you can have: geography, different types of institutions, cultural factors. Well, there’s places in sub-Saharan Africa that are 40 times poorer than the US. When you have that kind of income differential, there’s just a massive amount of variation to explain.

Nathan Nunn’s work on slavery and the role that that plays in explaining Africa’s long-run underdevelopment — he gets pretty large coefficients, but they still leave a massive amount of difference to be explained by other things as well, because there’s such large income differences between poor places in the world and rich places.

I think if persistence explains 10 percent of it, that’s a case where we see really strong persistence, and of course, there’s other cases where we don’t see much. So there’s plenty of room for everybody’s preferred theory of economic development to be important just because the differences are so huge.

COWEN: In Vietnamese history, do you think that the presence of ethnic Chinese or people who are partly ethnic Chinese is, in fact, the most important persistence for assessing how communities are doing in relative terms?

DELL: We didn’t look at that aspect specifically. More what we were able to focus on is the presence of institutions, and that holds true, like controlling for the percentage of the population that identifies as Chinese more recently. Certainly, the fact that you had ethnic Chinese there, setting up communities and bringing those institutions, was important to their persistence.

But I’m not sure, quantitatively, how important you would think that is to economic development, and it’s actually something that we can’t even really measure. A very small percentage of the population identifies as Chinese today. Of course, there were far more Chinese that settled and intermingled with Vietnamese, and we don’t really have a measure of that.

COWEN: But that may be driving the institutions because ethnic Chinese throughout Southeast Asia, as you know — it predicts household wealth pretty well. But there’s been varying degrees of discrimination, so people may not wish to identify too explicitly as ethnic Chinese, especially in Vietnam. Often they’re only part Chinese, and they want their children to have more or less normal lives and not be the target of discrimination.

So what do we think is actually driving the persistence effect across the different classes and villages in South Vietnam that you studied? Because the institutions are endogenous, right?

DELL: I think that an important part of it, just to step back and provide a little bit more history — the part of Vietnam that was a colony of China back a thousand years ago — it’s almost entirely in the north of Vietnam, and then across time, that state moved down and conquered more and more of the south, right? Because it was a much more capable kind of centralized state than the Cambodian state, so naturally, it started eating into its territory across time.

And when it did that, it was very good at bringing new territories into the Vietnamese state. If you look across other borders that were in effect for a couple hundred years, you don’t see any differences in income more recently. But finally, you get to the mid-19th century, and the French show up to colonize Vietnam.

If the French hadn’t shown up, the Vietnamese state would have probably kept going and conquered all of what is today Cambodia because they were a much more effective state, militarily, at that point in time. But that doesn’t happen. The French show up, and effectively, the very southernmost part of what is today Vietnam had just been brought into the Vietnamese state. The French essentially stopped that process of integrating them.

According to the historical literature, the local institutions, the local customs essentially persist under the French because the French actually don’t have a lot of power to change things on the ground. So the local norms at the time that the French show up essentially persist. Then we get forward to the period of independence, and that’s when we start observing all these differences in local governance.

I think, essentially, what happens is that the French don’t really have the capacity to actually change things at a village level on the ground. They can show up and take over the top level of the government, but they don’t have the capacity to go into a village and say, “Change how you govern yourselves.”

Essentially, that means that the way that things work in 1850 — that becomes pretty persistent over the next hundred years while the French are there, because there’s just not really a force to change the village government because the French say, “Do what you want at the village level. Pay taxes to us.” But they essentially don’t have the same capacity that the Vietnamese state had to conquer places, or they don’t have the desire to do it and bring them into a central state.

I think the persistence is largely about that because then, when you move into the independence period, the South Vietnamese state is also, at the national level, a chronically weak state. Again, the South Vietnamese state doesn’t have the capacity to go into the 18,000 hamlets that were in South Vietnam and say, “Change how you govern yourself.”

The persistence is really about there’s not a force to make a change, and maybe things drift a little bit over time. But going further back in time when the strong Vietnamese state comes in and really says, “Okay, you’re going to implement our institutions at the local level because this is the most effective way that we think we have to tax you and to make you contribute to the state,” then that’s a rupture, and the institutions change.

When there’s not the pressure to do that, then essentially, the structure that they have for governing themselves just reproduces over time. Places that had been part of this strong state historically — the village is the center of government. There’s a village government. They organize to tax. They organize to enforce laws, et cetera, and that continues.

Whereas in places that had not been part of the Vietnamese state, that were on the periphery of this very weak Cambodian state, there was no form of village organization to speak of. You might live near other people, but there wasn’t a village government that had been elected the way that it had in Vietnam. Instead, the way that that social organization worked is that there is a local lord who owns the land. It’s almost like a feudal system.

When the French come in, they don’t have an interest or they don’t have the capacity to essentially change that. And that power tends to reproduce itself until there’s a strong force to change. In the areas that had been part of this strong state, there weren’t strong landlords to speak of. There were these village governments, and the people who had the power in the village governments, the families — that tends to reproduce itself over time because people don’t want to let go of that power.

Where things were organized differently, where it was more about these lords who had power and you paid tribute to them, and then they passed that up — again, they’re not going to willingly let their power go. They want to reproduce that power. And that, essentially, is central leading to persistence. Of course, there can also be cultural attitudes and all those other things that underlie that.

Part of the dynamic of persistence is that when you have power, you want to do everything in your capacity to maintain that power. When you have economic resources, you want to do everything that’s in your capacity to maintain those resources.

Sometimes we get these radical ruptures that change things substantially. Other times, maybe there’s a gradual wearing away because the world economy has changed or things like that. But there’s going to be this push towards persistence because people that benefit from the way things are want to keep it that way.

COWEN: But again, how much does that matter in the aggregate? At least according to the CIA, in 1950, North and South Vietnam had the same per capita income, more or less, maybe poorly measured. But very different histories, right? Different ethnic groups, different methods of governance. And yet it seems to wash out. Isn’t that a puzzle for any of you?

DELL: I wouldn’t take the World Bank GDP data too seriously for anywhere in the colonial/postcolonial world after World War II just because it’s a very unusual time, right? There’s been these huge disruptions from being colonized by the Japanese. They caused a massive famine in Vietnam that killed millions of people. Then you have the independence war with the French, and how on earth any of this is measured, I don’t even know.

But if you go back, and you look, for example, at maps that were made in the 1930s by the French showing different types of infrastructure, which we think is likely to be correlated with economic development, you see more infrastructure in the north than in the south. So I think that we don’t have a great way to measure what economic output looked like historically, but I think it’s consistent with what we see more recently. I don’t think that there’s been a reversal in that case, despite what some GDP data may say. We have to be pretty suspicious of that data in that period.

COWEN: But there can be a lot of mechanisms that pretty simply undo a lot of the persistence.

DELL: Um-hmm.

COWEN: Say the villages in the south are poorer, so more people migrate to Saigon. The returns to urbanization, we know, are very high in development. Saigon is the center of the ethnic Chinese community, just anecdotally described as, by far, the richest place in Vietnam before the war really geared up. And maybe the south — because it was backward — precisely undid that through migration and caught up and became more urbanized — or not.

DELL: If you look across the border of North and South Vietnam — that’s not actually where the variation that we’re talking about is. The variation in what was part of the strong historical state before the French showed up — that border is much farther into the south of Vietnam.

If you look at data from household consumption surveys today, you still see a very strong effect. But again, that doesn’t mean that it’s permanent. And we were interested in looking at how that effect changed across time, and if anything, it does seem to get smaller across time, but then the problem is just your standard errors.

So we don’t have, really, enough data, and the changes across time aren’t large enough to really be able to establish if there’s some convergence. But this was something that we were actually interested in, in the context of this project. Are there forces that can make these effects grow smaller across time?

One of the forces is in these places that have had this long history of village government that have very effective local governments. They have high social capital. You see that they’re actually less likely to have loans from commercial banks because they don’t need that. If you need to get a loan, you can go get it from your friends or family. Whereas in the places that don’t have that high social capital, they’ve actually seen further inroads of commercial banks.

Or, if you look when the Vietnamese state provides formal property titles, people are actually less likely to go and get a property title in the places that have a strong history of village governance, again, because you don’t really need it. If your local government — if you trust them to enforce your property right, you don’t need to go get this thing from the provincial government, which people tend to trust less. It’s just not necessary. But that may matter as development continues.

I think that there’s certainly forces that could lead to reversal, and maybe in another 10 years, we’ll have more data that we can look at and try to understand more carefully. But at least in the data that you see through the present, you still see pretty strong economic differences. Maybe they’re getting weaker, but the differences across time that we’re able to observe — we don’t have, really, the power to say whether those relatively modest differences are convergence or they’re just noise in the data.

On what’s lacking economically in Vietnam

COWEN: What do you think is the institutional capacity lacking in Vietnam? You mentioned they had a great coronavirus response, and we all know they’ve done some good things with their education system, some very high math and other scores. Yet their per capita income — it’s about the same as Bolivia by some measures, right? It’s the same general category, which doesn’t seem so impressive institutionally. So what’s lacking in Vietnam?

DELL: I think they have been quite successful economically in recent years, relative to where they were historically. First of all, there’s the Vietnam War, which, if you look at the period between 1955 and 1975, 3 million people died in a country of 18 million. It’s just an enormously disruptive period. More bombs dropped, by a wide margin, per capita than in any other context in history.

The country was completely devastated by this conflict, which probably wouldn’t have happened without outside interference. They’re completely devastated by the conflict. And then they have this communist government with an ideology that doesn’t really work economically. They try to implement, but actually aren’t successful at implementing, some really economically harmful forced collectivization policies in the ’80s.

Then, following reform in China, they start reforming as well, and since then it’s been relatively successful. I think the questions that we have about Vietnamese development going forward, in some sense, are similar to the questions about Chinese development, and to what extent is it important to have a democratic state to promote innovation?

And these are questions — they’re very controversial questions that we don’t really have the answer to. Do we think that the policies of the Chinese government are actually bad for innovation because there’s not democracy and people are restricted in some ways? Or are they able to tailor their policies enough that the lack of democratic institutions may be harmful in other ways, but for economic outcomes, it’s not stifling innovation?

And that’s something that is a little bit hard to resolve at this level of development because, by and far, they’re not at the level of development where, to keep growing, they have to innovate. They still have some returns that can be made just by catching up with the global frontier of innovation.

COWEN: Here’s another way to put the question. Say we go back to the 1950s, before Vietnam made the list of mistakes you mentioned. Most of the Andean nations had higher or comparable per capita income than Vietnam would have before the war, before communism. What is it that Andean Latin America has had that Vietnam hasn’t, in terms of levels?

DELL: I wasn’t sure I quite understood. You mean why has Latin America done poorly relative to —

COWEN: Why has Latin America actually, in some ways, done better than Vietnam? It has a history of persistence and resource confiscation and landed elites — oligarchy, right? Made a lot of mistakes. Yet you get to 1950 or 1955, before the Vietnam War is so significant, and it seems that even poorer parts of Latin America, in general, are doing better than Vietnam, or for that matter, Cambodia, Laos.

What’s the way to think about what those parts of Latin America had that Vietnam didn’t? And this gets at the “what’s lacking in Vietnam institutional capacity” question.

DELL: Yeah, I think Latin America has a relatively high growth performance in the ’40s and ’50s, which I think is playing a role in why they’re doing a little bit better at that time. Whereas it’s a fiasco for Southeast Asia, in large part because of the war and Japanese colonization.

If you go back to 1900, Bolivia versus Vietnam are both going to be extremely poor because industrialization essentially hadn’t arrived yet. And part of that may be due to colonialism. Part of it is due to broader factors about why does it take time for industrialization to spread from England to the rest of the world?

So I think it’s a bit hard to take any specific date and say, “Okay, this place is doing better than that place,” because they all — especially for the middle-income-type countries today — they all tend to have patterns of up and down. Most of them haven’t been able to sustain growth over a long enough period yet that they’ve converged to the global frontier.

But if you look at Vietnam, it seemed to get on that path more recently. Whereas Latin America has this strong growth performance that happens earlier, and then they stagnate. And that goes back to this idea of a middle-income trap. People have talked about this comparison between East Asia and Latin America.

East Asia — their strong growth performance started later. Is China, is Vietnam going to get stuck in the same sort of middle-income trap that Latin America has seemed to be largely stuck in since the mid-20th century? Or is Vietnam going to be more like Taiwan or South Korea in that it can keep sustaining that growth across time?

I think that we don’t really have, from an empirical sense, a great understanding of why exactly Latin America got stuck. There’s some sense in the literature that it has a lot to do with political economy factors, that there were policies that were captured by elites.

If you compare import substitution industrialization — they had that in South Korea, too. But it was actually . . . the policy was wielded to try to promote economic development and to help companies to develop technologies with some protection. Then they were forced to go out and compete in global markets.

Whereas, when you look at how those policies were implemented in Latin America, they were directed towards enriching people connected to politicians. And I think there’s some sense that things like that, that the capture of policies plays a role in getting stuck in this middle-income trap.

But it’s not clear, at this point, what will happen in Vietnam or will happen in China, and institutions are a high-dimensional thing. And as I said, having a state that’s capable does not necessarily mean it’s going to be good or bad for economic development. It depends what that state does.

But if you don’t have a capable state at all — those places tend to be disproportionately in sub-Saharan Africa and parts of Southeast Asia that are extremely poor. And that really does seem to be a barrier to development because you need property rights or stability, public goods — those things to make it a worthwhile investment to invest in the things that promote economic growth, whether it’s investing in a business, investing in education, et cetera.

On Mexican villages

COWEN: Let’s say we take Mexican villages that were the beneficiaries of land reform during the time of the Mexican Revolution. What does that predict about their current status today?

DELL: If you look at places that benefited from land reform — which, importantly, is a very specific type of land reform — it didn’t give individual alienable property rights. It gave a community right to a certain amount of land, and then the community decided amongst itself how it was going to divide that. If you look at those villages that benefited more a hundred years ago, they are poorer today. They’re more agricultural today.

And a lot of that seems, again, to go back to this particular form of property rights where a community is given land. It can’t be converted from agricultural to nonagricultural uses. Households aren’t able to buy or sell the land. It belongs to the community. The community decides how it’s divided. That creates all kinds of political problems because the people with power in the community can sustain that power by holding control of land.

That has been bad for economic development, which again goes back to a point that when we make these statements about how history matters, it’s very important to be precise. If you were to make a statement like “Land reform is bad for economic development,” that wouldn’t really be consistent with the evidence because there’s been land reforms in East Asia following World War II that gave people individual property rights. Much of the land was land that had been confiscated from the colonizers in the case of Taiwan. And those seem to have been much more successful.

Whereas when you have this odd form of community property rights in Mexico that’s subject to these political problems, it seems to be pretty bad.

To use another example, I’ve done some work showing that extractive institutions in Peru — forced labor in mines — was bad for long-run development and traced out the reasons why that was the case. I have another paper looking at forced labor in Java in Indonesia, historically. And the conclusion there is that the places that were more affected by it are actually better off today. Why would that be the case?

I think we can all agree that forced labor was horrible for the people that were subjected to it and were mistreated, but it has very different long-run effects depending on exactly how it was set up. In Peru, this is forced labor in mines. They’re taking out silver. Once the silver disappears, that economic activity is gone. You don’t build much infrastructure because it’s just one mine. All you have to do is get that silver down to the coast where you can put it on a boat and send it to Spain, so there’s not much infrastructure built along with that.

In order to make it easier for the government to force people to work in the mines, you don’t let other economic activity come into those regions because you don’t want competition for labor. Part of that was not giving people secure property rights because they didn’t want landowners to come in and be competition with the state for this labor, and that has important effects.

Whereas in Indonesia, it wasn’t about mines. It was about sugar, and it was very large scale. It was actually a larger-scale system than what you saw in Latin America. It was a huge share of Dutch revenue. But sugar has a very different production process. Once you cut the raw sugar, you can’t put that on a boat and send it to Spain. It’ll spoil quickly. It’s very heavy to transport. You can’t even take it more than a few miles.

It needs to be processed, and that processing needs to happen in what, for the time, was a sophisticated food-processing factory. So what happens is you end up building a lot of infrastructure for factories, for transport because you have many of these factories scattered all out across Java. It’s not a single mine in one place. And for each of those factories, you need a road or a railroad to get that back to the port, to send it then to the Netherlands, which was the colonizing power.

Then you have these factories with sophisticated technology that’s set up. And the sugar that wasn’t good enough to be sent back to the Netherlands could be sold on the local markets. And you need to put that sugar in other food. Sugar has really dense linkages to a lot of other industries because you put sugar into everything. So you set up other food-processing factories nearby, and that’s a source of agglomeration.

They’re both forced labor. The objectives of both of those institutions were to make as much money as possible for the colonial powers. In both cases, they didn’t care much at all for the colonial population as long as they stayed alive and kept contributing to the first forced labor — that was the objectives of the colonizers.

But because there’s two very different technologies and associated sets of infrastructure that come along with mining versus sugar production, the long-run effects are totally different. I think the persistence is important in both cases, but the effects it has are very different. So we have to be specific —

COWEN: Take the Mexican results. Why should we trust the per capita income numbers from Mexican villages? I did field work in a Mexican village 20 years ago. I went around, and I asked everyone if they knew what a census taker was. They didn’t even know what I was talking about. They didn’t think any census taker had ever visited.

They’re supposed to be sent teachers from down in the city. The teachers never come. They say, basically, no one ever comes up to see the village. So if we look at per capita income numbers from 1920, 1930, even 1980, 1990, why do we trust them much? Aren’t they just numbers that Mexican workers made up so they didn’t have to go visit the village?

DELL: Historically, we wouldn’t have per capita income numbers at all. There’s different proxies we can use. For example, rather than looking at a household survey, you could look instead at electrification or access to water. People have even done that using satellite data, which is kind of crude, but we don’t think that that is subject to manipulation. And these proxies tend to be pretty highly correlated.

So I think you’re exactly right to say that we should be suspicious of the data and to try to find multiple indicators and make sure that they are going in the same direction. That’s going to be enormously important. And the sampling is going to be enormously important. If you only get the poorest households or the richest households, that’s going to look very different. All these issues are absolutely critical.

COWEN: Why is Enid, Oklahoma, the grain storage capital of America and, indeed, number-three city in the world?

DELL: Number-three city in the world, in terms of grain storage.

COWEN: Correct, according to Wikipedia.

DELL: I just learned that today. [laughs] I assume you’re asking that because that’s where I grew up, but I actually don’t know the answer to that. Maybe I should write more papers about the persistence of development in Oklahoma.

COWEN: Why is James Michener an interesting author?

DELL: I think it is very intriguing to think about the long-run history of places and how that matters.

COWEN: And you think he does that about Chesapeake?

DELL: Yeah, I loved to read his books growing up. I haven’t read them in a long time from the standpoint of evaluating him as a professional historian, but I think his work has really made people intrigued by history and the role that it plays in people’s lives.

COWEN: How much do you think there’s persistence in you in the sense that did growing up in Enid, Oklahoma, shape your later ideas and research? Or is that just orthogonal?

DELL: No, I think it’s definitely not orthogonal. If we were anthropologists, you’d spend half your paper or half your book talking about your own perspective and your own history and how that influences the way that you see things. We don’t really do that as economists. We like to think of ourselves as being 100 percent objective and all of that. But obviously, the questions that we’re interested in are priors. The way we approach them depends on our backgrounds.

I think that’s part of the reason why having diversity is super important, because if you have more people with different backgrounds, they’ll ask different questions, which are all quite important. So in my case, a lot of the questions that I’m interested in — probably the reason I find them interesting does go back to things that happened in my childhood.

I have written about issues related to security — a paper on the Vietnam War and the effects of bombing versus other strategies, and whether or not those were effective for achieving US objectives. Or I’ve done stuff on the drug trade in Mexico and effects of trying to crack down on the drug trade.

I think part of my interest in security goes back to the fact that my parents both worked at an air force base when I was growing up, and I was surrounded by people in that world. There’s lots of examples like that that I could point to, that in many ways, the things that we as researchers find interesting are a product of our experiences and, oftentimes, the experiences that happened as a kid or in college.

COWEN: If we read the Old Testament, are those fundamentally stories of cultural persistence? After all, Judaism is still with us, right? That’s quite remarkable. Wonderful that that’s the case, but it seems to be a whole book about cultural persistence.

DELL: Yeah, I think that that is definitely true. And there’s a question of why certain aspects of culture are particularly persistent, whereas others disappear. Economists have looked at this a little bit. There’s obviously a huge literature beyond that with historians and people in other fields who’ve tried to understand what gives some things lasting power, whereas others seem to fade away.

On standardized tests

COWEN: If I’m taking the SATs, and I don’t know the answer to a question, should I just guess? You’ve written an essay on how to take standardized tests. That’s the same Melissa Dell, right?

DELL: That’s true. Certainly, you don’t want to leave a blank.

COWEN: It’s a great essay.

DELL: If you leave a blank, it’s wrong. Although I’ve heard that the SAT is on its way out anyway.

COWEN: With COVID-19 coming, so many top schools have abandoned the requirement to standardized tests. Do you think we should make that permanent? Or do you think it’s bad?

DELL: I think that essentially the reason that they have it is, when you look at an application, it’s really hard to know what to make of grades or other things. It’s a metric on which everybody can be compared, but I think it’s a metric that ultimately holds a fairly limited amount of information. Part of what it does is to proxy for socioeconomic status because, ultimately, the SAT is a test that can be prepared for very well.

COWEN: More so than extracurriculars, right? Isn’t it, at the margin, somewhat egalitarian?


COWEN: Yeah, compared to the other standards that are used.

DELL: Yeah, I think that it depends on the way you use it, right? Let’s say that you were to have a cutoff, where “We’re not going to look at anybody who doesn’t have a 1600 on the SAT because that’s already more applicants than there are spots.” Then you end up missing a lot of people who are interesting in larger dimensions. But if you look at it within the context of the rest of the application, it does have information.

If you look at two students who are similar in other ways, and one of them has a much stronger SAT score than the other, that probably tells you something. But we have to appreciate that the SAT alone conflates a lot of different things and not just that student’s potential to thrive. Again, it’s one of those things, as you point out, what do we really have that’s better? Extracurriculars — maybe to an even greater extent, some students have more opportunities than others.

COWEN: Should we still call them Rhodes scholars?

DELL: I think that the concerns that people bring up are very reasonable. Obviously, from the standpoint of the Rhodes Trust, they say, “Well, this is the person that made it.” Other people say, “Well, look at all these horrible things that he did and his horrible views.” Which are true. The important thing is that we have conversations about this, and the conversations that we have are very useful.

The more concerning thing is not, to me, so much the name but in what sort of esteem is he held? And if they decided to change the name, I personally would be fine with that. But I think it’s also important that we have conversations about his legacy and his views and the role of colonialism more generally, rather than trying to pretend like it didn’t happen.

COWEN: What was your sport?

DELL: I did cross country.

COWEN: Running?

DELL: Yes.

COWEN: What is it that outsiders are least likely to understand about the joys of running, say, a hundred miles?

DELL: I think that the reason that I really like running long distances are you can get out, explore nature. And in some ways, it’s like a very, almost a very meditative activity. In modern life, we tend to be very focused on what we have to do next, on where we’re going to.

The reason that I really like running long distances are you can get out, explore nature. And in some ways, it’s like a very, almost a very meditative activity. In modern life, we tend to be very focused on what we have to do next, on where we’re going to.

Whereas, when you run long distances, all of that, in some sense, fades away, and you’re just focused on the activity of running, on where you are, and you stop worrying about what you’re going to do next, what happened previously. That, in some ways, is a very liberating experience.

COWEN: Growing up in Enid, Oklahoma — that’s about a hundred miles from Tulsa. How much were you taught about what’s called Black Wall Street and the 1921 race massacre? Was that a big thing in their curriculum or it was simply glossed over?

DELL: I honestly never heard of it until I went to college. I took Oklahoma history, and it wasn’t mentioned ever. We’d talk a little bit about the Trail of Tears, and a lot of it was spent on talking about the land run, which is what happened after they had forced the Native Americans off the land. They had, essentially, a race where white settlers . . . They shot off the gun, and whoever got to the piece of land first became the owner. We spent a lot of time talking about that. About African American history — totally, totally nonexistent.

COWEN: Now, you have a famous paper on temperature and economic growth. If I understand correctly, you have in there the result that an average temperature of one degree higher centigrade correlates with a growth rate of 1.3 percent lower, which is a significant effect, as you pointed out in the paper. What do you think is driving it that makes the effect so large? And it doesn’t, as you point out, seem to just be agriculture, right? It’s also manufacturing.

DELL: Yep, we find that effect in poorer countries, right? Countries that — the way we define that is they’re below median GDP at the start of the sample in 1970. Certainly, a part of it is agriculture, and agriculture is a very large share of those countries’ economy.

Even if you look at the US, hotter years are really bad for agriculture, even though we have things like irrigation and more resistant seeds and things like that. Still, weather has a very big impact on agriculture. But agriculture is a tiny share of US GDP, so it doesn’t really matter for the overall picture that we see. Whereas in poorer countries, agriculture is a much bigger deal.

But as you say, it’s not just about agriculture. If you think about T-shirt factories in Bangladesh, people are working in pretty bad conditions. There’s certainly not air conditioning. Oftentimes, there’s not even a fan.

There’s been numerous studies, both in the field and in labs, that when it’s extremely hot, you’re less productive at physical activities, and you’re less productive at cognitive tasks. If all of our air conditioning was to go out on a hundred-degree day, we would be less productive. And so like that comes into play as well.

Finally, there’s an impact on political stability. If agriculture has much lower output, food prices will go up. Oftentimes, there’s food-price riots, and that can spill over into more generalized political instability. And that matters for economic outcomes as well.

COWEN: So should foreign aid subsidize air conditioning?

DELL: That becomes a tricky question. We think that there are potentially huge productivity effects of air conditioning. People have done some randomized control trials. But if you go and propose that, then the environmental people get very concerned because it’s bad for the environment.

I think, even short of air conditioning, which is very expensive, even just things like fans or having access to electricity, having more efficient building standards can make a big difference and is probably more realistic, at the same time, than getting air conditioning to everyone in sub-Saharan Africa now, which would be very, very costly.

COWEN: In these poorer countries, do we see higher test scores in the cooler areas or the areas with more altitude? If we compare, say, Quito and Gayaquil or Lima and Cusco, whatever, there are comparisons you could make, right?

DELL: I think there are. There’s a lot of things that would be, say, different between Lima and Cusco anyways, so I don’t know people have done that.

But there are actually studies — more enriched countries — looking at if you have the school standardized tests on a hot day, does that matter for scores? And does it matter if your school has air conditioning or not? And people find a relationship there, that if you’re taking a standardized test in Massachusetts where most of our schools don’t actually have air conditioning, and it happens to be really hot, students don’t do as well.

COWEN: What did you learn from your grandmother?

DELL: I learned a lot in terms of what my interests are and what motivates me. My grandmother was a pilot during World War II and was always very, very interested in learning. Back in that time, she didn’t ever have an opportunity to go to college, but she just read books her entire life and was so curious and would give me all kinds of books, especially related to history and related to cultures in different places, and really encouraged me to be curious and to pursue things and inspired me that women can do that.

COWEN: How do you now decide which history books you want to read? Obviously, many will be project-related, but you know a lot of history about many different areas. What’s your selection algorithm?

DELL: As you say, a lot of it goes back to, I have a whole list of questions that I’m interested in and that I would really like to work on, but I don’t have, essentially, the right perspective to answer them because I need more knowledge of the institutions. I need data to be available. So a lot of it comes back to being project driven or sometimes just related to current events.

Something that I read recently was looking at the development of the polio vaccine, and there’s really a lot of parallels to our current situation with the coronavirus. Polio is a disease that — many people don’t know — if you get polio, about 80 percent of people are asymptomatic, but then it can be really serious. It’s not as if you get polio, you’re automatically paralyzed. It’s probably just a mild stomach virus in most people, but with some small chance that it invades your central nervous system and you get paralyzed.

But it was very difficult to control with things like contact tracing because so many people are asymptomatic. I was reading about the push for the polio vaccine in the 1950s, which is this super interesting case because over a hundred million Americans donated to the drive by the National Foundation for Infantile Paralysis to develop a polio vaccine. It was developed almost entirely with this private donation–based funding.

So they develop a successful vaccine, but then they’re in a big rush to push it to market. And the federal government’s committee that’s supposed to license vaccines spends less than two hours reviewing who to license, which in comparative perspective, they spend years and years today, typically.

And it ended up that there was a company that wasn’t really qualified to produce vaccine, and they produced 400,000 [correction: 40,000] doses of polio vaccine that had live virus in it. I got interested in reading about that history through the current events.

But then I realized, you look at where they sent this live vaccine, and effectively, it seems almost as if it’s more or less random because some lots were good; some lots — they didn’t successfully kill off the polio virus, which means they gave children who were injected with it, polio. And where exactly that went seems to be something that you could actually exploit. So we’re interested now in looking at how did that affect attitudes towards vaccination in the longer run? Did it affect it at all, or did it not?

COWEN: Two last questions. First, say a student comes to you, an undergraduate, and they’re thinking of trying to become a top economist. What is it you look for in that student that might make you think he or she is actually a good candidate to become a top economist?

DELL: I think, first of all, just having a curiosity for the world and a drive to discover facts is enormously important, to look at the world and say, “I want to understand this. And I want to be able to use logic, whether it be with data or with theory, to make sense of what’s going on.” To try to look at things that are complicated and find a way to simplify them enough that we can understand what’s going on and having that curiosity is, I think, enormously important.

A lot of emphasis gets placed on things like, “What’s your math background?” But I think that if you’re interested in the right questions, there’s a lot of capacity to make up potential deficits. What additional math classes do I need to take? Or what additional technical skills do I need to learn?

But I think almost part of it depends on how you look at the world. Do you look at things and then try to think, “Okay, how can I simplify this? How can I take something that’s incredibly complex and narrow in on a specific dimension and really understand that in a way that can shed light on the world?” And then along with that, also being humble about the fact that that doesn’t mean that you’ve understood everything. You’ve looked at one dimension. It sheds light on things, but you need to, really need to know what you don’t know.

COWEN: And other than, perhaps, doing work on the history of the polio vaccine, what can you tell us about what you might be doing or working on next?

DELL: I have a couple of broad projects which are, in substance, both about unlocking data on a massive scale to answer questions that we haven’t been able to look at before. If you take historical data, whether it be tables or a compendia of biographies or newspapers, and you go and you put those into Amazon Textract or Google Cloud Vision, it will output complete garbage.

It’s been very specifically geared towards specific things which are like single-column books and just does not do well with digitizing historical data on a large scale. So we’ve been really investing in methods in computer vision as well as in natural language processing to process the output so that we can take data, historical data, on a large scale. These datasets would be too large to ever digitize by hand. And we can get them into a format that can be used to analyze and answer lots of questions.

One example is historical newspapers. We have about 25 million-page scans of front pages and editorial pages from newspapers across thousands and thousands of US communities. Newspapers tend to have a complex structure. They might have seven columns, and then there’s headlines, and there’s pictures, and there’s advertisements and captions. If you just put those into Google Cloud Vision, again, it will read it like a single-column book and give you total garbage. That means that the entire large literature using historical newspapers, unless it uses something like the New York Times or the Wall Street Journal that has been carefully digitized by a person sitting there and manually drawing boxes around the content, all you have are keywords.

You can see what words appear on the page, but you can’t put those words together into sentences or into paragraphs. And that means we can’t extract the sentiment. We don’t understand how people are talking about things in these communities. We see what they’re talking about, what words they use, but not how they’re talking about it.

So, by devising methods to automatically extract that data, it gives us a potential to do sentiment analysis, to understand, across different communities in the US, how people are talking about very specific events, whether it be about the Vietnam War, whether it be about the rise of scientific medicine, conspiracy theories — name anything you want, like how are people in local newspapers talking about this? Are they talking about it at all?

We can process the images. What sort of iconic images are appearing? Are they appearing? So I think it can unlock a ton of information about news.

We’re also applying these techniques to lots of firm-level and individual-level data from Japan, historically, to understand more about their economic development. We have annual data on like 40,000 Japanese firms and lots of their economic output. This is tables, very different than newspapers, but it’s a similar problem of extracting structure from data, working on methods to get all of that out, to look at a variety of questions about long-run development in Japan and how they were able to be so successful.

More broadly, I’m really excited about unlocking data that not just us, but that lots of other people can use to understand lots of questions, not just about persistence, but maybe there’s some interesting variation in the past, and you want to use that to understand something that happened in the past. And you’re not concerned about if it persisted, but that variation is interesting. Hopefully, with these tools, it will unlock data. Other researchers can use them to unlock their own data. And that will open the door to lots of interesting questions.

COWEN: Melissa Dell, thank you very much. It’s been a pleasure.

DELL: Thank you so much.

COWEN: And look forward to continuing to read your work.

DELL: Thank you.



Mercatus Center
Conversations with Tyler

The Mercatus Center at George Mason University is the world’s premier university source for market-oriented ideas.