Why The Nobel Prize Fails

The people versus Acemoglu, Johnson, and Robinson

There is no wisdom on economic development, and there are no wise men. There is only economic theory, imperfect as it is, and empirical evidence; we should try to use them.

Paul Krugman, “Cycles of Conventional Wisdom on Economic Development” (1995)

Daron Acemoglu, Simon Johnson, and James Robinson received the Nobel Prize in Economics last week (read more here), and a lot of people were very mad about it - in particular, with three flavors of criticism: why is this Nobel-worthy work? why do they not engage with the broader literature more? and why did they give a Nobel Prize to such flawed research?

This works better if you’ve read my general piece on the Nobel Prize.

Not ready for primetime

The main argument against Acemoglu, Johnson, and Robinson (AJR) receiving a Nobel Prize is that their work is not transformative enough to merit a Nobel Prize. I glossed over this in my main piece, but the intellectual history of economics at large and development economics in particular is very important here. For example, Noah Smith makes the case here, and pseudoerasmus has a bunch of (very illustrative) tweets to the topic.

Basically, originally development economists were mostly people who thought that investment in the right industries produced prosperity. As a result, they advocated for large transfers of funds from rich to poor countries in order to jumpstart growth via investment, which was directed into the correct industries. This approach faced multiple issues: the first is that, in a standard growth model, basically no growth can be attributed to more capital or labor, and most if not all is related to productivity and human capital (education and skills basically). Similarly, economists also didn’t find any relationship between economic aid and growth, and the models used to explain this relationship were antiquated and flawed.

Basically, what happened was that economists saw the evidence and changed track - in particular, what they saw was that in many cases government did not work properly in developing countries, and so policies were easily captured by local elites to suit their own ends. Avoiding rent-seeking, as this pattern of behavior was called, was the name of the game. This followed a generalized reckoning with broader issues in Keynesian macroeconomics that emphasized microeconomic rationalism and deemphasized active steerage of the economy. As a result, the new consensus was that free markets, free trade, and the free flow of capital was the right way to go. These were said to be the correct way to raise productivity.

The problem here is that this approach did not work. Very few countries actually developed through this approach, and the ones that did were remarkably controversial - lots of papers were written about whether the “Asian Tigers” were export-driven and thus proved the Washington Consensus types right, or were state-and-investment driven and proved the Old Consensus right, or were human-capital driven and thus proved nobody in particular right. But then, a bunch of financial crises happened, and shook everyone’s faith in the Washington Consensus - basically, sound government and free finance did not ensure safety from catastrophic economic meltdowns. Plus, macroeconomics finally reconciled some Old Keynesian insights about discoordination with the new insights on microeconomic rationality, so a series of concerns over market failures made their way to economic development too.

The main item that most experts were coming towards, besides bringing back the Big Push and arguing about geography, was that government policy was the most important thing - in particular, that governments be able to provide public services, a stable macroeconomy, and protections to property rights. Put together, these factors should be able to provide both a stable, predictable climate for private investment, but also a credible and trustworthy government capable of providing social services and building infrastructure. So the question became, why did governments not respect private property and not be trustworthy and not be good at public services?

So we’re in the 1990s, and three important things happen. First, development economists had become increasingly interested in incorporating human capital and explaining productivity into their models, which was proving hard and contradictory. Secondly, economists began inventing cool new ways to prove causal relationships - away from big clunky regressions across countries, and towards slick and cool exploits that show “cleaner” relationships with causality and not correlation.

The last bit is that economic history as a discipline had, at that point, spent the previous twenty years becoming like, a real thing with actual rigor - in particular, it had been developing solid statistical foundations that allowed for actual, credible work to be done with historical statistics. This meant that, instead of being left to the devices of historians, questions like “was slavery profitable” could be answered with the tools of standard micro and macroeconomics. Economics was also experiencing another big shift in the 70s and 80s, this time in ideas: Douglass North, an economist from Berkeley, revived the notion of “institutions” and provided them grounding in European history - in particular, grounded them with the uber-neoclassical framework of public choice theory. In his most famous paper, North and frequent coauthor Barry Weingast made the case that England had been the most developed country in the 17th century because the Glorious Revolution, where a bunch of pro-Dutch parliamentarist types deposed the British king, created a system of government where the monarchy was bounded by clear rules - in particular, respecting private property. These two approaches (“cliometrics” and “New Institutional History”) received a joint Nobel Prize in Economics in 1993.

So, put together, you have an intellectual climate that is thirsty for answers about why governments adopt certain courses of policy, econometric techniques that allow for cleaner causal relationships than prior work, a very developed and refined understanding of the processes of government with a focus on history, and a branch of economic history devoted to finding data and using it. This meant that it could take someone with a Big Brain to put everything together and take a stab at answering the big question of Why Nations Fail - and it was, in fact, Acemoglu, Johnson, and Robinson, who tried building a map from economic class interests to political participation (through game theory, another “new” field) to economic development.

Now, you can debate the merits of “being really influential” to award someone a Nobel Prize, but the last 25-ish years of economic development discussions, as well as other related topics in growth and economic history, have been ruled over by this single paradigm. There are specific criticisms of the papers in question, which I’ve jotted down a few times before and I’ll just put together at the end to not derail the whole thing with discussions of the expropriation risk index. But it’s worth noting that every major Nobel-worthy paper has received criticism: there’s plenty of research finding no gender wage gap in certain circumstances (versus what Claudia Goldin said), there’s lots of disagreements over the credit channel in macroeconomics and how bank runs work (verus what Bernanke and Diamond/Dybvig said), there’s lots of controversy about the minimum wage and immigration (contra David Card), and RCTs are themselves extremely controversial as a research tool.

Lemmy versus Alphaville

A recent article for the Financial Times Alphaville section criticizes the AJR Nobel for a variety of reasons: not engaging with the history literature enough (true of most economists except econ historians), neglecting the connections between slavery, capitalism, and colonialism (wrong!), and not touching slavery enough anyways.

The problem is that the Alphaville article is wrong, and guilty of exactly the same sins it accuses AJR of committing - not engaging with the literature one is critiquing. Beyond minor sins like mixing up development economics with economic history and calling AJR “New Institutionalists” (they were not), and various other criticisms - mostly about ommitting the impact of slavery. Of course, this is all a symptom of the author (a historian) ironically enough refusing to engage with the literature he is criticizing - many of the shortcomings of the AJR corpus are in fact not short at all, and many others have been pointed out by other economists. James Robinson in particular engages with less orthodox literature very often: he wrote a well-regarded paper about what a Marxist would call “primitive accumulation”, and has many citations to Dependency Theorists and to critics of “neoliberalism”.

But slavery is present in the AJR corpus writ large: it is an extractive institution imposed on the colonies that empowered the merchant/capitalist class in Europe, and deeply damaged economic prospects in the New World. While the AJR approach to slavery leaves much to be desired, it is present in their canonical papers, and it is part of the malign extractive institutions. In fact most of the problems with AJR 2001 are precisely because slavery is so important to the causal channel between settler mortality and institutions and economic development!

Likewise, you also have critiques like this one from Branko Milanovic, which state that the paper ignores the historically contingent nature of institutions and that you couldn’t traspose, say, English parliamentarism into Uganda and have a new world power just like that. But Acemoglu, Johnson, and Robinson actually do pay very close attention to how certain institutions can get “locked in” from the past (called path dependence in economics, a wildly influential concept), and do pay a lot of attention to cultural factors, at least in their recent work and to the detriment of institutions.

In general, critiques of the AJR paradigm along these lines fail because they tend to overstate certain aspects - in particular, they tend to promote a “vulgar AJR” where economic institutions are all that matters, they can be easily implanted, and colonialism was only bad because of the property rights violations. But in fact, they have quite a nuanced view of institutions, which they think can’t be uprooted and replanted easily, and prioritize political over economic explanations consistently, which means that property rights are usually given a very underdeveloped treatment relative to political bargaining and elite processes.

One final point that is made is that AJR is “Eurocentric” and promotes European institutions and values as the ideal. I don’t think this is straightforwardly true: the cases of countries with institutions that developed are all in the Global North or adjacent because those are the only countries that managed to develop. Additionally, I think that directly blaming European colonialism and exploitation for the underdevelopment of the Third World is quite radical for public perception of the economics profession, and a lot of this is just people mad that they don’t blame colonialism in the same way they do. The main Eurocentrism is actually that not enough thought is given to precolonial institutions, which are extremely important - especially for the Reversal of Fortune thesis.

Why is it always you three?

Now, onto the three canonical AJR papers and their critique

“The Colonial Origins of Comparative Development” (2001)

The most famous paper in the bunch is “The Colonial Origins of Comparative Development: An Empirical Investigation” (2001), which launched AJR as a phenomenon on par with the Beatles.

The issue in question is that, while better institutions and better GDP are also correlated to each other and to variables that aren’t measurable, you need a middleman between the two. This middleman is called an instrument, and it’s only correlated to the cause variable (institutions) but not to any of the consequences (GDP) or other unobservables. You can test the first step of the relationship, which is somewhat strong in AJR 2001, and the other two are more or less just leaps of faith.

The instrument AJR use is the mortality rate of European settlers, which are caused by various geographical conditions (mainly, climate) and which are variable across developing countries and not systematically correlated to economic outcomes. The higher settler mortality rates would have thus incentivized the adoption of practices like slavery and forced labor, which in turn resulted in extractive institutions. Meanwhile, countries with better conditions of life created inclusive institutions similar to the ones in Europe, which over time resulted in a more representative government with more respect for private property. The regressions used by the paper, between settler mortality rates in the XIXth century and present GDP show, unsurprisingly, that this relationship holds.

The paper has a number of criticisms made later on. First, there are major concerns with the data, which interpolates unavailable mortality rates with either posterior ones or those from other countries; using better data, the relationship only holds for the US, Canada, Australia, and New Zealand. Plus, the measurement of institutions (an index that estimates the risk of expropriation) is also itself extremely flimsy and contentious on its own.

Secondly, there appears to be a problem with the instrument: the estimate for the institutions - GDP correlation is smaller than the one for the mortality - GDP one; this means that, rather than taking out the problematic unobservables, the instrument is adding more of them. This is because geographical variables make for bad instruments, particularly since disease, rainfall, and terrain are both tied to the mortality of European settlers and to other variables and to GDP, directly and indirectly. For example, settler mortality could also measure human capital, which is not related to institutions very straightforwardly - particularly, because the human capital of the settlers may have both caused and been a cause of said underdevelopment. And, as mentioned above, economic organization also has massive confounding variables with natural resources (tropical climates are good for growing tropical crops), and thus with slavery, per Engerman and Sokoloff, which also simultaneously caused underdevelopment in Africa.

Additionally, the history in question is somewhat important, particularly since the colonization of the “New World” took place over the better part 400 years: Africa was colonized substantially later and by different countries with different institutions to the Americas. Because of this, medical technology was different as a function of higher overall technological advancement, and more importantly, settler countries had better institutions in the 1850s than in the 1500s (the Glorious Revolution happened in 1688! Good institutions just didn’t exist in 1492): simply put, the time between the Age of Discovery and the Scramble for Africa was around as long as between the First Crusade and the Age of Discovery. Plus, Africa in particular was significantly affected by slavery in the Americas, meaning that data points for institutional quality are correlated to each other and to an unobservable.

“Reversal of Fortune” (2002)

“Reversal of Fortune” is a paper with a pretty simple claim: countries that had higher population density, and/or bigger urban populations, in 1500 (before European colonization) now have lower GDP per capita. Population density and urbanization are (generally) considered positive signs of development, so what gives?

This paper also uses instrumental variables, this time using pre-colonial population density as an instrument for resource abundance, and is overall fairly similar in its procedings. The relationship is pretty simple: countries with higher urban populations also had either more resources and/or more organized governments¹.

This meant that there were more resources to extract and more people to indenture into extracting them. In consequence, European colonizers imposed much more extractive institutions on the originally richer regions, while less settled areas were left to their own devices, over time developing more inclusive institutions. For example, the the wealthy and well organized Aztec Empire has a suitable climate for plantations and had mineral wealth; thus, the Spanish enslaved most of the Aztecs’ subjects a lot more easily.

The issues here are also multiple. Firstly, in pre-colonial societies the abundance of land is closely tied to economic prosperity, meaning that the countries with less land (which were not unified polities back in the day) would have been poorer anyways, because of the Malthusian trap. Additionally, if Reversal of Fortune is true, then Colonial Origins can’t be: countries that were richer pre-Europeans are poorer now, and vice versa. But why were those countries richer? Well, both papers say institutions + geography, but geography also had to be shaped by institutions - Tenochtitlan was, after all, built on a swamp. But if you control for pre-colonial institutions, then the of the correlation between geography and post-colonial institutions just disappears.

“The Rise of Europe” (2005)

The third and last paper focuses not on the impacts of colonialism on the colonies, but on the colonizers themselves. The claim AJR make in this paper is fairly simple and straightforward: the countries in Western Europe that had more inclusive political institutions before colonizing other nations developed stronger property rights due to the greater influence of merchants on the political process. Thus, trade and colonialism in the Atlantic was beneficial to colonizers (a group made up of Britain, France, the Netherlands, Portugal, and Spain), and who kept those gains was either a small elite cadre or a larger group of merchants and seafarers, depending on how “tyrannical” the government was.

The importance of the Atlantic trade and of colonialism cannot be overstated: Western Europe was not significantly richer than Eastern Europe or Asia before the 17th century, but began steadily diverging ever since - but only for nations involved in trade, as the rest of Western Europe (think Italy or Belgium) did not benefit. Citing North & Weingast (1989), AJR posit that political, not economic, institutions were the key to growth: the less tyrannical a nation during the Age of Discovery, the likelier it was to later develop a capitalist regime with strong property rights, since merchants could simply petition their relatively weaker sovereigns.

The issue isn’t the relatively straightforward thesis, but rather the much weaker empirics: GDP per capita isn’t reliably available, so urbanization is assumed to be a good proxy for it (using data from the Reversal of Fortune paper). The regression is pretty simple: urbanization as a product of geography, time, some fixed characteristics of countries, and the potential for Atlantic trade through various measures - particularly comparing Spain, England, and non-traders.

The main issue here concerns the identification strategy, which is comparing Western countries that traded to those that didn’t. If the differences in trade status or trade volume were caused by factors unrelated to institutions, then the two are a valid counterfactual to each other (for instance, think geography); however, if there are systematic institutional differences between traders and non-traders then you don’t have a valid counterfactual, and therefore aren’t comparing apples to apples. Thus, instead of random assignment, what you end up measuring is that countries that are favored to trade more, do actually trade more - astonishing! This is especially concerning considering that the polity scores are not very transparent themselves.

A second issue here is simply that the sample is very small: there’s only a handful of countries considered, and a handful of time periods. Thirdly, the quality of urbanization data isn’t very good (more on this later), the division between traders and non traders is a bit arbitrary, and the way institutions are accounted for is also arbitrary and weird (“Protestant” and “Roman”). Lastly, the history is a bit shaky: Spain and Portugal acquired most of their colonial possessions in the 1500s, while England and the Netherlands did so a century later, falling into another “legal origins” trap.

Conclusion

So, why did AJR win their Nobel? Not because their research is the be all, end all of the literature, but because it opened the gates to a lot of discussion that still continues to shape the literature - the entire field of development is, in some way, responding to their highly imperfect work.