Comment on Smith’s Critique of Why Nations Fail

Nathan Smith (via Bryan Caplan and Ryan Murphy) describes Daron Acemoglu’s and James Robinson’s Why Nations Fail as “one of the most over-rated books [he] has ever read.” I first read the book in the summer of 2012, so over a year has elapsed since then. I remember that there were parts of it I disliked. However, I don’t think Smith is being entirely fair in his critique.

Summing up Smith’s three main complaints,

  1. There is a lack of rigor (no formal theory or econometric evidence);
  2. Acemoglu and Robinson present a naïve view of democracy;
  3. The “inclusive” and “exclusive” terminology is “bizarre” (e.g. property rights are considered to be part of an inclusive set of institutions, but they’re actually designed to exclude).

Starting with (3), I think it’s best to interpret Why Nations Fail through a Buchananite (? Buchananian?) lens. When Acemoglu and Robinson refer to institutions, I have the feeling that they’re specifically thinking about political institutions — as opposed to things like property rights, the price system, et cetera. An inclusive institution, in this constricted context, is essentially synonymous with a pluralist institution. In public choice theory terms, inclusive institutions are those which approach unanimity at the constitutional level (and maybe also the decision-making level). This is what the authors most likely mean when they define inclusive and exclusive institutions in terms of the distribution of power.

Before discussing (2), I think we should quickly address (1). Why Nations Fail is not a scholarly book; it was not written for an academic audience. Rather, the book is intended to present the authors’ academic research to a non-expert audience. Looking at the authors’ academic work, there is an obvious difference in approach: the bulk of book is made up of historical illustrations of the authors’ main points, which are watered down versions of what they already presented to the scientific community. The academic piece that comes closest to the scope of the book is “Institutions as the Fundamental Cause of Long-Run Growth” (ungated), co-authored between Robinson, Acemoglu, and Simon Johnson. Other academic work that goes beyond the simple model presented in Why Nations Fail, includes: “Why Do Voters Dismantle Checks and Balances” (co-authored with Ragnar Torvik); “Why Not a Political Coase Theorem?” (ungated; a much simplified outline of the argument here); and (relevant to the authors’ discussion of political centralization) “The Monopoly on Violence: Evidence from Colombia” (ungated), co-authored with Rafael Santos. It’s also worth mentioning Acemoglu’s textbook, “An Introduction to Modern Economic Growth,” where the last 50 pages or so (not counting the appendixes and index) develop a mathematical model of institutions and growth. If we’re going to judge Acemoglu’s and Robinson’s contributions to growth theory (Smith calls Acemoglu “a disaster to the field”), we should look at their academic work, not at a book designed to popularize what otherwise would be unintelligible to the general public.

Smith might be right that Acemoglu and Robinson present a naïve theory of democracy in their book, but the value of the model has to be interpreted within the context of the objective of the book. It’s hard to think of relatively pluralistic political institutions that differ from democracy (which, I think, takes on a more general definition that includes representative rule), because society hasn’t experienced them yet. They do hint — although, perhaps, never explicitly tie together — towards several nuances which inform the readers’ understanding of what makes one set of political institutions more inclusive than another (which seem to support the interpretation I laid out above, when addressing point (3)). Take, for example, their discussion of the origins of Mexico’s and the United States’ respective constitutions (pp. 28–32 [all pages refer to the 2012 hardback edition]). Not only do they distinguish between the degree of unanimity at the constitutional-level, but they also compare the northern U.S. states with their southern counterparts. Democracy is a class of institutional sets, but the specific elements are relevant to the comparative success of one set over another. Consider also their reference to political instability in Latin America, despite democracy (pp. 37–38).

Consider the following,

Political institutions include but are not limited to written constitutions and to whether the society is a democracy. They include the power and capacity of the state to regulate and govern society. It is also necessary to consider more broadly the factors that determine how political power is distributed in society, particularly the ability of different groups to act collective to pursue their objectives or to stop other people from pursuing theirs (pp. 42–43).

Smith makes it seem as if Acemoglu and Robinson hold democracy as the ultimate standard of political institutions that induce (or, do not restrain) economic growth. But, in reality the authors’ argument is much more nuanced than that. Another example that contradicts Smith’s summation of their position is the reference to Syngman Rhee and Park Chung-Hee, who were autocratic rulers of South Korea (pp. 71–72). The authors mention the superior rate of growth, as compared to North Korea, despite both countries being autocracies.

The causal relationship between democracy and growth, however, is not clear, as Smith points out. He links to Robert Barro’s “Democracy and Growth.” The paper suggests that higher standards of living may lead to democracy, rather than the other way around. But, democracies don’t arise suddenly, being products of a spontaneous development of political institutions that often originate at the local level. It may be that a certain income level has to be attained to support a modern democracy, but this doesn’t mean that gradual changes in political institutions — that make them more inclusive — don’t positively affect standards of living. My guess is that the two phenomena are mutually reinforcing. Barro also argues that democratization, at some point, seems to have diminishing returns. (I’d say that widening the scope of collective action suffers from diminishing returns.)  This is consistent with the findings in the paper that Ryan Murphy cites.

Why Nations Fail does not discuss these problems. Here Smith is right. There is no discussion of knowledge problems that impact democratic decision-making. Neither is there an in-depth discussion of the advantages of smaller pluralistic governments versus larger ones, or the interaction of political institutions that make up a Federalist system. This was one of the problems I had with the book, specifically with the authors’ reference to “centralization.” They argue that centralization positively impacts growth. My understanding is that they’re comparing a state that governs over a territory with competing states (or governments) that violently compete for territory — in other words, territorial consolidation. But, this doesn’t really speak to the decision-making decentralization that characterizes most successful democracies. Neither does it consider that the optimal (territorial) size of nations may be a dynamic function.

However, I don’t think it’s fair to criticize Acemoglu’s and Robinson’s position on democracy by pointing to the public choice literature. Problems that weaken democratic decision-making may also be relevant when looking at autocratic decision-making. Similarly, while there are problems with democracy that may not exist in autocracies, there are problems with the latter that aren’t relevant to the former. The public choice literature, some of which is still in dispute (e.g. rational ignorance), does not say that monarchies, dictatorships, or other forms of past governance are superior, or equal, in economic performance to democracies. Neither do I think that Acemoglu and Robinson would deny that modern democratic institutions are improvable.

Finally, consider that the authors’ main argument is to explain the difference in national income between modern nations. It may be that the wealthier countries may need to develop new, more inclusive, more constraining institutions to improve their own standard of living after a certain point (or to breakthrough some growth rate threshold), but the typical undeveloped country has yet to transition from relatively exclusive governance to pluralistic, and institutionally-strong, democracy. Thus, their pro-democracy message, understood within a proper context, shouldn’t be as controversial as Smith makes it out to be.

Is Why Nations Fail overrated? Maybe; it depends on what the average sentiment is. If the average person thinks it’s the best resource on economic growth, then, yes, it’s overrated. Someone looking for rigorous theory and empirical evidence is better off looking at the authors’ academic work. But, it’s still a very good book. If you are a non-expert, it serves as a good overview of the authors’ research in the field. Is the book perfect? No. There is a lot missing. For instance, the authors consider institutions the fundamental cause of long-run growth, but they deal almost exclusively with political institutions. There is no discussion of economic institutions that have, at best, an indirect relationship with their political counterparts. But, I don’t think that the criteria Nathan Smith mention are good ones to judge the book by.

15 thoughts on “Comment on Smith’s Critique of Why Nations Fail

  1. truth

    To be fair, the empirical justifications behind a lot of AJR’s work is pretty much discredited in the literature (i.e. which is basically the final nail in the coffin of a lot of their settler mortality stuff [Albouy and others had been critiquing that work since the mid-2000s. One paper worth reading on the subject is Glaeser at al (2004): which attacks institutionalism in generam), which underpins the institutional framework they’ve consistently developed over the last ten years). If you want an institutionalist theory (which is fine, as long as you acknowledge a lot of the work done by Sachs/others on the effect of diseases (Carstensen and Gundlach (2006): is probably among the best examples) or remember to isolate the pure effect of geography) then there’s a very convincing argument that you should look to Engerman /Sokoloff (2002: is one example) instead of AJR.

    Honestly, take a class or two on the determinants of long-term economic growth or Latin American/colonial economic history – your continued support for AJR might erode a bit.

    1. Ryan Murphy

      But what is interesting, to me at least, is that AJR are on “our” side. The people in the mainstream attacking their line of research are those who support the human capital theory of economy growth. We don’t have tools to effectively distinguish between human capital and institutions, and when there is evidence presented against institutions (even if the measure incorporates democracy), it makes the stories about the links between human capital and economic growth comparatively better.

      1. truth

        Well, the mainstream in general, I’d give you, but not within the subfield itself. Many of the best economic historians I know (and it’s worth noting that my background is Latin American economic history, so I’m undoubtedly biased), are proponents of substantially more nuanced positions than just pure human capital (namely, “at times, look to Engerman/Sokoloff [which is 2000, I’m not sure why I said 2002], at times look to Douglass North or William Summerhill, but most of the time, let’s go to country and leader-specific effects”). If anything, a general theory of one factor of growth is dangerous to advocate.

    2. JCatalan

      For what it’s worth, I have taken classes on Latin American history and the Latin American economy. Not that classes are the ultimate standard by which to judge whether someone is well read or not. For future consideration, we should just assume that there is a reasonable disagreement between us, and not that one of us is better educated.

      Thanks for the Albouy and Cartensen/Gundlach papers. I haven’t seen those before. But, the Albouy paper doesn’t “discredit” Acemoglu’s research. It just argues that the empirical model Acemoglu used in one of his papers was inadequate for what it was designed to do. That happens all the time. There is a lot of criticism, for example, of econometric work on monopsony and wages, but we don’t say that these criticisms of empirical models discredit the entire approach (or theory). My understanding is that Acemoglu’s work is heavily contested, but that’s different than saying that it has been discredited.

      Nevertheless, I do agree with you that there is better work out there on the role of institutions in economic growth. That doesn’t mean that we have to hate on Acemoglu, though — especially hate on him for poor reasons.

      1. truth

        Well, it’s not *one* paper, it’s the underlying regression for most of AJR’s empirical basis for their theory. If you look to the Acemoglu chapter you reference above, for example, they call AJR 2001 as their empirical justification. Cut down the base and the tower crumbles -> and Albouy destroys the base.

  2. M.H.

    That study which uses structural equation modeling methods

    shows some causal pathways from earlier democracy levels to future GDP (look at the standardized path (β) coefficients, figure 2 and 3) of modest size (see table 1). However, education and intelligence are much better predictors of later GDP.

    With regard to this “[Barros’] paper suggests that higher standards of living may lead to democracy, rather than the other way around” Rindermann’s paper reports the following (note : demo stands for democracy, 1 for time1, 2 for time2) :

    National wealth had no positive influence on democracy (βGDP1→Demo2=−.12). Bivariate analyses with education and democracy (N=85 countries) or with GDP and democracy (N=98 countries) support these results (βSY1→Demo2=.67, r=.79, N=85; βGDP1→Demo2=−.28, r=.69, N=98; no figures). Education has always a stronger impact on democracy than vice versa; and democracy depends more on education than on wealth. A positive influence of GDP on democracy is completely attributable to education.

    Given this, I believe that democracy is much more a product of education/GDP than the reverse. Of course I am not rejecting the possibility of mutual effects. It seems likely to me. For example, democracy seems to predict cognitive abilities, and the latter is a good predictor of later GDP.

    1. JCatalan

      I remember skimming a (I think recent) paper that argued that more education does not improve democratic outcomes. But, there is other research that does support what you say (e.g. Murtin & Wacziarg [2013] — saw it recently referenced on Freakonomics, if I remember correctly).

      But, the paper you link to reads,

      The more important influences on cognitive abilities have been economic wealth, economic freedom, rule of law and (less stable) political liberty.

      They mention that the evidence doesn’t show that democracy has a direct relationship with improvements in education. But, what about an indirect relationship (by improving wealth, freedom, rule of law, et cetera — the latter two seem to be directly related to democracy and political institutions)?

  3. Nathan Smith

    I was a history major as an undergrad, and was considering going to grad school in history. One reason I didn’t was the influence of various neo- or quasi-Marxist schools in many universities. These people were laboring hard not to be disillusioned by the disastrous communist episode in the Soviet Union and the failure of Third World countries to flourish when freed from the imperial yoke. They REALLY wanted the poor of the earth, somehow, to be innocent victims, and the rich and powerful bourgeois capitalist West to be to blame for everything. Since such stories are not consistent with the actual facts of history, these people learned to cloak their thoughts in elaborate jargon, which protected them against refutation. They could operate like mafias in academic faculties, supporting one another, excluding nonbelievers in their theories and nonusers of their jargon.
    What impressed me about economics is that it is kept honest by two forms of rigor: (a) the logical rigor of economic theory, and (b) the econometric rigor of empirical research. To publish economic theory, your logic has to be impeccable. Unless you just made a mistake somewhere, your theory will demand assent even from those who hate your conclusions, at least in the “given your assumptions, your results follow” sense. Although it is always possible to question assumptions, theoretical integrity matters. It limits what you can say, and does so, to a considerable extent, in the interests of truth. At any rate, it leaves neo-Marxist jargon-mongers high and dry. To publish empirical claims, you have to have significant regression results. While data mining can be a problem, fundamentally it’s pretty hard to get significant regression results, and the need for them does keep people honest.
    The reason Acemoglu and Robinson are a disaster for the field is that they’ve become just about the most influential people in the field while abandoning both these kinds of rigor. Yes, AJR (2001) did get significant regression results. But it’s still pretty bogus econometrically, as Albuoy shows (thanks truth). I didn’t realize before reading him how shaky the settler mortality data itself is, but it was always obvious to me that using settler mortality as an instrument for institutions was loony. There are a million ways settler mortality could be “correlated with the error term” in a bivariate regression, the most important being that settler mortality is just as strongly linked with a human capital channel of causation as with an institutional channel of causation, and that settler mortality is correlated with the modern disease environment. Development economics should have rejected AJR (2001) as evidence for institutionalism quickly and thoroughly, as a possibly interesting data pattern interpreted with extraordinary incompetence and failing the methodological individualism standard at the theoretical level. And moved on.
    Institutions have become the dominant explanation of development largely by default, as theories that were more clearly stated are more easily tested and found wanting. That’s not to say that there’s nothing in institutions; rather, “institutions” becomes an umbrella under which a lot of theories of varying quality huddle. North is guilty of “Whig history” to a considerable extent, but is quite subtle and shrewd compared to Acemoglu and Robinson. Fundamentally, we need to get back to methodological individualism as a standard of rigor in theory and econometric integrity in empirics. Acemoglu and Robinson totally fail the first test and mostly fail the second test; at any rate, what little solidity there is in their empirical results is nowhere near enough to justify basing a book as fat as *Why Nations Fail* on it. If we can’t purge that kind of humbug from economics, economics will lose the kind of rigor that kept us honest where other social sciences have been prey to leftist ideologues.

  4. Nathan Smith

    The public choice literature has plenty of flaws, but it at least ATTEMPTS to cross-apply methodological individualism to public sector institutions. It is a good antidote to the kind of sloppy thinking that could frame hollow concepts like “inclusive” and “extractive” institutions.

  5. Nathan Smith

    re: “Finally, consider that the authors’ main argument is to explain the difference in national income between modern nations. It may be that the wealthier countries may need to develop new, more inclusive, more constraining institutions to improve their own standard of living after a certain point (or to breakthrough some growth rate threshold), but the typical undeveloped country has yet to transition from relatively exclusive governance to pluralistic, and institutionally-strong, democracy. Thus, their pro-democracy message, understood within a proper context, shouldn’t be as controversial as Smith makes it out to be.”
    Er, the problem is that in no sense is “pluralistic, institutionally-strong democracy” either a necessary or a sufficient condition for economic development, and the evidence provides little if any support for the idea that it’s even helpful. Protection of property rights, probably, though much more would have to be said. Democracy, not at all.

  6. Nathan Smith

    re: “The causal relationship between democracy and growth, however, is not clear, as Smith points out. He links to Robert Barro’s “Democracy and Growth.” The paper suggests that higher standards of living may lead to democracy, rather than the other way around. But, democracies don’t arise suddenly, being products of a spontaneous development of political institutions that often originate at the local level. It may be that a certain income level has to be attained to support a modern democracy, but this doesn’t mean that gradual changes in political institutions — that make them more inclusive — don’t positively affect standards of living. My guess is that the two phenomena are mutually reinforcing.”
    That’s fine, as a GUESS. It’s a guess that seems empirically testable, in principle, and evidence for it has been elusive. I doubt that it’s really true. My guess– again, just a guess– is that democracy has little relevance for growth, other than maybe avoiding extremes on both ends: democracies rarely do either really well or really badly at it. But the point is that Acemoglu and Robinson’s theory and evidence are weak enough that they are entitled to call their theory nothing more than a guess, yet they expound it with arrogant certainty, dismissing rather disdainfully a lot of clearer and better supported theories than their own, and then using a tendentious rambling history to support their claims, a style of argument which could be used to support just about any claim equally well.

    1. JCatalan

      I think I agree with pretty much everything you’ve said (although, I’m still skeptical that democracy — although, I suppose I really have in mind pluralism — is not strongly related, at least historically, with economic growth). However, I guess that I’m more inclined to see the value in AR’s research, despite whatever flaws it may have. And, while they need to develop a better model to test — especially if they’re going use it as the evidence for their argument –, at the same time critics aren’t making comparable tests to disprove their theory.

      Regardless, I still disagree with you that: (a) they lack formal, rigorous theory (to judge this we have to look at their academic work, not at a popular book, IMO); and (b) the “inclusive” and “exclusive” terminology is “bizarre.”

      1. Nathan Smith

        Let’s just say that “inclusive” and “extractive” are very crude and unilluminating terms.
        The problem is partly semantic. Much of what AR mean by “inclusive economic institutions” is respect for property rights. But property rights consist precisely in the power to EXclude. We didn’t really need a new term for “respects property rights,” but if we wanted one, “inclusive economic institutions” would be a singularly obtuse choice. Still, if you want to say, “Inclusive economic institutions mean that everyone is included in the property rights regime, in the sense that everyone’s property (if they have any) will be protected,” then fine, that makes sense. Bad word choice, useful concept.
        Only AR don’t want to mean just that property rights are protected. By “inclusive economic institutions” they want to mean a sort of vague mish-mash of whatever they happen to like and whatever happens to have worked. They don’t want to define their terms clearly enough that their claims can be properly checked, and adverse data can get in the way of their tendentious storytelling.
        “Extractive” seems to refer partly to mining (e.g., settlers in disease-ridden environments were often there to mine something) and partly to forcible exploitation of subject populations (which of course is quite a different matter). Again, if AR meant by “extractive economic institutions” simply high tax rates, that would be an interesting, and more or less testable, hypothesis. But they don’t want to mean anything so straightforward as that. They don’t want to apply their term to rich democracies with top marginal tax rates of 90%, though the term would apply very fittingly. They do want the term to apply to low-tax regimes that presided over economic stagnation, like early modern China.
        The whole idea of a DICHOTOMY between inclusive and extractive institutions makes little sense. Logically, a state could refuse to “include” you, in the sense of protecting your rights, without “extracting” from you. It could also be both inclusive and extractive, protecting your property rights, but exacting high taxes.
        The inclusive/extractive terminology seems to be chosen to suggest that inclusive political institutions should naturally be connected to inclusive economic institutions, and extractive political institutions to extractive economic institutions. In fact, there seems to be little or no theoretical reason why these should be connected. “Inclusive political institutions,” in the sense of democracy, can quite easily lead to extractive economic institutions, as majorities use their power to take wealth from the rich or from unpopular minorities and redistribute it. “Extractive political institutions,” in the sense of dictatorship, might very easily choose to respect property rights in “stationary bandit” fashion, so that there will be more wealth to tax– or simply for ideological reasons.
        History does not suggest, actually, that there is much connection between “inclusive” political and economic institutions. You can hand-pick historical episodes and narrate them tendentiously, as AR do, to suggest this. In fact, I think you could make a strong-seeming case for almost any historical hypothesis using this wholly invalid method. But it’s equally easy to find counter-examples. Consider the case of Britain. There was pretty good protection of property rights in Britain going far back into the Middle Ages. Royal monopolies in the 17th century were an irritant, but Parliament engaged in plenty of stupid micromanagement of the economy after 1688, too. What do you think Adam Smith was protesting against in *The Wealth of Nations?* And of course Britain was far from being a democracy after 1688. Not until well into the 19th century does the British constitution deserve to be called inclusive. And note what happened then. Laissez-faire capitalism and democracy coexisted for a generation or so, but then the British state grew and became much more extractive in every sense, from soaring tax rates to nationalized industries.
        Or consider Pinochet’s Chile. Or the contrast between China and India today. Or the rather depressing experience of the transition to capitalism in post-Soviet Russia. Or consider the oddity of AR’s claim that colonial institutions cast such a long shadow, when the imperialists are long gone and many revolutions have come and gone since. Or consider that most variables that purport to measure institutions are quite volatile.
        AR do not provide a formal, rigorous theory in their academic work. Their thinking is too sloppy to be useful. Yet at the same time, it seems pretty clear that if they did manage to formalize their theory enough to make it properly testable, it would not perform well. Everything that’s true in their writing was known long before and has been said better by others. All they do is add a lot of arrogant overconfidence and muddled thinking. The sooner we can purge AR’s influence from development economics and forget about them, the better.

  7. Pingback: Introduction to Institutional Economics | Open Borders: The Case

Leave a Reply

Your email address will not be published. Required fields are marked *