Acemoğlu and Robinson
March 4, 2024
I’ve been reading quite a bit of Acemoğlu and Robinson recently: their three books – The Economic Origins of Dictatorship and Democracy (2005); Why Nations Fail (2012); The Narrow Corridor (2019) – as well as various articles. There are many more articles I could and should read, but I feel I’ve read enough to give a high-level summary and assessment of their project.
Broadly speaking, I think you can say that Acemoğlu and Robinson have three ‘big ideas’.
- Political-economic institutions are the major determinant of economic growth.
- Political-economic institutions are themselves largely a function of the balance of class power within a society.
- The single largest determinant of the global distribution of wealth is the legacy of European colonialism.
I think all of these three ideas have a great deal to recommend them, and so at this level of abstraction I am strongly on board with Acemoğlu and Robinson’s project. Getting into slightly more detail, though, I do have some points of difference or criticisms. In the rest of this post I want to discuss a few specific dimensions of Acemoğlu and Robinson’s work.
First up: Acemoğlu and Robinson’s (hereafter A&R’s) relation to the existing literature. One of the oddities of A&R’s approach is how thoroughly it takes on board a set of moves that are historically associated with broadly Marxist approaches to political economy, and yet how little A&R reference that tradition in their discussions of prior scholarship. I don’t entirely know what to make of this. Are A&R quasi-concealing the Marxist influences on their work for respectability reasons? Are they unaware of those influences? A&R are not politically Marxists – they are, as far as I can tell, basically left-neoliberal social democrats, in favour of a moderately redistributive dynamic market economy. But their framework imports huge amounts from relatively orthodox historical materialism. Douglass North describes himself somewhere as a right-wing Marxist, which I think is accurate – and would be an accurate description of A&R as well. A&R take it that the appropriate goal of mass revolutionary politics is democratic political institutions plus broadly distributed and stable private property rights. Obviously this is a major difference from orthodox Marxism. But if you make that substitution, A&R’s account of class struggle as the route to achieving this outcome is quite ‘orthodox’ – essentially they just substitute “more redistributive and democratic capitalism” for “communism” as the historic goal of mass politics. In this respect, I think A&R can plausibly be seen as the ‘inheritors’ of the more econ-aligned wing of the short-lived [though maybe undergoing a mini-revival?] analytical Marxist tradition (and indeed they cite Roemer, Przeworski and Michael Wallerstein at various moments). This isn’t how they present themselves, though, nor does it seem very central to their reception. I guess people care more about whether someone’s politics is Marxist than whether their analytic framework is.
Anyway, that’s one oddity of A&R’s relation to the existing literature. Another is that they are not reliable summarisers of that literature. One example: they differentiate their position from that of Eric Williams, whom they seem to understand as emphasising “the plunder of the colonies by the Europeans” as the major driver of differential core/periphery development. A&R, by supposed contrast, argue that the most important difference between “a society like the Caribbean colonies during the 17th and 18th centuries” and the post-industrial-revolution European core was different categories of capitalist institution: extractivist and monopolistic versus industrial and dynamic. But this claim is central to Williams’ argument in ‘Capitalism and Slavery’ – the second half of that book is about the eclipse of the mercantilist-slaver model of capitalism by the free-trade-industrialist model, the slave colonies ‘left behind’ by the new, dynamic form of capitalism rising to power in the core. A&R seem to be overstating the novelty of their argument by contrasting it with a position that seems to me more like a precursor.
Another, stranger, example of this phenomenon is their treatment of W. Arthur Lewis. In their most-cited paper, ‘The colonial origins of comparative development’ (2001), A&R cite Lewis as someone who has lucidly articulated their most central claim about the damaging effects of the persistence of extractivist colonial institutions into post-colonial independent states:
Arthur Lewis gives a succinct statement of the issues (1965, pp. 32-33):
“… for most of [the governments of newly-independent African states] independence means merely that they have succeeded to the autocracy vacated by British and French civil servants. They model themselves on the arrogant and arbitrary pattern set by Governors and district commissioners.”
Here Lewis is cited as an influence on – or at least precursor to – A&R’s argument, and A&R’s original contribution is providing econometric support for a position already articulated, at least in broad brush strokes, by Lewis. By the time we get to ‘Why Nations Fail’, though, A&R seem to have forgotten this dimension of Lewis’s work. In chapter nine of ‘Why Nations Fail’ they contrast their institutional approach to Lewis’s ‘dual economy’ theory of development. They summarise Lewis’s dual sector model and argue that development theorists, building on Lewis, have failed to understand the political origins of the dual economy in many underdeveloped countries:
This perspective has much truth to it but misses the entire logic of how the dual economy came into existence and its relation to the modern economy.
Later:
The dual economy of South Africa did come to an end in 1994. But not because of the reasons that Sir Arthur Lewis theorized about. It was not the natural course of economic development that ended the color bar and the Homelands. Black South Africans protested and rose up against the regime that did not recognize their basic rights and did not share the gains of economic growth with them.
The strong, and strange, implication here is that Lewis is somehow unaware of or inattentive to the political-economic institutional features – including apartheid! – that constrained South African economic development, or the role of political contestation in transforming such institutions. A&R do not cite Lewis’s 1985 book ‘Racial conflict and economic development’, where he writes that “after testing the water in Kenya” Britain learned:
that even small guerrilla groups could reduce government almost to a standstill. France took longer to learn, and bloodied herself in Algeria and Indochina; but she learned. … The Union of South Africa will be the last to learn, but its time has not yet come.
I think it’s clear that Lewis cannot legitimately be used as a foil for A&R here, as an example of a theorist unaware of conflictual or structural elements of inherited colonial institutions. But this is how A&R use him in ‘Why Nations Fail’, even though they quote him appropriately elsewhere. I don’t really know what’s gone wrong here, but I don’t think one can rely on A&R to accurately summarise the literature they are engaging with.
Ok. This is all pretty much in the domain of ‘scholarly quibbles’. Moving more onto the substance.
I’ve said that one (big!) difference between A&R’s apparatus and that of classical Marxism is their actual politics. Another difference is methodological. A&R are frequently translating into the language of mainstream economics positions that have been widely held for a long time in other subfields. (There’s nothing wrong with this at all, in my view.) They have two toolkits for doing this: game theory, and econometrics.
On the game theory side of things, A&R’s first book – ‘The Economic Origins of Dictatorship and Democracy’ – is essentially presenting a traditional class conflict model of the economic determination of political institutions in game-theoretic terms. This is extremely good, in my opinion. Moreover the game-theoretic packaging (plus, no doubt, the authors’ political vibes) seems to make the framework entirely palatable within mainstream economics. Anyone who wants to make a class-conflict base-superstructure-type argument within mainstream economics can just pick up this framework and run with it. The framework can easily be integrated with other elements of mainstream economics, which is also good. At least, it’s good for those of us who dislike the siloing of critical political economy and mainstream economics, and think that each has much to learn from the other.
On the econometric side of things, several of A&R’s papers are doing quite laborious econometric analysis to make the case that different kinds of colonial institutions have differential ongoing effects on nations’ economic development. I’m no econometrician, so I’m not really in a position to evaluate this side of things. I do have some tentative methodological worries about this approach – but I will aim to circle back to those later in this post.
Ok. But what about the actual claims A&R are making? Here I think it’s useful to distinguish three different elements of their argument.
- Their emphasis on conflict theory.
- Their distinction between inclusive and extractive institutions.
- Their argument about the legacy of colonial institutions.
I am on board with (1) – the emphasis on conflict theory. I can imagine two broad objections to this approach. The first objection is the traditional objection to class conflict accounts of history – that it’s too crude and simplistic, missing out the countless factors that cannot be explained in terms of class interest. My attitude to this critique is sort of doubled. On the one hand, this critique is clearly correct – this kind of analytic approach is very simplistic. I think it’s entirely fair to object that much of the substance of history and of social life goes missing in this kind of account. On the other hand, however, I feel like a lot of supposedly more nuanced and sophisticated frameworks in fact do worse than vulgar class conflict analysis when it comes to explaining lots of important political-economic dynamics. I regard ‘vulgar class conflict’ analysis as a sort of baseline that it is surprisingly easy to fall below. So I regard this kind of approach as providing a valuable starting point to work from, rather than as something to be dismissed.
The second common objection to the kind of approach that A&R pursue is an objection to its formalism: isn’t game theory itself simply too crude a tool to provide any really useful analytic purchase on complex social reality? Here, again, my attitude is sort of doubled: yes, in principle that’s a perfectly valid objection. In practice, however, I find that many people who object to the crudity of formal approaches themselves have frameworks that are, if anything, even cruder – while also lacking the virtues of clarity and transparency that come from spelling out one’s commitments in simple formal terms. So, again, I think we should be extremely open to legitimately more sophisticated approaches, but I also think there’s much to be said for the formalisms. There are categories of handwaving and bluster that are basically incompatible with developing a simple toy model to illustrate one’s claims – and that in itself is a strong virtue of toy models.
Anyway, as I say, I like the conflict theory dimension of A&R’s approach, and I like that they integrate their conflict theory with mainstream economics via the use of game theory, albeit with appropriate caveats and cautions.
Point (2) – the distinction between inclusive and extractive institutions – is more complex. I think it’s clear that A&R are onto something with this distinction: it’s obvious that many historical colonial institutions were oriented toward the extraction of a region’s wealth for the benefit of overseas interests, whereas many core economies are much more “inclusive” both politically and economically. Similarly, it’s clear that lots of political-economic institutions are set up for the benefit of a ruling elite rather than for the benefit of society as a whole. So this distinction definitely captures something. But I also think this distinction is clearly much too crude to do much explanatory heavy lifting. To be sure, A&R distinguish between political versus economic extractive and inclusive institutions. Similarly, they are clear that there is a spectrum of inclusivity and extractivism. For example, on their (broadly Whiggish) account, post-Glorious Revolution British state institutions were much more inclusive than those that preceded the civil war. But those institutions were still non-inclusive in all kinds of ways. The degree of inclusivity they exhibited, however, (on A&R’s account) facilitated a “virtuous circle” whereby the inclusivity of British institutions increased overtime, largely via the expansion of the franchise. For A&R, in other words, political and economic inclusivity and extractivism aren’t a simple on-off switch – their account is more nuanced than that. Likewise, in academic papers they aim to distinguish between different categories of institution – for example, distinguishing “property rights institutions” and “contracting institutions” as different elements of economic inclusivity.
Nevertheless, these categories still seem very crude to me. Unlike the crudity of the basic conflict theory apparatus, I don’t (perhaps inconsistently or incoherently) find this crudity to be very productive. Rather, I feel like a great deal more “unbundling” is needed to clarify what we’re actually talking about when we talk about ‘extractive’ or ‘inclusive’ institutions. In particular, it seems to me that there are many different dimensions of political-economic institutions, and it’s not at all obvious that moving in an “inclusive” direction in one respect cannot also be moving in an “extractivist” dimension in another. To take an example that A&R discuss: the enclosure of the English commons (that Marx also thematises in ‘Capital’) is clearly part of the shift towards a capitalist property-rights-based economy. And yet it involved the expropriation of de facto property and the concentration of property rights in the hands of a smaller number of economic actors. Was this move “inclusive” or “extractive”? Different moments in A&R’s texts seem to imply different answers to this question. In discussing expropriation of the commons in some developing contexts they seem to label this move as part of an extractivist shift; but their account of early modern Britain is an account of increasing inclusivity of economic institutions, with the move away from feudal towards capitalist structures. Because the distinction between ‘inclusive’ and ‘extractive’ institutions is so crude, and so vague, it’s not clear that A&R have the theoretical resources to analyse institutional shifts that are inclusive in one sense or from one perspective, and extractive in another.
Similarly, the category of ‘extractive’ institutions has to cover vast historical ground, on A&R’s account. The Mayan civilisation falls under this category; so do capitalist slave economies; so does the USSR and Mao’s China. These are three very different political-economic institutional set-ups. It’s not that it’s wrong to call them all extractive – it’s just not clear to me that ‘extractive’ is doing much explanatory work here. At a minimum, I think we need to be able to break down these ‘inclusive’/’extractive’ categories in a much more fine-grained way. As stands, it’s hard not to feel that the vagueness of the categories gives A&R a lot of wiggle-room to classify ambiguous political-economic institutions as extractive or inclusive depending on what best serves their argument at any given moment.
So this is one of my big issues with A&R’s apparatus – the lack of clarity around their core distinction between ‘extractive’ and ‘inclusive’ institutions.
My final criticism or worry concerns A&R’s treatment of the dynamics of the international system. European colonialism is central to A&R’s argument. Their account of differential global development is essentially as follows. Where Europeans set up European-style institutions in colonised countries (basically, settler colonies) those colonies have gone on to prosper even after independence, by virtue of their growth-oriented institutions. By contrast, where Europeans set up extractivist institutions – slave societies, or structures oriented to the extraction of raw materials or the exploitation of local labour entirely for the benefit of the European core – those extractivist institutions have also often persisted in some way post-independence. Those extractivist institutions have tended to make post-independence societies also extractivist, now largely for the benefit of a local elite, to the detriment of both economic growth and broad public welfare.
I think this argument is correct as far as it goes (albeit obviously requiring very many caveats). One worry, though, is that A&R underrate the role of ongoing power relations in the international system for differential development. A&R’s approach is to treat European colonialism as an exogenous intervention in local governance institutions, which beyond a certain point no longer exists. In A&R’s model, post-independence nations’ differential economic performance can then mostly be explained by their (now local) institutions. This account does not really leave room for ongoing power dynamics within the international system. A&R’s framework is all about power dynamics within nation states, but it does not meaningfully discuss power dynamics between nation states.
This problem has consequences for both sides of A&R’s methodological approach. On the game-theoretic side, I think A&R’s framework needs to be supplemented by much more analysis of international conflictual dynamics. This is not entirely absent from their framework – for example, they discuss capital flight, and they discuss the role that powerful states may play in encouraging democratic movements. But, to take the most straightforward example, they do not discuss coercive interstate actions – either military invasion and regime change, or the orchestrating of coups by powerful states to topple the governments of weaker ones. Nor do they discuss the support of some group or groups within an internal conflict by another power. Nor do they discuss simple economic power or leverage that some political-economic actors have over others within the international system. All of this would be analysable game-theoretically, just as A&R analyse conflict within a state between a ruling elite and the rest of the populace. But A&R largely haven’t turned to this side of things. This seems like a useful way to extend their framework. (Of course, it’s likely that lots of people have already done this, and I simply haven’t read them.)
This problem is arguably even more consequential for the econometric side of A&R’s methodology. It’s central to A&R’s approach that states can, beyond a certain historical point, be treated as independent entities for statistical comparative purposes. Comparative institutional analysis operates by looking at the differential impact of different institutions on different political-economic entities. But what if the appropriate unit of analysis – at least for some purposes – is not the state, but is rather the world-system? What if these are not independent entities at all, but are complexly interwoven within a system that has its own dynamics, including dynamics that impose different functions on different subcomponents of the system? In this scenario, the degree of independence of cases that is assumed by standard statistical techniques just wouldn’t hold. At a minimum, one would need to be more cautious than A&R in applying these kinds of comparative analytic techniques.
It seems to me, then, that A&R’s approach should take world-systems analysis much more seriously than it seems to. I don’t know what that would look like yet, but it feels like a fruitful potential line of inquiry. (Again, it seems likely that people have already done this, and I just need to find them and read them.)
To sum up. A&R’s work has several through lines: the importance of institutions, specifically the distinction between ‘extractive’ and ‘inclusive’ institutions; the importance of class conflict in forming those institutions; the role of colonial-era institutions in explaining differential economic development internationally. For me, A&R’s most significant contribution is probably the way in which they ‘operationalise’ conflict theory accounts of institution formation with a set of developed game-theoretic models. This is valuable in itself, but it is also valuable because it makes it easier to integrate important elements of critical political economy with mainstream economics.
I do have some criticisms of A&R, however. I think they make it harder than it ought to be to draw connections between their work and the work of scholars in critical political economy, because their discussions of other scholars’ work isn’t always reliable. More substantively, I think there are two big areas where A&R’s framework needs extension or modification. First, A&R’s distinction between ‘extractive’ and ‘inclusive’ institutions needs much more fine-grained unpacking. Second, A&R are insufficiently attentive to the interstate dynamics of the world-system as a whole. Hopefully, as I keep reading in political economy, I’ll find people who have done some of this work.
Regularism and game theory
January 20, 2024
In this post I want to contrast Brandom’s apparatus in (especially) ‘Making It Explicit’ with Lewis’s game-theoretic approach to convention (in ‘Convention’). It’s now quite some time since I read MIE, and I’ll revisit it in detail at some point. But for now I want to discuss some connections between these two works, as part of my general effort to veer the Brandomian orientation of this blog around towards more political-economic themes. (Lewis is, of course, a philosopher, not a political economist – but ‘Convention’ is a foundational work in the canon of formal economics.)
Regularism, recall, is the idea that norms can ultimately be identified with regularities of social behaviour. The regularist move is: norms are socially instituted; we defer to our social milieu for judgements about correct and incorrect practice; how are we to understand the judgement of ‘the social’ to which we are thereby deferring?; maybe we can simply understand it as a regularity of practice. If everyone in a milieu does such-and-such, they have instituted the norm that such-and-such.
I take it that Lewis’s game-theoretic account of the emergence of conventions as regularities of strategic behaviour, in ‘Convention’, is a paradigmatic example of such an account. Lewis is interested in ‘coordination games’, in which (I’m being a bit loose here, but this is the general idea) every player benefits if they coordinate their actions successfully with the other players, but coordination is non-trivial: there are multiple available coordination equilibria. A classic example is the practice of driving on the right (or left) side of the road. It doesn’t matter which side of the road car users drive on, so long as everyone drives on the same side of the road, but disaster ensues if coordination is not achieved.
Lewis analyses conventions in these terms. His ultimate target is linguistic convention, and he’s preoccupied with linguistic convention because of the debates arising out of Quine’s sceptical critique of Carnap’s mobilisation of the concept of analyticity. This is another example, for what it’s worth, of the way in which Carnap’s project’s influence looms over this entire intellectual tradition in a way that I hadn’t appreciated until recently. But the point is that Lewis is interested in analysing a particular category of norm – conventions – in terms of strategic regularities within a game-theoretic model.
Now, I want to discuss two Brandomian critiques of this regularist project.
The first is the critique from ‘gerrymandering’. This critique is basically: what counts as a regularity? ‘Gerrymandering’ can take place at the level of the actions that ‘count’ towards the regularity: which actions are included in the set of actions analysed will influence one’s understanding of the regularity in question. But ‘gerrymandering’ can also take place at the level of any given set of actions: because any action can be described under multiple perspectives, it is always possible to construe any given set of actions as compatible with multiple regularities.
Lewis discusses this latter issue in ‘Convention’ – and basically concludes his discussion by saying “yes, this is all true, but let’s not be silly here”. I think this is a legitimate approach to the problem of gerrymandering for most practical purposes, but it’s not clearly a legitimate approach if the point at issue is sceptical critiques of the very idea of convention. Brandom, in his discussion of regularism, pushes on this sceptical point about gerrymandering. Brandom draws (I think correctly) the following conclusion: gerrymandering means that there is always a normatively interpretive element to the specification of any regularity; therefore normativity cannot be reduced to regularities of practice. You can’t simply explain norms in terms of regularities of practice, because specifying the relevant regularity of practice is already an interpretive (therefore normative) task.
I think this is all correct as far as it goes. In terms of the ‘two cultures of the social sciences’ distinction, that I discussed a few posts ago, I think this argument shows that you cannot entirely eliminate the ‘hermeneutic’ or ‘interpretive’ approach to social science from the ‘reductionist’ or ‘objectivist’ approach. Even formal game-theoretic models of the kind Lewis lays out are still ‘interpretive’ and ‘internalist’ in some sense. In terms of the vocabulary of contemporary philosophy of science, even our apparently ‘reductionist’ theoretical toolkits are always ‘value-laden’. This blog endorses the broad philosophy of science approach articulated by Helen Longino, and this result seems compatible with that approach, to me.
Ok. But we should also, in my view, take care not to draw overly strong conclusions from this argument. The facts that theory is value-laden and that game-theoretic models are interpretive is only a problem with Lewis’s analytic toolkit if we are trying to use that toolkit to bolster a very strong reductionism. These aren’t objections to the tools; they are only objections to the attempt to use these tools to fully reduce norms to regularities of practice, with no explanatory remainder.
In fact, Brandom’s whole project is about the interpretive act of ‘explicitating’ the norms implicit in practice, and thereby specifying the practice itself under one particular description. And this project is also grounded in Brandom’s critique of ‘regulism’, or Brandom’s broader pragmatism. Brandom is clear that we do rely on grounding our norms in regularities of practice, as long as all parties agree in practice on the tacit implications of those practices. The practical problem of ‘gerrymandering’ appears when there is a disagreement over the normative content of those practices, and therefore different interpretive perspectives on the ‘same’ set of practices. But such disagreements (while they undermine the strong reductionist project of wholly explaining norms in terms of regularities) are themselves reliant on background uncontested norms implicit in currently unproblematic (that is, unchallenged) regularities of practice. Brandom’s approach is therefore pointing in two directions simultaneously – rejecting strongly reductionist regularism, while still grounding the interpretive disagreements that prompt the gerrymandering problem within a larger context of background regularities of practice.
So Brandom’s larger problem, as I see it, is how to explain these disagreements – the interpretive disagreements that allow the gerrymandering problem to become live even to participants in Lewisian coordination communities – in terms of the social practices that are also generating the regularities in question. And this problem is what Brandom’s complex scorekeeping apparatus addresses. Brandom’s core account, as I see it, says: yes, norms are generated by regularities of practice, but the regularities of practice have to themselves include practices capable of both generating and resolving the gerrymandering problem for participants in those practices.
Now, as I see it the distinction between de dicto and de re tracking of entitlements and commitments gives Brandom the resources to explain how social actors within a scorekeeping game can themselves perceive the gerrymandering problem. The game of asking for and giving reasons then gives those social actors the resources to resolve the gerrymandering problem – to reach an explicit discursive consensus rather than implicit practical consensus around the relevant norm-forming regularities, once any given gerrymandering problem has been rendered problematic by a ‘challenge’ to the ‘default’ background consensus of practical coordination. (Of course, the game of asking for and giving reasons and the practices of scorekeeping are themselves mutually implicated or constitutive, on Brandom’s account, but even so.)
To sum up the post so far, then. My first point is that Brandom’s critique of regularism shouldn’t be seen as a global challenge to Lewis’s account of convention, which is useful in all kinds of ways. Rather, it should be seen as a challenge to the attempt to derive a reductionist account of normativity from Lewis’s account. My second point is that if we want a model of social practice that can deal with the gerrymandering problem, the social practices in question need to incorporate Brandomian scorekeeping.
What does this mean for the relation between Brandom and game theory? Well, I’m focusing in this post on only one dimension of that relation – the problem of regularism (ignoring, notably, issues around instrumental reason). But in terms of the problem of regularism, I think it means two things.
First: we need to remember that game-theoretic models of the kind Lewis is articulating are greatly simplified, and can’t themselves fully ground our account of normativity. I think this is a relatively uncontroversial point – that our models are greatly simplified and do not provide a comprehensive or foundational social ontology is sort of a methodological banality in the modelling literature. Second, though: Brandom’s own fuller account does provide resources that could themselves potentially be interestingly incorporated into game-theoretic approaches. What would Lewis’s models of convention look like if they were extended to incorporate scorekeeping? I don’t know. Nor do I know whether such an extension would be useful: another rightly emphasised methodological point in the modelling literature is that greater modelling complexity is not necessarily desirable. But still, I’d like to explore this possibility, if I can, eventually.