The Rise and Fall of Piketty Critiques
Daren Acemoglu and James Robinson (AR) have a new paper critiquing Piketty. They argue that Piketty’s search for the “general laws” which govern capitalism fails, as it failed with the classical economists, because it is both ahistorical and ainstitutional. AR argue that Piketty’s famous ‘r > g’ does not explain the dynamics of income inequality as well as their own alternative framework, which puts political institutions at centre stage. To demonstrate the effectiveness of this framework over Piketty’s they use the examples of South Africa and Sweden.
Won’t somebody please think of the institutions?
The claim that Piketty’s work is ahistorical and ainstitutional is an odd one which is easily belied. For a start, Piketty states that the truth of r > g “depends, however, on the shocks to which capital is subject, as well as on what public policies and institutions are put in place to regulate the relationship between capital and labor.” Piketty’s obvious awareness of institutions is presumably the reason he spends four chapters documenting the kinds of political institutions that might be put in place to counteract a rise in inequality: a wealth tax, financial transparency & regulation, a very high top income tax, and numerous other suggestions. Yet AR spend a non-trivial amount of time documenting the rise of the welfare state in Sweden as if it is a counterexample to Piketty’s arguments, rather than one of the major examples of his proposed reforms working.
AR’s other major example of institutions at work is how apartheid in South Africa affected income distribution, both within as well as between races. It’s true that Piketty doesn’t deal with apartheid in South Africa explicitly, but I’m confused as to why on earth AR think he would disagree with them that institutions like apartheid matter. First, Piketty spends a considerable amount of time detailing the effects of slave ownership on economic development and income distribution in the USA. Second, he justifies his heavy use of data from France partly because “the French Revolution…quickly established an ideal of legal equality in relation to the market.” Such a condition clearly excludes apartheid, making his 'laws' institution-specific.
This kind of selective reading pervades the review. At one stage, Piketty argues that wealth inequality has not yet increased because “not enough time has passed since 1945”. In a footnote, AR quote the rest of Piketty’s sentence as “This is no doubt part of the explanation, but by itself it is not enough,” and churlishly add that this leaves the reader “to ponder what else is required for the claim to be true.” Yet if they had managed to read to the end of the paragraph, they would have seen that Piketty immediately makes two arguments: one, labour inequality has risen rather than capital inequality; two, significant redistribution has taken place. Surprise surprise, AR then attempt to use both of these as points against Piketty.
As the title of AR’s paper is “The Rise and Fall of General Laws”, it is worth briefly considering their (mis)treatment of Marx, Malthus and Ricardo’s ‘general laws’, too. This shares some of the same characteristics as their mistreatment of Piketty, in particular the idea that the “laws” identified were not contingent on at least some factors and did not have any offsetting tendencies. For example, AR argue that Malthus “deduced that increased fertility would always drive wages back to subsistence”, but this turned about to be wrong. They fail to note, as Grant McDermott has previously detailed, that Malthus was asking what would happen if both population and food were completely unmanaged and unchecked. Over on Marginal Revolution, commenter Ross Emmet helpfully directs us to his paper discussing Malthus’ other writings, which explored the institutions (!!) which might counteract the tendencies Malthus had in mind in his initial essay.
Similarly, AR state Marx’s Tendency of the Rate of Profit to Fall (TRPTF) somewhat glibly (or dishonestly) as “as capital accumulates, whatever the path of technology, the rate of profit [RoP] falls”. Yet in chapter 14 of part 3, volume 3 of capital Marx identifies six “counteracting tendencies” which work against the TRPTF, one of which is “Cheapening of Elements of Constant Capital”, i.e. a reduction of costs, one of the results of technological progress. AR then go on to reference work showing the RoP wasn’t falling when Marx was around; they might also want to check more recent empirical work by Andrew Kliman showing the opposite for the contemporary USA. Economist Branko Milanovic has commented that AR read like “Wikipedia entries with regressions”; frankly, reading the Wikipedia entry on Marx would have been a great improvement.
As for Piketty himself, he is simply not seeking to uncover general laws of capitalism. What he is doing is identifying the conditions under which inequality will tend to increase, asking whether they are empirically reasonable, and making predictions based on this framework. His first law is just an identity; his second law is an “asymptotic law”, subject to a number of qualifiers, which describes the direction in which the capital/income ratio will evolve at any one time. As for r > g, he himself states that it “is a contingent historical proposition, which is true in some periods and political contexts and not in others.” And we have already seen that he places institutional considerations front and centre of his analysis.
The strangest thing about the choice of theme is that AR dispute the idea there are general laws of capitalism by...offering their own general laws of capitalism:
“The quest for general laws of capitalism or any economic system is misguided because it is a-institutional. It ignores that it is the institutions and the political equilibrium of a society that determine how technology evolves, how markets function, and how the gains from various different economic arrangements are distributed.”
So there is no general theory of how capitalism will work, except a general theory in which everything flows from institutions. Cool.
Piketty’s formal framework
AR’s misinterpretation of Piketty’s formal framework is, unfortunately, predictable. They follow Krussel-Smith (KS) in reconstituting Piketty’s ‘second fundamental law’ as a straightforward equality, rather than as he intended it: an asymptotic, dynamic relationship. AR write the law as:
Capital share of national income = r x s/g
Which is just the basic β = s / g multiplied by the rate of return on capital. They go on to argue that r and s are required to be constant as g falls in order for capital’s share of income to increase. But this is only true if the equality holds at all times, which Piketty makes clear it does not. Piketty’s simple point is that with a high r and low g, capital’s share of income will tend to move upwards, depending on the starting point. The higher the savings rate, the faster this will happen, but it is not required that the steady state equilibrium described by the second law ever materialises, and there is no reason to expect it would do so in reality.
AR make a similar mistake with Piketty’s r > g. They seem to interpret it as the proposition that, the greater the difference between r & g, the worse subsequent inequality. They then run regressions of r – g on subsequent inequality at various time lags, find no significant association and conclude that r > g is not what’s driving inequality. But this is not Piketty’s argument: as Dan Kervick has documented, Piketty is always careful to add qualifiers such as "distinctly and persistently"; “markedly and durably,” “significantly and durably” when describing the gap between r & g. It is especially odd that AR do not realise this, since they quote two of the relevant passages in full.
In other words, Piketty is not arguing that inequality will increase to the degree that r is greater than g; he's arguing that if there is a significant gap between r & g for a significant period of time, inequality will begin to increase. Subsequently, we would not expect to see a linear correlation between r > g and subsequent inequality, no matter the chosen lag – the correlation for low values of r – g would be too weak. It is also strange that AR do not include controls for policies and institutions in their regressions, considering this is the major thrust of their paper. I’m not dismissing the econometric approach outright: it is one of the better ideas in the paper. However, the execution is poor.
To sum up,
AR claim that Piketty seeks to uncover immutable ‘laws’ of capitalism which
transcend politics, culture and institutions and which result in an inexorable
rise in inequality. However, Piketty is well aware that there are offsetting
tendencies to his laws, which themselves are historically contingent and which
he thinks can and should be directly
counteracted by political institutions. AR also misinterpret the laws
themselves as straightforward mechanistic relationships, the kind which can be
put into steady-state growth models and linear regressions (is there any other
way to do economics!?), but Piketty’s primary concern is uncovering the dynamic
behaviour of the economy, given certain conditions, at any one time. Finally,
it’s not clear how AR’s main choice of ‘counter examples’ to Piketty – Sweden and
South Africa – have any bearing on his point whatsoever. I genuinely hoped for
better when I first saw the authors of the review.
Keep up to date with the latest thinking on some of the day's biggest issues and get instant access to our members-only features, such as the News Dashboard, Reading List, Bookshelf & Newsletter. It's completely free.