## July 31, 2014

### Books to Read While the Algae Grow in Your Fur, July 2014

Attention conservation notice: I have no taste.

Stephen King, Eyes of the Dragon
Mind candy. I really liked it when I was a boy, and on re-reading it's not been visited by the Suck Fairy, but I did come away with two thoughts. (1) I'd have been very interested to see what a writer with drier view of political power would have done with the story elements (the two princes, the evil magician, the exiled nobles) — Cherryh, say, or Elizabeth Bear. (2) Speaking of which, it's striking how strongly King's fantasy books (this one, The Dark Tower) buy into the idea of rightfully inherited authority, when his horror stories are often full of healthy distrust of government officials ("the Dallas police"). I don't think he'd say that being electorally accountable, rather than chosen by accident of birth, makes those in power less trustworthy...
Charles Tilly, Why?
Tilly's brief attempt to look at reason-giving as a social act, shaped by relations between the giver and receiver of reasons, and often part of establishing, maintaining, or repairing that relationship. Tilly draws two distinctions: between reasons which involve cause-and-effect and those which use a logic of "appropriateness" instead, and those between which require specialized knowledge and those which don't. "Conventions" are common-knowledge reasons which are invoke appropriateness, not causal accounts. (Think "Sorry I'm late, traffic was murder".) "Stories" give causal explanations which only invoke common knowledge. Tilly is (explicitly) pretty Aristotlean about stories: they involve the deeds of a small number of conscious agents, with unity of time, place, and action. Codes are about matching circumstances to the right specialized formulas and formalities --- are your papers in order? is the evidence admissible? Technical accounts, finally, purport to be full cause-effect explanations drawing on specialized knowledge.
The scheme has some plausibility, and Tilly has lots of interesting examples. But of course he has no argument that these two dimensions (generalist vs. specialist, causation vs. appropriateness) are the only two big ones, that everything (e.g.) the "codes" box really does act the same way, etc. So I'd say it's worth reading to chew over, rather than being deeply illuminating.
Elliott Kay, Rich Man's War
Sequel to Poor Man's Fight, continuing the same high standard of quality mind-candy. — Sequel.
Alexis de Tocqueville, Democracy in America
Yet another deserved classic read only belatedly. Volume I is actually about de Tocqueville's observations on, and ideas about, democracy in America. This is interesting, mostly empirical, and full of intriguing accounts of social mechanisms. (I see why Jon Elster is so into him.) Volume II consists of his dictates about what democracy and social equality will do to customs and character in every society. This is speculative and often the only reference to America comes in the chapter titles. (I see why this would also appeal to Elster.)
I would dearly like to find a good "de Tocqueville in retrospect" volume. Some of his repeated themes are the weakness of the Federal government, the smallness of our military, the absence of serious wars, the relative equality of economic condition of the (white) population, the lack of big cities among us. So how have we managed to preserve as much democracy as we have? For that matter, how does the civil war and its outcomes even begin to make sense from his perspective?
— Rhetorical observation: de Tocqueville was very fond of contrasts where democracy leads to less dispersion among people than does aristocracy, but around a higher average level. He either didn't have the vocabulary to say this concisely, or regarded using statistical terms as bad style. (I suspect the former, due to the time period.) He was also very fond of paradoxes, where he either inverted directions of causal arrows, or flipped their signs.
Maria Semple, Where'd You Go, Bernadette?
Literary fiction about Seattle, motherhood, marital collapse, aggressively eccentric architects, and Antarctica. Very funny and more than a bit touching.
Thomas Piketty, Capital in the Twenty-First Century [Online technical appendix, including extra notes, figures, and spreadsheets]
Yes, it's as good and important as everyone says. If by some chance you haven't read about this yet, I recommend Robert Solow, Branko Milanovic and Kathleen Geier for overviews; Suresh Naidu's take is the best I've seen on the strengths and weaknesses of the book, but doesn't summarize so much.
Some minor and scattered notes; I might write a proper review later. (Why not? Everybody else has.)
1. Perhaps it's the translation, but Piketty seems wordy and a bit repetitive; I think the same things could have been said more briskly. Perhaps relatedly, I got a little tired of the invocations of Austen, Balzac, and American television.
2. The book has given rise to the most perfect "I happen to have Marshall McLuhan right here" moment I ever hope to see.
3. Attempts to undermine his data have, unsurprisingly, blown up in his attackers' faces. Similarly, claims that Piketty ignores historical contigency, political factors and institutions are just bizarre.
4. Granting that nobody has better point estimates, I wish he'd give margins of error as well. (A counter-argument: maybe he could calculate purely-statistical standard errors, but a lot of the time they could be swamped by nearly-impossible-to-estimate systematic errors, due to, e.g., tax evasion.)
5. His two "laws" of capitalism are an accounting identity (the share of capital in national income is the rate of return on capital times the ratio of capital to income, $\alpha = r \beta$ ), and a long-run equilibrium condition (the steady-state capital/income ratio is the savings rate divided by the economy-wide growth rate, $\beta = s/g$ ), the latter presuming that two quite variable quantities ($s$ and $g$) stay fixed forever. So the first can't help but be true, and the second is of limited relevance. (Why should people keep saving the same fraction of national income as their wealth and income change?) But I don't think this matters very much, except for the style. (However, Milanovic has an interesting defense of Piketty on this point.)
6. He gets the Cambridge Capital Controversy wrong, and while that matters for our understanding of capital as a factor of production, it's irrelevant for capital as a store of value, which is what Piketty is all about. Similarly, Piketty doesn't need to worry about declining marginal returns to capital in the economy's aggregate production function, which is good, because aggregate production functions make no sense even within orthodox neo-classical economics. (The fact that orthodox neo-classical economists continue to use them is a bit of an intellectual embarrassment; they should have more self-respect.)
7. The distinction between "income from labor" and "income from capital" is part of our legal system, and Piketty rests a lot of his work on it. But It seems to me an analytical mistake to describe the high compensation of a "super-manager" as income from labor. While it isn't coming from owning their corporation, it is coming from (partially) controlling it. In some ways, it's more like the income of an ancien regime tax farmer, or an Ottoman timariot, than the income of a roofer, nurse's aid, computer programmer, or even an architect. (Actually, the analogy with the timariot grows on me the more I think about it. The timariot didn't own his timar, he couldn't sell it or bequeath it, any more than a super-manager owns his company. Officially, income in both cases is compensation for services rendered to the actual owner, whether sultan or stock-holder.) It would be very nice to see someone try to separate income from labor and income from control, but I have no clue how to do it, statistically. (Though I do have a modest proposal for how to reduce the control income of super-managers.)
8. p. 654, n. 56: For "Claude Debreu", read "Gerard Debreu". (Speaking of economists' "passion for mathematics and for purely theoretical ... speculation"!)
ETA: Let me emphasize the point about production functions, marginal returns on capital, etc. It cannot be emphasized enough that capital, for Piketty, is the same as wealth, assets, stores of value which can be traded in the market. He does not mean non-human factors of production, "capital goods". (Cf.) Capital goods can work fine as assets, but a much more typical asset is a claim on part of the product achieved through putting capital goods and labor to use. Because he is looking at wealth rather than capital goods, the appropriate unit of measurement, and the one he uses, is monetary rather than physical. One consequence is that Piketty can legitimately add up monetary amounts to get the total wealth of a person, a class, or a country. (Whereas adding up capital goods is deeply problematic at best; I don't think even the dullest Gosplan functionary would've tried to get the total capital of the USSR by adding up the weight or volume of its capital goods.)
This also has implications for the "marginal product of capital" question. If a capital good is measured in physical units, it's not crazy to imagine diminishing marginal returns. If some line of work needs tools, equipment, a proper space, etc., to be carried out, then the first crude tools and the shack which allow it to get started increase output immensely, then having a bit more equipment and a decent space helps, and after a certain point one extra spanner or crucible or square foot of floor-space, with no extra worker, does very little. (Not crazy, but also not obviously true: see the work of Richard A. Miller [i, ii], which I learned of from Seth Ackerman's piece on Piketty.) Some critics of Piketty's forecasts point to this, to argue that his vision of widening inequality will fail on these grounds. They equate the rate of rate on capital, Piketty's $r$, with the marginal product of capital, and, believing the latter must decline, think $r$ must shrink as well. We thus have the curious spectacle of apostles of capitalism claiming it will be saved by a falling rate of profit. (I believe Uncle Karl would have savored the irony.) This intuition, however, is based on physical units of capital --- spanners, crucibles, servers, square meters of buildings. What about in monetary units?
Well, what price would you, as a sensible capitalist, pay for a marginal increase in your supply of some capital good? Its value to you is the present value of the increased future production that makes possible. (One buys a stock of capital and receives a flow of product.) A \$1 marginal increase in the capital stock has to produce at least \$1 of present value in extra production. If it augmented the present value of production by more than \$1, you'd be happy to buy it, but the price of that same physical capital good would then presumably be bid up by others. (Or, alternately, if not bid up, you would then buy another \$1 worth of capital, until the diminishing returns of physical capital set in.) At equilibrium, a marginal extra dollar of capital should always, for every enterprise, increase the present value of production by \$1. Under the simplest assumption that the extra product is constant over time, this means a marginal \$1 of capital should increase production in each time period by \\$ $\delta$, the discount rate. (Again, we're using monetary and not physical units for output. Also, I neglect small complications from depreciation and the like.) In symbols, $r = \partial Y/\partial K = \delta$. ($K$ has units of money, and $Y$ of money per unit time, so the partial derivative has units of inverse time, as $\delta$ should.) It is surely not obvious that the discount rate should fall as capital accumulates.
Expressed in other terms, the elasticity of substitution between capital and labor thus ends up being the elasticity of the marginal product of labor ($\partial Y/\partial L$) with respect to the ratio of capital to labor ($K/L$). Again, this may or may not fall as $K$ increases, but I don't see how diminishing returns to physical capital guarantees this.
However, the fact that the measured real rate of return on capital (which Piketty puts at roughly 5% over all periods and countries) is so much higher than any plausible discount rate suggests that the whole enterprise of trying to relate returns on capital to marginal products is ill-conceived. Indeed, Piketty rightly says as much, and his claim that $r > g$ is just an empirical regularity, true for most but not all of his data. So it's clearly not immutable, and indeed his policy proposal of a progressive tax on capital is designed to change it!
Manual trackack: A Fine Theorem (on aggregate production functions).
Charles Stross, The Rhesus Chart
Mind candy. Of course this is what would happen if some City quants happened to find themselves turning into vampires... (Before; after)
Susan A. Ambrose, Michael W. Bridges, Michele DiPietro, Marsha C. Lovett and Marie K. Norman, How Learning Works: Seven Research-Based Principles for Smart Teaching
An excellent guide to what psychological research has to say about making college-level teaching more effective --- that is, helping our students understand what we want them to learn, retain it, and use it and make it their own. I'd already been following some of the recommendations, but I am going to consciously try to do more, especially when it comes to scaffolding and giving rapid, targeted feedback. Following through on everything here would be a pretty daunting amount of work...
Disclaimer: Four of the authors worked at CMU when the book was published, and one is the spouse of a friend.

Posted at July 31, 2014 23:59 | permanent link