Book review: Levy and Murnane, The New Division of Labor (2004)

I wrote a review of Frank Levy and Richard Murnane's new book _The New Division of Labor: How Computers Are Creating the Next Job Market_ (2004) for the _International Review of Social History_ and thought I'd post an excerpt below, since I haven't had time to blog on much else this week.

Economists Frank Levy and Richard Murnane have written an engaging and accessible introduction to the political economy of a very specific but very important type of ‘‘information labor’’: that subset of work which is amenable to ‘‘computerization,’’ which in some cases means outright substitution of computer algorithms for human labor (a classic ‘‘deskilling’’ argument), and in other cases means careful augmentation of human labor through interactive software (a classic ‘‘upskilling’’ argument). The main point that the authors make is that these two simultaneous paths to what might be called the ‘‘digitalization of labor’’ are quite distinct, in both the kinds of tasks they encompass and the kinds of workers they affect. As computers colonize more and more industries and occupations, Levy and Murnane present a detailed analysis of what these electronic tools can and can’t do to predict that certain workers will continue to benefit while others will increasingly suffer in a ‘‘hollowing-out of the occupational structure’’ (p. 4) – a nuanced ‘‘digital-divide’’ scenario which can only be addressed, the authors conclude, through state intervention and educational reform.

Levy and Murnane begin by noting that, although ‘‘all human work involves the cognitive processing of information’’ (p. 5) there are many different kinds of information processing, only some of which are easily and affordably coded as computer algorithms. For example, the pattern-recognition (and consequent tactile dexterity) performed by even the most low-wage service workers remains uncomputable – don’t expect to see robot janitors any time soon. Similarly, complex communication tasks, such as those used by middle-income salespersons and educators, remain out of the computer’s reach. And finally, tasks that require novel and open-ended problem-solving, often called ‘‘symbolic analysis,’’ are restricted to human creativity (though computers are often used as productive tools by such high-wage workers). But any task which may be broken down into a discrete and finite set of steps and ‘‘rules’’ is potentially computable, and thus jobs which consist in whole or in part of such tasks will be increasingly endangered as the capital cost of computing power continues to fall. And crucially, ‘‘A task, once computerized, is potentially easy to replicate and so invites intense competition’’ (p. 54) with such information technology penetrating quickly through whole industries and occupations.

Levy and Murnane then move from a consideration of what kind of tasks favor computer substitution vs computer complementarity to what kind of workers will see their jobs eliminated by computers vs enhanced by computers. Not surprisingly, education is the key intervening variable. ‘‘Rapid job change raises the value of verbal and quantitative literacy’’ (p. 101), the authors argue, because reading and mathematics skills are ‘‘enabling skills’’: skills that are ‘‘necessary but not sufficient for economic success’’ (p. 103), especially in an increasingly information-based economy. Thus labor-market entrants who have had the opportunity to hone these enabling skills (e.g. college graduates) should fare much better than labor-market entrants without such skills (secondary-school dropouts or, sadly, even many secondary-school graduates, according to the authors).

Levy and Murnane back up these claims using historical labor market data from the United States. ‘‘In 1979, the average thirty-year-old man with a bachelor’s degree earned just 17 per cent more than a thirty-year-old man with a high school diploma. Today, the equivalent college–high-school wage gap exceeds 50 per cent, and the gap for women is larger’’ (p. 6). Similarly, they point out, while only 24 per cent of US workers used a computer on the job in 1984, now over 50 per cent of US workers do so (p. 105). These parallels represent a causal link, argue the authors – though they leave many of the details out of this book, instead referring readers to a 2003 paper in the Quarterly Journal of Economics, written with David Autor, which details the quantitative data and formulae that ground these assertions.

Given this increasing divide in ‘‘enabling skills,’’ wages, and occupational choices, what is to be done? Limiting their prescriptions to the US context, Levy and Murnane do not shy away from the obvious policy questions here, but instead assert that ‘‘the nation cannot rely on for-profit firms as the primary institutions responsible for teaching the enabling skills needed to excel at complex communications and expert thinking tasks’’. Instead, ‘‘America’s schools will continue to be the critical institutions responsible for teaching American children the enabling skills’’ (p. 130). While Levy and Murnane in general recommend a social policy where ‘‘the better-off pay compensation through taxes or charity’’ (‘‘[c]ompensation will not come through the market since the market is creating the winners and losers in the first place’’ (p. 155)), their most specific proposal revolves around a vision of ‘‘standards-based education’’ – setting clear goals for student progress, standardizing instruction to meet these goals, and measuring student progress toward these goals ‘‘frequently’’ enough to make sure they are attained (pp. 134–135).

The rest of the review critiques their final recommendations a bit, but I'll save that for folks who want to search out the original. It's a good book.