The rise of inequality in rich countries is way over-explained. Because income inequality (evaluated at the level of households or individuals) is such a complex variable, outcome of a vast number of technological, political, demographic and behavioral factors and its neat decomposition into these various factors is impossible, we shall always have a plethora of potential explanans. This was a point nicety raised in a recent post by Chico Ferreira from the World Bank.
But not all explanations are equally powerful or make sense. A couple of days ago at a conference at Northwestern University, I listened to the explanation proposed by Gerald Davis from University of Michigan. Davis argued, at first very counter-intuitively (so much so that at first I thought I misunderstood his point), that the rise of US inequality coincides with the decline of large companies that used to employ hundreds of thousands or even millions of workers and by their substitution by much smaller companies. As shown in a graph from Davis and Cobb (“Corporations and economic inequality around the world”, 2010) the share of large employers in total US employment went down simultaneously with the increase in US income inequality. (What Davis and Cobb call the “ratio of top 10 employers to labor force” is, I suppose, the percentage share of all US workers employed by ten largest companies.)
Why did it seem counter-intuitive? Because there is an earlier influential literature going back to Herbert Simon and Thomas Mayer that saw the behemoths of the 1960s with their heavy hierarchical structures as precisely driving inequality up. The more ladders you have in a hierarchical structure, the more pay grades there are, and the more likely are the top managers who lord it over thousands of employees to have high salaries. Hence greater inequality. At least that was the theory.
But Davis argues the opposite (and was helped by the data that go this way). His argument is that large hierarchical structures have to engage in some evening out of salaries (internal redistribution) in order to keep cooperation, needed for the success of the enterprise, going. On the contrary, if the big bureaucratic machines get divested, and jobs hitherto performed within company get outsourced to different contractors, there is nothing to keep the new small, lean and mean company from extracting all the surplus from each and every contractor the way that Amazon is credited (or “credited”) of doing it. So take a General Motors and break it into thousands of independent companies producing components, then outsource cleaning, marketing, food catering and legal services, offshore accounting and customer relations, and you end up with today’s enterprise structure. Those who have remained at the core can pay themselves huge salaries since they do not depend on the goodwill and cooperation of the outsourcing companies. If the accountants in Chicago feel they are paid too little, GM will gladly hire accountants in Calcutta.
This is basically, I think, Davis’s idea (and I hope I presented it accurately). But my explanation for this change went, I thought, one step further. I think that the deconcentration of which Davis writes is only a proximate cause of the rise in inequality. The “deep cause” was technological change, combined in an inextricable way (as I argue in my “Global inequality”) with globalization. What happened, I think, was that advances in technology such as stock management (just-in-time), ability of speedy “bespoke” production, and crucially advances in telecommunications made “broken up” (devolved) production more efficient. The concentration of workers in one place that, at the origin of the Industrial Revolution, was made indispensable by the importance of the energy sources available at discrete physical points, could now be reversed. We could go back to a type of production that predates the Industrial Revolution, a kind of a modernized “putting out” system.
In doing so companies were helped by globalization because the area over which the putting out system can now extend was the globe itself, not only villages 10 miles in perimeter from Manchester. Only then, all the elements mentioned by Davis, could follow. The companies did not need to ensure the collaboration of employees by dint of redistribution of some profits across the labor force. The broken-down units were too small to allow for any meaningful unionization, and the trade union density declined. Further, there are no labor conditions to negotiate with a company to which you sell your goods or services. At the extreme, imagine a structure where the core company divests practically all tasks to companies composed of one individual each. Inequality within each company will be zero while total inequality could be quite high (because between-company differences in average wages will be high).
My view is indeed one that may be called “technological determinism”, but that determinism is playing itself out on an ever expanding field made possible by globalization. In other words, technological determinism is itself a function of globalization (heaving off some activities clearly could not be equally efficient in a world where you had to hire US workers only) and technological determinism in turns helps globalization. So the two go together. Moreover even things that seem to be policy-related (say, decline of trade unions) may in many instances be driven by the combination of new technology and globalization.
It is for that reason that I am skeptical that “the happy days of the 1960s” will ever come back. This is the idea that I somewhat jokingly called “Make America Denmark Again”. That world made sense with the technology and the policy as it was in the 1960s but not today. The world of large-scale manufacturing, homogeneous working class, trade unions, capital controls and a quasi-closed economy (US exports and imports combined were less than 10 percent of American GDP in the 1960s; they are more than 30 percent today) is over. When we think of how to address inequality today we should move from the ideas that worked half-a-century ago.