The Dismal Science

Talking About a Revolution

How the late Kelvin Lancaster helped transform economics.

Kelvin Lancaster died last month. He was an economist’s economist, famous within the profession (when an economist uses the adjective “Lancastrian,” he isn’t talking about the Wars of the Roses), but largely unknown beyond it. I can’t claim to have known him personally: I was 3 years old when his famous paper on the theory of the “second best” was published (click here to read more about it), and we probably met only four or five times. But nonetheless there was a time, a couple of decades ago, when we were comrades-in-arms–when he and I, along with several dozen other people, helped make a revolution in economic theory.

That revolution went unnoticed among the wider public, even among those who follow (or think that they follow) economic thinking. David Warsh of the Boston Globe wrote an excellent series of articles about some of the revolutionaries, and he’s in the process of finishing a book-length treatment. But other journalists ignored the story, or if they did write anything got it wrong, preferring camera-ready fantasies in which heroic outsiders challenged an obtuse Establishment. I’ve tried elsewhere to correct some of the myths (click here to read my article on the power of biobabble and here to read about the legend of Brian Arthur), but maybe Lancaster’s death is a good occasion for me to wax nostalgic, to recall what actually did happen.

To understand the revolution, you need to grasp two related dichotomies. One is that between constant and increasing returns; the other between perfect and imperfect competition. Constant returns is the assumption that if you increase your inputs, your output will grow by the same amount–e.g., if you double your inputs you will also double your output. Increasing returns, on the other hand, says that doubling inputs will more than double output. Perfect competition is the assumption that producers are like wheat farmers, who take the price of wheat as a given–and not like, say, Apple, which must decide what to charge for an iMac and can choose within limits to raise that price if it is willing to accept a reduction in sales.

Perfect competition and constant returns go together like cookies and milk; without constant returns, the assumption of perfect competition becomes very hard to swallow. The reason, basically, is that when there are increasing returns an industry will tend to become dominated by at most a few large players, and these players are bound to realize that they have some price-setting power. They are also likely to realize both that it is in their common interest to agree, at least tacitly, to set prices high, and that it is in their individual interest to cheat on that agreement and undercut their rivals. Is the eventual result a stable cartel, a perpetual price war, or an irregular alternation between the two? Hard to say. But what has long been clear to economists is that increasing returns normally lead to imperfect competition, and that imperfect competition can be a messy and intractable subject.

That recognition, in turn, led the profession to spend about a century and a half–from David Ricardo until the 1970s–concentrating its theoretical energy on models that assumed constant returns and perfect competition, and economists tended to avoid questions where increasing returns or imperfect competition were self-evidently crucial. In so doing they were neither foolish nor dogmatic: Most economists, I think, understood that increasing returns are sometimes important, and a few people did try to take them into account. (In my specialty, international trade theory, increasing-returns analysis is usually dated from a 1925 paper by the Princeton economist Frank Graham; the first fully worked-out mathematical model was published by R.C.O. Matthews in 1950.) But useful theorizing in complex subjects such as economics is always a matter of choosing the right strategic simplification, and for a long time it seemed that the clarity of constant-returns/perfect-competition analysis justified its violence to reality. Even now, one can say–and I did, back in 1995–that 95 percent of the time, it would be a blessing if politicians could understand what’s right about the constant returns model, not what’s wrong with it.

By the 1970s, however, patience with constant-returns economics was wearing thin. Exactly why is hard to say. I don’t think you can claim that returns were less constant or competition less perfect in the real world of 1975 than they had been in 1955, or even 1925. More likely, the driving force was the field’s internal intellectual logic: Economists had answered most of the interesting questions they could ask in the old framework and found that constant-returns economics was running into, well, diminishing returns. And so they were finally ready to try something different.

Kelvin Lancaster was one of those who was driven to increasing returns. In the 1960s he had introduced a seemingly obvious but highly useful twist to the analysis of consumer behavior by pointing out that what consumers often want is not so much a specific product as a particular bundle of characteristics. To take a modern example, what business travelers care about in their notebook computers are low weight, long battery life, and high computing power, rather than the logo on the case. There are trade-offs among these good things; what differentiates one notebook from another is where in this “characteristics space” they are located. But in that case, why doesn’t the market produce every possible notebook? (Much as I love my Hewlett-Packard Jornada handheld, I’d prefer a machine with a slightly better word processor, for which I would happily sacrifice something else.) The answer, of course, is increasing returns: To proliferate varieties (and hence to produce each variety at a smaller scale) means to increase costs.

N ow at this point Lancaster found himself up against the usual problem: Increasing returns mean imperfect competition, and in general imperfect competition is nasty stuff. But somehow, circa 1974, economists went through a shift in mindset. My colleague Robert Solow likes to say that there are two kinds of economists: those who look for general results and those who look for illuminating examples. And more or less suddenly fell into the second group; they decided that while a general theory of how imperfect competition must work was never going to happen, it was OK to focus on interesting examples of how it might work. How does the market for an industry with Lancaster-type differentiated products function? It could be dominated by a single firm that proliferates products to deter potential competitors–OK, Dick Schmalensee wrote up that story. Or it could be “monopolistically competitive,” each variety produced by a different firm–OK, Steve Salop wrote that up in one version, Mike Spence in another, Avinash Dixit and Joe Stiglitz in yet another. The point was to find stories that hung together, not determine once and for all which was right.

It’s hard to convey, if you weren’t there, just how liberating this was. Once they decided it was OK to tell illustrative stories rather than produce theorems, economists could write about exciting topics that had been off limits: predatory pricing, strategic investment to get the jump on competition, technological races, struggles to define industry standards. By 1988, when Jean Tirole published his landmark textbook The Theory of Industrial Organization, just about every idea about the “new economy” that trendy writers proclaim as a radical departure from conventional economic thought was, well, already in the textbook.

A mong other things, someone was bound to notice that the interaction between increasing returns and product differentiation could help explain some puzzles about international trade–like why most trade is between seemingly similar countries. In the late ‘70s three people independently wrote up that insight: the Norwegian economist Victor Norman, Lancaster himself, and yours truly; and the “new trade theory” was born. A few years later economists such as Paul Romer and Philippe Aghion applied related ideas to technological change and economic growth, giving birth to the “new growth theory”; and the ripples spread ever outward.

Alas, golden ages do end. By the early 1990s, the thrill of increasing-returns economics was fading. It wasn’t just the inexorable working of the law of diminishing disciples. There was a deeper problem: The new ideas were immensely liberating, but at some point you can get too liberated. In international trade, people started to joke that a smart graduate student could come up with a model to justify any policy; similar sentiments were felt in many fields. In short, we all got tired of clever analyses of what might happen; and throughout economics there was a shift in focus away from theorizing, toward data collection and careful statistical analysis.

But it was a golden age–a time of innovation and intellectual excitement, when all of economics seemed up for reinvention–and Kelvin Lancaster was one of those who made it so. Let us honor his memory.