Letter from America - the NIH; the Supremes and the Economists

In his latest letter, Angus Deaton shows how economists, perhaps inadvertently, have helped to limit the levels of inequality in some aspects of US life.

It is hard to think of the American economics profession as a defender of equality. Economists, like most of the public, are deeply divided on whether or not current levels of inequality are a problem, and along predictable lines.

Yet here are two areas where I believe that economists, not through any political commitment, but through their regular activities, have been helping keep inequality under control, and where, as is often the case, the realisation comes only when they stop doing so, or when the forces on the other side win a victory.

Economics helps restrain health inequality…
The first case comes from the National Institutes of Health, in Washington. As readers of this Letter will know, the NIH has over the years become a major funder of economics-and other social science-both through research grants, and through the provision of large scale data sets, including the Health and Retirement Survey, the SHARE in Europe, and other clones around the world. The number of health-related titles in the NBER working paper series is one indication of the importance of funding for health economics. There are two NBER programs on health, as well as programs on children and on aging, both of which have large health components. Economists now regularly publish in the top medical journals, sometimes jointly with doctors, and sometimes on their own. A recent well-publicised case, published in the New England Journal of Medicine is an analysis of the effects of providing public health insurance to a random sample of low-income Oregonians on an oversubscribed waiting list.

The Republicans in the House of Representatives are determined that this funding should stop. Eric Cantor, the House Majority Leader, argued that, ‘funds currently spent by the government on social science — including on politics of all things — would be better spent on trying to find the cures for diseases.’ (It is hard not to admire a politician who can use the public’s detestation of politicians to enlist public support.) In the short run, the target seems to be any research that might support Obamacare. In the long term, the deeper force is likely the richly funded lobbying machine of pharmaceutical firms and medical equipment manufacturers who see cost effectiveness analysis as potentially leading to cost control, the ultimate enemy. Economists, who might discover that public health insurance is good for people, or who might argue that some treatments enrich suppliers without helping patients, are decidedly unwelcome at the party, particularly when financed by public money.

What will happen to NIH funding for social and behavioral research, including economics, is currently unresolved. Economists and other social scientists have organized in opposition, but it was difficult to get politicians to support the effort, though in the end eighty-three (Democratic) congressmen, led by Representative Lucille Roybal-Allard from California, signed a letter to Francis Collins, the Director of the Institutes, making the case for health economics and urging continuing funding. Of course, NIH already faces budgetary difficulties because of the automatic cuts from the ‘sequester’ and the social science budget may be relatively weakly defended inside the Institutes.

…while econometrics battles gender inequality
My second example will be familiar to anyone who has taught-or even taken-a basic econometrics class. One of the workhorses of teaching regression analysis is the ‘discrimination’ regression. Data on wages show that women in the Plastic and Elastic Shoe Company earn less than men. P & E admit the fact, but argue that they do not discriminate, and the difference in means comes from the fact that the men and women that they employ are different, that the men have more experience and are better educated. So the regression is run again, now including, not just a constant and a gender dummy —which showed the difference — but also including years of education and years of experience in the firm. If the gender dummy is now insignificant, P & E is innocent: if it remains negative and significant, P & E is guilty.
Class-action suits by women, or minorities, are regularly brought against firms, and economists make a modest consulting income arguing the case for the plaintiffs (or somewhat less modest if they work for the defendants.) The argument often focuses on what variables can legitimately be included in the regression. For example, if the gender effect can only be eliminated by adding a dummy for attendance at an all-female college, or a measure of hair length, or heavily disguised versions thereof, then P & E should still be guilty. More contentious is whether to control for job assignment, clearly illegitimate if women are systematically denied access to better-paying jobs for which they are qualified.

These procedures are currently under threat from a Supreme Court decision on a discrimination class action suit by 1.5 million women against Wal-Mart, the largest employer in the US, alleging systematic discrimination in wages, training, and promotion. The case, Wal-Mart v Dukes et al, was settled in June 2011 when the Supreme Court unanimously decided that the class-action suit could not go ahead. The grounds are technical, but the problem appears to have been that the suit was too large, and that the women who were party to the suit did not have a common complaint that, if proven, ‘will resolve an issue that is central to the validity of each one of the claims in one stroke.’ The court demanded significant proof that Wal-Mart operated under a general policy of discrimination, and dismissed ‘a sociologist’s analysis asserting that Wal-Mart’s corporate culture made it vulnerable to gender bias’ as ‘worlds away from significant proof,’ particularly given that Wal-Mart’s own announced policy forbids sex discrimination. If some individual managers discriminated, but others did not, then there was no valid class-action suit against Wal-Mart as a whole.

Reports from the courts suggest that Wal-Mart v Dukes is undermining the operation of the regression conviction in the courts. Even if the gender dummy remains negative, that is no longer enough to convict P & E; instead, there must be direct evidence of a discriminatory policy by the company, for example a company document, e-mails, or instructions to managers. This much more severe standard, makes it much harder for those who face discrimination to obtain redress.

That the current Supreme Court — with its conservative majority — should be an engine of inequality is no surprise. Nor is it a surprise that the bloated healthcare sector should seek to bloat itself further through its influence on politicians. Yet who would have thought that the economists would be on the other side?

Page Options