Home Mail Articles Stats/current Supplements Subscriptions Links

The following article appeared in Left Business Observer #84, July 1998. It retains its copyright and may not be reprinted or redistributed in any form - print, electronic, facsimile, anything - without the permission of LBO.

There was a discussion of this piece on the lbo-talk mailing list. Click here to see the transcript.

Unemployment, pay, and race  by Heather Boushey
Click on the MS Word icon for the original version of this paper, which is longer and more technical. Format is Word 95/6.0.  Boushey paper

Over the past few years, non-traditional economic theory has been receiving empirical validation from an unlikely source: mainstream economists. For example, in their recent book about the effects of the minimum wage on employment, David Card and Alan Krueger showed that a rise in the minimum wage did not lead to decreasing employment and in some places employment actually rose. That contradicts the predictions of textbook neoclassical theory, which declares that any increase in wages should be accompanied by a fall in employment.

Similarly, those textbooks also assume that high wages go along with high unemployment; that's the standard explanation of high European jobless rates, for example. But two other esteemed mainstreamers, David Blanchflower and Andrew Oswald, have developed evidence showing that the textbooks have the relation exactly backwards: all other things being equal, unemployment depresses wages. This is no surprise to students of Marxian theory, where the unemployed serve as a "reserve army" of workers to keep the employed pliant, but it is a shocker to the orthodox.

Blanchflower and Oswald call their discovery the "wage curve." [Click the title to see info about their book, The Wage Curve, from Amazon.com. If you choose to order the book, LBO will get a small commission on the sale.] Through surveys of nearly four million people from 16 countries, they found that local unemployment rate affects pay level such that "A worker who is employed in an area of high unemployment earns less than an identical individual who works in a region with low joblessness." They estimate that a region with an unemployment rate one percentage point higher than another region will have wages that are ten percent lower (or, in the jargon of the trade, the elasticity of pay with respect to unemployment is ­ 0.10). A stylized version of the relationship is shown in the chart nearby.

In conventional theory, wages are supposed to adjust to balance the supply and demand for labor. Unemployment is a sign that something is wrong - that wages aren't falling fast enough or far enough to bring the demand for workers into line with the supply.

Theoretically, though, Blanchflower and Oswald try to explain their results within the bounds of orthodox theory. They don't look at all to nontraditional, heterodox models, those that begin from the premise that unemployment is endemic to the capitalist economy. Unemployment serves to discipline labor. For example, the work of Samuel Bowles and Juliet Schor emphasizes the influence of the "cost of job loss" on wages - how much of a drop in pay a worker would experience if she lost her job and had to look for a new one. The higher the prevailing unemployment rate, the harder it is to find a new job, which raises the cost of job loss. (So does a cut in unemployment insurance benefits.) The higher the cost of job loss, the less likely workers are to strike, or join unions, or to resist employer demands. The wage curve supports this theory.

Finer grain

Unemployment, however, does not occur in the abstract, but happens to actual workers. Historically, certain groups of workers - women, African-Americans, and other "minorities" - have had a rougher labor market experience than others. Mainstreamers are often reticent about calling this by its proper name - discrimination - but that's what it is. For most of the postwar era, female unemployment was greater than male unemployment and African-American unemployment was greater than white unemployment. However, in the early 1980s, women's unemployment began to dip below that of men and since that time women's unemployment has cycled just beneath male unemployment. African-American unemployment has not undergone such a transformation; it's been stuck at about twice the white rate for decades. Further, black unemployment usually rises more sharply than white in recessions.

Blanchflower and Oswald didn't address sex and ethnic differences in their analysis. Since discrimination is such an important feature of labor markets, it's worth asking just how these differences in unemployment might be related to differences in pay. Does their magic ­0.10 number hold for all groups of workers, or does the curve bend harder against the discriminated-against?

To answer that question, I looked at U.S. data for 1994, covering all workers between the ages of 18 and 64 in the fifty largest metropolitan areas in the U.S. Workers were coded by sex, race, marital status, union membership, part-time status, educational level, age, industry, occupation, and state. In general, the effect of unemployment on wage levels is much stronger for men than women, and for African-Americans more than whites. Specifically, the effect of unemployment on black wages is nearly twice the Blanchflower and Oswald estimate (that is, the elasticity is ­0.25 relative to black unemployment rates). So, a male African-American in a high unemployment region can expect pay 25% lower than an identical one in a region with a black jobless rate 1 percentage point lower. Elasticities for black women are ­0.22; for white men, ­0.20; and white women, ­0.13. Pay is much more sensitive to racial than to gender unemployment rates.

So, African-Americans who live in large urban areas with high black unemployment experience a double whammy: not only are they more likely to be jobless, those who are working also suffer substantially lower pay caused by that high unemployment rate.



What about regional pay levels as a whole? Is there some central unemployment rate that sets the tone for the whole region? It seems that white men play this role: pay levels are far more sensitive to unemployment rates of whites and men (that is, the elasticities are greater) than to those for blacks and women. In other words, the unemployment rate of the groups that are hypothesized to be in the reserve army of labor does not have a strong effect on the earnings of the aggregate population.

These findings confirm suspicions that labor markets remain highly segmented along lines of sex and race - that they operate in part as worlds of their own. Increases in unemployment for discriminated-against workers lowers all earnings, but to a lesser extent than the unemployment of non-discriminated against groups. Thus, in 1994, with all else being equal, individuals who lived in large urban areas with relatively high unemployment rates for whites or males experienced lower pay than individuals living in large urban areas with a relatively high unemployment rate for African-Americans or females.
These findings support the argument that it is in the interest of whites and men to maintain their privilege because it sustains their higher earnings. In bad times, when even whites or males lose their jobs, all groups suffer in terms of pay, but when African-Americans or women lose their jobs, there's not much fallout for other groups. These findings suggest why there is so little discussion in the media or in academia of the excruciatingly high unemployment rates for African-Americans in urban areas. Persistently high black unemployment rates don't really affect the pay or employment of white workers. No wonder they don't seem to care.

Heather Boushey is an economist in New York ; she works for the New York City Housing Authority.

There was a discussion of this piece on the lbo-talk mailing list. Click here to see the transcript.

Home Mail Articles Stats/current Supplements Subscriptions Links