The Wage Curve. - book reviews

"THE LABOR MARKET IS UNLIKE ANY OTHER"

Monthly Review normally publishes its book reviews in the back of the bus, so to speak, in rather small type. However, the following review by Michael Yates of three books on the determination and distribution of wages in capitalist economies seems to the editors to merit a first-class seat.

I

The neoclassical economists take great pride in debunking "common sense" or "popular opinion." This is especially true for the orthodox analysts of the labor market. Most people in the United States think that the minimum wage should be higher. But the economists tell us that raising the minimum wage will lead to considerable losses of employment, hurting the very group of workers that the minimum wage is supposed to help. Most people believe that labor unions win higher wages for working people. But the neoclassical economists tell us that higher wages for union members can be won only at the expense of unemployment among union members and lower wages for nonunion workers. The woman in the street can see that there doesn't seem to be any rhyme or reason to the distribution of wages. People who appear to have the same abilities often receive widely different wages. But the orthodox economist tells us that this is an illusion. Other things equal, equally productive workers will earn the same wage rate. The workers in my labor education classes readily embrace the notion that a large reserve of unemployed workers puts downward pressure on wage rates. But the purveyors of the dismal science argue that the opposite must be true. A region with a high unemployment rate will not be a good place to live, so employers will have to pay relatively higher wages to attract workers to it.

Unfortunately for the neoclassical analysis, recent studies using exceptionally detailed data and sophisticated statistical techniques have called each of the above predictions into question. The first two of the three books under review demonstrate conclusively that 1) a higher minimum wage will not decrease employment and may even increase it, 2) the labor markets do not generate equal wage rates for equally productive workers, and 3) high unemployment rates are associated with low wage rates. The first book also implies that a union may win higher wage rates with either no employment loss or an employment gain.

Of course, the fact that the predictions of a theory are not borne out by the evidence does not by itself destroy theory as such. What is required is a more powerful theory, one whose logic gives rise to predictions consistent with the evidence. The third book under review offers just such a theory, a modern version of classical Marxian economics.

II

Myth and measurement, by Card and Krueger, has created a minor sensation among economists. This is because its conclusion counters one of the strongest predictions of neoclassical economists, namely that a higher wage rate must result in lower employment. If we assume that business firms seek to maximize profits, then they will hire workers only if they produce output more valuable than their wages. Some workers will produce output worth only a little more than their wage rate; therefore, if the wage rate rises above the worth of their output, they will have to be discharged. If in the short run, it is not possible to do this, then in the long run, firms will seek to substitute capital for the now more expensive labor. Many previous empirical studies appeared to confirm this prediction, with the consensus view being that a I 0 percent increase in the minimum wage would lead to about a 3 percent loss of employment.

Card and Krueger lay waste to the neoclassical prediction that a higher minimum wage must inevitably result in less employment. First, they conducted a number of studies which used a close approximation to the experimental method used by scientists. In this method, one group of persons is designated the "treatment" group and one group of persons is designated the "control" group. The treatment group is, for example, given an experimental drug, while the control group is not (its members may be given a placebo). Statistical tests are then applied to account for differences between the groups other than the use of the drug (age, for example). Differences in outcomes for the two groups can then be attributed to the effects of the drug.

Such experimental design is not usually possible in economics. However, Card and Krueger take advantage of what they call "natural" experiments, situations in which an independent variable, such as the minimum wage, changes for one group of persons but not for another. Their prototypical study examined the effect on employment in fast-food restaurants of the April 1992 increase in New Jersey's minimum wage. This increase put New Jersey's minimum wage at $5.05, while the minimum wage in neighboring Pennsylvania remained at the federal minimum of $4.25. The fast-food restaurants in New jersey were then the treatment group, while those in Pennsylvania were the control group. Card and Krueger discovered that, after controlling for an exhaustive list of variables besides the different minimum wage which might have accounted for differences in employment between the two state's fast-food restaurants, the hike in New Jersey's minimum wage rate actually led to a slight increase in employment in New Jersey's fast-food restaurants.

The New Jersey study's method was also applied to Texas fast-food restaurants with similar results. Another study covering all minimum wage workers in California again failed to yield a negative relationship between the minimum wage and employment. So, too did a study of the employment effect of the last increase in the federal minimum wage. In this study the natural experiment method could still be used because the fifty states differ with respect to the fraction of the labor force which is affected by the increase in the minimum wage. We would expect high-impact states, those with a high ratio of minimum wage workers to the total labor force, to show a larger (and negative) employment effect. This was not the case.

The authors further buttress their findings by reexamining the major minimum wage studies previously done and by extending their studies beyond the United States. They argue convincingly that most of these had fatal weaknesses such as inappropriate or insufficient data, inadequate or inappropriate choice of independent variables, and failure to correct for statistical problems. Once these weaknesses are corrected, the negative wage-employment relationship either disappears or becomes much weaker. This is also true for studies done of the effect of the minimum wage on employment in Puerto Rico. The Puerto Rican results are especially significant because the minimum wage there is much higher relative to the average wage rate than is the case in the United States. This suggests that even a relatively large increase in the minimum wage might not reduce employment by much.

An increase in the minimum wage might have other than employment effects, and Card and Krueger do not neglect them. Some of their studies (but not all) show that an increase in the minimum wage results in a small increase in prices. Higher minimum wages have reduced wage inequality and poverty, although the effect on poverty has been small due to the modest increases which have actually been made in the minimum wage. In the fast-food studies, they found no evidence that increases in minimum wages caused any increase in the rate of restaurant closings. Similarly there is no strong evidence that higher minimum wages reduce the profitability of firms or lower their stock prices. Finally, they discovered that there is a spike in the wage distribution right at the minimum wage, that is, a disproportionate number of workers earn exactly the minimum wage. When the minimum wage increases, the spike moves to the new and higher minimum wage. This fact makes questionable the neoclassical assertion that workers are paid according to their productivity. If this were so, we would expect that many of the workers paid a wage between the old minimum wage and the new one would be discharged because of their low productivity. Instead, employers simply pay them the new minimum wage. Indeed, workers who previously made just above the new minimum wage are also likely to receive higher wages as a result of the higher minimum wage.

In their reexamination of prior studies, the authors offer evidence of two practices which should make social scientists skeptical of all mainstream empirical work. These are "specification searching" and "publication bias." The first occurs when the researcher manipulates a model until it yields results consistent with the expectations of the neoclassical model. The second occurs when a professional journal disproportionately accepts for publication articles which contain such results. Taken together, these practices have yielded studies which are biased in favor of the result that an increase in the minimum wage leads to reduced employment.

Considering how thorough and professional is Card's and Krueger's empirical work, it is remarkable that their theoretical speculations are rather shallow. Since the standard predictions of neoclassical economics are based upon a model which assumes that markets are perfectly competitive, that is, no market participant can influence the price of any input or output, they suggest that their findings must be due to a lack of competitive conditions. Employers must have some control over the wages that they pay. If this idea is coupled with the notion that, contrary to the neoclassical belief, wages can influence productivity, then it is possible that a wage increase could lead to higher employment. Higher wages could make workers more productive, so a firm's costs per unit of output could remain the same or even fall. Another possibility is that higher wages make it easier for businesses to recruit workers, lowering hiring costs. Or, higher wages may reduce turnover and make it less necessary for employers to monitor employees. If any of these things happen, the negative employment effect of a minimum wage increase will be reduced or reversed.

There are two problems with this line of reasoning. First, the premises of many of the noncompetitive theories are dubious, a point developed thoroughly in the Botwinick book. For example, it is well-known that, other things equal, plants with large work forces pay higher wages than those with small ones. It has been argued that in work places with many employees, it is costly to monitor workers and prevent them from shirking. Higher wages presumably reduce the need to monitor employees because workers are less likely to shirk when they are paid higher wages. That is, the cost of shirking rises as the wage rate increases. Why it is that monitoring is more difficult in larger plants is left unexplained, and with good reason, since in many large plants, the pace of work is driven by machinery, making shirking more rather than less difficult.

Second, as Botwinick emphasizes, blaming perverse outcomes upon noncompetitive markets leaves perfectly competitive markets as the ideal against which all other economic arrangements are measured. This might not be so bad if perfectly competitive markets were used by theorists as a first approximation of capitalist reality. However, this is not the case. Perfect competition is posited as a (reactionary) ethical ideal toward which society ought to strive. So long as this is true, it will never be possible to penetrate the true nature of capitalism. For example, neoclassical economists only consider profits to be a surplus in noncompetitive markets. In competitive markets there are no profits at all, only a return to the risk-taking or the willingness to forego consumption of the capitalists. Therefore, there might be a justification for a higher minimum wage or a union presence in noncompetitive markets but not in the ideal markets of perfect competition. Liberal economists like Card and Krueger might argue that the facts justify a higher minimum wage, but the true believer will still argue that this would not be the case if markets were competitive. And Card and Krueger cannot respond that profits are a surplus in all markets, the result of the exploitation of wage labor, because, like their neoclassical colleagues, they do not think that this is true.

III

The Wage Curve shares important similarities with Myth and Measurement. It relies on natural experiments, large samples from many parts of the United States and other countries, and sophisticated statistical techniques. The authors conclude that their results directly contradict the predictions of the standard neoclassical model. In the textbook case, higher wages must be associated with higher unemployment rates. In any nation as a whole, a wage rate above the "equilibrium" wage rate (above the cross point on a supply-demand diagram) generates a labor surplus, and the more above equilibrium it is, the greater is the surplus. If we consider unemployment within different regions of a nation, regions with higher unemployment rates must offer higher wages if they are to attract the requisite number of workers. This is because workers contemplating moving to a high-unemployment region will know that they face a greater risk of losing their jobs in such a region and will therefore demand higher wages as a condition for making the move. In neoclassical parlance, workers must be paid a "compensating wage differential" for the "disamenity" created by the region's relative lack of employment opportunities. This logic is the same as that offered by neoclassical economists to show that jobs with higher than average risk must be paid a higher wage rate.

Blanchflower and Oswald, after conducting several rigorous statistical investigations, reach exactly the opposite conclusion. In all of the countries that they studied (United States, Great Britain, Germany, Austria, Italy, Netherlands, Ireland, Switzerland, Norway, Canada, and South Korea), they found an inverse relationship between unemployment rates and the level of wages. In fact, this relationship appears to be amazingly stable. The data show that a doubling of the unemployment rate is associated with a ten percent decline in the level of wage rates, regardless of the country.

Like Card and Krueger, these authors develop noncompetitive models which can yield predictions matching the empirical results. While they do not offer specific tests of these models, they do indicate what types of results each predicts. The one which, on the surface at least, gives the best match is a bargaining model. Employees, whether they are organized in unions or not, will try to get as much of the monopoly profits of their employers as they can. Their power to do so will depend upon the ability of their employers to replace them which in turn will depend upon the level of unemployment. This implies that wages should be lower in areas of higher unemployment. The bargaining model also predicts that given the level of unemployment, wages will be positively correlated to profit levels. That is, the more profits there are, the more of this surplus workers will be able to win. The authors' empirical findings support this prediction.

III

It struck me when I read the first two books that their conclusions fit very nicely into a Marxian framework. After all, one of Marx's most important insights was that the labor market was unique because labor power was no ordinary commodity. To examine the labor market properly, it was necessary to center the analysis on the relations of production in a context of accumulating capital. Needless to say, neither Card and Krueger nor Blanchflower and Oswald consider a Marxian economic model. Card and Krueger make no mention of Marx, and Blanchflower and Oswald simply note in passing that Marx in Capital also predicted a negative wage rate-unemployment relationship. Of course, this is not surprising. Established neoclassical economists do not become radical economists just because of troubling empirical findings.

Fortunately, a radical economist, a proponent of classical Marxian economics, has developed a comprehensive theory of labor markets which easily accommodates the discoveries of the first two books. Labor organizer turned economist, Howard Botwinick, has written a seminal book in labor economics. Drawing upon the theoretical work of his teacher, Anwar Shaikh of the New School for Social Research, Botwinick in Persistent Inequalities has built what has eluded radical economists, namely, a fully determinate model of labor markets.

As we have seen, reality in capitalist labor markets confronts neoclassical economics with many anomalies. These provided radical economists with an incentive to devise nonneoclassical theories which could explain them. Many radicals theorized that labor markets were of two types: primary and secondary. In the former, firms enjoyed monopoly power and this gave workers room to improve their wages and working conditions. Employers, too, had an incentive to pay higher wages, to avoid turnover and the accompanying hiring and training costs associated with the greater skill requirements of work in the primary sector. In the secondary market competition among employers made profits and wages lower. Skilled workers were not required, so turnover was almost costless for both employers and employees. Secondary market workers moved from job to job developing habits which made them unemployable in the primary sector. Unions would not be effective here because of the low profits, intense employer competition, and high employee turnover.

This theory helps to explain some of the neoclassical anomalies. For example, workers of equal potential productivity will earn different wages if they are in different markets. If we suppose that minority workers and women are concentrated in the secondary market, this theory can explain race and gender wage differences. In the primary market, higher wages may not reduce employment, because they may increase productivity. Wages, then, can be dependent upon the class power of workers, which, of course, will be greater in the primary market. A ready reserve of workers in the secondary market combined with lower profits reduce labor's power and keep wages low.

Unfortunately, a dual labor market (labor segmentation) theory cannot explain the success of unions in competitive markets such as trucking and longshoring. Nor can it tell us the upper limits of wage rates in the primary sector, since they depend upon the ambiguous concept of class power. It cannot tell us why there are many minority and female workers in the primary sector nor how some previously secondary markets such as steel and autos became primary labor markets.

Some prominent radical economists, notably Samuel Bowles, Herbert Gintis, and Michael Reich, have moved decisively toward the noncompetitive models espoused by Card and Krueger and Blanchflower and Oswald. These are built essentially upon the neoclassical assumption that social outcomes are the result of the self-interested choices of individual market participants. A fuller accounting of the factors which enter into these choices gives outcomes more in accord with radical sentiments: higher minimum wages and labor unions may not be as socially costly as the traditional neoclassical theory has implied. However, none of these formulations indicts capitalism itself, with whatever liberal reforms are appended to it, as incapable of the liberation of working people. Therefore, it is legitimate to ask whether these are radical theories at all.

Botwinick begins with the observation that the postulates of neoclassical economics are not assumptions but rather idealizations, ends toward which society must strive if it is to reap efficiency in its system of production. An objective theory must begin with assumptions which abstract from reality's complexity but which capture essential features of the reality to be examined. Following Marx, Botwinick uses the method of successive approximation. He begins with an economy in which there are no differences between individual capitals (firms) and then moves to one in which there are only interindustry capital differences and finally toward one in which there are both interindustry and intraindustry differences. This method allows him to examine all of the causes of wage inequality, from the most general to the most specific.

In the most abstract model, firms accumulate capital on the basis of the surplus labor time which they extract from their workers. Workers can win higher wages if they organize, but if their wages increase faster than the rate of accumulation, profits will fall. To prevent this, firms introduce the detailed division of labor and mechanization to create a reserve army of labor. Since different groups of workers will be unequally threatened by the reserve army of labor, because of skill or location for example, unequal wages may develop for workers otherwise equal. The same result can be seen if we look at accumulation from the perspective of the members of the reserve army. Otherwise equal potential workers will have different access to employment; for example, some will get part-time jobs and others full-time depending on their location or luck, and therefore they will have different wages. Unionized workers may foster wage inequalities by devising mechanisms which protect them from competition from the reserve army. At this level of abstraction it is important to understand that wage increases are possible, but they are limited by the rate of accumulation and technical progress which lowers the value of labor power. In other words, labor unions can force wages up without risking a loss of employment, but their ability to do so is not unlimited.

When we move to a lower level of abstraction, we must consider capitalist competition. In neoclassical theory, this means that capitals constantly move from low to high profit markets, and workers move from low to high wage markets. As a result, in the long run firms are predicted to have equal profit rates and workers of equal abilities equal wage rates. Botwinick shows clearly that Marxian economics makes no such predictions. Production in different industries naturally involves different technologies, that is, different levels of capital intensity. And technologies naturally develop at different paces in different industries. Within any given industry, firms will also have different technologies as new inventions are introduced by new firms and as different managers introduce different work organizations. These facts mean that, at any point in time, profit rates will vary between and within industries.

Furthermore, capital moves from market to market in response to the profit rates made by the most advanced capitals, what Botwinick calls "regulating capitals." These are the most efficient capitals in an industry, those using the best technology and organizational techniques. Within any industry, it is not possible for each firm to replicate the cost structure of the regulating capitals, because firms have fixed capital to depreciate and managers with different amounts of skill and luck. Therefore, within each industry, different profit rates will prevail; the regulating capitals will have the highest rate and the "subdominant capitals" various lower rates. Among different industries, unequal profit rates among the regulating capitals will elicit capital movements toward the industries with higher profit rates. There will thus be a tendency toward the equalization of profit rates among regulating capitals. This equalization is only a tendency, however, because capitalist competition continuously changes the techniques and organization of production.

Different profit rates within and between industries imply the possibility of different wage rates for workers with the same "productive" characteristics and willingness to work. Wage rates will depend upon the productiveness of the capital with which the workers labor, the strength of the threat which the workers face from the reserve army, and the degree to which workers organize to fight for higher wages. For example, a college-educated and highly motivated worker earning a high wage in a regulating firm in an above average profit industry will likely suffer a large wage cut if she is forced to relocate to a subdominant firm in a lower than average profit industry. She may even earn less than an unskilled, unionized worker in another regulating firm in a high-profit industry.

The beauty of Botwinick's analysis is that, while class struggle enters into wage determination, wage rates are not completely indeterminate. There are concrete limits to wage increases. Suppose that the economy is experiencing a period of capital accumulation. Workers situated in the firms with regulating capitals could, with organization, win higher wage rates up to the point at which the regulating capital's cost advantage disappears. As the regulating capital's costs rise, the entry of new capital into the industry will slow down, increasing the relative price of its products. Workers in subdominant capitals will not be in so advantageous a position; if they win higher wages, nothing will happen to selling prices, so their profit rates will fall. Workers may still win higher wages, depending upon how costly it is for their employers to resist their wage push, and this, in turn, will depend upon the firms' access to the reserve army of labor. For example, unionized workers in a declining domestic industry with large fixed capital costs, for example steel, may keep their wages well above average as long as they can prevent the hiring of replacement workers during strikes and as long as their employers find it more costly to move their capital than to pay the higher wages. Botwinick provides readers with clear numerical examples which illustrate the various limits to wage rate increases.

Botwinick's work has important implications for the labor movement. At any given time, there will be excellent organizing opportunities. Many of our service industries today are likely to be regulating capitals and, therefore, good targets for unionization. So are low-wage workers in highly capitalized industries. Unions need to pay attention to the cost structures and technologies of industries and firms and target those with the highest wage limits. Concessions should never be offered to regulating capitals, a fact realized by the Local P-9 workers at Hormel in Austin, Minnesota but not by the leaders of the national union. Today, many regulating capitals are in other countries, so it will be necessary to organize workers across borders and for each country's workers to show solidarity with those in every other nation. Finally, any threats to profit rates elicit division of labor, mechanization, and capital flight, all of which expand the reserve army. Independent political organization will be absolutely essential to force the governments of all capitalist countries to keep unemployment low and to provide the unemployed with adequate incomes.

IV

Card and Krueger's book has already generated considerable discussion, including predictable attacks by diehard neoclassicals. Radicals should use it to enhance their critique of orthodox economics and to teach working people. The Wage Curve does not deal with a subject as topical as the minimum wage and it is not nearly as accessible to noneconomists as is Myth and Measurement, so it has not had much impact either within the profession or in the larger society. However, its results are just as critical of neoclassical economics and just as deserving of widespread publicity by leftists. The merits of these books notwithstanding, the Botwinick book is the most important. It shows that Marxist economic theory can explain those phenomena of the labor market which have confounded orthodoxy, including the findings of both of the other books. None of the theoretical suggestions of Card and Krueger and Blanchflower and Oswald are nearly as plausible as the theory developed by Botwinick. Unfortunately, just when classical Marxism shows its amazing resilience, many former Marxists have rejected it and gone over to the enemy. This is a shame because it will make it more difficult for radicals to inform a resurgent labor movement with an adequate theoretical foundation for its program and the strategies to achieve it.

Michael D. Yates teaches economics at the University of Pittsburgh at Johnstown and is the author of Longer Hours, Fewer Jobs: Employment and Unemployment in the, Unit,d States (Monthly Review Press, 1994). The books included in this review are David Card and Alan B. Krueger, Myth and Measurement: The New Economics of the Minimum Wage, Princeton, N.J.: Princeton University Press, 1995. 402 pp. $29.95; David G. Blanchflower and Andrew J. Oswald, The Wage Curve, Cambridge, Massachusetts: The MIT Press, 1994. 481 pp. $39.95; Howard Botwinick, Persistent Inequalities, Princeton, N.J.: Princeton University Press, 1993. 300 pp. $49.50.

COPYRIGHT 1996 Monthly Review Foundation, Inc.
COPYRIGHT 2004 Gale Group