Skills and the Wage Collapse

Despite the record economic expansion and near full employment, wages for the bottom fifth of the work force are still far below their 1979 levels. Well over one-fifth of the male work force earns poverty-level wages (22.5 percent in 1997), almost twice as high as in the early 1970s (12.8 percent in 1973). This wage collapse is a big part of the sharply rising inequality experienced by American workers over the past two decades.

The conventional explanation rests on the view that wages mainly reflect the interaction between the demand for and supply of skills. The collapse at the bottom, in this view, is simply a consequence of the failure of individuals to provide the skills employers require in an increasingly computerized global economy. The remedy follows directly: We need a more highly skilled work force. This story is very plausible--and substantially wrong.

To be sure, more highly skilled workers do earn higher wages, other things being equal. But other very important things are not equal. While advanced technology in a globalized and deregulated economy has surely changed the skills sought by employers, there is little evidence that the rising demand for skills in the American workplace was any greater in the years after 1980 than in preceding decades, when the wages of low-skill workers were rising.

So what is different now? Beginning in the mid-1970s, public policy and management strategies have reflected a profound national ideological shift toward laissez-faire markets. At the same time, we have witnessed a revolution in information technologies, which has greatly facilitated communication and transportation. These two concurrent developments have made it far easier, and indeed necessary, to shop around for the most productive, lowest-wage workers in the world. Globalization, with its dislocations, has also brought to our shores millions of new, low-skilled immigrant workers willing to work for even lower wages.

The collapse in earnings is not mainly a skills problem caused by the demands of a new computerized economy. After all, there has been no wage collapse for the least skilled in other developed countries. Rather, the problem is declining worker bargaining power in increasingly global and deregulated labor markets.

Who the New Economy Hurt

Only those at the very top of the earnings distribution have seen real wage gains over the past two decades. Most have experienced declines. Median earners, adjusted for the costs of living, made $11.46 in 1979, $11.18 in 1989, and just $10.82 in 1997 (in 1997 dollars). Even those nearer the top of the earnings distribution, at the 70th percentile, saw their earnings fall from $15.69 to $15.08. While the economy has grown, productivity has increased and the number of millionaires skyrocketed. Even those near the top, at the 80th percentile, only broke even over this 18-year period. Only in the past two years have median wages begun a very modest recovery.

Of course, it was precisely the workers who could least afford lower take-home pay who were hurt most by the "new economy." Those in the 20th-40th deciles lost about 8 percent between 1979 and 1997. Those in the 10th saw a collapse of 15 percent, from $6.42 to $5.46. These figures cover all workers. The picture looks far worse for male workers at all points in the wage distribution. Between the 20th and 40th deciles, male workers in 1997 earned about 18 percent less than they did in 1979. There has been a very slight reversal of these shifts since 1996, but it is too recent and too dependent on abnormally tight labor markets to be a reliable trend.

The earnings collapse has been concentrated among those with the least education and experience, particularly men. For high school graduates with one to five years of work experience, earnings fell by a staggering 27 percent between 1979 and 1995. Even college graduates with less than six years of work saw earnings decline by 11 percent. Except at the very top, only female college graduates with six or more years of experience have thrived in this booming economy, with earnings increasing by 21 percent. It's also true that American workers with less than a college degree are, by standard measures, far less skilled than most northern European workers. For example, among those aged 16-25, 55.5 percent in the United States placed "below an adequate threshold of literacy," defined as level 1 or 2 on the International Adult Literacy Survey (IALS) tests in 1994-1995. This contrasts with 44.4 percent in the United Kingdom, 34.2 percent in Germany, 23.6 percent in Belgium, 22.9 percent in the Netherlands, and 19.7 percent in Sweden.

Bad Timing

At first blush, this all seems quite consistent with the conventional wisdom. American workers indeed need higher skills. But the evidence does not point to technological shifts or skills deficits as the main source of the earnings problem. For starters, the timing is off. Most of the shift in the skill-intensity of employment (conventionally measured as the white-collar share of employment) between 1979 and the present occurred between 1980 and 1983, well before computers could have had large, economy-wide impacts.

Nor do wage trends seem to fit the story. The wage collapse took place between 1979 and 1994. The lowest wage workers (10th decile) fared the worst in the early 1980s, but their wages stabilized between 1984 and 1998, just when information technology began having a widespread impact in many workplaces. In contrast to the popular view that the lowest-skilled workers have experienced the greatest wage cuts, between 1984 and 1994 it was those in the middle of the distribution that fared the worst (in percentage terms). And if skill mismatch in the face of computerization is the main story, why the improvement since the mid-1990s? Are low-skill workers now much better matched to workplace technology than in the late 1980s, the end of the last boom? Hardly.

Moreover, many of the occupations that show large wage declines do not fit the conventional story. If computer-based technological change led to a massive demand shift against the least skilled, and this is the principal cause of the wage collapse, data on occupations and wages should reflect that. But the data present us with a much more complex and ambiguous picture. There turns out to be no statistical association between wage growth and employment growth across some 450 occupations in the 1984-1997 period (or for either 1984-1992 or 1989-1997), and only a quite modest positive link between skill levels and wage growth.

In fact, some of our largest low-skill occupations have experienced very large wage declines (compared to the economy-wide average) at the same time that they've experienced large employment increases (relative to the economy-wide norm) over the 1984-1997 period. These included cashiers, cooks, nurse's aides, orderlies and attendants, groundskeepers and gardeners, janitors, and truck drivers. Further confounding the conventional wisdom, all of these large, rapidly growing, declining-wage occupations showed increasing shares of workers with more than a high school degree, which is particularly striking given the large shares of immigrant workers in many of them. Indeed, five of these eight occupations showed increases in this measure of skill that were far above the average for all occupations.

And finally, the recent scholarly literature offers little evidence of particularly large or accelerating "skill-biased" demand shifts that could plausibly account for the unprecedented collapse of low-skill male wages since the 1970s. Nor have studies found a strong causal link between computerization and wages. (It turns out that pencil use has had about the same statistical effect on wages as computer use.)

So it is fair to ask: How did the technology/skills story become the conventional wisdom, and why has it persisted?

What Does Skill Have To Do With It?

The skills story is attractive intuitively: We are experiencing dramatic technological changes, and those with the highest educational credentials are unquestionably doing the best. But it is a simple supply-and-demand story that omits a lot of messy details about labor market institutions, social norms, management strategies, and bargaining power.

Economists wedded to the usual story read the increase in overall earnings inequality in the 1980s and 1990s as just an increase in the "return to skill." Between the 1970s and 1990s, the share of college-educated workers increased dramatically, yet these workers saw huge gains relative to those with high school degrees or less. In a simple demand-and-supply world, a relative shift in the demand for highly skilled workers "must have" overwhelmed this increase in supply to produce the observed relative wage increases. Hence, the standard explanation for rising inequality.

There are two problems with this "must have" explanation. First, efforts by economists to actually measure demand and supply shifts by decade have produced no support for the requirement of this simple model that labor demand shifts in fact overwhelmed labor supply shifts in the 1980s, the decade in which most of the wage collapse occurred.

The second problem with the supply-demand logic is that other things matter. The minimum wage, union coverage, and the deregulation of a number of industries played key roles in the wage collapse. From a value of $6.29 in 1979 (1997 dollars), the minimum wage fell to $4.34 in 1989, a 31 percent decline. Although there were two increases between 1989 and 1991, by 1995 its value was only slightly higher, $4.48. Two additional increases between 1995 and 1997 made a big difference, increasing the 1997 minimum wage to $5.15. Along with declining unemployment, these increases were probably a big part of the reason for the real-wage increase shown in the hourly wage figure between 1994 and 1998.

Another institutional shift has been the precipitous decline in union membership, from 24 percent of the work force in 1979 to just 13.9 percent in 1998. Perhaps the best measure of union power, the number of days lost to strikes, also shows steady declines since the early 1970s. And concerning deregulation, we can return to our example of truck drivers: Deregulation in the late 1970s facilitated a shift toward nonunion drivers. The result? Sharply falling average real wages for truck drivers between 1984 and 1997, a period of rising demand for drivers, sharply increasing educational attainment among drivers, and declining overall unemployment. Deregulation also undercut the pay of workers in industries as diverse as airlines, public utilities, and banks, all of which were once more intensely regulated than they are today. According to the best estimates in the economics literature, deregulation has cost workers up to $5.7 billion in trucking, $3.4 billion in airlines, and $5.1 billion in telecommunications.

It's About Bargaining Power

To understand the role of lost bargaining power in rising inequality, it pays to revisit the work of the early postwar labor relations specialists, such as Clark Kerr and John Dunlop, part of a long tradition of social scientists who look beyond the intersection of demand-and-supply schedules to account for real-world labor market outcomes. Work in this more "structural" tradition suggests that employers in fact offer a wide range of starting wages and wage increases for a given level of "skill," depending on the industry, the occupation, and the firm's management philosophy and competitive strategy.

Within a broad range set by demand-and-supply forces, by social norms, and by legal constraints, relative wages substantially reflect the outcomes of bargaining between workers and employers. In this view, several other institutional factors will all influence wage-setting and contribute to different wage outcomes for similarly skilled workers in similarly attractive jobs across establishments. These include imperfect information about worker performance, the importance of teamwork in production, the degree of market power in product markets, the share of labor in total costs, the collective power of workers, managerial preferences over competitive strategy, and government regulations.

With this institutional vision of the way wage-setting works, a very different story of the rising wage inequality emerges. It begins with the 1970s, a decade marked by declining productivity growth, rising unemployment, and inflation. On top of these difficulties, workers and unions were faced with a marked ideological and institutional shift in the late 1970s away from collective/public and toward competitive market solutions, much like earlier laissez-faire, antigovernment episodes in the 1920s and 1950s. Partly spurred by this ideological shift in the United States and in the United Kingdom, but also by rapid technological advances in communications and transportation and a growing role for Wall Street in corporate governance, the pace of globalization in production, trade, and financial movements accelerated, which in turn led to a rise in price competition.

In this new ideological and competitive context, labor markets were deregulated, trade barriers dismantled, the minimum wage allowed to decline sharply in real terms, and union power undermined. The available global supply of low-wage workers increased. These new conditions not only allowed, but required, firms to cut labor costs through the adoption of such strategies as wage concessions, the use of low-cost temporary labor and immigrant labor, and the relocation and/or outsourcing to low-labor-cost regions. Consequently, as economists Dani Rodrik and Adrian Wood have long argued, increasing trade and capital flows reduced U.S. worker bargaining power, especially for those least able to contest the wage cuts.

Indeed, legal immigration was allowed to reach its highest level since the first decades of the twentieth century, and current estimates put illegal immigration at six million. We have, in fact, made a political choice to pay large shares of immigrant workers' poverty-level wages for their hard work. This is our official government policy, supported in particular by the self-interests of the farmers, small businesses, and upper-income families who rely on the services they provide.

The consequence of these changes in the effective (often poorly measured) supply of labor, in the deregulation of wage-setting, and in the social acceptability of paying the lowest wage possible, is that changes in relative wages can be unrelated to changes in skill requirements. An alternative ideological, political, and institutional setting might have encouraged firms to take the "high road," in which costs are reduced not via wage and benefits cuts, but through the use of more advanced technologies and more "employee-friendly" (and productivity-enhancing) governance structures and human-resource policies.

Can firms in the "new global economy" make choices to follow the high road? While not easily put to a statistical test, there is some supporting evidence. Based on his analysis of manufacturing plants of 3,000 firms with fewer than 500 employees, economist Dan Luria concludes that "clearly, in most industries, firms can now adopt recipes with very different mixes of wages, skill, technology, training, and basic management discipline." Luria's account suggests that the problem is, perhaps, less an acceleration of skill-biased technological change than political choices that have increased employer incentives to cut labor costs via wage cuts instead of through investment in advanced technologies. Economist Paul Osterman's Securing Prosperity provides a detailed map of a "high road" strategy.

But won't the higher wages that come with the labor market institutions and social policies that shelter workers from virulent wage competition simply produce higher unemployment? Harvard economist Larry Katz has recently explained to a New York Times reporter that "[o]ne can say the Europeans have made a political decision that they are unwilling to tolerate as much income inequality as in the United States. But equity comes at the cost of job creation." And in his BusinessWeek column, Gary Becker asserts that the employment problems of Europe are explained by "conventional interventions in labor markets that discourage companies from hiring workers." The obvious solution is "cutting taxes, subsidies, regulations, and controls over employment, wages, and new businesses."

Despite the pronouncements of leading liberal and conservative economists and the widespread acceptance in policy circles, the welfare-state-as-culprit story does a poor job of accounting for the rise in European unemployment. To be sure, unemployment has been a serious problem in much of Europe. But unemployment trends vary so widely across the continent that it is almost impossible to generalize. Further, the distinctiveness of the U.S. unemployment performance has been quite recent. It also turns out that standard measures of protective institutions and social policies are for the most part not closely correlated with unemployment levels or trends across Organisation for Economic Co-operation and Development nations. Unemployment rates are now falling in much of Europe, despite welfare-states little changed from a decade ago, when the European unemployment crisis was at its peak.

And crucially, the skills dimension is missing. In the conventional story, the technology-driven shift in demand away from the least skilled is the shock that requires wage flexibility if unemployment is to be avoided. If this is so, the least skilled should be driving the upward movement of unemployment rates in Europe. But that has not been the case. Countries experiencing rapid increases in unemployment almost uniformly show rising rates across the skills distribution.

There is no question that in recent years the United States has been quite successful generating employment growth. But there is much more to the story than labor market flexibility. While the growing availability of a rising share of temporary, low wage, often immigrant workers has contributed to U.S. employment growth, as economists Alan Krueger and Jörn Steffan Pischke have argued, much of the U.S. employment advantage can be explained simply by much higher working-age population growth and greater product market flexibility in the United States. Until recently, the United States also had a much more expansionary macroeconomic policies (via deficit spending), while key European economies were constrained by extremely restrictive monetary policies of the German Bundesbank. Further, the business cycle contributed to the rising gap: As the United States came out of a modest recession in the early 1990s, much of Europe went into one, magnifying the differential in unemployment rates. Most experts now recognize that Europe's famous labor market rigidity is not the leading culprit.

Beyond Wage Competition

In sum, America needs more well-trained workers, but the skill-mismatch story does not explain our massive growth in wage inequality. In a long-run (1940-1990) perspective, there is little evidence of particularly large or accelerating skill-biased shifts in the demand for workers that could account for the unprecedented collapse of low and moderate-skill male wages since the late 1970s.

The skill-mismatch view leads either to a do-nothing strategy (leave it to the labor market!) or to policy solutions that put all eggs in the skills basket, or to redistribution programs such as the EITC [see Jared Bernstein, "Two Cheers for the Earned Income Tax Credit," page 66]. Skills and redistribution are part of a comprehensive make-work-pay program, but should not be the whole policy response. James Heckman, a respected University of Chicago economist, has calculated that relying on education and training alone to overcome the inequality trends of the past two decades would require astronomical outlays.

But it is not just a problem of cost. There are many jobs for which employment opportunities are expanding that happen to require low cognitive skill levels. Does greater educational attainment produce greater workplace-relevant skill? And in our current deregulated environment does such investment pay off? Take child care workers. The share of these workers with more than a high school degree increased at a spectacular rate between 1984 and 1997, from 18 to 42 percent. But their real wages declined. Other occupations that showed rapidly rising educational attainment and greater-than-average declines in wages were firefighting occupations (from 47 to 68 percent with more than a high school diploma), data-entry workers (from 38 to 58 percent), telephone operators (from 26 to 46 percent), telephone installers/repairers (from 35 to 55 percent), and electricians (from 33 to 51 percent). We should be wary about imputing much about changes in the supply of skill from educational attainment data, much less about assuming that more education, by itself, guarantees higher wages.

While low-skill wages have stabilized in the past two years, an economic slowdown will push wages again, absent policies that address the real problem: unfettered wage competition. However, once we appreciate that the bulk of the low-wage problem can be traced to a shift in bargaining power, policy options expand dramatically. Policy makers should aim both to improve productivity (raising the ability of employers to pay) and to strengthen wage-setting institutions (reducing destructive wage-competition), as well as to increase skills. The problem is not too much technology, but too little. We need higher wages to encourage higher-road (higher-productivity) competitive strategies by firms. That means a higher minimum wage and a higher share of workers covered by some form of collective bargaining.

In the United States, only about 18 percent of all workers are paid wages that are set collectively, in contrast to 85-95 percent throughout Europe. While there is no need to slavishly follow the path Europe has taken, the fact is that low- and moderate-skilled workers in other developed countries have not experienced real-wage declines, much less ones close to the magnitude experienced here.

We should recognize, applaud, and build upon our current economic success--and give credit where credit is due. While economists continue to hotly debate whether the minimum wage produces negligible or just very small negative employment effects, we have seen a 15 percent increase in the value of the minimum wage between 1995 and 1997. But employers still complain of labor shortages, and the Federal Reserve still thinks the economy is growing too fast. The 15 percent wage hike has apparently not caused many employment problems, but it has certainly helped those at the bottom of the wage scale. Male workers at the 10th decile finally experienced a 7.5 percent real-wage increase between 1994 and 1998, the first good news for that part of the distribution in three decades.

A higher minimum wage and a full-employment economy works. The fact is that U.S. policy makers have more latitude than most think to re-establish the modest constraints on wage competition that prevailed as recently as the 1960s and 1970s. The richest nation in the world can afford and should adopt a living-wage policy for the new millennium. ¤