Copyright 1995 by the Twentieth Century Fund. Readers may redistribute this article to other individuals for noncommercial use, provided that the text and this notice remain intact. This article may not be resold, reprinted, or redistributed for compensation of any kind without prior written permission from the author. If you have any questions about permissions, please contact The Electronic Policy Network, P.O. Box 383080, Cambridge, MA 02238,, or by phone at (617) 547- 2950.

Preferred Citation: Barry Bluestone, The Polarization of American Society: Victims, Suspects, and Mysteries to Unravel (New York: Twentieth Century Fund Press, 1995

Victims, Suspects, and Mysteries to Unravel

Barry Bluestone


From the end of World War II through the early 1970s, America reveled in its "Glory Days." Pent-up demand due to rationing combined with pent-up savings set aside during the war generated a massive consumption boom. Conversion of defense industry to civilian production required a flood of new investment. Urban renewal, suburban expansion, and family migration to the West and Southeast generated enormous increases in state and local government spending. Federal revenue lavished on the GI Bill, veterans' housing projects, the construction of the immense interstate highway system and then the Cold War and the missile race turned out to be more than adequate substitutes for the war spending of the early 1940s. Marshall Plan grants and loans to war-devastated allies and enemies alike fueled foreign spending on American exports. Imports, the only possible detraction from an unparalleled boom in national output and income practically disappeared from U.S. shores for nearly two decades.

The unprecedented coincidence of expanded consumption, investment, government spending, and exports (with no imports to worry about) generated a net increase of some 28 million jobs between 1947 and 1973. Moreover, the expansion of the trade union movement from about 15 percent of the labor force before the National Labor Relations Act of 1935 to more than a third of all workers by the mid-1950s contributed to extending the boom in wages and benefits for a quarter of a century. The glory days saw an increase in real weekly earnings by 60 percent and a virtual doubling of median family income. What is more, the cornucopia of growth was not limited to the most advantaged, the most educated, nor the financially well-to-do. The incomes of the poorest quintile of the economy -- the bottom one-fifth -- expanded faster than anyone else's (see Chart 1). America enjoyed "growth with equity".

For much of America, the Glory days came to an end in the early 1970s. For more than two decades now, real wages in America have been stagnating or falling, the distributions of earnings and income have become increasingly unequal, and the bulk of financial wealth has been accumulating among fewer and fewer families. Since 1973, real average weekly earnings for the more than eighty percent of the workforce who are counted as production or nonsupervisory workers have fallen by 19 pecent; median family income is no higher than twenty years ago; and the share of total marketable net worth accruing to the top 1 percent of all households has increased from about 20 percent to nearly 39 percent. Income gains have gone primarily to the already well-off, while those at the bottom of the income distribution have sustained real losses (see Chart 2).

This working paper has three purposes. The first is to present a reasonably succinct overview of U.S. wage, income, and wealth trends for the past quarter century. In doing this, we will try to identify some of the economic "victims" of this particular era in economic history. The second purpose is to identify some of the "suspects". What factors have been responsible for the success of the victors and the failure of the victims in this economic and social drama? Finally, we shall try to pinpoint a significant number of mysteries that still need unraveling if we are to better understand what caused the end of the Glory Days and divine what public policies might be used to stimulate wage and income growth and produce a society with less inequality and, presumably, less social tension.

We focus on wage, income, and wealth statistics for an important reason. When economists attempt to evaluate the success or failure of an entire economy, they often focus on summary indicators such as gross domestic product. GDP is by no means a perfect measure of the true value of national income -- disasters like the Los Angeles earthquake actually increase GDP in their wake. But the overall trend in GDP nonetheless accurately charts underlying strengths and weaknesses in the economy.

Likewise, wage, income, and wealth distributions are powerful summary indicators of a whole flood of nascent trends and events. If income is polarizing, something quite fundamental is likely emerging in society and the economy as a whole. If real wages shift from rapid growth to continuing stagnation, something quite elemental is likely causing this to occur. Moreover, while the academic debate over wage, income, and wealth trends often involves such lacunae as the optimal measure of income inequality and the correct specification of regression equations, the issues that give rise to such inquiries touch the very soul of the nation. Behind the trends are nettlesome questions regarding race and ethnicity, social class divisions, the survival of central cities, America's role in the global economy, and the challenge of making technology serve human need.


The first evidence of growing inequality -- at least among men -- was uncovered in the early work of Peter Henle (Henle 1972) . Earnings inequality, measured by the Gini index, showed growing variation in earnings between 1958 and 1970. In 1984, Robert Lawrence, then at the Brookings Institution and now at the Kennedy School at Harvard, demonstrated a decline in middle income jobs and a growth in the low-wage segment of the earnings distribution (Lawrence 1984). Lawrence guessed that this trend was benign, representing little more than the temporary crowding of an inexperienced baby-boom generation into the labor market. The economic journalist, Robert Kuttner, saw something much more ominous in the data -- an economy with a "missing middle" (Kuttner 1983).

Two years after Lawrence's work, Bennett Harrison, currently at Harvard University, and I presented our work on "The Great Jobs Machine" to the Joint Economic Committee (JEC) of the U.S. Congress (Bluestone & Harrison 1986). We confirmed that the American economy had generated millions of new jobs as the Reagan Administration was claiming. But we discovered by studying Current Population Survey (CPS) data for 1963 through 1984 that the rate of job growth was not quite as high as under President Carter -- and more importantly -- was comprised disproportionately of low wage jobs. Later research would document a "great U-turn" in the proportion of low wage jobs in the U.S. -- a decline in jobs paying half or less of the median real 1973 wage during the period 1963 to 1970; no change in the proportion during most of the 1970s, and then sharp growth in low wage employment thereafter (Harrison & Bluestone 1988).

The original JEC findings might have gone largely unnoticed except for the shrill reaction to them by the Department of Labor. Responding in the Washington Post to the "Great Jobs Machine" findings, Secretary of Labor William E. Brock sharply rebuked the authors for injecting "new life into this twentieth century flat earth society comprised of believers in the 'bad job myth.'" Other articles challenging our findings appeared from the American Enterprise Institute. Marvin Kosters and his colleagues at AEI massaged the basic CPS data until they had removed all traces of an increase in the low wage share of jobs -- except in the appendices to their papers (Kosters & Ross 1987)!

Such a lively debate inevitably attracts new research and in the years to follow, a spate of wage inequality and low wage employment studies appeared. It is fair to conclude that the overwhelming conclusion from this research was support for the rising inequality/increasing low wage job phenomenon. Fortune Magazine, which had joined the chorus of naysayers when the original JEC paper first appeared, finally threw in the towel in late 1992 conceding in its cover story, "The Job Drought: Why the Shortage of High Wage Jobs Threatens the U.S. Economy."

The best summary of the wage inequality/low wage employment results is found in a recent survey article by Frank Levy and Richard Murnane (Levy & Murnane 1992). They conclude that among men inequality in annual earnings rose moderately from the mid-1960s to the early 1970s; changed little during the 1970s; and then rose rapidly after 1978-80. This was true among all male workers as well as those who work year-round, full-time. Among women, annual earnings inequality rose from the mid 1960s to the early 1970s; fell through the remainder of the decade; and then rose moderately after 1979. Rising inequality among women was more pronounced among year-round, full-time workers than among all women workers. Finally, if men and women are combined as in the original JEC analysis in order to approximate the entire underlying job distribution, there is in the new research confirming evidence of the U-turn that Harrison and I first reported. In still another review of the evidence, Gary Burtless of the Brookings Institution found similar trends toward increasing wage inequality throughout the 1980s (A Future of Lousy Jobs? 1990).


To provide a more detailed and nuanced view of these trends, we present here a set of statistics reviewing what we know about individual wages, family incomes, and finally wealth. Here we rely heavily on the compendium of statistics produced by Lawrence Mishel and Jared Bernstein of the Economic Policy Institute and recent research of New York University economist Edward Wolff produced under the auspices of the Twentieth Century Fund (Mishel & Bernstein 1994; Wolff 1995}.

Wage Trends

As noted above, real average weekly earnings for production and nonsupervisory workers grew rapidly from 1947 through 1973 and then began to collapse. This was, however, not true for the roughly twenty percent of the U.S. workforce comprised of executives, managers, and supervisors. Their wages actually grew sufficiently to offset the losses of the other four-fifths of the workforce. Weekly wages for all workers combined managed to remain virtually the same in 1973 and 1992, the first clue pointing to a change in the overall earnings distribution (see Chart 3).

Gender differences in these overall trends are important. Between 1973 and 1993, the real hourly wage for the median male worker declined from $12.73 to $11.24 -- a 12 percent reduction. At the same time, the median female worker experienced a wage increase from $8.28 to $8.79 an hour -- a 6 percent improvement. As a result, the female-male median wage ratio increased from 65 percent in 1973 to 78 percent twenty years later. This presumably good news is tempered by the fact that nearly three-fourths of the improvement in the wage ratio is due to the decline in male earnings rather than improvement in women's wages.

Education appears to play an important role in explaining these wage trends. One indication of this is found in the widening wage ratio between college educated workers and high school dropouts. In 1963, the mean annual earnings of those with four years of college or more stood at just over twice (2.11) the mean annual earnings of those who had not completed high school. By 1979, this ratio had increased to 2.39. This was but a harbinger of things to come. The education-earnings gap skyrocketed to nearly three to one (2.91) by 1987 and continues to expand (Bluestone 1990).

In fact, the entire pattern of wage growth during the 1980s reflects a remarkable labor market "twist" tied to schooling as shown in Chart 4.

During this decade, the average real wage of male high school dropouts fell by over 18 percent while male high school graduates suffered nearly a 13 percent real earnings loss. At the other end of the distribution, men who completed at least a Masters Degree were the only real winners. Their earnings rose by more than 9 percent. Note that even men who had attended some college saw a serious erosion in their earning power while men who completed college discovered their undergraduate degrees did no more than prevent a decline in inflation-adjusted wages. Women fared better than men in terms of overall wage growth, but the imprint of a labor market twist is clearly discernible here as well.

That three out of four U.S. workers have not completed college provides some indication of how large a proportion of the entire labor force has been adversely affected by the new wage distribution. If we take some liberty with Labor Secretary Robert Reich's definition of "symbolic analysts" -- people such as research scientists, design engineers, and public relations executives whose work focuses on problem-solving, problem-identifying, and strategic brokering activities -- and limit the use of this term to those with two or more years of schooling beyond the B.A. degree, the successes in the new economy make up just 7 percent of the American labor force (Reich 1991). If we count men with the equivalent of at least the M.A. degree plus women with at least the B.A., we can push the proportion of real earnings winners to about 15 percent of the workforce. The extreme losers in this new "meritocratic" society -- those with no more than a high school diploma -- still comprise more than half of all U.S. workers.

In economic terms, what has been occuring is a sharp divergence between the "return" to education and its "rate of return." The distinction may seem academic, but it is not. The "return" to schooling refers to how much one earns with a given level of schooling. The "rate of return" refers to how much an additional year of education is worth. What we have seen is a reduction in the return to education -- a decline in earnings for high school graduates, for example -- while the increment in earnings due to a little more schooling pays off a whole lot. This is why the college degree for men has become a "defensive" good. It provided almost no wage growth during the entire decade of the 1980s, but at least it kept college graduates from suffering the nearly 13 percent loss sustained by those with only a high school diploma. For men, completing college during the 1980s became the equivalent of donning a brand new pair of running shoes for a group camping trip in bear country. You cannot outrun the bear. But you can survive if you can outrun everyone else.

Being less educated and young produces double jeopardy. Among high school graduate men, entry-level wages -- the wages for workers with 1-5 years of work experience -- has plunged by over 30 percent since 1973. Entry-level wages for male college graduates have also declined, but by only 8 percent. For women, entry-level wages have fallen by about half the rate for men in each education category.

Being less educated, young, and black (or Hispanic) places one in triple jeopardy. A recent study, comparing labor market outcomes in the 1980s with an equivalent period in the 1960s, found a dramatic increase in joblessness among 20 year old black men who had no more than a high school degree (Bluestone, Stevenson & Tilly 1992). In the mid-1960s, 20 year old black men who were high school graduates had virtually a zero "jobless rate". This was not surprising given the stringent definition of "jobless" used in the study: not a single hour worked; not a single dollar earned during the entire year. Twenty years later during the mid-1980s, the identically-measured jobless rate for similarly educated black 20 year old men was 14.3 percent. Among high school dropouts, joblessness skyrocketed from 6 percent in the 1960s to 36 percent in the 1980s. Equivalently-educated 20 year old white men also saw a rise in their jobless rates -- but only to 2.5 percent for high school graduates and 10.3 percent for dropouts.

Family Income Trends

So far we have looked at individual earnings. What do we know about family income? There is, of course, a close link between earnings and income because wages and salaries plus the value of job benefits comprise more than 70 percent of personal income. Nonetheless, because of the possibility of multiple earners and because of various forms of capital and transfer income, the trends in family living standards can diverge from the trends we have seen in the wages of individual workers.

Indeed, unlike the decline in real wages, average real per capita personal income -- total personal income divided by total population -- has increased nearly monotonically as shown in Chart 5. Yet this belies what has happened to the "typical" family -- the middle or median family in the entire household distribution arranged from poorest to richest. Median family income rose slowly after 1973, peaked in 1989, and has fallen ever since. The different between average and median suggests a substantial portion of the aggregate personal income generated since the early 1970s has gone to the most well-to-do families. This is precisely what more disaggregated data show (see Table 1). Between 1973 and 1993, median family income in the poorest fifth of all families declined by 14.6 percent; the middle fifth rose by only 1.4 percent; while the top fifth enjoyed more than a 25 percent increase in average family income. The top five percent of families received a 37 percent increase. This distribution more closely resembles the distribution of wages.

                               TABLE 1
               Real Family Income Growth by Income Group,

                    Lowest fifth        -14.6%
                    Second fifth        - 6.0
                    Middle fifth          1.4
                    Fourth fifth          9.9
                    Top fifth             25.4

                    Top 5%                37.2

               Source:  Mishel and Bernstein, The State 
                         of Working America.

As a result of this lopsided income growth, by 1989 the top fifth of all families was receiving more than half (51.8%) of all family income generated in the U.S. The bottom four-fifths of all families received the other half.

Here demographics matter. Young families have been clobbered by recent trends. Between 1979 and 1993, the real median family income of families headed by an adult 25 or younger plummeted from more than $25,000 per year to less than $18,000. Over the same period of time, the median for families with a 45 to 54 year old head rose from $49,400 to $52,000. Marital status is also crucial. Two-earner married couples saw their income rise from $41,800 to $43,300; married families where only the husband worked saw their median income decline by over $4,000 ($34,600 vs. $30,200). Both male and female single parent families experienced significant income loss. Essentially, to prosper in the 1980s and early 1990s you needed to live in an older intact family where both adults worked. Unfortunately, in 1993, less than half (47%) of all married couples had two earners and only one fifth of all married couples were headed by a 45 to 54 year old. Hence, only about one-tenth of all families were in the one demographic group which saw, as a group, its median income rise after 1979.

So far, we have been comparing "cohorts," not individuals. That is, we have been comparing means or medians for different people with the same demographic characteristics at different points in time. This conventional analysis, for example, compares 20-year-old young men in the 1960s with demographically similar 20-year-olds in the 1980s. These comparisons provide valuable information, but need not necessarily resemble what has happened to specific individuals over time.

Unfortunately, there are few "time series" (or "panel") data on individuals, and mining such data is arduous. Still, at least one researcher has been able to carefully track the earnings patterns for prime age individuals as they get older. Stephen Rose has followed individuals who were between the ages of 22 and 48 at the beginning of the 1970s to see what happened to their incomes by the end of the decade (Rose 1993). He then compared this group with 22 to 48 year olds at the beginning of the 1980s to see how they fared at the end of that decade. Normally, we expect younger people to improve their incomes over time as they gain more work experience and on-the-job seniority. What Rose found was that downward income mobility was much higher during the 1980s and upward income mobility was much lower. During the 1970s, only 9.2 percent of prime-age individuals suffered an income loss of 25 percent or more. In the 1980s, the number experiencing this kind of sharp loss in income jumped to 16.3 percent. At the opposite extreme, the proportion of this group experiencing an 80 percent gain in their incomes as they aged ten years fell from 30.7 percent to only 20.4 percent. Those with more education suffered fewer losses and had more gains, but even among college educated workers, income mobility slowed considerably in the 1980s. African-Americans suffered greater losses than whites and had a larger drop-off in terms of high income gains.

Thus, the income mobility data reveal similar patterns as we saw in the cross-section group data. The likelihood that prime age family members would lose ground increased during the 1980s, and the likelihood of this occuring was particularly high for the young, for blacks, and those without a college education.

Wealth Trends

Finally, we can look briefly at what has happend to family wealth over time. In general, we find that the trends revealed in the earnings and income data hold for wealth, but more so. As Ed Wolff demonstrates in his recent Twentieth Century Fund monograph (Wolff 1995, p.2):

The growing divergence in income distribution is even starker in wealth distribution. Equalizing trends of the 1930s-1970s reversed sharply in the 1980s. The gap between haves and have-nots is greater now than at any time since 1929. The sharp increase in inequality since the late 1970s has made wealth distribution in the

United States more unequal than in what used to be perceived as the class-ridden societies of northwestern Europe.

Chart 6, based on Wolff's data, shows a great U-turn in the share of net worth held by the top 1 percent of households in the U.S. From the mid-1920s through the late 1960s, wealth holding became significantly more equal as middle class families were able to save and invest in housing and even modest amounts of financial assets. Ever since the end of the 1960s, however, a disproportionate amount of the wealth created has gone to the very richest families. Note that since 1969, the share of net worth going to the top 1 percent of U.S. households has nearly doubled to 39 percent. By way of comparison, the share going to the bottom 80 percent of all households is now less than 16 percent.

Financial wealth -- limited to bank accounts, stocks, bonds, life insurance savings, and mutual fund shares -- is even more concentrated. In 1989, the top 1 percent of households held 48 percent of such assets; the bottom four-fifths only 6 percent. Perhaps most startling of all, while the richest families were adding immense fortunes to their existing assets between 1983 and 1989, 40 percent of American households actually saw their total net worth decline -- despite the fact that the Dow-Jones Industrial Average rose from 1,190 to 2,509!

There is a fairly simple explanation for this. During the 1980s, most lower income and many middle income families clung to their accustomed living standards by going deeper into debt. The total amount borrowed by consumers nearly doubled between 1981 and 1986, from $394 billion to $739 billion. Families expanded their use of "plastic money" even faster. Revolving installment credit, via Mastercard, Visa, and the likes of the Sears credit card grew from $55 billion in 1980 to more than $128 billion in 1986. The total, including loans for automobiles, passed the half-trillion dollar mark in 1985, and kept growing nearly a $100 billion a year (Harrison & Bluestone 1988).

Black families, in particular, have been left out of the wealth bazaar. In 1983, the median white family had 11 times as much wealth as the median black family. Six years later, this ratio had grown to twenty. No doubt a great deal of this was due to housing segregation. White families in America's suburbs saw their homes skyrocket in value during the 1980s. Black families, caught in inner city neighborhoods, saw their home values plummet.

One could go on with reams of additional statistics. But from the summary we have provided here, a number of things should be abundantly clear. For the past twenty years, earnings have stagnated, the education-earnings distribution has become vastly more polarized, younger workers and families are at greatest risk, black families continue to suffer in both income and wealth, and wealth has been concentrated in a smaller and smaller number of hands. The question we turn to now is why -- and whether these trends are transitory or endemic?


In just the past three or four years, the debate among social scientists has shifted away from whether income and wealth has become more unequal -- there is pretty much unanimity on that question -- to explaining the origins of such trends. Here the debate has become hot and heavy. For the purpose at hand, we can list ten suspects which have been put forward as possible culprits in this drama.

Suspect #1 Skill-biased Technological Change

Robert Lawrence of the Kennedy School at Harvard and Paul Krugman, now at Stanford, are the leading advocates for the position that the new information technologies are skewing the earnings distribution by placing an extraordinary premium on skilled labor while reducing the demand and hence the wage for those of lesser skill (Krugman & Lawrence 1994). This, they contend, is about all you need to explain current earnings trends, and by extension, income.

The problem is, no one has any direct measure of the skill content of technology and certain proof of this hypothesis requires not just skill-biased technological change, but evidence of a tremendous acceleration in new technology during the 1980s consistent with the rapid acceleration in inequality during this decade. To be sure, technological innovation continued to advance during the 1980s, but according to new studies by David Howell and by Mishel and Bernstein , there is little evidence that the pace of introduction was any faster than during the 1960s or 1970s. (Mishel & Bernstein 1994a; Howell 1994)

Essentially, most businesses are not introducing technology that requires vastly improved skill. Many are simply paying less for the same skills they have been using all along while others are hiring better educated workers at lower wage rates to do the work earlier relegated to lesser educated employees.

Suspect #2 Deindustrialization

Other researchers, including George Borjas of USC, have argued that a primary suspect is the shift from goods-producing sectors into services -- deindustrialization. In previous work, I have estimated that between 1963 and 1987, the earnings ratio between college graduates and high school dropouts working in the goods-producing sector (mining, construction, and manufacturing) widened from 2.11 to 2.42 -- an increase of 15 percent. In services, however, the school-related earnings ratio mushroomed from 2.20 to 3.52 -- a 60 percent increase (Bluestone 1990). That all of the employment growth in the economy during the 1980s came in the services sector which was polarizing four times faster than the goods producing industries seems a likely candidate for explaining at least part of the dramatic increase in earnings inequality.

Suspect #3 Deregulation

Goverment deregulation of the airlines, trucking, and telecommunications very likely has the same effect. In each of these industries, intense new competition has forced large wage concessions and employment downsizing. How much this has contributed to overall earnings inequality is the question.

Suspect #4 Deunionization

Reinforcing the inequality inducing shift in industry composition, Richard Freeman of Harvard and a number of his colleagues have noted the decline in unionization (Freeman & Katz 1994). Unions have historically negotiated wage packages that narrow earnings differentials. They have tended to improve wages the most for workers with modest education. Higher union density is one of the reasons for the lower wage dispersion found in manufacturing. That unions have made only modest inroads into the service economy may explain in part why earnings inequality in this sector outstrips that elsewhere.

Suspect #5 "Lean" Production

The restructuring of corporate enterprise toward "lean" production and the destruction of internal job ladders as firms rely more heavily on part-time, temporary, and leased employees is still another suspect in this mystery (Harrison 1994). The new enterprise regime creates a segmented labor force of "insiders" and "outsiders" whose job security and earnings potential can differ markedly.

Suspect #6 "Winner-Take-All" Markets

The heightened competitive market which forces firms toward lean production may also, according to economist Robert Frank, be creating a whole new structure of free agency "winner-take-all" labor markets in which "a handful of top performers walk away with the lion's share of total rewards" (Frank 1994). Today, the fields of law, journalism, consulting, investment banking, corporate management, design, fashion, and even academe are generating payoff structures that once were common only in the entertainment industry and professional sports. Just a handful of Alan Dershowitz's, Michael Milken's, Michael Eisner's, and Harvard Business School Michael Porter's can have a sizable impact on earnings dispersion within each of their own occupations. There is considerable evidence that inequality is not only rising across education groups, but within them, very likely reflecting such "winner-take-all" dynamics.

Suspect #7 Free Trade

Even more fundamental to the recent restructuring of the labor market -- and a likely proximate cause of deindustrialization, deunionization, lean production, and perhaps even the "free agency" syndrome is the expansion of unfettered global trade. According to "factor price equalization" theory, the simple fact of increased trade is sufficient (without any accompanying capital mobility or immigration) to induce the price of each input factor (for example, the wages of unskilled workers) to equalize across trading countries. As the global economy moves toward reduced tariffs and quotas, uniform consumer "tastes" and production techniques, and lower transportation and communications costs, factor price equalization is more likely to occur. But as each particular factor price equalizes, the wage gap between different factors is likely to increase. This will be particularly true in a world where there is a plentiful supply of unskilled labor juxtaposed to a relative scarcity of well-educated workers.

According to UCLA's well-respected trade theorist, Edward Leamer, freer trade, partly as a result of NAFTA, will ultimately reduce the wages of less-skilled U.S. workers by about a thousand dollars a year (Leamer 1992). If factor price equalization is a chief source of wage dispersion today, just consider the implications of when China and India with their immense unskilled workforces enter fully into global markets.

Suspect #8 Transnational Capital Mobility

That freer trade generally provides for the unrestricted movement of investment capital across borders inevitably accelerates the entire wage disequalizing process. Modern transportation and communications technologies, combined with fewer government restrictions on foreign capital investment, have clearly led to increased multinational capital flows between countries. To the extent that companies move to take advantage of cheaper unskilled labor, transnational investment adds to the effective supply of low-skilled workers available to American firms, thus reinforcing factor price equalization.

Suspect #9 Immigration

Increased immigration has potentially the same effect, if a disproportionate share of new immigrants enters with limited skills and schooling. At least among legal immigrants, we know this is true. In the U.S., the typical legal immigrant today has nearly a year less of schooling than native born citizens. Undocumented immigrants almost surely have less. The top ten sources of immigration into the United States in the mid-1980s were Mexico, the Philippines, Korea, Cuba, Vietnam, India, Dominican Republic, China, Jamaica, and Iran.

Suspect #10 Chronic Trade Deficits

Finally, the continuing deficit in the balance of merchandize trade may be culpable to the extent that the trade deficit is driven by deterioration in the trade performance of the manufacturing sector -- the one sector which historically has helped keep earnings inequality in check. Moreover, trade data indicate that the import surplus itself is disproportionately composed of products made with low and modest-skilled labor. This contributes indirectly to the effective supply of workers at the bottom of the education-earnings distribution and thus depresses the relative wages of U.S. workers with little schooling.


In this rogue's gallery, we have presented ten suspects: skill-biased technological change, deindustrialization, industry deregulation, de-unionization, lean production, winner-take-all dynamics, free trade, transnational capital mobility, immigration, and a persistent trade deficit. Quantitatively parsing out the relative impact of all of these forces on the wage distribution is fraught with enormous difficulty. Still, Richard Freeman and Lawrence Katz have attempted to do something like this, at least for the wage gap between men with a college degree and those with a high school diploma. Drawing on their research and that of others including George Borjas, we summarize their findings for some of these suspects in Table 2 (Freeman & Katz 1994; Borjas, Freeman & Katz 1991).

                              TABLE 2
                        Sources of Inequality
             Factors Responsible for the Increase in the Male 
          College/High School Wage Differential During the 1980s

                    Technological Change           7%-25%
                    Deindustrialization           25%-33%
                    Deunionization                    20%
                    Trade & Immigration       15%-25%
                    Trade Deficit                     15%

          Source:   Richard B. Freeman and Lawrence F. Katz,
                    "Rising Wage Inequality: The United States 
                    vs. Other Advanced Countries." 

What do these results suggest? If the Freeman and Katz estimates are anywhere near in the right ballpark, the answer to our "Who done it?" is the same denouement as Agatha Christie's in Murder on the Orient Express. They all did it! While it may provide a fascinating statistical exercise for econometricians, the most important fact about this mystery is that every major economic force in the American economy seems to portend the same fate for the distribution of earnings: greater inequality largely correlated with education and skill. What is more -- and this is critical -- at present none of these trends shows the least sign of weakening.

Is there anything that ties all of these diverse factors together? I believe there is. Each of these reflects a growing "marketization" of the economy and a diminution of institutional constraints on unfettered competition. Increased reliance on domestic market dynamics as the sole determinant of earnings appears to inevitably generate inequality. Heightened competition within these markets, as a consequence of fuller integration into the global economy, exacerbates wage dispersion. While it may be sinister, there is nothing conspiratorial about this phenomenon. It is embedded in the very nature of laissez-faire market dynamics. For this reason, the type of inequality developing in the U.S. is much harder to remedy than, say, overt forms of race and gender-based employment discrimination.


This said, inequality has become a serious enough challenge that even economists who tout the merits of the market have come to recognize the need to reverse current income trends. But given a long-standing reticence to most forms of public intervention in the marketplace, the search for solutions has been restricted to four types of countermeasures: macroeconomic stimulus; education and training; immigration reform; and direct tax and transfer policy.

Macroeconomic Stimulus

Historically, inequality has been countercyclical, growing during recessions and declining during expansions. During economy-wide contractions, lower-skilled workers usually bear the brunt of rising unemployment and stagnating wages -- leading to a wider gap in outcomes between themselves and more skilled workers. During past economic recoveries, these lower- skilled workers have benefited the most in terms of reemployment and rising wages. Family incomes have followed suit. The rapid economic growth sustained for the more than two decades after World War II is partly responsible for the equal growth in incomes we saw in Chart 1

Beginning in the late 1970s, however, earnings inequality has defied the cycle. It has increased during recessions as in times past but then has continued to increase right through recovery. In 1979, the ratio of real average hourly wages of workers in the 80th percentile of earnings to those in the 20th was 2.42. Throughout the 1980s, despite overall economic growth following the 1980 and 1981-82 recessions, the advantage of well-paid workers continued to accelerate. By 1989, the 80/20 percentile wage ratio had risen to 2.77.

Family income inequality did exactly the same. In the recession year 1980, the top one-fifth of all families enjoyed an average income 10 times that of the bottom fifth. After seven straight years of economic expansion, the richest fifth earned 13 times their poorer counterparts . While a "rising tide may have lifted all boats" during the Glory Days, its equalizing power appears to have been overpowered in recent years by the structural factors discussed above. Keeping unemployment low can be recommended for a whole host of reasons. But low unemployment is no longer a ready remedy for inequality.

Education and Training

This is one of the reasons that more attention has turned to education and training. In theory, equalizing education can offset the disequalizing effect of skill-biased technological change and factor price equalization. By upgrading less-skilled workers, the surplus of low-skilled workers is reduced and the shortage of skilled workers is relieved. If, by doing this, the overall level of education is increased, this strategy has the added advantage of improving overall labor productivity and ultimately real wages.

A number of education and training programs have widespread appeal. Expanding the Head Start Program for disadvantaged preschool children, charging a corporate tax levy to finance on-the-job training, instituting a national apprenticeship program, and converting current grant and loan programs into income-contingent loans for college and university students are among them. Setting national standards for school performance, introducing merit systems to reward successful teaching, establishing "charter" schools, and providing for teacher/parent centered control of schools have all been put forward as possibilities for education reform.

Although the exact means of augmenting education are contested, there is almost no disagreement within the economics profession and even less in political circles that schooling and training are the keys to ameliorating maldistribution. Presumably, if somehow we could produce a true glut of symbolic analysts in place of high school dropouts, "meritocratic" inequality would begin to resolve itself.


Legally restricting immigration is a second policy gaining momentum. Canada has a higher rate of immigration than the United States. But immigration laws in the two countries have produced very different effects on their respective labor markets. Since the 1960s, U.S. policy has stressed family reunification (Card 1994). Canada, in contrast, employs a "point system" designed to produce a more skilled immigrant labor pool. This approach has produced legal immigrants in Canada who average 1.3 more years of education than native Canadians (Borjas 1994). If we ignore the thorny ethical issues surrounding the rights of political refugees and judgments about the worthiness of individuals seeking to emigrate, one could imagine tilting immigration policy toward greater use of skill-based criteria.

Taxes and Transfers

Finally, if education and immigration control cannot do the job satisfactorily, there is the old standby of progressive tax and transfer policy to directly redress inequality. Instead of altering the distribution of labor inputs so as to affect the pre-tax and transfer distribution of earnings, this approach redistributes income after it is generated in the private market. Traditionally, most liberal economists have favored this method for it entails the least interference with market forces.


On the surface, this complement of liberal policies seems germane for coping with inequality. Not surprisingly, all four are at the top of the domestic policy agenda of the Clinton White House. Yet, given the powerful set of national and global forces we have outlined above, it is questionable whether any of these, separately or in combination, are up to the Herculean task of reversing the powerfully persistent trends we have explored in this paper.

A case in point is education and training. Counterintuitive as it may seem, greater equality in schooling does not necessarily produce more equal earnings. The fact is, the distribution of education has become significantly more equal over the past three decades. Among year-round, full-time workers, the overall variation in completed years of schooling has declined by more than 25 percent since 1963. This convergence in education is also reflected in the performance of black students and other minorities on the Scholastic Aptitude Test (SAT). In 1976, the average verbal SAT score for blacks stood at the 74th percentile of whites; by 1990 the average score was up to the 80th percentile. A similar six point improvement is found for black students on the Math SAT (Mishel & Bernstein 1994).

Unfortunately, the income equalizing effect of these schooling trends has been more than offset by the growing returns to education driven by our rogue's gallery of suspects. This is true for the entire income distribution. It is also reflected in black/white earnings ratios. For example, the proportion of black men age 25-34 with at least a high school degree increased from 76.7 percent in 1980 to 83.0 percent in 1990. Since the percentage among equivalently aged white men increased only slightly -- from 89.6% to 89.9% -- the black/white male schooling ratio increased from .856 to .923. Nevertheless, between 1979 and 1993, black men in this group fell further behind white men in relative hourly earnings -- from a ratio of .843 to .821 (Bernstein 1995).

In effect, schooling gaps are shrinking but increasingly smaller educational differentials have greater earnings clout. Recall the distinction between the "return" and the "rate of return" to schooling. As such, no matter what other benefits might flow from increased schooling, expanded education is not, by itself, a sure cure for inequality.

Typically, job training programs have had even less success at closing the earnings gap. While the federal government has experimented with a bevy of programs from the original Manpower Development and Training Program (MDTA) of the Great Society days to the Job Partnership and Training Act (JTPA) program in the 1980s, repeated evaluations suggest mixed results at best. Some programs like the Job Corps, which provide long-term training opportunities to disadvantaged youth, have been found cost-effective. The vast majority have dubious cost-benefit ratios. In any case, even when these programs are "successful", the earnings advantage they give participants produces only the slightest dent in the trends we have documented. James Heckman of the University of Chicago has estimated just how small this dent might be (Heckman 1993). Assuming a generous 10 percent rate of return on investment, he calculates a staggering $284 billion would have to be spent on the U.S. workforce simply to restore male high school dropouts to their 1979 real incomes. To restore education-based wage differentials to 1979 levels without reducing the real incomes of existing college educated workers would take more than $2 trillion dollars!

To be sure, further investments in human capital programs may have a somewhat better track record than past attempts, particularly if they are well-targeted. But one cannot ignore the enormous increases in inequality that have already taken place nor overlook the Alice in Wonderland nature of the problem. To counter the accelerating trend of wage disequalizing global competition and technological change, these programs would have to be expanded at a frenetic pace simply to keep inequality from growing worse. Just for the record, note that at the present rate, forty years from now, those with a Master's Degree or more education will comprise at most 15 percent of the U.S. labor force. This is far short of the glut of symbolic analysts needed to narrow wage differentials.

As for immigration reform, it too may have a marginal impact on the earnings distribution, but any improvement will be largely limited to regions of the country where immigration flows have been disproportionately large -- California, Texas, Florida, and perhaps a few states in the Northeast.

That leaves tax and transfer programs as the centerpiece for adjusting distributional outcomes. On paper, a suitably progressive set of tax rates combined with sufficiently generous transfer assistance can be constructed which radically redistributes income after it is earned in the market. But, in practice, even such hard-to-win liberal measures as the Clinton 1993 tax initiative produced a degree of redistribution so small it hardly deters the inequality trend set in motion by the phalanx of global and domestic forces examined above. In 1977, when the federal tax system was significantly more progressive than today, the ratio of pre-tax total income shares between the richest fifth and the poorest fifth of all families was 9.5 to 1. Federal taxes reduced the overall gap in relative shares by less than 20 percent; regressive state and local taxes wiped out this improvement. Given increased reliance on regressive payroll taxes and an aversion to any further increase in progressive income taxation, substantially more redistribution is unlikely through the tax system.

The same is true of public transfer programs. Over the past twenty years, the New Deal "social safety net" of unemployment insurance and welfare assistance has come under attack. Unemployment insurance (UIB) covered more than 60 percent of the jobless during the 1961 and 1975 recessions. Despite the greater severity of the 1982 recession, only 43 percent of the unemployed collected jobless benefits. During the 1991 recession, coverage was down to 40 percent. While there may be important reforms of the federal UIB system under the Clinton administration, it is unlikely the states or the federal government will greatly expand coverage of the unemployed. As for the traditional welfare system including Aid to Families with Dependent Children (AFDC), real benefit levels have been cut in many states and the system has imposed greater eligibility restrictions. Proposed reform of the AFDC program may change the dynamics of dependency, but it will do nothing to change the final distribution of income -- and it could, by forcing welfare recipients off the roles, make it worse.

In sum, education and immigration reform as well as redistributive tax and transfer policy could contribute to reducing inequality, but they are by themselves -- even under the best of political scenarios -- no match for the concerted forces now driving the labor market.


There is, however, an additional policy agenda which a less ideologically-constrained government could entertain. This agenda focuses specific attention on the forces generating greater inequality. First, there is direct regulation of the labor market. As the empirical evidence demonstrates, the growth in earnings inequality has materialized in part because of a serious erosion in wages at the bottom of the skill distribution and a sharp decline in unionization. Higher minimum wage standards are one way government can affect the distribution of employee compensation. While raising the mandatory wage minimum theoretically entails some tradeoff in the form of job loss, a batch of recent studies proves the positive earnings impact of modest increases in the statutory minimum far outweighs any employment effect (Card & Krueger 1995). Thus, the aggregate wage bill paid to less-skilled workers increases, improving the living standards of those confined to the bottom rungs of the earnings ladder.

Labor law reform, making it easier for unions to organize unorganized workers, provides an indirect method to accomplish the same objective. While there are many reasons why union membership is dwindling, the recent Fact Finding Report of the U.S. Commission on the Future of Worker-Management Relations found undeniable evidence that unions do not face a level playing field when it comes to organizing drives (Commission on the Future of Worker-Management Relations 1994). That employers can permanently replace striking employees reduces the ability of unions to organize and to freely negotiate collective bargaining agreements. That unions do not have free access to employees during membership drives, and that the penalties for employer "unfair labor practices" are so trivial, tilt the playing field toward management. Legislation that would ban permanent striker replacements, permit union organizers access to in-plant bulletin boards and public forums, impose more costly penalties on employers who violate the legal rights of union organizers, expedite legal remedies, and authorize binding arbitration for first contracts would all presumably augment union membership.

Finally, there are industrial and trade policies to consider. Advocates of industrial policy can point to the success of the U.S. aircraft industry and agriculture as examples where government R&D subsidies and government purchases helped create and maintain industries that dominate world markets. The Carter Administration's Chrysler Loan Guarantee, which provided an eleventh hour reprieve from certain bankruptcy for the then hapless automaker, played a critical role in turning around an old "smokestack" company and saving tens of thousands of well paying jobs -- not only at Chrysler but at hundreds of its suppliers. With a new lease on life supplied by the federal government, Chrysler has surged back as a world leader in automotive technology. There are, of course, many instances of failed industrial policy -- the government's ill-fated Synfuels Corporation, for example -- but there are ample cases on the other side of the ledger. Maintaining as much of the nation's manufacturing base as possible would no doubt have a salutary effect on incomes. Encouragement from the government might help.

The other measure that might be used to bolster the goods producing sector is implementation of "fair trade" language in trade agreements. One way of doing this is to use tariffs and trade barriers designed to give temporary protection to key industries so as to promote industrial revitalization and economic transition. Another form of managed trade ties the offer of reduced protection to a trading partner's compliance with certain environmental and labor standards. Critics of NAFTA argued for side agreements that would have linked the pace of tariff reduction to the rate at which Mexican wages caught up with Mexico's rapidly rising productivity. Government-imposed limits on trade can have detrimental effects on prices and therefore reduce average real incomes from what they might be under a free trade regime. Nevertheless, a carefully crafted set of trade policies that condones temporary protection of selected domestic markets and sets minimum labor and environmental standards could soften the distributional impact of factor price equalization. Linking the pace of trade liberalization to the rate at which foreign competitors close the productivity-wage gap in their own countries may even produce a "positive sum game." For example, encouraging increased wages in Mexico -- in line with the country's increased productivity (in part due to U.S. investment in that country) -- would boost living standards for the Mexican workforce, reduce downward pressure on U.S. wages, and increase overall aggregate demand and output in both countries. In this case, it is altogether possible that lower-skilled workers in both the U.S. and Mexico would benefit due to higher wages and consumers would not suffer major price increases because wage growth would follow productivity growth. The only groups possibly harmed would be high income investors in both countries who would see some decline in their profit shares. The trick is to keep any imposed tariff barriers from becoming permanent or prompting a trade war.

One might reasonably ask, is there any evidence that more aggressive structural policies can actally help? Critics like Mickey Kaus think not (Kaus 1992). In declaring that "the venerable liberal crusade for income equality is doomed," Kaus argues that cannot decide to keep all the nice parts of capitalism and get rid of all the nasty ones. You cannot have capitalism without "selfishness," or even "greed," because they are what make the system work. You can't have capitalism and material equality, because capitalism is constantly generating extremes of inequality as some individuals strike it rich ... while others fail and fall on hard times.

This may sound sensible, but it will come as something of a surprise to a large number of America's foreign "capitalist" competitors. A comparison of earnings trends across countries suggests that different institutional frameworks, all operating within a capitalist framework, produce substantially different distributional outcomes.

All nations now face nearly identical pressures from technological change and global competition. Yet not all are experiencing the same degree of growing income inequality. Those countries with stronger unions, national wage solidarity agreements, generous social welfare programs, and more vigorously pursued industrial and trade policies have much less inequality than countries pursuing pure free market strategies. Relying on an extensive review of comparative statistics, Richard Freeman and Larry Katz conclude that while educational and occupational skill wage differentials were growing rapidly in the U.S. and the United Kingdom during the 1980s, the experience elsewhere was quite different (Freeman & Katz 1994; Atkinson, Rainwater, and Smeeding 1994). The Netherlands saw an actual decline in inequality; France, Germany, and Italy experienced no noticeable change at all in wage differentials; and Australia, Canada, Japan, and Sweden suffered at most modest increases in wage dispersion.

Similar results for education-earnings differentials are shown in Table 3. The table compares the ratio of the mean (or median) college-equivalent wage to the high school-equivalent wage for the early 1970s and the mid-to-late 1980s. By far the largest increases in earnings inequality by education are found in the United States and in the U.K. There was very little or no increase in Sweden, Canada, and Japan. Both the Netherlands and West Germany actually experienced a decline in educational earnings differentials.

                          TABLE 3
          International Educational Differentials 
                 for Males during the 1980s

                     Ratios of Mean or Median College-Equivalent
                          Wage to High School-Equivalent Wage

                     1979 or            Mid-to            5-Year
  Country           Early 80s          Late 1980s         Change 

United States          1.37               1.51             0.09
United Kingdom         1.53               1.65             0.07
Sweden                 1.16               1.19             0.03
Canada                 1.40               1.42             0.02
Japan                  1.26               1.26             0.00
West Germany           2.00               1.94            -0.06
Netherlands            1.50               1.22            -0.35

Note: With the exception of the Netherlands and Germany, all
figures control for changes in the age composition of educational

Source:  OECD (1994); Abraham and Houseman (1994).

The "flexibility" of the U.S. market may be partly responsible for lower overall unemployment rates compared with these countries, but the price of this flexibility seems to be much higher levels of economic polarization and social inequality. Interestingly, recent research by Rebecca Blank suggests there is little empirical evidence that labor market flexibility is substantially affected by the presence of social protection programs nor does limiting social protection automatically enhance the speed of labor market adjustment (Blank 1994) . Instead, she finds by enhancing worker well-being, enhanced social protection may actually permit flexibility that would not otherwise be possible.


Can we have a society where not everyone is a symbolic analyst and yet have an equitable distribution of income? The answer is a qualified yes, but it almost certainly requires using the entire arsenal of policy weaponry, not just the munitions which appear regularly on the standard "safe" list. To shift direction toward a more equal wage and income distribution in the face of the powerful trends discussed here requires a careful consideration of new education policies and trade policies; new tax and transfer programs and new labor market regulations. Anything less will probably fail.

Modern advances in technology and increased global economic integration -- for all the benefits they provide -- have the unfortunate consequence of leading to greater inequality and social polarization in totally unfettered markets. In the absence of a much greater redistribution effort tied to human capital investment and tax and transfer policy, an industrial policy/managed trade strategy may end up as a sine qua non for the maintenance of civil society. Indeed, in the economic world we have described, a frontal assault on inequality will require major efforts on all sides of the market. Strong macroeconomic stimulus will be necessary periodically to maintain employment growth; education and training efforts targeted to the disadvantaged will be needed to keep skill-based earnings differentials from further widening, and demand-side policies will be needed to keep manufacturing employment opportunities from disappearing. The hard work -- both practically and politically -- is to devise policies that generate more equality without unduly sacrificing economic efficiency. Quite clearly, this issue will be at the core of political and economic discourse as we approach the 21st Century.


Go to interpretations page

Go to H. T. Reynolds page

Copyright 1995 New Prospect, Inc.