Book | Wages, Incomes, and Wealth

The State of Working America 2002-03

Introduction and Executive Summary

Read the press release

Purchase this publication in hardcover

Purchase this publication in paperback 

Table of Contents

Executive Summary
Introduction: What kind of recovery?
Documentation and Methodology

Chapter 1:

Family Income: Full employment reverses historic stagnation
Median income: slow recovery, then strong gains
Latter 1990s pays off for less-advantaged family types
An income ‘generation gap’
Strong growth among dual-earner couples and single mothers
Growing inequality of family income
Counter-arguments to the evidence on income trends
Are taxes the reason for rising inequality and disappointing growth in family incomes?
Is the increase in inequality sensitive to income definitions?
The role of mobility and demographics
Growth in inequality narrows pathways to prosperity
Expanding capital incomes
The ‘time crunch:’ married-couple families with children working harder than ever

Chapter 2:

Wages: Broad-based gains in the late 1990s
Contrasting hours and hourly wage growth
Contrasting compensation and wage growth
Wages by occupation
Wage trends by wage level
Shifts in low-wage jobs
Trends in benefit growth and inequality
Explaining wage inequality
Productivity and the compensation/productivity gap
Rising education/wage differentials
Young workers’ wages
The importance of within-group wage inequality
Wage growth by race and ethnicity
The gender wage gap
Unemployment and wage growth
The shift to low-paying industries
Trade and wages
The union dimension
An eroded minimum wage
The technology story of wage inequality
Executive pay soars

Chapter 3:

Jobs: Recession leads to employment losses
Unemployment and the earnings distribution
Work hours
Nonstandard work
Part-time work
Job stability and job security
Declining job stability
Job security

Chapter 4:

Wealth: Deeper in debt
Net worth
Racial divide
Low net worth
Home ownership
Retirement wealth and income adequacy
Debt service
Student loans

Chapter 5:

Poverty: Historic progress, but high rates persist
The course and composition of poverty, 1959-2000
Alternative approaches to measuring poverty
What’s wrong with the current poverty measure?
Poverty, growth and the inequality wedge
The impact of demographic change
The changing effects of taxes and transfers
Work and poverty: the policy thrust of the 1990s
Poverty and the low-wage labor market

Chapter 6:

Regional Analysis: Significant variation among the states
Labor markets
Poverty and low-wage shares

Chapter 7:

International Comparisons: More inequality, less mobility out of poverty
Incomes and productivity: U.S. lead narrows
Workers’ wages and compensation: unequal growth
Household income: unequal growth
Poverty: deeper and more enduring in the United States
Employment and hours worked: problems with the U.S. model
Evaluating the U.S. model

Appendix A: [To be included in final publisher’s release]

Appendix B: Wage analysis computations

Table notes
Figure notes

[return to top]

Executive Summary

A comprehensive review of the state of working America reveals three important developments.

First, the U.S. labor market has moved into a recession for the first time in a decade. The downturn has been underway in manufacturing since 2000, and, as of June 2002, unemployment remained high. The terrorist attacks of September 11, 2001 meant that, unlike in previous recessions, the services industry has experienced little or no employment growth, and transportation, retail trade, and wholesale trade are experiencing unusual weakness. So far (through June 2002), the unemployment levels are below those of the early 1990s, but job losses have been steep nevertheless. The percentage decline in private-sector employment is greater than that seen in the early 1990s, and the employment decline for women is double what it was then. Higher unemployment is beginning to take a toll on wage growth, which has begun to slow, and there are indications of an expansion in earnings inequality.

Second, this recession comes after years of persistently low unemployment. Although structural problems remained-the long-term decline in job quality and unionization, the deregulation of key industries, and the persistence of imbalanced trade and the ensuing loss of a manufacturing base-low unemployment brought rapid wage and income growth to families across the income distribution. Most notably, middle- and lower-income families, whose economic fortunes had stagnated in prior years, saw real income gains over the late 1990s. African American and Hispanic families also disproportionately benefited in terms of low unemployment and fast earnings growth during those years.

Third, the long-term trend of increased hours of paid work by America’s families continued through the late 1990s. The pace slowed, however, because, as wages were rising, families could work the same hours while bringing home more income. In any case, more time at work, a reduction in paid vacation and holiday time, and the lack of legislated paid family leave mean that families are under increasing time stress.

The living standards of most American families are determined by opportunities in the labor market. The majority of family income derives from earnings, and the loss of a job poses real hardship. In this regard, the recent recession and the ensuing slow-growth recovery are serious problems that have been underappreciated by many commentators who have judged the downturn to be mild based on macroeconomic measures such as overall growth in gross domestic product. Although production has begun to increase and the recession in output may have passed, unemployment continues to rise. As in the early 1990s recession, the United States appears to be in yet another “jobless recovery.” The lack of job growth during this recession is compounded by the fact that the traditional reliance on services to pull up employment has not been effective during this recession.

The data described in the following chapters provide a thorough examination of the trends affecting workers and their families over the post-World War II period. This history-in-numbers shows that falling unemployment in the late 1990s was critical for workers for two reasons. First, it provided workers with
the foundations upon which to bargain with their employers over wages and working conditions. Second, it provided a counterpoint to structural changes in the U.S. economy-the long-term decline in unions, industry deregulation, and continued declines in manufacturing-that had been undermining the security and bargaining power of workers for the past two decades. The low unemployment of the late 1990s was also important because it demonstrated that the economy could reach 4% unemployment without generating inflation, contrary to the long-held wisdom of the economics profession.

Family income: full employment reverses historic stagnation

The full-employment economy of the late 1990s made a large and positive difference in the growth of real income for low- and middle-income families. Whereas real median family income grew 2.8% annually between 1947 and 1973, growth slowed to 0.4% between 1973 and 1995. Between 1995 and 2000, though, growth accelerated to 2.2% per year. The least advantaged-younger families, minority families, and families headed by single mothers-benefited most from the tight labor markets that prevailed in the latter half of the 1990s. For example, between 1995 and 2000, the real median family income of African American and Hispanic families grew 17% and 27%, respectively, compared to 11% for white families.

The larger gains by lower-income families also meant that inequality grew more slowly in the 1990s. Inequality did not, however, stop growing, nor did it reverse course. The richest families continued to pull away from the pack over the decade: the income of the top 1% of taxpayers (including their realized capital gains) grew by 59% from 1995 to 1999 (the most recent available data of this type) while that of the bottom half grew by 9%. Thus, while full employment gave low- and middle-wage workers the bargaining power that was missing over prior decades of stagnant growth, it did not correct structural inequities that persist in the economy.

While these recent developments have lifted family incomes throughout the income scale, longer-term stagnation among low- and middle-income families led to large increases in the amount of time families spend at work. Over the last 30 years, workers in middle-income married-couple families with children have added an average of 20 weeks at work, the equivalent of five more months. Most of the increase comes from working wives, many more of whom entered the labor market over this period, adding more weeks per year and more hours per week. In fact, middle-income wives added close to 500 hours of work per year between 1979 and 2000, the equivalent of more than 12 weeks of full-time work.

Increases in family hours have been equally as large among families headed by high school graduates or minorities as they have been for high-income, highly educated families. On average, between 1979 and 2000, increases in annual hours worked were slightly greater for minorities than for white families: 14.7% and 14.4% for black and Hispanic families, respectively, compared to 11.7% for white families. Middle-income African American and Hispanic families worked significantly more hours than did white families in order to reach the same income levels (given the existing racial wage gaps, this is to be expected). By 2000, middle-income black families worked the equivalent of 12 full-time weeks more than white families.

Wages: broad-based gains in late 1990s

Wages make up the majority of income for most American families. Over the late 1990s, low unemployment played a critical role in boosting wage growth overall, but particularly at the bottom, by strengthening workers’ bargaining power with respect to their employers. Jobs were relatively easy to find, and many employers had to compete for workers. This in turn spurred strong wage and income gains over the latter half of the 1990s economic boom.

The era of stagnant and falling wages from the early 1970s to 1995 gave way to one of strong wage growth after 1995 as wages changed course, rising strongly in response to persistent low unemployment and the faster productivity growth relative to the 1973-95 period. However, despite the strong wage improvements in recent years, it was not until 1998 that the wage level for middle-wage workers (the median hourly wage) jumped above its 1979 level. The median male wage in 2000 was still below its 1979 level, even though productivity was 44.5% higher in 2000 than in 1979. One reason for this divergence is increased corporate profitability, which drove a wedge between productivity and compensation growth.

The trends in average wage growth-the slowdown in the 1970s and the pick-up in the mid-1990s-can be partly attributed to corresponding changes in productivity growth, as well as to changing macroeconomic conditions. Productivity accelerated in the late 1990s, and its growth continued into the current recession, helping to spur strong growth in average wages. Even so, the benefits of the faster productivity growth went disproportionately to capital, as income shifted from labor to capital in the 1995-2000 period.

Over the late 1990s, the pattern of wage growth shifted and growth in inequality decelerated, although it did not change course. In the 1980s, wage inequality widened dramatically and, coupled with stagnant average wages, brought about widespread erosion of real wages. Wage inequality continued to grow in the 1990s but took a different shape: a continued growth in the wage gap between top and middle earners-the 90/50 gap between high-wage workers at the 90th percentile and middle-wage workers at the median-but a shrinking wage gap between middle and low earners-the 50/10 gap. The positive trend in the 50/10 wage gap over the 1990s owes much to several increases in the minimum wage, low unemployment, and the slight, relative contraction in low-paying retail jobs in the late 1990s. Slower growth in wage inequality at the top, relative to the 1980s, was the result of the continuing influence of globalization, deunionization, and the shift to lower-paying service industries (“industry shifts”).

Explaining the shifts in wage inequality requires attention to several factors that affect low-, middle-, and high-wage workers differently. Low unemployment benefits workers, especially low-wage earners. Correspondingly, the high levels of unemployment in the early and mid-1980s disempowered wage earners and provided the context in which other forces-specifically, a weakening of labor market institutions and globalization-could drive up wage inequality. Significant shifts in the labor market, such as the severe drop in the minimum wage and deunionization, can explain one-third of the growing wage inequality in the 1980s. Similarly, the increasing globalization of the economy-immigration, trade, and capital mobility-and the employment shift toward lower-paying service industries (such as retail trade) and away from manufacturing can explain, in combination, another third of the total growth in wage inequality.

One explanation that does not hold up is that the growth of wage inequality reflects primarily a technology-driven increase in demand for “educated” or “skilled” workers. Economists have found that the overall impact of technology on the wage and employment structure was no greater in the 1980s or 1990s than in the 1970s. Moreover, skill demand and technology have little relationship to the growth of wage inequality within the same group (i.e., for workers with similar levels of experience and education), and this within-group inequality was responsible for half of the overall growth of wage inequality in the 1980s and 1990s. Technology has been and continues to be an important force, but there was no “technology shock” in the 1980s or 1990s and no ensuing demand for “skill” that was not satisfied by the continuing expansion of the educational attainment of the workforce.

Among other noteworthy trends in wages:

  • The long-term convergence of wages for men and women stalled, and
    the gap between men’s and women’s wages was about a wide at the end of the 1990s as at the beginning.
  • Benefits declined in the late 1990s. Although health insurance coverage increased after falling for more than a decade, employer costs for health insurance dropped in the 1990s. Employer pension contributions also fell. Since 2000, the amount of money that employers have spent on benefits has grown, as health care costs began to rise again.
  • As wages fell for the typical worker, executive pay soared. From 1989 to 2000, the wage of the typical (i.e., median) chief executive officer grew 79.0%, and average compensation grew 342%. In 1965, CEOs made 26 times more than a typical worker; this ratio had risen to 72-to-1 by 1989 and to 310-to-1 by 2000. U.S. CEOs make about three times as much as their counterparts abroad.
  • Unionization provides an 11.5% wage advantage to workers. However, the union edge is even greater for benefits, with union workers far more likely than non-union workers to receive health insurance and pension coverage from their employers. Moreover, union workers have better health plans with lower deductibles and less cost sharing, and are provided more paid time off, including three more days of vacation.

Jobs: recession leads to employment losses

Rapid economic growth combined with unemployment averaging below 5% improved the job prospects of American workers in the late 1990s. Employment opportunities expanded considerably, especially for traditionally disadvantaged groups such as women, African Americans, and Hispanics. However, the recession threatens to erode these gains. Unemployment began to rise in October 2000 and, through June 2002, it had risen by 2.0 percentage points, up to 5.9%. Both the percentage-point increase and the level of unemployment are smaller than during the recession of the early 1990s, when unemployment increased by 2.6 percentage points to 7.8%. However, the employment losses (in percentage terms) have been greater.

Typically, men’s unemployment increases are larger than women’s, but during this recession women and men have experienced a similar rise in unemployment. Although the total employment loss for men remains higher than for women, women’s employment losses have been twice as large as their losses during the early 1990s recession. Higher job losses for women are a result of the change in the industrial composition of job losses during this recession. Usually, the employment in the service sector (employing about one-third of all workers-two out of five women but only one out of four men) rises over recessions. During the 2000-02 recession, however, services rose only slightly. The terrorist attacks of September 11, 2001 appear to have played a large role in the composition of industries that lost employment over this recession, as transportation and retail trade saw major job losses after September 11.

Jobs losses have been spread fairly equally across educational groups, but not so among racial groups. African American workers have experienced nearly twice as large an increase in unemployment as have white workers; the increase among Hispanic workers is a third larger than among white.

Nonstandard work arrangements-part-time or contingent employment- remain widespread after the 1990s boom, even though this work is generally substandard. Compared to regular full-time work, nonstandard arrangements pay less for comparable work, are much less likely to provide health or pension benefits and, almost by definition, provide far less job security. Over the 1980s and 1990s, temporary work doubled each decade, even though it remains a small proportion of nonstandard work overall. As the economy moved toward full employment in the late 1990s, the share of nonstandard workers reporting that they would rather have full-time work fell, presumably the result of dissatisfied contingent workers finding full-time regular employment.

Job security fell in the 1980s and 1990s as workers began spending less time with one employer. The long-term trend in job stability is disconcerting for a number of reasons. First, workers who are displaced from their jobs often find new ones that pay less and are less likely to offer benefits. Further, many employee benefits, such as health insurance and pensions, are tied to employers. Workers who switch jobs not only tend to start at the firm’s minimal number of vacation weeks, but they may have to go through waiting periods for employer-provided health insurance or vesting of pensions. Since many employers use health maintenance organizations, job switching may also entail changing doctors. However, over the late 1990s, employment retention-whether or not a worker was able to maintain consistent employment over time-increased slightly. Thus, even though workers were switching jobs more rapidly, some of these changes may have been voluntary responses to more rewarding economic conditions.

Low unemployment was insufficient to reverse the long-term trend toward fewer fringe benefits. In 2002, workers were less likely to have employer-provided health insurance than they were 30 years ago, and those who have it pay more. Further, there has been little progress in expanding job flexibility and paid time off-vacations, holidays, and family and medical leave. Heightened job insecurity left over from the recession of the early 1990s along with lower rates of unionization may have limited workers’ capacity to bargain for on-the-job benefits, even when labor was scarce.

Wealth: deeper in debt

Like wages and incomes, wealth is a vital component of a family’s standard of living. Several key features about American wealth stand out. First, wealth distribution is highly unequal. The wealthiest 1% of all households control about 38% of national wealth, while the bottom 80% of households hold only 17%. The ownership of stocks is particularly unequal. The top 1% of stock owners hold almost half (47.7%) of all stocks, by value, while the bottom 80% own just 4.1% of total stock holdings.

Second, the total wealth of the typical American household improved only marginally during the 1990s. The net worth of the average household in the middle 20% of the wealth distribution rose about $2,200 in the 1990s-from $58,800 in 1989 to $61,000 in 1998. On the asset side over the same period, the value of the stock holdings of this typical household grew $5,500, and the value of non-stock assets increased $8,500. Home ownership-the most important asset for most American families-rose over the late 1990s, especially among non-white households. Meanwhile, on the liabilities side, typical household debt rose $11,800. The relatively modest gains in stock and non-stock assets combined with the explosion in household debt meant that the 1990s were far less generous to typical households than business-page headlines often suggested.

Third, only households at the very top of the income spectrum are likely to be adequately prepared for retirement. For many Americans, the 1990s boom left them less prepared for retirement than before: for two out of five Americans, income from Social Security, pensions, and defined-contribution plans will replace less than half of their pre-retirement income.

Fourth, even well into the decade-long boom in the stock market, most Americans had no economically meaningful stake in it. The most recent government data show that less than half of households hold stock in any form, including mutual funds and 401(k)-style pension plans. The same data reveal that 64% of households have stock holdings worth $5,000 or less. While this means that most Americans did not benefit from the stock market boom of the late 1990s, it also means that most have not been directly hurt by the crash of the stock market in 2000.

Finally, for the typical household, rising debt, not a rising stock market, was the big story of the 1990s. Household debt grew much more rapidly than household income in the last decade. By 2001, total household debt excee
ded total household disposable income by nearly 10%. Households in the middle of the wealth distribution absorbed the largest share of this run-up in debt. While low nominal interest rates have made it easier for households to carry the greatly expanded debt, many households appear to be straining. The most recent government data show that 14% of middle-income households have debt-service obligations that exceed 40% of their income; 9% have at least one bill that is more than 60 days past due. Meanwhile, despite the robust state of the economy, personal bankruptcy rates reached all-time highs by 2001.

For many, the debt run-up begins from the day they exit college. Student loans increased substantially over the 1990s as college costs rose and grant aid declined. Most of the increase in loans was in the form of unsubsidized federal student loans-loans that accrue interest while the student is in school. Although students from lower-income families continue to take out more loans on average relative to those from higher-income families, students from higher-income families saw their loan burdens-especially their unsubsidized loan burdens-increase over the 1990s. The result is that students are graduating with historically high levels of loan debt.


Over the last five years of the 1990s boom, significant progress was made in reducing poverty, particularly rates for minorities and children. Yet in 2000, the last year of available data, the share of the nation’s poor was about equal to its level in 1973. Despite positive and even robust income growth, poverty remains stuck at relatively high levels. This lack of progress occurred despite growth in productivity of 52% and growth in real per capita income of 60% over this period.

While levels of poverty are still high in historic terms, the trends of the latter 1990s were clearly positive, especially for the least advantaged. Low unemployment and slower-growing inequality helped drive poverty lower for demographic groups that have historically faced persistent high rates. While the overall poverty rate fell 2.5 percentage points from 1995 to 2000, the rate for African Americans fell 7.3 points, to a historic low of 22% (still twice the overall rate), and Hispanic poverty fell 9.1 points, close to two points per year. For minority children, the declines were even larger. For example, for African American children under age 6, poverty declined by 16 points from 1995 to 2000; for young Hispanic children it fell by 14 points. Poverty among single-mother families also fell steeply, by 9 percentage points between 1995 and 2000, though a third of these families still remained poor in 2000.

Economic growth post-1973 did not lead to as much poverty reduction as expected for a number of reasons. While demographics-the continued increase in female-headed households, in particular-and measurement issues play non-trivial roles, the more important set of factors are economic. The slower overall growth that has prevailed since 1973 (except for the 1995-2000 period) meant there was less income to be distributed throughout the income scale over this period. But equally important, especially over the 1980s, when poverty rates were particularly unresponsive to growth, was the increase in inequality. By preventing low-income families from benefiting from overall growth as they had pre-1973, higher inequality created a wedge between growth in the overall economy and the economic progress of low-income families

Various policy changes over the past decade have had a significant impact on the way we view poverty, both in terms of the problem and the solutions. Among the most important policy changes were the increase in the Earned Income Tax Credit (EITC) in the early 1990s and the vast changes in the welfare system signed into law in 1996. In both cases, the emphasis was on work in the paid labor market as the primary pathway out of poverty.

The impact of these policy shifts comes across clearly in the poverty data from this period. The poor, particularly families headed by a single parent, are working more and deriving much more of their income from the labor market than they were in prior years; work is playing a much larger role in the lives of the poor and near-poor now than at any time over the past few decades. For example, for single-mother families with incomes below the median, labor market earnings as a share of income climbed from 41% in 1979 to 73% in 2000. At the same time, considerably less government cash assistance (transfer income) is flowing to the poor.

Effective poverty reduction depends of both market forces and redistribution of economic resources. During the 1960s, when the United States most effectively lowered its share of poor, both the market and the tax and transfer system were working in tandem. Low unemployment, rising real wages, and broad-based, equally shared growth were complemented by transfers that helped to raise family incomes above the poverty line. Though some redistributive efforts were considerably expanded in the 1990s (such as the EITC), others, such as cash assistance, were sharply cut. Thus, the market and the tax and transfer system were working against each other. Market outcomes drove poverty down by 3.3 points in the latter 1990s, but the diminished effectiveness of transfers added 1.6 points back to the trend.

Measuring poverty presents numerous methodological challenges, and the official U.S. poverty measure has serious shortcomings in this regard. A new measure, implemented on an experimental basis by the Census Bureau, corrects many of the shortcomings in the official measure; it estimates that 15% of the nation was poor in 1999. This rate, much higher than the official rate of 11.8%, assigns an additional 8.8 million to the ranks of the poor.

Regional analysis: significant variation among the states

The national trends in income, wages, employment, and poverty vary extensively by state and region. For example, the 1990s boom took hold in the Midwest and South sooner than in the rest of the country, leading to faster income growth and bigger poverty declines there. Part of these gains stem from the fact that the early 1990s recession, mild by historical standards, was felt much more acutely in the Northeast and, to a lesser extent, the West.

While labor markets tightened everywhere in the 1990s (particularly in the second half), unemployment fell the most in the Midwest and the South, helping to fuel some of the impressive trends in income and poverty there. Yet middle- and low-wage workers in almost every state experienced gains in wages and income and declines in poverty in the latter 1990s, as unemployment fell throughout the country and low-wage labor markets tightened.

Still, as impressive as these gains were, they did not continue long enough to reverse the trend of rising inequality in most states. The gap between those at the top and the bottom of the income scale was significantly higher at the end of the 1990s than in the late 1970s in almost every state.

It is too early (as of this writing) to fully assess the geographical impact of the recession of 2001 and the subsequent slow-growth recovery. It appears, however, that the increase in unemployment and the loss of jobs were greater in the South and Midwest, the areas that most enjoyed the boom of the latter 1990s. In this regard, the pattern of the last recession, in which the East and West Coasts felt the brunt of the downturn, may be shifting.

International comparisons: more inequality, less mobility out of poverty

Because of high productivity growth and low unemployment in the United States during the 1990s relative to Europe, many have argued that Europe should emulate key features of the U.S. economy, including weaker unions, lower minimum wages, less-generous social benefit systems, and lower taxes. The international comparisons in this analysis can shed light on this ongoing debate about the advisability of exporting the “U.S.

Overall the 1990s were a period of slow growth in national income and productivity in most of the OECD economies. In the second half of the 1990s in the United States, however, both national income and productivity growth increased more so than in other OECD countries. But the recent good news for the United States must be considered along with other economic trends. First, income and productivity growth over the last decade have generally trailed the rates obtained in the 1970s and 1980s and are far below those of the “Golden Age” from the end of World War II through the first oil shock in 1973. Second, the above-average income and productivity growth in the United States in the late 1990s came after decades of consistent rankings in the middle or near the bottom among the OECD countries since the 1970s. Third, the U.S. economy has consistently produced the highest levels of economic inequality. Moreover, inequality in the United States (along with the United Kingdom) has shown a strong tendency to rise, even as inequality was relatively stable or declining in most of the rest of the OECD. Fourth, poverty is deeper and more difficult to escape in the United States than in the rest of the OECD. The lack of redistributive social policies only exacerbates the high levels of poverty and income inequality in the United States.


America’s working families responded to the strong economy of the latter 1990s by working harder and more productively, and, for the first time in decades, their efforts were rewarded with real gains in income and declines in poverty. Yet, while inequality grew more slowly in the 1990s than in the prior decade, the richest families continued to reap disproportionate gains. And while working harder meant higher incomes, it also came to mean increasingly more hours spent at work and away from the family. The challenge moving forward is to generate positive gains in living standards that reflect the multidimensional needs of families, both in terms of their incomes and family lives.

The historically long period of economic growth that began in 1991 ended in 2001, and it is likely that the broad-based income gains of the latter part of that recovery have slowed, stalled, or even reversed during the recession. What will happen to family income as the recovery gets underway? If past is a guide, then we might anticipate a return to the experience of the 1990s recovery. But which part? The slow growth early years, when unemployment continued to rise and family income fell? Or the later years, when the economy headed toward full employment and family income grew at historic rates? Unfortunately, the early indicators suggest a slow growth recovery, with unemployment lingering at relatively high rates and family income taking its expected hit.

The turnaround of the late 1990s has been a boon for workers’ wages and incomes, but now all eyes are on the future. If productivity growth continues to be strong (say, 2% or higher), even if not as strong as in recent years, then real wages may continue to expand for most of the workforce. However, unless we head back to full employment-with unemployment in the 4% neighborhood-low-wage workers are unlikely to have the needed bargaining power to claim their fair share of the added growth. Continuing pressures from globalization, more losses of manufacturing jobs, deunionization, high average unemployment, or a fall in the real value of the minimum wage could weaken wages and exacerbate inequality. The near-term answers depend on the extent of the recession and the extent to which the recovery is characterized by growth too slow to lower unemployment. In the longer term, our willingness to alter these structural constraints in the labor market will determine the future living standards of working America.

[return to top]

Introduction: what kind of recovery?

With the recovery of the 1990s officially at an end and a new business cycle struggling to begin, the American economy is facing considerable and unexpected challenges. The recession of 2001 appears to have been short lived, and by some measures, not particularly deep. But even as the gross domestic product expanded through the first half of 2002, the unemployment rate remained well above its level of a few years back, job growth was virtually zero, wage growth was slowing, and the share of the long-term unemployed was hitting historically high rates. At the same time, corporate accounting scandals and diminished investor confidence sent financial markets tumbling.

With those dynamics in play, it is easy to forget that the last half of the 1990s boom was, from the perspective of America’s working families, truly exemplary. In those years, a number of economic factors aligned in a way not seen in decades, and the results were real increases in living standards for the millions of middle- and lower-income American families who had hitherto seen their economic prospects stagnate while the American economy expanded.

In this period, lasting roughly from 1995 to 2000, full employment masked a set of structural problems in the labor market that have evolved over the past few decades. These structural imbalances-the long-term decline in job quality and unionization, the deregulation of key industries, the persistence of imbalanced trade and the ensuing loss of the manufacturing base, not to mention the erroneous belief that unemployment itself could not fall below 6% without generating spiraling inflation-served to shift bargaining power away from minority, blue-collar, and non-college-educated workers, and thus drove the fastest increase in inequality of economic outcomes since such data were collected over the last half century. Now, as full employment fades, recent data suggest that inequality’s growth is resurfacing.

Our main conclusions are that:

  • While the recession that officially began in March 2001 may ultimately be recorded as relatively short and mild, both the downturn itself and the slow growth recovery that arrived in its wake have created a slack labor market. Unemployment as of mid-2002 was two percentage points above its level two years before, significant “hidden unemployment,” exists, job growth is scant, wage growth has slowed (especially at the bottom of the wage scale). There are early signs of a return to the pattern of widening wage inequality seen in the 1980s.
  • In the latter 1990s, the labor market reached full employment for the first time in 30 years. At the same time, productivity growth accelerated. Together, these two trends meant that more economic growth was available to be distributed and that those whose weak bargaining power had previously prevented them from receiving their fair share of growth were now in a much better position. The tight labor market meant that, for the first time in decades, employers had to raise wages to keep or expand their workforce.
  • The other side of this picture is that nothing fundamental has shifted in the economy to ensure that the negative trends that loomed so large over the 1979-95 period won’t return. By lifting the bargaining power of low- and middle-wage workers, historically low unemployment was a potent antidote against the structural imbalances noted above. But these negative forces remain in place, and unless they are addressed, in tandem with efforts to return to full employment, the broad-based prosperity of the 1990s is unlikely to return any time soon.
  • Families are spending more time than ever at work. Over the last 30 years, workers in middle-income, married-couple families with children have added an average of 20 more weeks at work, the equivalent of five more months. Most of the increase comes from working wives, many more of whom entered the labor market over this period, working more weeks per year and hours per week. In fact, middle-income wives added close to 500 hours per year to their work schedules between 1979 and 20
    00, the equivalent of more than 12 weeks of full-time work. Contrary to any notion that increased work in the “new economy” falls mostly to busy professionals, increases in annual hours worked have been larger among middle-income families than among those at the top of the income scale, and increases in annual hours worked by middle-income families headed by a high school graduate surpass those of families headed by a college graduate.

The economic trends of the late 1990s managed at least to hold these forces at bay. What can we learn from this period about reconnecting economic growth to the progress in living standards for middle- and low-income families? To what extent did those boom years repair the damage of prior years, when growth was concentrated at the top and wages fell for large sections of the working class? The answers to these questions will be important as we evaluate developments in the current, fledgling business cycle.

The impact of the recession

What determines whether an economy is in recession? The National Bureau of Economic Research, to which policy makers and academics turn for questions about the timing of business cycles, use a variety of indicators, including industrial production, real aggregate income growth, retail sales, and payroll employment, in making its widely accepted judgment. To best evaluate the impact of the recession from the perspective of working families, however, we employ a different dating procedure that focuses on what might be termed the “labor market recession.” In this dating method, the downturn begins at the end of a recovery when unemployment begins to rise and ends when, as the new recovery takes hold in the labor market, unemployment begins to fall. Since unemployment can sometimes rise well into an official recovery, labor market recessions can last longer than NBER-measured recessions.

For example, the recession of the early 1990s was officially dated as lasting from July 1990 through March 1991, despite the fact that unemployment continued to rise until June 1992. In the current context, since unemployment began rising after October 2000 and was continuing to rise through June 2002 (the last month for which we have data), we consider the labor market recession to have begun in October 2000 and to be continuing as of this writing. By contrast, NBER’s official start date is March 2001.

The fact that the unemployment rate had not yet surpassed 6% by June 2002 has served as evidence for some that the recession has been mild. But just as important as the level of unemployment is the size of the increase, which tends to influence how the downturn is perceived by working families. Since its low point of 3.9% in October 2000, the unemployment rate has climbed two percentage points, adding 2.9 million to the jobless rolls. The increase for various groups of job seekers, particularly minorities, has been more severe. For example, the unemployment rate for African Americans was in double-digits by June 2002 (10.7%), an increase of 3.3 points over this period. The increase for black women, also 3.3 points, was only slightly lower than the 3.5-point increase over the prior downturn (see Chapter 3). And, given projections for relatively slow growth throughout the rest of this year, these rates are likely to climb higher before they reverse course.

While the rates for minorities have gone up the most, it is also the case that this recession, much like that of 1990-91, has been more broad-based than prior downturns. In prior downturns, unemployment grew most among less-educated workers. However, between the fourth quarter of 2000 and the second quarter of 2002, unemployment grew fairly equally for workers with less than a high school degree (2.0 percentage points), workers with a high school degree (2.1 percentage points), and those with some college (2.3 percentage points). As usual, workers with a college degree have experienced a smaller increase in unemployment-1.4 percentage points-but this increase is also greater than that of past recessions.

The fact that unemployment has fairly equally affected workers with different levels of education is driven largely by the industries and occupations where jobs have been lost. Unlike in previous recessions, the service industry has failed to grow over this recession, and retail and wholesale trade have both seen more job losses than usual. In addition, both this and the prior recession (1990-91) were more “white collar” than previous downturns, suggesting that one downside of the “new economy” is that higher education and occupational status provide less insulation from negative market forces.

The increase in unemployment has been considerably dampened by another factor at work in this recession: the slowing of growth in the civilian labor force. Had the trend in labor force growth continued along its path of the early 1990s, in 2002 there would be 2.2 million more workers in the labor market than was actually the case (Figure A). Such a gap is symptomatic of a weak labor market in which potential job seekers avoid even attempting to enter the workforce because they are pessimistic about their employment prospects. It is likely that the national unemployment rate would have been higher in 2002 if these “missing workers” had instead opted to seek work. Had the labor force continued to grow at the rate that prevailed before the slowdown, and if instead of leaving or not joining the labor force half of these missing job seekers had unsuccessfully sought work over this period, then the unemployment rate would have been 6.6% in June 2002 instead of 5.9%. This represents “hidden unemployment” of over a million workers.

Figure A

Another key symptom of the slow-growth recovery is the lack of private sector job growth compared to earlier recessions and recoveries. Figure B plots private sector employment over the last three downturns. In each case, the trend is indexed to its level in the first month of the downturn and tracked for 21 months, the length of time between October 2000 and June 2002. In the 1980s downturn, payroll employment grew slowly at the beginning, fell sharply as the economy “double-dipped,” and then, about 14 months out, began to rise briskly. But in both this last downturn and that of the early 1990s, payrolls contracted initially and then stagnated, as the economy grew too slowly to generate jobs-the hallmark of a jobless recovery. Moreover, while payroll employment was clearly declining before the terrorist attacks in September 2001, the impact of the attacks is evident in the figure, as employment begins to fall more quickly beginning around Month 12, or September. As of June 2002 private sector payroll employment was still 2.2 million, or 2%, below its peak in February 2001.

Figure B
Slack in the labor market affects not only the workers who are unemployed or considering whether to seek jobs; it also reduces pressures on employers to raise wages. When the economic boom began in the mid-1990s and labor markets approached full employment, wage growth picked up, particularly for lower-wage workers (Table 1). Nominal earnings growth for the typical (median) male worker increased from 2.0% in the 1995-96 period (below the rate of inflation) to 5.4% in 1999-2000 (2% above inflation). Low-earning males saw a similar increase, but their wages began to slow in 1999-2000. This pattern of acceleration is less clear for female workers, especially at the median.

Table 1
The data through the first half of 2002 show a fairly clear pattern of deceleration in wage growth in response to weaker labor markets. Both male and female wages decelerate at each percentile shown between 2001 and 2002. Sinc
e inflation was low over that period (1.3%), these increases still translated into real gains, but it appears that, much as falling unemployment led to sharply higher earnings earlier in the boom, increasing unemployment is reversing this effect.

More evidence of the negative impact of the recession and the slow-growth recovery on wage growth is evident in the path of growth in average nominal hourly wages for production workers in manufacturing and for non-supervisory workers in services (Figure C); these workers make up about 80% of the workforce. In the beginning of 1995 these wages started out growing at an annual rate of 2.6% and accelerated to around 4% by 1997 (the average growth rate through 2000 was 3.9%). Growth rates slow in 2001, and decelerate sharply in 2002. In fact, the 2.6% growth rate for the final quarter (second quarter of 2002) is the slowest growth in nominal wages since the first quarter of 1995, the first point in the graph.

Figure C

Higher unemployment is also driving a return to a more unequal pattern of nominal earnings growth, at least for males (Figure D). Before falling unemployment boosted wages, earnings growth for males formed a staircase, with lower wages growing more slowly than those at the median, and the median growing more slowly than the top. By 1999, when the labor market was much tighter, not only were men’s earnings growing faster overall, they were also growing at the same rate for low and middle earners (though still more quickly at the top). By the last period, the staircase pattern had returned.

Figure D

There is no such pattern for women workers. The nominal earnings of the lowest-wage female workers grew more slowly in the first half of 2002 than it did for other wage groups, but there is no evidence of a more unequal pattern of growth accompanying the higher unemployment rates of the 2001-02 period.

The extent to which the pattern of wage growth among men persists and spreads to women will be indicative of whether the factors that fueled the growth of inequality over the 1980s and early 1990s—lower minimum wages, less union coverage, loss of manufacturing employment, high unemployment—were suppressed by the tight labor market. In this sense, the lowest unemployment rates in decades acted to temporarily return bargaining power to less-advantaged workers. As the unemployment rate rises, this macroeconomic source of bargaining power dissipates. Barring a quick return to full employment, less broadly distributed growth is likely to resume as time progresses.

Since most families depend on their earnings as their primary income source, it follows that the lower wages and fewer hours of work that result from higher unemployment would also take a toll on family income. Statistical analysis of this relationship (discussed in both Chapters 1 and 3) finds that a 1% increase in unemployment results in a loss of 1.8% of family income (i.e., family income would be that much higher had unemployment not gone up). By 2002, unemployment was up 2 percentage points, implying a loss of about $1,800 in 2002 to a middle-income family. Rising unemployment is also associated with rising poverty, and so, after falling steeply over the latter 1990s, poverty is unlikely to fall further over the next few years.

Thus, by the middle of 2002, evidence of the recession and the ensuing slow-growth recovery could clearly be observed in the growth of unemployment, the lack of job creation, the slowing of earnings, and the threat of resurgent inequality. In addition, these negative developments point to the probable slowing of family income growth and the lack of progress against poverty in coming years. In this sense, those who look solely at GDP growth or recent gains in industrial capacity may fail to understand the much more direct impact of the downturn on the lives of working families, an impact that is all the more pronounced given its contrast to the positive trends that prevailed over the latter 1990s. We turn now to an analysis of this important period.

Moving toward full employment, 1995-2000

The living standards gains of the 1990s, the latter 1990s in particular, came in two main forms. First, the key determinants of the economic well being of working families—wages, incomes, poverty rates—all improved in real terms. The real wages and incomes of families whose fortunes had stagnated reversed course around the mid-1990s, and the gains were often greatest for the least advantaged. For example, after increasing by 1.0 percentage points between 1989 and 1995, the overall poverty rate fell 2.5 points from 1995 to 2000 (Table 2). Poverty rates for minorities (not shown in this table, but discussed in Chapter 5) fell even more quickly over this five-year period, by 7.3 percentage points for African Americans and 9.1 points for Hispanics. Income for minorities also grew quickly at the median between 1995 and 2000, both in historical terms and relative to that of whites (though by no means fast enough to close persistent racial income gaps). The ratio of black-to-white median income stood at 56% in 1989; by 2000 it reached a historic high of 64%.

Table 2

Second, though it continued to grow, inequality grew more slowly over the 1990s, and its shape changed as well.

The synergistic combination of faster productivity growth and low unemployment was a central factor generating the broad-based gains of the latter 1990s. Productivity grew more quickly over the 1990s than in the 1980s, and the acceleration occurred in the latter half of the decade (Table 2). Similarly, unemployment fell 1.3 points over the 1990s, hitting 4% in 2000, compared to a half percentage-point decline over the 1980s to a much higher 5.3% in 1989.

In terms of income, growing inequality is evident over both the 1980s and, to a lesser extent, the 1989-95 period, as family incomes grew faster at the top (95th percentile) than at the median and faster at the median than toward the bottom (20th percentile) (though in the 1989-95 period both low and median income growth were stagnant). But over the last five years of the 1990s, real family income growth sped up, and it grew at almost the same rates for lower- and middle-income families (12.1% at the 20th percentile and 11.7% at the median). Thus, unlike the fanning out of income growth at each level over the 1980s, inequality did not increase between middle- and low-income families in the 1990s, primarily because the strong economy boosted their earnings, and, as discussed below, their hours of work as well.

Income growth at the 95th percentile, however, also was faster in this period; it rose 15.5% between 1995 and 2000. In fact, growth among the wealthiest families was remarkably consistent over the 1980s and 1990s business cycles, expanding by about 20% in each period. Data presented in Chapter 1 that include realized capital gains provide further evidence that, even with the strong growth spurt at the low end of the income scale, the wealthiest American families continued to pull away from the rest of the pack throughout the 1990s.

Hourly wage growth, unsurprisingly, follows a pattern similar to that of income, though the sharp reversal toward positive real wage growth for the lowest-wage workers was even more plain. Hourly wages at the 20th percentile were down 14.1% over the 1980s and median wages were stagnant; high wages rose 8.1%. Low wages stopped falling in the first half of the 1990s (in part due to minimum wage increases in 1990-91), but they sharply accelerated in the 1995-2000 period, growing more quickly than at any point since the mid-1970s. Median wages grew less quickly; a discus
sed in Chapter 2, they too respond to lower unemployment, but they do not do so as quickly or dramatically as those of the lowest paid. Thus, the gap between middle- and low-wage workers narrowed in the 1990s. For the most part, high wages grew most quickly of all (the slight exception being the latter 1990s), and pulled away from median wages even more quickly in the 1990s than over the 1980s.

Along with higher wages, the 1990s saw a continued trend in rising hours in the paid labor market. The annual weeks and hours worked by married-couple families with children provide compelling evidence of some of the stresses facing families trying to manage their work and family lives. Analysts who focus exclusively on individual average weekly hours have missed this trend, because that measure fails to reflect the combined work efforts of all family members (in fact, when a part-time working wife joins the workforce, average weekly hours will fall while family hours will rise). Due to more wives working more weeks per year and more hours per week, the average middle-income, married-couple family with children is now working 660 more hours per year than in 1979, the equivalent of more than 16 extra weeks of full-time work.

Contrary to conventional wisdom, it is not the case that high-income, highly educated families are driving this trend toward more work. Rather, increases in family hours have been equally as large among families headed by high school graduates or minorities. On average, between 1979-2000 increases in annual hours worked were slightly larger for minorities than for white families: 14.7% and 14.4% for black and Hispanic families, respectively, compared to 11.7% for white families. Middle-income black and Hispanic families worked significantly more hours than did white families in order to reach the same income levels (given the fact of racial wage gaps, this is to be expected). By 2000, middle-income black families worked the equivalent of 12 full-time weeks more than white families.

The increase in wives hours of work was driven by the influx of women into the labor market since the early 1960s, as job opportunities opened up and discriminatory hiring and promotion practices were outlawed. Women not only have increased their presence in the labor market, they are also now less likely to drop out when they marry or have children. At the same time, most male workers experienced stagnant or falling real wages in the 1980s and early 1990s. Wives’ contributions to family income thus grew in importance. Had wives not increased their labor supply, the real incomes of middle-income families would have declined by 2.5% between 1979 and 1989 instead of rising by 6.3%.

Increases in work were by no means restricted to married-couple families. The strong expansion of demand for low-wage workers, in tandem with the welfare-to-work emphasis of welfare reform, led to historically large increases in the employment rates of low-income single mothers with children. While their hours and earnings increased, their welfare benefits contracted sharply, dramatically reshaping the sources of their income. In 1979, labor market earnings and cash assistance each made up about 40% of the income of a low-income single mother (the rest came from other miscellaneous sources). By 2000, about 70% of her income came from labor market earnings and less than 10% came from cash transfers.

Clearly, the labor market, and the low-wage sector in particular, is much more of a factor in the lives of these low-income families than at any time in the recent past. In this sense, low-income families are also more exposed to market forces, for better or worse. Thus, when the market stumbles, as it did in 2001, and when the recovery is too slow to move unemployment back down, they are likely to experience both higher increases in joblessness and less protection from a safety net that has become less effective at preventing hardship among non-workers.

The United States stands apart from other advanced economies in the hours of work individuals devote to paid labor. The U.S. economy employs a greater share of its working-age population, and, in 2000, its workers worked an average of 1,877 hours per year, more than in any other rich, industrialized economy. Even as Americans work more hours, they are not seeing respite through increases in the number of vacation days or holidays. Further, the increased hours of work for families mean that there is less time available for family members to care for their families. Unlike almost every other advanced economy, the United States provides no paid family leave for mothers and fathers. Press reports indicate that the stress on working families has been palpable; the numbers reported here show that families must indeed have difficulty balancing work and family.


The labor market is the primary determinant of living standards for working families. True, more families now hold stock than in the past (about half of households hold some stock, either directly or through a pension plan), but for low- and middle-income families the magnitude of these holdings is small (about $4,000 for the bottom 60% in 1998, the most recent data available), and the inherent risk in such assets is glaringly clear in the current economy. The earnings and employment opportunities in the job market will continue to be the defining factor in the economic lives of the majority of families.

Prior to the mid-1990s, many working families were struggling to avoid getting caught by an economic undertow that eroded their bargaining power and led to lower earnings and more inequality. But a unique period took hold in those years, characterized by a potent combination of low unemployment and fast productivity growth. The economic lessons of those years may well be called the most important of the past half-century: the United States can run a full-employment labor market, generating broad-based gains in living standards without fear of overheating, spiraling inflation, or whatever distortion economic theory incorrectly predicted would befall us if unemployment fell below 6%.

Of course, nothing in this lesson contradicts the precept that the economy is cyclical, and that recessions are endemic. And the recession of 2001 and the current slow-growth recovery are proving to be considerably more difficult for working families than the conventional wisdom suggests. Much of this current discussion focuses on a set of macroeconomic indicators that seem to suggest that the recession was historically mild and that the recovery is, if not particularly strong, well underway and building momentum. As the proponents of this view increasingly maintain: “the fundamentals are all in place.”

Perhaps they are. But if so, the recovery they are driving is a slow-growth one, and it is a certainty that unemployment will fail to fall back anywhere near the levels of the latter 1990s within a year or even possibly two. This is problematic for two related reasons. First, the loss of the tight labor market also spells the loss of the strengthened bargaining power that lower-wage workers enjoyed for the first time in decades. And as this dissipates, so will the unique earnings gains from that period. In fact, earnings’ growth has already decelerated.

Second, full employment masked a set of structural problems that have evolved over the past few decades in the labor market. Prior to the latter 1990s, these structural imbalances served to shift bargaining power away from minority, blue-collar, and non-college-educated workers, and thus drove a historic increase in inequality. As full employment fades, inequality growth is much more likely to resurface, and indeed may have done so by mid-2002.

These issues need to be a top concern of policy makers at all levels of government. Federal policy makers, including but not limited to central bankers, must acknowledge the importance of full employment and chart a course back as soon as possible. The Federal Reserve
and other national policy makers need to recognize the difficulties inherent in a jobless recovery and consider appropriate demand stimulus and safety-net measures. State policy makers can complement these actions by avoiding contractionary spending cuts and also by pushing the federal government—which, unlike the states, can engage in deficit spending—for grants to spend on local stimulus projects.

The chapters that follow make this case in great detail. By examining hundreds of economic trends relevant to the lives of working families, we build a strong case for tapping the benefits of full employment. But our view is a historical one, and we are mindful of the limits of five years of strong growth in the context of decades of eroding institutional supports and diminished regulations. In prior decades, these protections, in tandem with full employment, helped to ensure that the benefits of growth were fairly shared with the families of working America. There is no better goal for economic policy.

Read the press release

Order a copy of The State of Working America 2002-03

Return to SWA home page