Spatial and labor market contributions to earnings inequality: an overview - Special Issue: Earnings Inequality

Earnings inequality has increased markedly in the United States in recent decades. In 1979, full-time workers at the 90th percentile of the wage distribution earned about three times as much as those at the 10th percentile. By 1992, high earners were being paid about four times as much as low earners. Inequality has risen along various dimensions - by education, by age group, and among similarly educated workers with similar length of experience. Moreover, the trend toward greater earnings inequality looks more pronounced when one takes account of persons who work less than full-time or less than year-round. And the degree to which inequality has risen has been greater in the United States than in other advanced economies.

Increasing inequality has occurred against a backdrop of increasing economic segregation. In the 1960s, middle-class and wealthy Americans began to move out of central city neighborhoods to the suburbs in greater numbers, leaving poor Americans increasingly isolated in poverty-stricken urban areas. Although segregation by race remains at very high levels, the exodus of middle- and upper-income blacks from central cities has accelerated. This growth in segregation by income has left many poor (and often minority) households living in neighborhoods that lack positive role models or established job networks, have high crime rates, and are isolated from fast-growing suburban employers.

Some economists have explained trends in earnings inequality by looking exclusively at changes in job skills, job requirements, and labor market institutions. Others focus on the deteriorating quality of many urban neighborhoods and the increasing economic isolation of their residents that inhibits the accumulation of skills by youth and reduces their access to jobs. In November 1995, the Federal Reserve Bank of Boston convened a symposium of experts representing these alternative perspectives, with the goal of fostering a more integrated understanding of the causes of inequality and generating fresh insights into possible policy responses.

The symposium began with two overviews of the largely separate literatures on labor market and spatial contributions to earnings inequality, written by Yolanda Kodrzycki and Christopher Mayer. A consensus appears to be growing within each of these bodies of research that the rise in inequality has been multifaceted. Discussions at the symposium expanded on this theme and, appropriately, considered a broad array of potential remedies.

The participants generally agreed that increases in earnings inequality during recent decades are a source of concern. They also indicated that the distribution of earnings is likely to become even less equal in the future. Although some experts (particularly those specializing in labor economics) expressed support for measures that would have the effect of reducing inequality in the upper half of the income distribution, most policy proposals appeared motivated by a specific concern for low earners. One prominent focus of such efforts would be to reduce concentrations of poverty, both by reducing barriers to residential mobility for the urban poor and by enhancing their educational opportunities.

I. Spatial Contributions to Earnings Inequality

For years, economists have recognized that residential location plays an important role in the labor market outcomes of individual workers, but have debated about which specific locational factors are most relevant.

How Residential Location Matters

Katherine O'Regan and John Quigley analyze employment and "idleness" (not employed and not in school) outcomes for a large sample of "at-home" urban youth. They find evidence in favor of two broad explanations for why location is so important. First, the "spatial mismatch hypothesis," developed in the 1970s by John Kain, posits that racial segregation of minorities in inner-city neighborhoods and increasing job suburbanization have reduced the incomes and employment opportunities of blacks. More recently, William Julius Wilson's work on the "urban underclass" suggests that the social isolation resulting from the concentration of minorities has a negative effect on individuals generally, and on their labor market performance specifically.

The authors' analysis is based upon an unusually rich sample of 28,000 youths in four New Jersey metropolitan areas, matched to detailed census tract demographic information and specially constructed measures of access to employment. Even after controlling for a variety of human capital characteristics, O'Regan and Quigley find that neighborhood factors matter consistently in explaining both "idleness" and employment, although they cannot distinguish whether the effect is due to informal job networks, role models, or peer influence. Measures of access to jobs, while not consistently significant across metropolitan areas, are positively related to employment in some areas, especially for minority youth. Access to jobs appears to play essentially no role in determining youth idleness.

Simulations using these results demonstrate quite clearly that the constellation of factors that distinguish "good" from "bad" neighborhoods affects teenage employment in profound ways. For example, predicted employment rates for a neighborhood with characteristics encountered by the average white youth are up to one-third higher than for the average black youth's neighborhood.

In his discussion, Edward Glaeser comments that O'Regan and Quigley have made significant improvements in distinguishing the impact of proximity to employment from that of neighborhood characteristics on employment outcomes. Nonetheless, the paper still faces possible problems of interpretation because attributes of the neighborhood may be related to unobserved individual characteristics, and individuals might choose a neighborhood based on the productivity of that neighborhood for that specific person. Glaeser notes that the reliability of O'Regan and Quigley's results depends critically on two assumptions: first, that parents of at-home youth choose a neighborhood based on their own job concerns rather than the employment prospects of their children; and second, that these two factors are independent of each other. On the policy front, Glaeser suggests that the role of neighborhood factors in influencing individual outcomes could provide an efficiency argument for government policy intervention since individuals cannot directly control the actions of others. He warns, however, that government may exacerbate rather than improve existing market failures.

Glaeser concludes that policies designed to promote and subsidize human capital accumulation and alter the patterns of family responsibility are most likely to be successful, while enterprise zones, subsidized transportation, or mobility strategies are unlikely to provide many long-lasting benefits. He draws this conclusion by noting that measures of job access are less significant in the O'Regan and Quigley paper than the measures of neighborhood composition. Glaeser agrees with the authors, however, that it is difficult if not impossible to separate out specific neighborhood influences (such as the crime rate, the percentage of poor residents, or the employment rate). Nonetheless, identifying such specifics is a necessary first step in designing government policies to address neighborhood externalities.

Employers and Location

Much of the empirical evidence and theoretical support for spatial mismatch comes from studying individuals, but Harry Holzer and Keith Ihlanfeldt argue that an investigation of firm behavior is also important. Previous studies have found strong evidence that access to jobs affects labor market outcomes for minorities, and many researchers have proposed policies to improve access from predominantly minority urban areas to suburban job locations or to encourage minority households to move to the suburbs. Yet, if the newly located minority households do not have the skills that firms demand, or if firms discriminate in the hiring process, such policies are doomed to failure.

Holzer and Ihlanfeldt use data from a new survey of over 3,000 employers in four major metropolitan areas to investigate the determinants of black employment and wages at the firm level. While considering a wide variety of other factors that might affect black employment, including employers' skill needs, the race of customers, and the race of the person responsible for hiring, the authors find that employers' proximity to black residences and to public transit increases the likelihood of hiring black workers. They also find that wages are lower at employers located close to the black population. Holzer and Ihlanfeldt argue that such results are consistent with the predicates of spatial mismatch in which labor demand shifts away from black areas, but the labor supply response is limited by housing segregation.

Holzer and Ihlanfeldt conclude that policies designed to encourage the mobility of minority households or to subsidize reverse commuting would likely raise employment levels and wages for black workers. The authors support programs to improve the skills of minority workers, but they view such programs as complements rather than substitutes for transportation programs, because one type of policy enhances the effectiveness of the other.

The discussant, James Rosenbaum, accepts Holzer and Ihlanfeldt's empirical findings, but takes issue with their policy recommendations. In assessing the empirical findings, Rosenbaum notes that sample selection is still a potential problem - that is, black workers hired in white suburbs might somehow be different from other black workers in ways that are not captured by the controls. Yet Rosenbaum goes on to present evidence supporting Holzer and Ihlanfeldt from the Gautreaux experiments in Chicago, in which low-income black applicants to public housing were randomly allocated to the city or the suburbs. In a subsequent study of 300 adults, movers to the suburbs had significantly higher employment rates than those who moved within the city, while employed members of the two groups had similar wages and hours of work.

In developing policy recommendations, Rosenbaum is less optimistic than Holzer and Ihlanfeldt about transportation policies unless they are combined with some programs to equip black workers with the attributes that many employers seek. He cites evidence showing that employers are often concerned with personality issues in addition to directly measured skills, using such imperfect indicators as a firm handshake or a traditional hairstyle. In fact, employers often use residential location as another signal of whether job candidates will be "good" workers, a practice that casts doubt on the possible success of transportation programs. Programs must be developed to give employers more accurate signals of a potential worker's quality. One advantage of residential mobility programs like Gautreaux is that they allow potential workers to tap into informal job networks through local schools or churches, which provide signals that employers trust. Rosenbaum concludes by describing the significant long-term advantages of encouraging the residential mobility of blacks: The largest benefits of mobility occur in the second generation.

Inequality in Local Schools

John Kain and Kraig Singleton focus on differences in the access to educational inputs among children in five racial and ethnic groups. The authors pattern their paper after the Coleman Report, a path-breaking study prepared nearly 30 years ago, which found that the nation's schools were highly segregated by race but that schools of various racial composition had quite similar access to direct school inputs. Kain and Singleton focus on schools in Texas, where they have collected enrollment data and assessment test scores for 1.8 million students during the five-year period between 1990 and 1994.

The authors find that important changes have taken place since the Coleman report was published. The most obvious change has been a significant reduction in the severest types of racial and ethnic segregation. In fact, only 16 percent of Anglo students in Texas attend schools that are more than 90 percent Anglo, with similar results for other groups. While segregation may have diminished, Kain and Singleton show that test scores for black and Hispanic students are significantly lower than for Anglo and Asian-American students, even controlling for school fixed effects and the family's poverty status. Finally, the authors find significant differences in the within-district provision of several important school inputs. Teacher ability, as measured by proficiency test scores, is inversely related to a school's percentage of black and Hispanic students. In addition, teachers employed in schools with a high percentage of disadvantaged minority students have fewer years of education and less experience than average, although they also have slightly smaller classes.

Kain and Singleton view their paper as a first step in a longer-term project of quantifying the impact of educational inputs on student performance. However, even if policymakers acknowledge only a limited relationship between inputs and performance, the results in this paper suggest additional opportunities to equalize inputs across schools. Texas educators appear to have narrowed differences in class size at schools within a district, but they have yet to equalize teacher quality.

Eric Hanushek reviews the Kain and Singleton paper with few criticisms, but a different interpretation of the results. He considers the most important finding in the paper to be that estimated racial and ethnic differences in student performance appear to be independent of school inputs. Even if school inputs do vary by a school's race and ethnicity, such differences will neither harm nor help the performance of minority students. This conclusion supports Hanushek's views (formulated from previous research) that measured school inputs are, for the most part, unrelated to student performance.

Nonetheless, Hanushek concedes that many studies have found that teacher test scores are positively correlated with student performance, suggesting the possibility that differences in teacher quality could have some measurable effect. He also notes that the racially biased distribution of inputs that conventional wisdom links to student performance could imply systemic racial discrimination.

II. Labor Market Contributions to Earnings Inequality

The labor market studies presented at the symposium concentrated on three themes: demand and supply trends for college-educated workers, the effect of workplace innovations on wage structures, and institutional influences on earnings inequality.

Responses to the Rising College Wage Premium

The earnings of college graduates and non-graduates diverged sharply in the 1980s and early 1990s, after showing little relative change during the 1970s. John Bishop's paper explores whether the earnings premium associated with a college education will increase, given likely labor market trends and demographic and institutional constraints. He concludes that opportunities for college-educated workers will continue to expand and will lead to a continuation of recent trends in their relative earnings.

Bishop criticizes U.S. Bureau of Labor Statistics (BLS) calculations that appear to predict rising underemployment among BAs. The BLS derives projections of the demand for college-educated workers from forecasts of industry output, using data on occupational staffing ratios by industry and on the prevalence of college graduates within occupations. This methodology historically has produced poor forecasts of the direction of the college job market. According to Bishop, the BLS attempt to divide jobs into some that require a bachelor's degree and others that do not is flawed conceptually. In many occupations that do not require a B.A., productivity nevertheless rises with academic skills. Furthermore, over time many sectors of the U.S. economy have been increasing their demand for professional, technical, and managerial workers, so historical relationships between industrial composition and occupational mix provide an unreliable basis for making forecasts.

Bishop proposes an alternative, regression-based model of the demand for different occupations. This model, which has been more accurate than the BLS methodology in the past, now projects that professional, technical, and managerial jobs will account for 60 percent of total job growth between 1990 and 2005. Although Bishop believes that college enrollments will respond to these work opportunities, he foresees a decline in the ratio of BAs awarded to total employment because relatively few individuals will be in the age group that typically attends college. Thus, government should intervene to improve access to colleges and universities. To this end, Bishop suggests raising high school standards, increasing student financial aid, making tuition tax-deductible, and halting tuition increases at public institutions.

Discussant Richard Murnane agrees with Bishop's prediction that college will continue to be a good investment in the years ahead. He explores possible reasons for the observed anxiety over the fate of college-educated workers, despite the considerable evidence that this group has done and will continue to do better relative to their less-educated counterparts. One observation relates to successive cohorts of BAs. Recent male college graduates are not earning any more, adjusted for inflation, than men of similar age and educational background 15 years ago. (The growing earnings premium of male college graduates results from a decline in the real earnings of high school graduates rather than any increase in their own earnings.) And because of real tuition increases, the recent BAs have a higher debt burden than their predecessors. Second, an increasing proportion of older male college graduates are now in low-paying jobs, compared to the situation in the late 1970s. This trend presumably relates to earnings patterns for workers who lose their jobs in corporate restructurings and other layoffs. Finally, Murnane indicates that highly educated workers often spend a couple of years in low-paying jobs immediately after graduation. The popular image of college graduates wasting their education working in coffee bars and the like is based largely on such temporary employment.

Murnane shares Bishop's concern about improving access to higher education, but differs somewhat in how to achieve this goal. He argues that states should allow tuition at public institutions to continue to rise, but they should allocate more funds to need-based financial aid.

Technology in the Workplace

A number of studies have argued that technological change has been raising the demand for highly educated and otherwise skilled workers, while lowering the demand for the less educated and the less skilled. Some of these studies do not measure technology directly but attribute unexplained trends in wage patterns to technological change. Others include only very general measures of technology, such as capital intensity or computer investments by industry. In contrast, Peter Cappelli's paper uses the responses to a national survey of about 3,000 employers to take a closer look at the influence of technology on skill requirements and relative wages for production workers and their supervisors. He uses the term "technology" to encompass changes in workplace organization, such as changes in how jobs are defined and who controls their content.

Cappelli reviews findings from a range of social sciences disciplines on how technology alters skill requirements. In line with the themes of these earlier studies, his regression analysis indicates that workplaces with more widespread computer usage, higher capital intensity, a research and development center, and a total quality management (TQM) program are more likely to require upgraded skills among production workers.

Cappelli then examines the links between technology and wage structures by comparing pay at establishments that make intensive use of the new technologies with others that do not. The introduction of computers and new work practices appears to increase inequality within occupations. At the same time, more advanced technologies tend to reduce inequality within establishments, as indicated by the relative pay of production workers and their supervisors.

Cappelli warns that these findings do not provide clear guidance on future wage or employment patterns for the economy as a whole. If employers have difficulty hiring skilled workers (or if wage requirements for these positions become too high), they may take compensating actions, such as redesigning jobs to have lower skill requirements. Because they lack knowledge of what measures employers are likely to take, government policymakers cannot know what interventions are needed to achieve societal earnings distribution goals.

In his discussion, John Bound questions the ability of largely cross-sectional studies of work establishments such as Cappelli's to shed light on trends in inequality. Despite isolated examples to the contrary, technological change undoubtedly has increased the demand for skills over the past two decades. This conclusion is based on indirect, but irrefutable, evidence that the relative utilization of skilled labor has been rising despite increases in its relative cost. Furthermore, a focus on the skills and wages of production workers ignores the decline in the fraction of the work force in production jobs. This shift in the composition of jobs is an important component of the skill upgrading that has been taking place.

Bound acknowledges, however, that Cappelli's examination of production workers results in some new insights. The paper is the first to find that computer usage and workplace innovations such as TQM raise the pay of these employees. Bound calls for a more detailed examination of exactly how these capital investments and human resources management practices affect skill requirements.

Influences of Labor Market Institutions

The trends toward a higher share of professional, technical, and managerial employees in the work force, and toward the adoption of new technologies, are common across a number of countries. Yet the United States has experienced a greater increase in earnings inequality, and a greater decline in the earnings of low-end workers, than other industrialized countries. Richard Freeman's paper explores the degree to which international differences are due to differences in labor market institutions, notably collective bargaining and the mandated minimum wage. He finds that the effects of institutional factors are at least as strong and arguably stronger than the effects of more commonly cited market factors.

A higher degree of inequality in the United States than in other countries is sometimes attributed to our greater degree of ethnic and cultural diversity. Freeman disputes this allegation by comparing workers of different backgrounds in the United States and Sweden. Americans of Swedish descent and Swedes of non-Nordic descent have earnings distributions similar to those prevailing in their country of residence - not their country of ancestry. Freeman concludes that cross-country differences are due to institutional factors, which he describes.

Freeman cites several studies showing that declining unionization has accounted for about one-fifth of the increase in U.S. earnings inequality over the past two decades. He notes the remarkable unanimity in the findings despite considerable diversity in methodologies, but argues that the studies underestimate the true impact to the degree that union wage settlements have spillover effects to nonunionized sectors. Freeman's own regressions, which explain earnings inequality in 12 OECD countries as a function of collective bargaining coverage and unionization, suggest that as much as one-half to two-thirds of U.S.-European differences in the degree of inequality can be traced to these institutions. Finally, Freeman finds that countries with the largest declines in collective bargaining coverage generally had the largest increases in inequality.

Freeman emphasizes that the cures for problems do not have to be linked directly to the causes. Thus, a more progressive fiscal policy or measures to increase the access of poor children to technology are solutions worth considering. But in light of the seriousness of the rise in inequality in the United States, Freeman believes that institutional interventions such as a rise in the minimum wage should not be ruled out. Expansions in collective bargaining coverage may have unique benefits, such as increasing the degree of worker influence in workplaces, which in turn has the potential to raise workplace efficiency.

In his comments, Peter Gottschalk agrees that institutions matter; one cannot hope to explain U.S.-European differences in inequality by market forces alone. But he calls for a more rigorous assessment of the relative role of institutions and markets in determining changes in inequality over time. Gottschalk finds that roughly one-half of the countries with centralized wage-setting institutions also experienced education- or age-related changes in labor supplies that are consistent with earnings patterns. By not including explicit tests of the role of markets in earnings inequality, Freeman's analyses offer only upper-bound estimates of the role of institutions.

Turning to policy, Gottschalk argues that regardless of whether labor market institutions were or were not an important cause of the increase in inequality, the United States could have done more to offset this trend. He urges consideration of an expanded earned income tax credit in addition to Freeman's proposals.

III. Policy Implications: A Panel Discussion

The panelists were asked to reflect on three broad issues:

* the nature of the "problem" posed by rising inequality and, in particular, its implications for economic growth;

* the urgency of the problem; and

* the appropriate policy actions to take.

Anita Summers observes that development of appropriate policies can occur only when consensus is reached that inequality has risen to unacceptable levels. In her view, inequality is currently too high because of its consequences for those at the bottom of the income distribution - the poor - and it is the poor on whom policy concern should focus. Comprehensive policy programs, not single-pronged strategies, are needed to address the problems of those at the bottom. Such programs must deal with the spatial concentration of the poor as well as with their low incomes per se.

Any consideration of policy must recognize the ongoing devolution of responsibility from the federal government to state and local governments, Summers points out. Increased emphasis on sub-federal governments' decision-making will complicate development or implementation of policies focused on labor markets and transportation, and especially any policies aimed at spatial dispersion of the poor, because they may cross the relevant jurisdictions' borders. For example, analysts are already concerned about interstate competition in pursuit of welfare reform. In the context of devolution, the biggest U.S. cities - in which the poor are highly concentrated - may lose out because they are outvoted in their states by suburban and rural interests.

Ann Schnare carries forward the theme of "people versus place," focusing on the housing market. After rising rapidly in preceding decades, home ownership rates have held steady in the last two decades as a result of income stagnation among the middle class. And income declines at the bottom have made well-maintained rental housing unaffordable for many, resulting in physical decay in those residential areas in which low-income renters are concentrated - central cities. This decay has, in turn, increased the impetus for those who have a choice to live elsewhere, further augmenting spatial segregation by income.

Discussing mortgage programs of Freddie Mac as well as housing policies of the Department of Housing and Urban Development (HUD), Schnare reports some discouraging developments. Special low-equity mortgage programs aimed at low-income potential home-buyers have a weak record because many poor individuals cannot maintain payments on their properties. For mainstream mortgage programs, by contrast, loans to low-income individuals appear to be no riskier than loans to middle-income borrowers, but neighborhood income exerts a strong independent effect: Loans on properties in distressed inner-city neighborhoods are noticeably riskier, regardless of the income of the borrower. By the same token, Schnare argues, HUD programs have become less successful as they have attempted to allocate their increasingly limited budgets to both providing housing assistance to very poor individuals and addressing the community development needs of distressed neighborhoods. She suggests that policies centered around vouchers and mixed-income developments would suffer less from, and indeed might help to reduce, the negative neighborhood externalities that plague current subsidized housing programs.

Frank Levy focuses on the potential responsiveness of labor supply - via education and human capital development - to the sizable shifts in labor demand that underlie much of the recent increase in U.S. earnings inequality. While he believes that in the long run rising inequality slows the nation's economic growth, in the shorter run the causation runs in the opposite direction. Restructuring in the manufacturing and services sectors has raised the education premium, and thereby increased inequality, by reducing the demand for semi-skilled labor much faster than these workers can acquire new skills. The policy challenges are twofold: In the short run, he argues, we need "emergency" policies to ease the impact of these economic adjustments on those hardest hit. In the long run, the supply of well-qualified workers must increase via education, particularly elementary and secondary schooling because it is difficult for mature workers to upgrade their educational attainment later.

Spatial issues enter the labor policy discussion because of the local character of public schooling in the United States. Schools should adjust their curricula and their teaching routines to provide a better fit between what they teach and what graduates will need when entering the labor market, Levy argues. The need for these adjustments is greatest in the communities that have historically prepared children for the jobs that are in decline. Moreover, parents in these communities have often been on the wrong side of the rising educational wage premium. Thus, because of increased geographic homogeneity by income, the communities needing the greatest school adjustments are generally those with the fewest resources to explore and finance them. Information about labor market opportunities and skill needs is critical: even as we keep upgrading educational standards, we also need to disseminate information to children and parents early enough to influence tracking and other educational decisions.

Larry Katz also focuses on the shrinking demand for disadvantaged workers, defined as those with limited education or skills, or from poor families or impoverished neighborhoods. Neither this labor demand shift, nor the reinforcing changes in other supply and demand factors that have accompanied it, show any signs of abating, he notes. While strong macroeconomic growth improves disadvantaged workers' prospects, the recent "twist" in the wage structure has been so massive that specific initiatives are needed to address the structural barriers preventing a speedy supply response. In addition, because the supply response takes time, the government may want to intervene in the short run on the demand side of the market (via wage subsidies, expansion of the earned income tax credit, or raising the minimum wage, for example) to ensure that work pays.

A number of lessons can be drawn from our experience with training and wage subsidy programs, Katz says. One is that early intervention pays off; once a student drops out, it is difficult to turn things around. And mature workers are much less likely than those still in school to continue their schooling. Thus, provision of mentoring and more general information to improve the matching of skills and demands, as well as the possibility of financial support for higher education, must be in place early enough to provide young students with a realistic expectation of realizing the payoff to further education. The returns to higher education are particularly large for those from disadvantaged backgrounds, suggesting that borrowing constraints are binding in some cases. Financial constraints also keep many poor families in public housing, where access to jobs is poor and where they feel their children's futures are threatened by crime and poor schooling. To overcome the particular difficulties of the poor who are spatially concentrated, the government should try "place-based people policies," which would target training and human resource funds on the areas with the greatest needs.

IV. Overall Themes and Policy Conclusions

A key issue that both motivated the symposium and animated much of the discussion is the question of what current trends imply about future patterns of earnings inequality in the United States. Virtually all participants agreed that the increase in inequality in recent decades is a source of concern, even where consensus is lacking about "appropriate" or "acceptable" levels of inequality. Thus, a critical question is whether the forces that caused the recent rise in inequality are likely to reverse themselves, or be offset by other equalizing forces, without policy intervention.

Space Matters

One basic conclusion that emerged from the conference is that "space matters." This is not a very encouraging message when evaluated in the context of future inequality trends. Access to jobs is critical to any individual's labor market success. And access has a geographic aspect - the length of the commute to employment opportunities - as well as an individual job preparation (education/skills) aspect.

In a generally suburbanizing economy, residents of core parts of many of the nation's largest cities have become increasingly isolated from the functioning of the broader economy. Neighborhood externalities cause this isolation to foster further isolation, both cultural and geographic, as employers make location decisions considering the availability of the existing work force, for example, and children pattern their behavior and aspirations on what they observe in their own neighborhoods. Transportation policy could help to overcome some spatial barriers to equal job access, but effective policies do not appear to be operating today. And transportation policies cannot offset the broader isolating forces operating in the markets for housing and schooling.

The Labor Supply "Problem"

Individuals have begun responding to the incentives offered by the rising educational premium by obtaining more education. But these investments take time, and for some students they may be limited by financial constraints. Therefore, demand for educated workers may continue to grow faster than supply for a considerable period. Policy interventions may be needed in the near term to enhance the supply adjustments.

Critical to the demand for educated workers is how technology influences skill requirements. In this regard, the symposium highlighted changes in the structure of work relationships, in addition to the investments in computers and research and development that are more typically thought of as "technology." Despite evidence that selected changes have served to equalize pay within workplaces, most participants appeared skeptical that technological advances or employer actions would reverse the trend to greater inequality.

Changes in the Policy Context

A recurring theme in the policy discussions was the prospect of a change in the U.S. system of federalism. Partly as a result of initiatives designed to encourage competition and innovation among sub-national jurisdictions, the federal government is devolving responsibility for policy development and implementation in a number of functional areas toward states and localities. As responsibility moves to smaller geographic entities, direct redistribution or indirect policies targeted at improving specific geographic areas become more difficult to implement. This difficulty happens virtually by definition, and also because increased residential sorting is likely to occur as revenue-raising and service provision are made more dependent on geographic location. Thus, existing federal policies that may have helped reduce inequality in the past cannot be expected to have greater redistributional effects in coming years. Future policy initiatives must be evaluated in light of this shift in the decision-making locus.

Future Directions for Policy and Research

Both because "space matters" and because labor supply and demand forces show no signs of turning the tide, the trend toward increased earnings inequality is very likely to continue. Most participants view government policy as having an important role to play in alleviating some of the perceived market failures that contribute to this trend. While the symposium did not develop a consensus regarding which policies should be adopted, it did raise some useful questions and - by spelling out where we are, how we got here, and where we seem to be headed - it provided a context in which to consider policy proposals.

One key question is the degree to which concern should focus on the bottom of the income distribution rather than on inequality across the entire income distribution. Broad policy interventions are not needed, for example, if market-responsive shifts in the supply of labor can be counted on to shrink the educational wage premium over the longer term. Some labor economists at the symposium indicated that a decrease in the relative pay of high earners would improve the international competitiveness of the U.S. economy. But a range of participants noted that market failures pose obstacles to those at the bottom. Thus, most of the policy interventions proposed at the symposium focused on raising incomes at the bottom of the income distribution.

Institutional arrangements and traditions have limited the extent to which U.S. governments intervene in labor markets. A few participants expressed support for a greater role for government or labor unions in setting pay, more supported at least stopgap measures that would help to ensure that "work pays" at the bottom, but most assigned highest priority to education, housing, or general fiscal measures.

Several participants recommended policies that would encourage residents of poor and minority inner city neighborhoods to move to the suburbs. As a decentralized, democratic society, the United States has not historically undertaken policies to disperse the residents of urban neighborhoods (and some government policies may even have helped to create these neighborhoods). Anti-discrimination statutes, however, have been enacted to provide more equal access to housing and credit markets. While mobility strategies patterned after the Gautreaux program in Chicago can provide some hope for a few residents of disadvantaged neighborhoods, such programs are likely to leave many remaining residents of poor neighborhoods facing even more severe neighborhood conditions (especially in the absence of alternative programs designed to encourage reverse migration by middle-class households). This discussion suggests an important role for policies designed to reduce the broader consequences of neighborhood isolation, such as government efforts to raise the quality of schools in poor districts or to directly improve neighborhood conditions or the skills of existing residents.

Education emerged as a critical arena for policy, especially since a significant governmental role already exists. Symposium participants generally appeared to view schools as the most appropriate place for providing students with information about workplace technologies and labor markets, in addition to teaching them traditional academic skills. Increases in residential sorting make the equalization of educational opportunity at the elementary and secondary level a difficult task. In addition, rising real student costs and declining real incomes have reduced access to higher education for students from lower and middle-income households. While addressing the causes and consequences of inequality is difficult enough for the generation currently in the labor market, the problems of inequality will become even more intractable if not addressed for future generations. Viewed in this context, redoubled efforts to assure all children access to good-quality schools makes sense as an early intervention strategy.

Katharine L. Bradbury, Yolanda K. Kodrzycki, and Christopher J. Mayer Vice President and Economist, Senior Economist, and Economist, respectively, at the Federal Reserve Bank of Boston. The authors are very grateful to Joan Poskanzer and Rebecca Hellerstein for their excellent editing of the symposium papers and panelists' remarks for publication.

COPYRIGHT 1996 Federal Reserve Bank of Boston
COPYRIGHT 2004 Gale Group