Two trends have helped transform the New York metropolitan labor market in recent decades. On the demand side, there has been a massive shift in the job structure from manufacturing to services. Between 1980 and 1998, employment in the metropolitan area increased by about 1.1 million workers, from 4.1 to 5.2 million (+27%), reflecting a gain of about 1.24 million (+43%) in the service sectors and a loss of nearly 287,000 (-33%) in manufacturing.  On the supply side, a similarly dramatic shift has occurred, away from a predominantly white native-born to an increasingly foreign-born workforce. Between 1980 and 1998, the foreign-born share of all employed workers rose from about 22% to 33%, balanced by nearly the same percentage point decline for native-born whites: from 56 percent to less than 47 percent. According to Census Bureau data, fully one-fifth of all workers with jobs in the New York area during 1998 arrived from a foreign country in the previous 15 years.  Since many recent immigrants are undocumented and less likely to be counted in the official statistics, these figures certainly understate the true growth in the foreign-born share of metropolitan employment.
    With sectoral and demographic shifts of these magnitudes, how have various native- and foreign-born workers fared? In this paper, we focus on three sets of questions related to recent changes in labor market status for native- and foreign-born workers differentiated by race, ethnicity and gender. First, we consider changes in the composition of employment across the metropolitan job structure. Has there been a general upgrading in the job quality mix – the ratio of high-wage, high skill jobs to those requiring low skills and paying very low wages? What has happened to the middle of the job structure? Which groups, distinguished by race/ethnicity, nativity, and gender, are increasing their shares of the best and worst jobs? Second, we address differences in educational attainment. With this standard measure of workforce preparation for good jobs, how do each of these demographic groups compare? Our third topic is relative earnings. What has happened to the buying power of each demographic group’s earnings over the past two decades? How have earnings differentials between these groups changed? Has overall earnings inequality risen, and if so, is it mainly due to the increasing presence of immigrants?
    Answers to these and related questions can help provide an indication of what employment restructuring, social networks, the educational system and increasingly unsheltered (more competitive) labor markets have meant for the labor market status of native- and foreign-born workers in the metropolitan area. This has obvious implications for social policies concerning education and workforce development, health and social service needs, and the strength of social networks and communities – all topics central to this volume.
    We focus on employed workers in 14 demographic groups: native-born white, black and Hispanic workers, and foreign-born whites, blacks, Hispanics and Asian workers, separately for males and females. We group individual level data from the Census (1980 and 1990) and from the Current Population Survey (Merged Outgoing Rotation Groups for 1997 and 1998) into 389 jobs, defined by occupations cross classified with industries, and further aggregate these into three job quality “tiers.” Our aims are limited to mapping changes in the labor market status of these demographic groups as measured by trends in employment and earnings across these job tiers. We do not attempt to explain employment and earnings outcomes, but some clues are provided by the kinds of jobs that are playing pivotal roles.
    We begin by describing our data and defining the demographic groups and job tiers. Section 2 addresses trends in the mix of employment across and within the job tiers. We then describe levels and changes in educational attainment for each of the 14 demographic groups in Section 3. Finally, Section 4 considers changes in real earnings by demographic group, changes in the wage gap between each group and native-born white workers, and changes in overall earnings inequality for the metropolitan area – with and without recent immigrants. We summarize and make concluding remarks in Section 5.

1. Data, Demographic Groups and Job Tiers
The decennial census is the best source of employment and earnings data at the city or metropolitan level, particularly when broken down by type of employment (occupation and industry) and demographic group (race, ethnicity, nativity, and gender). However, the most recent Census data available was for 1990. While data from the Current Population Survey was available through 1998, a consistent time series with CPS data from 1980 onwards was not possible since there is no identifier for immigrants or citizenship status in the survey before 1994. Our solution was to examine changes in employment and earnings trends over the past two decades with data from the 1980 and 1990 Public Use Microdata Samples (5%) of the U.S. Census and the merged outgoing rotation group files of the Current Population Survey for 1997/98.  This gives us 3 points in time for employment (1980, 1990 and 1997/98) and earnings (1979, 1989, and 1997/98).
    While care needs to be taken when Census and CPS data are used together, we are confident that the comparisons are meaningful at the levels of aggregation used in this paper. According to a recent study by the Community Service Society of New York (CSS), the demographic characteristics of New York City using the 1990 PUMS and the 1990 March file of the CPS were nearly identical (Hispanic persons were 23.5% of the population in the CPS and 23.6% in the PUMS; black non-Hispanic persons were 25.6% in both). The poverty rate was also quite similar (21% in the CPS and 20.4% in the PUMS). However, the CSS found substantial differences in personal and household income between the two data sources ranging from 7 to 23 percent across deciles (10th, 20th, and so on). Upon closer inspection it appears that this discrepancy was due almost entirely to differences in self-employment income, capital income (interest, dividends and rental) and social security income.  The measure we are concerned with, wage and salary income, was nearly identical.
    We relied on CPS outgoing rotation group data, both for sample size and because individuals are asked for their hourly and weekly earnings. Unfortunately, there is no directly comparable earnings measure with the Census, which asks only for annual earnings. Since there is no weeks worked variable in the CPS rotation group data, to compare earnings between the two sources, the annual figures from the Census must be converted to weekly (or hourly) earnings by dividing the annual figure by weeks (or weeks times hours) worked. Because of this difference, the magnitude of our estimates of changes in weekly earnings between 1989 and 1997/98 must be viewed with some caution. But our main concern is  less with absolute magnitudes of wage changes than with relative earnings across demographic groups, and we have no reason to believe that this measurement issue should affect one group more than another.
    We were able to achieve reasonably large sample sizes with CPS data (1997/98) 1) by using the outgoing rotation groups for 24 months (1997 and 1998); 2) by including workers who live in the metropolitan labor market rather than in New York City; 3) by limiting our focus to (14) broad demographic groups; and 4) by grouping detailed jobs (defined by occupation and industry) into just three job quality tiers.

Demographic Groups
    We limit our focus to employed wage and salary earners 18 to 64 years old in five metropolitan Primary Metropolitan Statistical Areas (PMSAs) in New Jersey and New York: Bergen-Passaic, Jersey City, Newark; Nassau-Suffolk, and New York (New York City, Westchester and Putnam counties). We classify these workers into 14 demographic groups. Separately for males and females these are: native-born non-Hispanic white; native-born non-Hispanic black; native-born Hispanic; foreign-born non-Hispanic white, foreign-born non-Hispanic black; foreign-born Hispanic, and foreign-born Asian.
    In 1998, there were approximately 1.7 million foreign-born workers employed in the New York metropolitan area. Among white foreign-born workers, most came from Russia (11.6%), Poland (11.3%), and Italy (8.7%). Black foreign-born workers came mainly from Jamaica (27.8%), Trinidad, Tobago and Guyana (19.8%), and Haiti (15%). For Hispanic foreign-born workers, the largest countries of origin were the Dominican Republic (24.8%), Ecuador and Colombia (22.7%), and Mexico (14.6%). Almost one-third of Asian foreign-born workers arrived from China (27.9%), followed by India (14.4%) and the Philippines (11.5%). Those born in Puerto Rico comprised about 30% of the Hispanic native-born category.
    Table 1 presents selected summary statistics of workers for 1997/8. The first two columns of Table 1 show the number of observations (employed individuals between 18 and 64 living in the five PMSA’s) and estimated numbers of employed workers for each demographic group (which is derived from weights on the observations). Most of the analysis presented below is conducted at the level of our three job tiers. For this purpose, even the smallest demographic group, foreign-born black workers (1,082 cases) seems adequate. For example, there were 549 cases for employed non-Hispanic foreign-born black women, of which 89 were in top tier jobs, 125 held middle tier jobs, and 335 held bottom tier jobs.
    The first row of Table 1 shows that by the mid-late 1990s, white native-born workers represented less than half (47%) of all employed workers in the metropolitan area. Black native-born and Hispanic foreign-born workers were the next largest groups, with 616,000 (12.1%) and 591,000 (11.6%) of the workforce respectively. About 47 percent of those employed in the metropolitan area were female, but this ranged from 55 percent for native-born blacks to about 42 percent for foreign-born white, Hispanic and Asian workers. Among the black foreign-born, approximately half, 51%, were female. There were also substantial differences in the share of young workers: those ages 18-24 ranged from just 6.6 percent of all employed for the white foreign-born to more than 21 percent for native-born Hispanics. Education differentials also vary greatly across both native- and foreign-born demographic groups. The share of employed workers with more than a high school degree was far higher for white native-born workers (69%) than for either native-born black (53%) or Hispanic (45%) workers. Similarly, among the foreign-born, Asian workers (61%) had far higher levels of educational attainment than the white (53%), black (43%) and Hispanic (29%) groups.

Job Quality Tiers
    One of our objectives is to get a better understanding of changes in the kinds of jobs in which immigrants are employed in the metropolitan area. Following earlier work by Howell (Howell and Mueller, 2000; Gittleman and Howell, 1995), we define a job as a detailed occupation cross-classified by industry. Using cluster analysis, Gittleman and Howell (1995) evaluated 621 jobs (94% of the nonagricultural workforce) on the basis of 17 measures of job quality and found that they could be grouped into three “job segments,” each with two component “job contours”.  Each of these contours employed between 11 and 21 percent of total employment, with each segment accounting for roughly one-third of the national workforce in 1980. Average annual earnings ranged from $17,400 in the highest contour (the Private Independent-Primary contour), where 86 percent worked full-time, to $4,700 in the bottom contour (the Low-Skill Service contour), where only 37 percent were full-time. The segments and contours conform nicely to the divisions in the labor market described in the labor market segmentation literature (e.g., Gordon, Edwards and Reich, 1982).
    In this paper, we restrict our focus to these three job segments and, in an effort to use more transparent terminology, refer to them here as “job quality tiers.”   Table 2 shows that, while most jobs (154) were located in the middle job tier in 1997/8, they accounted for only 24% of total employment. Reflecting several decades of a “declining middle” in the metro area (see below), by the late 1990s most employment was located in the top and bottom tiers (40 and 36 percent, respectively). At $850, the typical top tier job had median full-time weekly earnings that were more than twice as large as median bottom tier wages ($375), with the middle tier at $576. The share of employees with more than a high school degree also varied sharply across the tiers: just under 81 percent at the top, 53 percent in the middle, and just 34 percent for bottom tier jobs. The bottom row shows that by the mid-late 1990s, foreign-born workers comprised about one-fifth of top tier jobs, more than one-quarter of middle tier jobs, and almost half of all bottom tier jobs.

2. Immigrants and the Metropolitan Area Job Structure
    We address three questions in this section. We begin with an overview of recent employment changes: Has the distribution of employment across the 389 jobs for demographic groups converged or become more dissimilar? Second, we group jobs into three job tiers: What has happened to the employment mix for native- and foreign-born workers as measured by the distribution of share of their employment across the tiers? Third, we describe changes in the demographic composition of each of the job tiers: Which demographic groups are increasingly concentrated in the top, middle, and bottom tiers?

Employment Trends and Demographic Groups: Increasing Convergence?
    The metropolitan area has experienced strong growth in employment since 1980, with the number of employed workers (ages 18-64) increasing from slightly less than 4.1 to more than 5.1 million. Remarkably, Table 3 shows that three-quarters of this increase was accounted for by foreign-born workers (796,600). These aggregate trends in the employed workforce reflect sharply divergent patterns across our demographic groups. Asian and Hispanic immigrants show the greatest employment increases in both absolute and percentage terms (Table 3). Asian foreign-born workers tripled their numbers, due mainly to increases in the 1980s. In contrast to every other group, Hispanic immigrants grew far more rapidly in the 1990s, more than doubling their size between 1980 and 1997/8. While native-born white and black workers show large increases in the 1980s (150,000 and 65,000, respectively), both declined in number in the 1990s, leaving native-born blacks with negligible employment growth and a declining employment share (from almost 15% to 12%), only slightly higher than that of Hispanic immigrants (11.6%).
    How did these divergent trends affect the distribution of the demographic groups across the job structure? More specifically, between 1980 and 1997/8, did the mix of jobs held by foreign-born workers look increasingly more or less similar to that of native-born workers? To answer this question, we use a standard measure of segregation (the “dissimilarity index”) which provides a summary indicator of the extent to which two distributions are different.  The index ranges from zero (identical distributions) to 100 (complete segregation). The top panel of Table 4 shows that the native- and foreign-born job distributions are substantially more alike for men than women in 1980 (26.3 vs. 35.1). But the male index rises sharply, so that by 1997/8 the degree of dissimilarity between the native-born and foreign-born is similar by gender (33.5 for men compared to 36.3 for women).
    Among male workers, the table shows a large increase in dissimilarity over the two decades for each comparison except foreign-born Asian workers. The job mix of foreign-born Hispanic males was the most different from that of native-born whites and rose dramatically (from 44.8 in 1980 and 57.5 in 1997/8). Interestingly, both native-born Hispanic and foreign-born black workers were similarly unlike the native white distribution, both rising sharply from 39 to 51-54.
    Among female workers, the adjacent panel indicates that there are two levels of dissimilarity relative to white women – more dissimilar for foreign black, Hispanic and Asian workers and more alike for foreign-born white and native-born black and Hispanic women. Comparing the male and female panels, it is also worth noting that the 1997/98 index is nearly identical for both foreign-born Hispanic female and male workers (57.2 and 57.5) and foreign-born black women and men (55.6 and 53.7).
    These dissimilarity results show that for both men and women, the job mix for foreign white workers are most similar to that of native whites, while the employment distributions of foreign Hispanic workers are least similar. Second, in every case there is either stability or increases in dissimilarity – there is certainly no evidence of convergence to the white distributions. If the white male and female employment distributions are the standards by which the quality of a group’s employment mix is judged, these results suggest growing inequality in labor market status in the metropolitan area.

Changes in the Job Structure
    We can begin to assess the nature of recent changes in the quality of metropolitan area jobs by examining employment trends for each of the three job quality tiers. In 1980, the tiers were nearly identical in size: 1.3 million were employed in the top tier; 1.36 million in the middle tier; and 1.42 million in the bottom tier. By 1997/8, the distribution looks quite different. The top tier gained over 700,000 workers (to 2.05 million), the middle tier declined by more than 100,000 workers (to 1.24 million), and the bottom gained about 500,000 (to 1.88 million). Figure 1 documents changes in the composition of the three job quality tiers for native and foreign born workers. This figure highlights two key trends. First, measured by the distribution of employment both across and within the three job quality tiers, native-born workers experienced a strong upgrading in employment mix. Second, the employment distributions of both native- and foreign-born workers show a distinct “hollowing of the middle.”
    Figure 1 shows that the share of these workers employed in the best jobs increased from about 35% to almost 46%, while the proportions of these workers employed in both the bottom and middle tiers declined. The bottom panel indicates that the implications of recent employment shifts regarding job quality have been mixed for immigrant workers. While the distribution of foreign-born workers also shifted towards the top tier (an increase from 21% to 26%), immigrants also became slightly more concentrated in the bottom tier (from 50.7% to 53.5%).
    In the top tier, the job with the most rapid growth was one of the least well paid: sales supervisors in the retail and eating/drinking establishments industries. Employment in this job increased from 28,800 to 126,500 between 1980 and 1997/8, an increase of almost 98,000 workers. About one-third of this increase (32,000) was foreign born. The median weekly earnings for full-time prime age workers in these jobs were just $560 for native-born and $553 for foreign-born in 1997/8, which compares to the overall median for workers in top tier jobs of $850. For example, while both native- and the foreign-born workers saw rapid growth as managers n.e.c. (not elsewhere classified) in retail and eating and drinking establishments and health diagnosticians, other fast growing top tier jobs strongly advantaged native-born workers. For example, among the top five fastest growing jobs for native-born white workers were lawyers ($1,423) and managers n.e.c. in transportation, communications and public utilities (TCPU) ($991).
    Figure 1 clearly indicates a strong decline in the shares of both native- and foreign-born workers employed in middle tier jobs by the late 1990s. To be sure, some middle tier jobs grew strongly: information clerks (medical services and hospitals), nurses, health technicians, and writers and artists for women; truck drivers and electricians for men. But these gains were more than offset by stagnation or declines in good blue-collar manufacturing jobs (machine operators, in particular) and good office jobs (typists and financial record keepers in high paying service sectors: FIRE, TCPU, and Public Administration).
    Bottom tier jobs with the largest employment increases for both native- and foreign-born workers included cashiers and miscellaneous health service occupations, which paid about the same weekly wage to each group. While the wages paid to native- and foreign-born workers were similar in the same rapidly growing jobs, the mix of jobs held by each group in the bottom tier has become less similar, with higher quality jobs (measured by both educational attainment and earnings levels) going to native workers. For example, native workers experienced their largest gains as teachers (teacher aides, kindergarten and pre-k, and child care workers) and substantial declines in machine operators (light manufacturing – textile, apparel, wood, and leather). In contrast, immigrants increased most rapidly as cooks, bus/taxi drivers, and household workers.

Employment Growth Within Job Tiers-Male Workers
    Figure 2a shows that top tier employment increased for every male demographic group and most notably for native whites and foreign-born Asians. The middle tier shows just the reverse – modest or declining male employment across the board. The greatest range in employment trends appears in the bottom tier with a small decline for native-born white men, even smaller declines for foreign-born whites and native-born blacks and native Hispanics, but large bottom tier gains for foreign-born black and Asian workers and exceptionally large gains for Hispanic immigrants.
    As noted above, the fastest growing job for men in the top tier was sales supervisors. A relatively low paying job, its stature varies widely across demographic groups. Among men, white native born workers received median full-time weekly wages of $673 in 1997/8, which compares to $700 for Asian foreign-born and $634 for black foreign-born, but black native-born and foreign-born Hispanic workers received only $519 and $475 respectively.  For native-born white workers, the second fastest growing job was lawyers ($1,538). For foreign-born Asian men, it was sales supervisors in wholesale trade sector, which coincidentally paid the same ($1,538). On the other hand, for native-born black and Hispanic men, the second largest growing job was protective services (exc guards), which at $840 and $865, paid well below the mean paid to all top tier men in 1997/8, $923. These results suggest that because of both differences in pay for the same job title and differences in the kinds of jobs showing large employment growth, the quality of employment in the top tier has shifted in favor of white native-born and Asian foreign-born workers.
    The stagnation of male middle tier employment was accounted for in large part by sharp declines in native-born white (-17%) and black (-28%) workers, offset in part by foreign-born gains. The job with the greatest decline for native white men was financial records processors in the heavy manufacturing sectors. Black native men experienced their largest declines in three relatively high paying manufacturing jobs: precision workers, welders and assemblers. Perhaps the most striking development in the middle tier has been the replacement of native- with foreign-born truck drivers. White and black native born truck drivers in the wholesale sector declined by half between 1980 and 1997/8, from about 11,200 to just over 6,000, and were almost exactly replaced by Hispanic and Asian immigrants (+5,700). The largest absolute growth in employment in middle tier jobs for black, Hispanic, and Asian immigrants came as truck drivers in TCPU and retail trade.
    In the bottom tier, this same divergence between declining native and growing foreign-born presence is even more evident. In addition, shifts in the composition of jobs suggests declining median wages and increasing differences across demographic groups.  Foreign-born Hispanic men experienced by far the largest numerical growth in bottom tier jobs, from 72,000 in 1980 to almost 220,000 in 1997/8. Foreign-born black and Asian men also increased their numbers substantially here (56,000 and 57,000). Native whites, foreign-born Hispanics and Asians grew fastest as cooks, with average weekly earnings of $292, $300, and $347, respectively (which compares to a $400 median weekly wage for all bottom tier men). But while the next largest growth jobs for white native men were carpenters ($382 per week), for Hispanic immigrants it was miscellaneous food occupations ($260). Black foreign-born workers increased most dramatically as bus and taxi drivers ($480). While native black men declined slightly in the bottom tier, the jobs in which they grew most rapidly here paid quite poorly: miscellaneous health service occupations ($250), cashiers ($140) and materials/scheduling clerks (medical services and hospitals) ($320). In contrast, the jobs in which native black men declined most were in two relatively well paid (for native-born workers at least) machine operator classifications in the textile, lumber, furniture, printing and miscellaneous manufacturing sectors.

    Female Workers
    Figure 2b shows a similar pattern for female workers: strong growth in the top tier for native-born whites; strong growth in the bottom tier for foreign-born Hispanics. The percentage increases in top tier employment were greater for female than male workers in every demographic group, particularly for white native-born women, but they were driven by some of the same jobs (sales supervisors and sales reps.). In addition, fast growing top tier jobs for white native women included administrators in the health sectors and specialized teachers (librarians, counselors). Whereas male employment in the middle tier was stagnant (an increase of about 7,000 positions), female employment collapsed (a decline of 127,000 jobs), caused by an extraordinary decline in employment for native-born white women (about 159,000) and driven by large reductions in typists and financial records processors.
As Figure 2b shows, the most striking changes in bottom tier employment for women were the increases in the number of black and Hispanic foreign-born workers. The increase for black immigrants is almost entirely comprised of miscellaneous health services positions (in the medical services and hospitals sectors), an increase of 32,300 (from about 14,000 to 46,300). For female Hispanic immigrants, the growing concentration in bottom tier jobs is due to growth in three jobs: misc. health services (+14,700), private household workers (+13,900), and cashiers (+11,900), all extremely low paid jobs. While white native-born women were paid an average of $400 in bottom tier jobs in 1997/8, the misc. health services positions paid $350 to black immigrants and $280 to Hispanic immigrants. The other jobs in which foreign-born Hispanic women grew strongly paid average weekly wages of $168 (cashiers) and $200 (household workers).
    The employment trends that appear in Figures 2a and 2b provide some insight into which demographic groups are driving the results shown in Figure 1. The “hollowing of the middle” is a consequence of a massive shift in native-born white male and female employment from middle to the top tier jobs, while foreign-born Hispanic men and women have led the increase in bottom tier employment.

Distribution of Demographic Groups Across the Job Tiers
    These employment trends have also changed the distribution of employment across the job tiers for each of the demographic groups. Table 5 shows substantial increases in the proportion of native-born demographic groups – male and female – employed in top tier jobs.  For native-born white and black workers, this reflected sharp declines in the middle tier. In contrast, the declining shares of foreign-born black and Hispanic men in middle tier jobs were balanced by an increasing concentration at the bottom. Between 1980 and 1997/8, the share of employed black immigrant men in bottom tier jobs rose from 45.4 to 52.1 percent, while the share of all employed foreign-born Hispanic men in bottom tier jobs increased from 55.3 to 64.8 percent. On the bright side, native Hispanic men show a pronounced decrease in concentration at the bottom (from about 52.9% to 45.6%). For black native men, almost all of the improvement at the top (23.2% to 34%) was compensated for by declines in the middle (32.6% to 23.9%).
    For female workers, Table 5 reports that employment shifts ended up looking much like those for men: small increases in female concentration in the worst jobs and larger declines in the middle and increases at the top. What is distinctive is the uniformity: all female demographic groups improved their position in the top tier and declined in the middle. Substantial improvement at the bottom was limited to native Hispanic women (like their male counterparts) – due largely to their reduced numbers as machine operators in light manufacturing (a decline of over 8,000 positions). The only group showing a substantial worsening in employment mix was black immigrants, whose concentration in the lowest tier increased from 55.5% in 1980 to 61.1% in 1997/8 (although there is a curious large dip to 47.5% in 1990). As noted above, this growth was due mainly to the increase in health sector employment.
    Measured by shifts towards the top tier, these employment trends have unequivocally improved the composition of employment for native-born male and female workers in all three groups (white, black and Hispanic), but the evidence for foreign-born workers is mixed. Table 5 shows that foreign-born black and Hispanic men became increasingly concentrated in bottom tier jobs, while all foreign-born female groups show high concentrations in these jobs and little if any improvement over time.
    These trends in the distribution of the demographic groups across the job structure led to big changes in the demographic composition of the tiers. While native-born white and black workers became much more concentrated in the top tier, Table 6 shows that their share of top tier employment declined. Indeed, as the top row shows, the native-born white share of employment fell by about 10 percentage points in all three tiers – from 71.7% to 62.3% in the top tier, 59.7% to 48.4% in the middle, and 39.3 to 29.5% in the bottom tier. But the reverse characterized foreign-born black and Hispanic workers in the bottom tier: as they became more concentrated there, their share of bottom tier employment exploded.  Foreign-born black and Hispanic workers increased their shares of bottom tier employment from 5.6% and 10.6% in 1979 to 10.4% and 21.1% in 1997/8. However, despite growing representation of foreign-born black and Hispanic women and stable shares of  their male counterparts in the top tier, black and Hispanic immigrants held just  3% and 4.9% of top tier jobs in the late 1990's (up from 1.7% and 2.6% in 1979.
These employment trends can be summed up in four main findings. First, the employment distributions tended to diverge over the period. In particular, foreign-born black and Hispanic workers (as well as native-born black workers) held jobs that looked increasingly unlike those held by native-born whites. Second the job structure shifted away from the middle towards top and bottom tier employment Third, native-born males and both native- and foreign-born females have become much more concentrated in top tier jobs, while foreign-born black and Hispanic men and foreign-born black women increased their concentration in the bottom tier. And fourth, as a result of these trends, foreign-born black, Hispanic and Asian workers doubled their share of all bottom tier jobs, from less than 20% to almost 40%.

3. The Educational Attainment of Immigrant and Native-born Workers
    Although a crude indicator of work-related skill, a high school degree has become a necessary credential, effectively precluding all but the lowest paying, manual jobs in the bottom.  But there are several problems with using educational attainment as a measure of workplace relevant skill. It is a credential whose substantive content varies within schools (among students), across schools (standards differ), and over time (reflecting changes in norms as to what is a minimally acceptable level of attainment). Thus, unlike test scores, which have remained stable, national data show that the share of those with more than a high school degree has increased sharply (Mishel, Bernstein and Schmitt, 1998). This certainly characterizes the New York metropolitan area employed workforce as well: those working with more than a high school degree increased from about 1.6 million to nearly 3 million (from 39 to 57%) between 1980 and 1997/8.
    However, there remain almost as many with a high school degree or less employed at the end of the 1990s as in 1980: 2.21 million compared to 2.49 million. In the bottom tier, the number of these poorly educated workers has actually increased, from 1.16 million to 1.24 million. As Table 2 indicates, in 1997/8 only about one-third of workers in bottom tier jobs had achieved more than a high school degree. The only substantial change in the number of those with low educational attainment appears in the middle tier, where these workers declined from 907,500 in 1980 to 400,000 in 1997/8.
    Not surprisingly, immigrants represent a rapidly increasing share of those employed with low education. Among workers in the New York metro area with a high school degree or less, one quarter were foreign-born in 1980. This share increased to more than two fifths (43%) by 1997/8. This increase was accounted for mainly by the Hispanic foreign-born. While 48% of all workers with a high school degree or less were native-born whites in 1980, this fell to 34% by the late 1990s. In contrast, foreign-born Hispanic workers increased their share of the poorly educated from 8 to 19%.
    However, while those with low education levels are increasingly foreign-born, educational attainment levels vary widely across demographic groups and there is no overall tendency for the foreign-born groups to have lower levels of attainment. As Figure 3 indicates, in 1997/8 the white native-born and Asian foreign-born groups had the lowest shares of low education workers (31% and 39%), followed by white foreign-born and black native-born (both 47%), black foreign-born and Hispanic native-born (57% and 55%). Among the employed, Hispanic foreign-born workers were by far the least educated – 71% had no more than a high school degree in 1997/8.
On the other hand, within each race/ethnicity group, foreign-born status appears to matter a great deal. Among white workers, 31% of the native-born had only a high school degree in 1997/8, but 47% of the foreign-born had just this level of education. Among black workers the differential was 10 percentage points (47% vs. 57%) in 1997/8, but it should be noted that there was little or no gap in 1980 and 1990. The gap among Hispanic workers was even larger, 16 points (55% vs. 71%), and again, this is a recent development (in 1980 native-born Hispanics had the higher low education share!). In sum, in the New York metropolitan area, foreign-born demographic groups have lower educational attainment only within each race/ethnicity group, although for blacks and Hispanics the gaps developed after 1980.
    Figure 3 also shows that the substantial educational upgrading that characterizes most groups does not appear for either foreign-born Hispanic or Asian groups with large immigration into the region to staff bottom tier jobs. It is instructive to compare the results for white native-born and Asian foreign-born workers in Figure 1. While these two groups show the smallest shares with low education levels, their trends over the last two decades have been quite different. There has been a sharp reduction in the share of white native-born workers with low education levels, from 52 to 31%, as poorly educated whites exited the middle tier of the labor market (from 528,600 in these jobs in 1980 to 259,500 in 1997/8) at the same time that college educated native-born whites increased their numbers in the top tier (from 508,000 to 806,400). Asian foreign-born workers, in contrast, show the lowest share with poor education levels in 1980 at 45.7%, but this falls only modestly to 39.5% in 1997/8. The reason is that the impressive rise in those with college degrees or more in the top tier by Asian workers (from 46,800 to 154,500) was in large part offset by the increase in poorly educated workers in the middle and bottom tiers (from 44,900 to 116,800).
A similar tale can be told for African-Americans and Hispanic immigrants. Although their 1980 figures are not greatly different, African-Americans (black native-born) show an extraordinary decline in the share with a high school degree or less, from 72 to 47%, while foreign-born Hispanics show only a small improvement, from 79.4% to 71.5% over these 18 years. The educational upgrading of native-born blacks reflects declining numbers employed in middle and bottom tiers and a remarkable increase of those with at least a college degree in the top tier: from 17,700 in 1980 to 92,400 in 1997/8. These data indicate that, among those African-Americans with jobs, there was a substantial upgrading in their labor market status over this period (from the middle to top job tier). The continued high share of Hispanic immigrants with low education is a reflection of the draw of bottom tier jobs: Hispanic foreign-born workers with a high school degree or less and employed in the bottom tier rose from 133,800 in 1980 to 327,500 workers in 1997/8.
    These results indicate a clear widening of the educational attainment gap between native- and foreign-born workers. This finding probably reflects the retirement or relocation out of the metropolitan labor market of low-skill native-born white and black workers at the same time that large numbers of less educated black, Asian and Hispanic immigrants have entered it.
    Even workers employed in the worst jobs have seen dramatic increases in educational attainment. Figure 4 shows that the share of bottom tier workers with more than a high school degree increased dramatically for all seven demographic groups over these two decades. The greatest percentage increases were for native-born Hispanic workers (from 8.3% to almost 30%), followed by foreign-born whites (12.9% to 33.2%). The share of more highly educated workers more than doubled for both native- and foreign-born blacks, and more than half of all native white workers in bottom tier jobs had more than a high school degree by 1997/8, an increase of almost 25 percentage points from 1980. Part of the reason for this improvement for native-born whites appears to be the shift towards jobs requiring higher education among female workers, for example teachers (Kindergarten and Pre-K) and Teachers Aides. These gains in educational attainment appear across bottom tier jobs for workers in each of the demographic groups. The same low skill jobs (e.g., cooks, cashiers, food service, health care aides and assistants, bus and taxi drivers, garbage collectors and materials handlers) are being filled by workers with much higher educational attainment than they had in 1980. But as the next section will show, these additional years of schooling have not yielded higher real earnings.

4. Earnings Patterns
    We have seen that native-born white workers have experienced relatively large improvements in job quality (employment shifts towards the top job tier and away from the bottom) and educational attainment. At the same time, we have found relatively slow growth in educational attainment for Hispanic and Asian immigrants and the increasing concentration of the foreign-born in bottom tier jobs. In this section, we compare earnings trends for each of our demographic groups across the three job tiers and three education levels for the 1979-1997/8 period. We conclude the section with a discussion of several indicators of trends in overall earnings inequality.

Male Earnings by Job Tier
    Figure 5a presents real earnings trends by demographic group for each job tier.  It documents a striking collapse in real median weekly earnings for those most attached to the labor market – prime-age, full-time workers. We underscore the caution made earlier: the large declines may reflect in part the comparison of adjusted annual earnings figures from the census (1980) with the reported weekly earnings in the Current Population Survey (1997/8). Our concern is with the pattern of wage declines across the demographic groups in each of the three jobs tiers, and the identification of the jobs that appear to be driving these declines.
    While the earnings declines were clearly least bad in the top tier, even here black men as a group experienced substantial declines (-16% for native black men, -8% for foreign-born black men). The best performance at the top of the job structure was achieved by foreign-born Hispanic and Asian workers, reflecting both the relatively low earnings for Hispanic men in top tier jobs in 1979 and the high skills of recent Asian immigrants employed in top tier jobs.
    Among men in middle tier jobs, the largest wage declines were experienced by foreign-born whites (-29%), native blacks (-17%), and foreign-born blacks (-28%). White immigrants show large wage declines in a number of middle tier jobs in which they are heavily concentrated: truck drivers and vehicle equipment mechanics in TCPU and wholesale trade, and precision workers and electrical equipment repairers in manufacturing. Native-born wage declines in the middle tier appear due in large part to their concentration in the health and TCPU sectors. Real weekly earnings for full-time prime-age native black men in three health technician positions fell by 58%, 23% and 31% over the two decades. Similarly, in the TCPU sectors, earnings declined in all the jobs in which they were most concentrated: information clerks (-50%), material scheduling clerks (-10%), vehicle equipment repairers (-30%), and most significantly, truck drivers (-23%).  The sources of the middle tier wage collapse for foreign-born black men (Figure 5a) are broadly similar.
    As for the bottom tier, Figure 5a shows that all but Asian men experienced large wage declines, ranging from 14 to 25 percent. Unlike the middle tier, the wage declines for each demographic group reflect collapsing wages in the fastest growing bottom tier jobs.  Foreign-born white workers, for example, grew fastest in the two jobs in which they were most concentrated and saw large wage declines in each: bus/taxi drivers (-41%) and sales, hardware and parts (-49%). Other construction trades in which white immigrants have a strong presence and lost considerable ground in pay were masons and tilesetters, carpenters, and painters/plasterers. Hispanic immigrants show the same concentration in jobs with the greatest wage declines. The five bottom tier jobs with the greatest increase in foreign-born Hispanic workers accounted for one-third (33%) of all their bottom tier employment and four of the five show large real weekly wage declines: cooks (-18%), misc food occupations (-28%), garbage collectors and materials handlers (-31%), and painters/plasterers (-34%).
    Not only did native black men increase their employment most in jobs in which they experienced huge declines in real weekly earnings, but these jobs were also among those with the lowest pay. They increased most rapidly (again, in absolute terms) in miscellaneous health occupations even as real earnings fell by 55 percent, leaving the median weekly pay for full-time prime-age black men at just $240 in 1997/8 (which compares to an overall bottom tier median of $394). The bottom tier job with the next greatest employment gains by black men over these two decades was cashiers, where they saw a 30% decline and a 1997/8 wage of $312.
    In sum, for men, white native-born and Asian foreign-born workers show by far the best wage performance over these two decades. The wage collapse was greatest for native-born black and foreign-born white, black and Hispanic men. Two trends appear to play central roles. First, there were large declines in real earnings in formerly “good” middle tier sectors (manufacturing, construction, TCPU). Second, in the bottom tier these four groups experienced large cuts in real pay for jobs in which they were increasingly concentrated. At least in the bottom tier, the earnings collapse has been the result of sharply falling pay for the fastest growing low-wage jobs.

Female Earnings by Job Tier
    Among female workers, the earnings trends shown in Figure 5b are more diverse, ranging from generally positive in the top tier to quite negative in the bottom tier, although even in the latter the declines were not nearly as large as experienced by their male counterparts. A couple of important trends stand out. First, native-born white women do relatively well – far better than foreign-born white women. The reverse is true for black women: native-born black women did relatively poorly – and much worse than foreign-born black women – in each job tier. Finally, foreign-born Hispanic women show large real wage gains at the top (out performing native Hispanic women), but report the greatest wage declines in both the middle and bottom tiers. What happened at the jobs level that can help account for these differences? We will focus on the women with the poorest earnings outcomes – foreign-born whites, native-born blacks, and foreign-born Hispanics in the middle and bottom tiers.
    As Figures 5a and 5b indicate, native-born white women in top tier jobs enjoyed by far the greatest increase in real earnings over these two decades. The jobs in which they showed the most rapid gains in employment showed mixed changes in real earnings: sales supervisors in retail (-9%), sales supervisors, other sectors (+7%); sales representatives, FIRE (+34%), health sector administrators (-18%), and teacher specialists (e.g., librarians, counselors) (+10. Perhaps more than others, native-born black women increased their top tier employment most rapidly in jobs with large real earnings losses. This not only contributed to the fall in real earnings shown in figure 5a, but to extremely low 1997/8 earnings levels as well. For example, earnings of native black female adjusters and investigators in the FIRE sectors declined by 35% to $480. Similar patterns appear for sales representatives in FIRE (-15%, $360) and protective service workers in public administration (-20%, $600). The median weekly wage for full time prime age women in top tier jobs was $759.
In the middle tier, foreign-born white, foreign-born Hispanic and native-born black women all show similar declines in real weekly earnings. The sources of the white and Hispanic results are difficult to identify – there is no set of jobs that appear to contribute greatly to the declines. But in the case of native black women, the data do reveal a story. Much of the earnings problem is located in the health care sectors. Over 11 percent of black women working in middle tier jobs were employed as nurses, physician assistants, dieticians and therapists, and another 5.3 percent were health technicians. Their weekly earnings in these two jobs declined by 12 and 14 percent, respectively.  Another 11 percent of middle tier native-born black women were employed as typists in three formerly high paying sectors – FIRE (where earnings declined by 6%); medical services and hospitals (-26%); and public administration (-12%).
    In the bottom tier, the same three female demographic groups also show the largest earnings declines. Two jobs – miscellaneous health service positions (nurse aides) in the medical services and hospitals sector and cashiers (all sectors) – go a long way towards accounting for these declines. For example, over 12 percent of foreign-born white women employed in the bottom tier worked in these low skill health services positions and had weekly earnings declines of 29 percent. The same held true for foreign-born Hispanic and native-born black women.
    In sum, foreign-born white, foreign-born Hispanic, and native-born black female workers  experienced large earnings declines in both the middle and bottom tiers. Most notably, native-born black women saw large earnings declines in moderately skilled clerical and technical positions in sectors that used to pay fairly well (health care; finance, insurance and real estate; and public administration). In the bottom tier, foreign-born white, native-born black, and foreign-born Hispanic women also had large earnings declines, particularly in the health care and retail sectors, where they are heavily concentrated.
    Taking male and female workers together, a key consequence of these earnings trends has been the substantial increase in the earnings advantage of native-born workers within each job tier. The ratio of native-born to foreign-born earnings in the top job tier (median earnings for full-time prime age (25-54) workers) grew from 11% in 1979 to 14% in 1997/8. The native-born advantage and its growth was even greater in the middle job tier, where the +differential rose from just 4% to 15%, and in the bottom job tier, where it shot up from 15.5% to 24.3%.

Earnings Trends by Education Level
    An alternative way to view earnings trends for each of our demographic groups is to group the data by level of educational attainment. Table 6 shows that most demographic groups in most education categories experienced declining real earnings between 1979 and 1997/8. The last row shows that only female workers with at least a college degree saw positive real earnings (+9.6%), whereas male college graduates who experienced an earnings decline of 3.5%. The positive trend for the most highly educated female workers was a reflection of gains for native-born whites (+23%), foreign-born Asians (+13%), foreign-born whites (+8%), and foreign-born Hispanics (4.4%). Foreign-born women did substantially better than their male counterparts. For example, while foreign-born white male workers with a college degree saw an earnings loss of 14%, their similarly educated white female immigrants gained over 8%.  Among college educated native-born blacks, women show a decline of almost 4% but men lost over 14%.
    Perhaps the most troubling result is the poor earnings performance of black male and female workers, whether foreign- or native-born, both at each education level and in each job tier.  For example, men with no more than a high school degree, native- and foreign-born blacks show the greatest earnings declines (-23% and –25.8%). Among the most educated male workers, it was again native- and foreign-born black men who fared worst (-14%, -11%), and for female workers with at least a college degree, native- and foreign-born blacks are the only groups exhibiting declining wages (-3.9% and –6%).

Race/Ethnicity Earnings Differentials
    The earnings trends described in Figures 5a-5b and Table 7 show that, with one exception, white native-born men and women have experienced the highest earnings growth over these last two decades (by these measures, Asian men have done at least as well as white native men). We now turn to a more direct measure of the relative performance of our demographic groups. Table 8 presents demographic group earnings differentials, measured as the ratio of the median weekly earnings for full-time prime-age workers between each demographic group and native white workers, calculated separately for men and women, for 1979, 1989 and 1997/8. For each comparison, we present three statistics: the earnings ratio for all workers, for just those for workers with a high school degree or less, and for these low education workers employed in the bottom job tier.
    Among male full-time prime-age workers, the race/ethnicity wage gaps were concentrated in the 60-70 percent range over the course of these two decades. White native-born workers tended to have a wage advantage of 30-40 percentage points, ranging from a high of 54 points over foreign-born Hispanics in 1997/8 (.46) to 14 points over foreign-born white men in 1979 (.86). The overall native white wage advantage remained stable over both native-born black (.65 - .69) and native-born Hispanic (.60) workers, despite large increases in the white advantage among the least educated  (the 2nd row for each group). But the third row shows that among those with just a high school degree or less, the growing disadvantage for native black and Hispanic workers was not located in bottom tier jobs (where the wage gap was stable), a conclusion consistent with Figure 5a.
Foreign-born Asian men also show stability in their earnings relative to white men, and among the least educated in the bottom job tier, Asian men actually narrow the gap substantially (from .52 to .67). In contrast, three foreign-born groups show large increases in the wage gap: foreign-born whites (.86 to .71); foreign-born black men (.63 to .55); and foreign-born Hispanic men (.60 to .46).
     For male workers, then, two conclusions stand out. First, while the earnings advantage of native white men over other native-born men was stable, it grew substantially over foreign-born men (with the exception of Asians). Second, there is no evidence of a growing wage gap among poorly educated men in the bottom tier for any demographic group.
    Among female workers, the wage gap grows sharply for all demographic groups except native Hispanics. Table 8 shows that the relative earnings of black women fell substantially for both the native and foreign-born: a racial wage gap of 15% for black native-born women grew to 29% in 1997/8; for black foreign-born women this differential rose from 19% to 33%. Among just those with the least educational attainment, both native- and foreign-born black women narrowed the wage gap with native white women. This suggests that for those with more than a high school education, white women increased their earnings advantage over black women. Certainly this is consistent with Figure 5b, which shows that in top tier jobs (which almost always require more education) native-born white women had 15% real wage increases over these two decades, compared to -6% and +2% for native- and foreign-born black women. Despite a much larger initial wage differential for foreign-born Hispanic women (38%), the native white advantage grew to 52% by 1997/8. Again, this increasing earnings inequality does not appear to be due mainly to growing differences among women with the lowest educational attainment (row 2). This also holds true for the growing gap between Asian and native-born white women. The higher earnings of white women with at least some college appears to be generating growing racial/ethnic wage gap with nearly every female demographic group.

Overall Earnings Inequality
    Have earnings trends generated growing earnings inequality across the metropolitan area? If so, is it generated mainly by the influx of recent immigrants? A widely used measure of inequality is the ratio of the median earnings at the 90th (high earners) to the 10th percentile of the earnings distribution. We can then look at the 50/10 and 90/50 ratios to determine where in the distribution changes are taking place. Table 9 presents these ratios for prime-age full-time workers (top panel) and all employed workers (ages 18-64) for both male and female workers.  We also calculate these ratios without recent immigrants (those arriving in the previous 15 years) to get some idea of what these newcomers contribute to recent  inequality trends.
    Panel A of Table 9 shows a huge increase in overall earnings inequality (the 90/10 ratio) in the 1980's for both male and female workers. While the 90th percentile male worker earned less than 4 times that of the 10th percentile worker in 1979, this differential grew to a multiple of almost 5.4. Although lower in level, the magnitude of the increase for female workers was similar: from just under 3.4 to about 4.2. But while inequality continued to grow in the 1990's for women (to almost 4.6), it declined slightly for men (to about 5.1). The narrowing for male workers in the 1990's was due to small inequality declines at both the top (the 90/50) and the bottom (the 50/10). Similarly, the inequality increase for female workers appears both at the top and bottom. The key point is that full-time prime age earnings inequality was far greater in 1997/8 than in 1979 for both men (5.13 vs. 3.94) and women (4.56 vs. 3.35).
    We have seen that immigrant workers have become increasingly prominent, with nearly half of all bottom tier jobs by the late 1990's and generally declining earnings relative to white native born workers. It seems reasonable, then, to suppose that much of the growth in earnings inequality is attributable to the recent large flows of immigrants into the local labor market. Columns 2 and 3 show that, for full-time prime-age workers, the inclusion of recent immigrants increases inequality (90/10) slightly in 1979 but more substantially in 1989 and 1997/8 for both male and female workers. Excluding recent immigrants in 1997/8 reduces male inequality from 5.13 to 4.86. The effect for female workers is even greater; by excluding immigrants, inequality drops from 4.56 to 4.08. On the other hand, even without recent immigrants the data show extraordinary increases in earnings inequality for such a short period – from 3.87 to 4.86 for native-born men and from 3.25 to 4.08 for native-born women. So while these results confirm that recent inflows of foreign-born workers are an essential part of the story of growing earnings inequality in the metropolitan area, it is certainly not the case that rising immigration can directly account for most of it.
    Panel B of Table 9 presents the same calculations for all workers (including the younger, older, and part-time workers excluded in Panel A). Here we also find a large growth in earnings inequality both with and without recent immigrants. We also observe much higher levels of inequality and inequality growth for both male and female workers in the 1990s. The important point is that using this much more inclusive population of workers produces the same result: large increases in earnings inequality whether or not recent immigrants are included.
    We can check this result by using an entirely different measure of inequality – one based on educational attainment. Table 10 confirms that for full-time prime-age workers there has been a substantial growth in wage inequality for both for native- and foreign-born male and female workers. The ratio of college to high school median weekly earnings rose from 1.54 to 1.76 for native-born men and from 1.63 to 2.07 for foreign-born men. Interestingly, the growth in this measure of inequality was even greater for women. Among the native-born the female ratio was nearly identical to the male ratio in 1979, but rose to 1.9 (compared to 1.76 for native-born men). Similarly, by 1997/8, the college/high school wage differential grew to 2.22 for foreign-born women, compared to 2.07 for their male counterparts.

5. Conclusions
    The U.S. labor market has been distinguished recently by massive increases in earnings inequality and by an equally exceptional collapse in the real earnings of the least skilled (Gottschalk, 1997). At the same time, the New York metropolitan labor market experienced a large-scale shift from manufacturing to service sector employment and a huge inflow of new immigrant workers. In this context of rising national inequality and local industrial and demographic transformation, how have native- and foreign-born groups fared in the New York metropolitan labor market? We address this question by focusing on recent trends in the employment, educational attainment and earnings of immigrants and native-born workers with jobs in the New York metropolitan area.
    The three job tiers were roughly similar in size in 1980, but by the end of the 1990’s middle tier employment had collapsed – its share of total employment fell from 27.4% to 21.6% for men and from 39.8% to 26.4% for women. This “hollowing of the middle” was a consequence of a substantial shift in native-born male and female workers (mainly white) from middle to top tier jobs as large numbers of foreign-born workers entered bottom tier jobs.
    These employment shifts across tiers, compounded by changes in job composition within them, suggest a growing gap in the quality of jobs held by native- and foreign-born workers. While the composition of employment has unequivocally improved for native-born male and female workers in all three groups (white, black and Hispanic), the evidence for foreign-born workers is mixed. Foreign-born black and Hispanic men showed increasing concentration in bottom tier jobs, while all foreign-born female groups maintained high levels of concentration in the bottom tier. By the mid-late 1990s, foreign-born workers comprised about one-fifth of top tier jobs, more than one-quarter of middle tier jobs, and almost half of all bottom tier jobs. Our analysis of job dissimilarity across demographic groups, measured by comparing one group’s distribution of employment across jobs compared to the native-born white distribution) showed that in every case there are either increases in dissimilarity or stability – there is certainly no evidence of convergence to the white distributions.
Turning to educational attainment, we found that while workers with low education levels are increasingly foreign-born, there was no overall tendency for the foreign-born groups to have lower levels of attainment. In 1997/8, the share of employed workers with more than a high school degree was far higher for both native-born whites (69%) and foreign-born Asians (61%) than for either native-born black (53%) or native-born Hispanic (45%) workers. At the same time, foreign-born white workers had the same share of more highly educated workers as native-born blacks. All demographic groups show increases in the share of workers with more than a high school degree and gains in educational attainment can be seen across bottom tier jobs for workers in each of the demographic groups.
    These employment composition and educational attainment trends appear to favor native-born workers, particularly white native-born. Our results for median weekly earnings (full-time, prime-age) confirmed this. For men, white native-born and Asian foreign-born workers show by far the best wage performance over these two decades, while the largest real earnings declines were experienced by native-born black and foreign-born white, black and Hispanic men. Two trends appear to play central roles. First, there were large declines in real earnings in formerly “good” middle tier sectors (manufacturing, construction, transportation, communications and public utilities), as foreign-born workers replaced native-born whites. Second, in the bottom tier these four demographic groups experienced large cuts in real pay for fast growing jobs in which they were increasingly concentrated.
    Similarly, female workers experienced earnings declines in both the middle and bottom tiers. Most notably, native-born black women saw large earnings declines in moderately skilled clerical and technical positions in sectors that used to pay fairly well (health care; finance, insurance and real estate; and public administration). In the bottom tier, foreign-born white, native-born black, and foreign-born Hispanic women also had large earnings declines, particularly in the health care and retail sectors, where they are heavily concentrated.
    Taking male and female workers together, these results point to two conclusions. First, a key consequence of these earnings trends has been the substantial increase in the earnings advantage of native-born workers within each job tier. The ratio of native-born to foreign-born earnings in the top job tier (median earnings for full-time prime age workers) grew from 11% in 1979 to 14% in 1997/8. The native-born advantage and its growth was even greater in the middle job tier, where the differential rose from just 4% to 15%, and in the bottom job tier, where it shot up from 15.5% to 24.3%. Second, evidence of large decreases in real pay in the fastest growing bottom tier jobs challenges the conventional demand shift (against the least skilled) explanation for the growth in earnings inequality (e.g., see Johnson, 1997). The problem was not declining demand for low skill workers, but fast growing jobs paying lower wages.
    Turning to earnings trends by demographic group, we find that the earnings advantage of native whites was substantial in 1997/8, ranging from a ratio of .76 (native-born Hispanic to native-born  white women) to .46 (foreign-born Hispanic to native-born white men). Between 1980 and 1997/8, the earnings advantage of white workers increased over most groups, but there is no evidence of a growing wage gap among the least educated in the bottom tier for any demographic group. Where the differential favoring white native-born earnings has grown, it appears to have been driven by a growing gap at higher skill levels (in the middle and top tiers).
    Finally, we explored inequality trends for the metropolitan area. Earnings inequality grew substantially for both men and women between 1979 and 1997/8. Excluding recent immigrants noticeably reduced male inequality, measured by the 90/10 ratio for full-time prime age workers, from 5.13 to 4.86. The effect for female workers was even greater; by excluding immigrants, inequality drops from 4.56 to 4.08. Our findings were similar when we calculated the 90/10 ratio for all workers with earnings. We also found substantial growth in inequality among both native- and foreign-born workers using the earnings ratio of college to high school degree holders.
    So the good news is that there has been substantial growth in top tier jobs and foreign-born workers have captured an increasing share of them. But there is plenty of bad news. Trends in employment composition, educational attainment, and earnings were most favorable for native-born whites and least favorable for foreign-born blacks and Hispanics. As native-born white workers have become more concentrated in the top tier and in the higher paying jobs within each tier, they have maintained or increased their earnings advantage. The poor earnings performance for black and Hispanic workers (both native- and foreign-born) appears to be linked to their concentration in bottom tier health and retail jobs, which expanded in number and fell sharply in quality, as well as in low skill middle tier jobs that used to pay well – blue-collar and clerical positions in public administration; transportation, communications and public utilities; and finance, insurance and real estate.
    We do not attempt to explain these trends, but our findings are consistent with our view that supply shifts and the de-institutionalization of the labor market have played central roles. The explanations may vary by job tier. The large wage declines in middle tier jobs appears tied to changes in wage-setting practices (such as the outsourcing of low-skill work to smaller "peripheral" firms and the declining power of unions) that reflect increasing pressures to reduce labor costs, particularly in the public and formerly regulated private sectors. The poor earnings performance of bottom tier workers may be due in part to collapse in the real value of the legal minimum wage at the same time that the metropolitan area has received a massive inflow of poorly educated immigrants willing to work for wages that most native-born workers find unacceptably low (Fortin and Lemieux, 1998; Bean and Hamermesh, 1998). These institutional changes and supply shifts reflect public policy choices. Effective policy interventions to improve the employment status of lower skilled workers requires more research aimed at developing convincing explanations for the disturbing trends we have documented.

References
Gittleman, Maury B. and David R. Howell (1995).  “Changes in the Structure and Quality of Jobs in the United States:  Effects by Race and Gender, 1973-1990.”  Industrial and Labor Relations 48(3):  420-440.

Gordon, David M., Richard C. Edwards, and Michael Reich (1982).  Segmented Work, Divided Workers:  The Historical Transformation of Labor in the United States.  Cambridge:  Cambridge University Press.

Gottschalk, Peter (1997). "Inequality, Income Growth, and Mobility: The Basic Facts," Journal of Economic Perspectives 11, no.2 (Spring): 21-40.

Hamermesh, Daniel S. and Frank D. Bean (1998). Help or Hindrance: The Economic Implications of Immigration for African Americans. New York: Russell-Sage.

Howell, David R. and Elizabeth J. Meuller (2000).  “Immigration and Native-Born Male Earnings:  A Jobs-Level Analysis of the New York City Metropolitan Area Labour Market, 1980-90.”  Journal of Ethnic and Migration Studies 26(3): 469-493.

Johnson, George E. (1997).  “Changes in Earnings Inequality:  The Role of Demand Shifts.”  Journal of Economic Perspectives 11(2): 41-54.

Mishel, Lawrence, Jared Bernstein and John Schmitt (1999).  State of Working America:  1998-1999.  Ithaca, New York:  Cornell University Press.

New York City Department of City Planning (1999). "The Newest New Yorkers: 1995-1996." DCP #99-08 (September).

Waldinger, Roger David (1996).  Still the Promised City?:  African-Americans and New Immigrants in Postindustrial New York.  Cambridge, Massachusetts:  Harvard University Press.