How Have Hispanics Fared in the Jobless Recovery?

Transcription

1 How Have Hispanics Fared in the Jobless Recovery? William M. Rodgers III Heldrich Center for Workforce Development Rutgers University and National Poverty Center and Richard B. Freeman Harvard University and NBER

2 How Have Hispanics Fared in the Jobless Recovery? William M. Rodgers III Heldrich Center for Workforce Development Rutgers University and National Poverty Center And Richard B. Freeman Harvard University and NBER January 2006 This paper was prepared for the Center for American Progress and was presented at the January 6th, 2006 American Economic Association s Session Economic and Policy Issues Facing the U.S Hispanic Community. We thank Sue Stockly and panel participants for the comments and suggestions. We also thank the Center for American Progress for their financial support.

3 Executive Summary Since the end of the recession in November 2001, U.S. workers have experienced slow job growth and stagnant wages. These conditions have hurt the economic security of all Americans. Hispanics, the nation s fastest growing minority group, have been no exception. Hispanics have had fewer jobs, lower wages, less health insurance coverage, and declining pension coverage in recent years. Instead of catching up to other groups, Hispanics have remained behind. Some Hispanics, particularly Mexicans, have fallen even further behind. The main findings from this paper show that: Since 2001, the share of employed Hispanics, especially Mexican Americans, has declined. The employed share of Hispanic men fell by 1.8 percentage points, about the same as white men (1.9) and less than blacks (3.8). Mexican-American men fared worse than Hispanics in general, with a 2.4 point decline. Hispanic women saw their employed share drop by 1.4 percentage points, more than white women (0.9) and less than black women (2.7). The employed share of Mexican-American women decreased by 2.4 percentage points, larger than the decline for Hispanics overall. Wages for Hispanics stayed below those of other groups, and in some ways have fallen further behind. By 2004, Hispanic men s wages were 44.7 percent less than white men s wages, after growing 1.3 percent since This difference is after a period when wages for Hispanic men grew faster than those of white men, whose wages increased by only 0.4 percent. In addition, Hispanic men s wage growth has been less than that of black men, whose weekly wages grew by 2.4 percent. Hispanic women are the only group for which weekly wages declined from 2001 to Wages for Hispanic women were 18.1 percent lower than those of white women, after falling by 0.6 percent. During that same time period, wages rose for black women by 0.2 percent and for white women by 0.1 percent. Hispanics have low work-related benefits, such as health care coverage and pensions. American-born, as well as foreign-born, Hispanics have lower health insurance and pension coverage rates than i

4 other groups. Three-quarters of white men have private health insurance, compared to 59 percent of black men and 48 percent of Hispanic men. Similar differences exist for women. In part, these differences are due to low education and are thus unlikely to be eliminated even in a jobs boom. During the 1990s expansion, private health insurance coverage among new Hispanic job entrants increased by 3.3 percentage points, less than one-half the white increase, and one-fifth the black increase. Employment patterns for non-hispanics are not different in areas with large undocumented immigrant populations. Changes in the employed share of the population and of hourly wages of whites, blacks and Hispanics in the states with the 15 largest undocumented immigrant populations are comparable to the patterns observed elsewhere. When it comes to jobs for Hispanics, geography matters. The drop in the employed share of Hispanics was smaller in metropolitan areas with larger Hispanic populations. The employed share of Mexican-American men in areas with Hispanic populations of less than 3 percent fell by 6.3 percentage points, compared to a 3 point decline in areas with Hispanic populations of 30 percent or more. Manufacturing s decline particularly hurts Hispanic men. The manufacturing sector s decline lowered potential Hispanic male employment by an estimated 2.3 percent by This is larger than the estimated declines of 1.7 and 1.8 percent for white and black men. The employment shift reduced Hispanic female employment by 0.1 percent, compared to increases of 1.5 and 1.6 percent for white and black women. Among Hispanics, Mexican Americans exhibit the greatest vulnerability to slow job growth. Mexican Americans have the lowest wages and the least health insurance and pension coverage out of any Hispanic group. Amid weak job growth, Mexican Americans saw some of the sharpest increases in economic insecurity. From 2001 to 2004, Mexican-American men were the only group of men for whom weekly wages fell. Mexican-American women had fewer jobs, falling wages, dropping health insurance coverage, lower pension coverage and higher food stamp usage in 2004 than in 2001, making them the only group of women for whom all measures of economic security deteriorated. ii

5 Compared to Mexican Americans, Puerto Ricans trade off less full-time employment against higher wages and benefits. Full-time employment among Puerto Rican men and women dropped by 0.9 and 5.5 percent from 2001 to It stayed flat for Mexican-American men and dropped by 1.8 percent for Mexican-American women. With respect to wages and benefits Puerto Ricans did universally better than Mexican Americans. Cubans have the largest difference in wage trends by gender and have larger wage movements than Mexican Americans and Puerto Ricans. Weekly wages for Cuban men rose by 9.0 percent, compared to an increase of 4.6 percent for Puerto Ricans and a drop of 0.9 percent for Mexican-American men. In contrast, wages for Cuban women fell by 12.8 percent, compared to a 2.2 percent drop for Puerto Rican women and a 7.5 percent decline for Mexican-American women. Unionization is lower among Hispanics than among blacks or whites. Hispanic union members earn higher wages than nonunion Hispanics. The median nonunion weekly wage among Hispanics is $428, compared to the union median weekly wage of $679. This advantage is almost twice the advantage that black and white union members experience. Similarly, unionized Mexican-American women earn 70 percent more than nonunion Mexican-American women. Only 10.1 percent of Hispanics were union members in 2004, compared to 15.1 and 12.2 percent for blacks and whites. Hispanics benefit from a tight labor market. A booming economy with a strong labor market increases economic security. Over time and across localities, strong job growth is associated with more full-time and full-year work, higher health insurance and pension coverage rates, and lower food stamp usage. iii

6 Introduction Despite solid economic growth, this recovery has been marked by the slowest job growth of any recovery since World War II. Because the labor market is playing catch-up with past recoveries, many minorities and workers with the least skills who benefited from the 1990s boom are having difficulty maintaining their gains. This is true for African Americans and new job entrants (people with 0 to 10 years of potential experience). It is also true for the nation s fastest growing minority group, Hispanics. The lack of strong job creation that typically accompanies resurgent economic growth has given rise to growing economic insecurities for Hispanics. Depending on their particular demographic characteristics, this has meant fewer jobs, lower wages, less health insurance, and declining pensions for various Hispanic groups. In particular, our results show that: Job growth has been weak in this business cycle. By November 2005, 3.4 million new jobs were created, compared to 8.3 million during the 1990s recovery and 12.8 million during the recovery that started March The lack of employment growth has affected all Americans, but especially Hispanics and other minorities and new job entrants. There are important differences in employment opportunities by location. For instance, the deterioration in the employed share of the Mexican-American population from 2001 to 2004 was smaller in areas with larger Mexican-American populations. Manufacturing s decline has adversely affected Hispanics, especially men. For Hispanic women, their disproportionate presence in the education and health services sector helps to offset job losses in the wholesale and retail trade. Changes in the employment-population ratios of Hispanics, whites, and blacks who reside in the states with the 15 largest undocumented immigrant populations are not significantly different from the patterns observed in the general population. Rising economic insecurities for Hispanics manifest themselves in various ways, depending on particular demographic characteristics. Lower wages, less health insurance coverage, fewer pensions, and increased usage of food stamp programs are prevalent among all Hispanic minorities, even among U.S.-born Hispanics. These deteriorations come on top of already lower wages, health insurance and pension coverage than whites. 1

7 Hispanics have the lowest union membership of major groups and thus benefit least from what unions do for workers. Only 10.1 percent of Hispanics belong to a union, compared to 15.1 percent of blacks and 12.2 percent of whites. Among Hispanics, unionization ranged from 19.8 percent for Puerto Ricans to a low of 2.9 percent for Cubans. Hispanics who are union members earn significantly higher wages than nonunion Hispanics. The median nonunion weekly wage among Hispanics is $428, compared to the union median weekly wage of $679, generating a 59 percent advantage to union membership. This advantage is almost twice the size of the advantage that black and white union members experience. Hispanics benefit from a tight labor market. A booming economy with a strong labor market increases economic security. Over time and across localities, strong job growth is associated with more full-time and full-year work, higher health insurance and pension coverage rates, and lower food stamp usage. The Most Recent Business Cycle: The Catch-Up Economy Even with the acceleration in job creation since August 2003, the 2000 recovery, which started in November 2001, had slower employment growth than all prior recoveries since 1960, including the 1990s recovery, when employment also took a long period to recover (Figure 1). After 49 months of this recovery, just over 3.4 million new jobs were created, compared to 11.5 and 12.8 million during the recoveries that followed the 1980s and early 1970s recessions. Number (In Thousands) 15,000 10,000 5, ,000 Recoveries Figure 1: Cumulative Employment Growth During the Six Most Recent Recoveries Number (in Thousands) Months Since Start of Recovery Months Since Start of Recovery Nov-01 Mar-91 Nov-82 Jul-80 Mar-75 Nov-70 Feb-61 Source: Nonfarm Payroll Establishment data. U.S. Department of Labor, Bureau of Labor Statistics ( Each series is Source: Nonfarm Payroll Establishment data. U.S. Department of Labor, Bureau of Labor Statistics ( Each series is benchmarked to the start of its benchmarked recovery as defined to the by the start NBER of its Business recovery Cycle as Dating defined Committee. by the Figures NBER are Business through the Cycle September Dating 2005, Committee. the 49th month Figures of the current are through recovery. the September 2005, the 49th month of the current recovery. 1 2

8 Employment in many private sector industries, such as manufacturing and trade, in which Hispanics are concentrated, remained below that at the start of the recovery. By November 2005, employment was 9.4 percent lower in durable manufacturing and 10.6 percent lower in nondurable manufacturing than when the recovery began (figure 2). In contrast, by the 49th month of previous recoveries, nondurable and durable manufacturing had typically expanded 3.7 and 9.4 percent. Even with the recovery employment remained 11.0 percent lower in the broad information sector, which was supposed to produce good jobs to replace declining employment in traditional manufacturing. During earlier recoveries this sector had grown at an average rate of 7.4 percent (figure 2). Figure 2: Cumulative Employment Change by Industry (in percent) Total Nonfarm Mining Construction Durable Manufacturing Nondurable Manufacturing Wholesale Trade Retail Trade Transportation Utilities Financial Activities Leisure Government Other Services Information Professional Business Services Temporary Help Services Education Health Services

9 Employment growth in other sectors is slower than the average over the last six recoveries. This is true for wholesale and retail trade and even for interest rate sensitive industries, such as construction and financial activities. Construction grew by 8.4 percent, compared to 13.8 percent during the previous recoveries. Financial activities expanded by 5.9 percent during the current recovery, while growing by over 11.2 percent in previous recoveries. In contrast, employment has grown strongly in the education and health services sector, where many Hispanic women are employed. The slow jobs recovery shows some variation across states (table 1). Looking at the past three recoveries to 2005, 1991 to 1995, and 1982 to 1986 we see that average state employment growth was 10 percent in the two previous recoveries, while during the current recovery employment growth stagnated, falling at 0.3 percent. Looking at states with large Hispanic populations, Arizona and Florida had positive employment growth but that growth was two to three times slower than past recoveries. Texas, California and New York, which also have large Hispanic populations, had drops in employment during the current recovery of between 1 and 3 percent. Table 1: 2004 Hispanic Population and Growth in Total Nonfarm Employment by State and Recovery Hispanic Population Distribution Actual Change in Employment (Percent) State Within Across New Mexico 43.3% 2.0% 3.7% 16.6% 11.0% California 34.7% 30.1% -0.6% 0.5% 13.0% Texas 34.6% 18.8% -0.7% 11.8% 4.8% Arizona 28.0% 3.9% 6.3% 20.4% 29.9% Nevada 22.8% 1.3% 11.7% 25.0% 16.7% Colorado 19.1% 2.1% -2.3% 18.7% 7.0% Florida 19.0% 8.0% 6.1% 13.3% 22.3% New York 16.0% 7.4% -3.1% 0.1% 9.0% New Jersey 14.9% 3.1% -0.7% 2.9% 12.8% Illinois 14.0% 4.3% -5.0% 6.9% 4.3% Connecticut 10.6% 0.9% -2.6% 0.4% 11.9% Utah 10.6% 0.6% 2.5% 21.8% 13.1% Rhode Island 10.3% 0.3% 0.5% 4.4% 13.3% Oregon 9.5% 0.8% 0.0% 13.9% 10.1% Idaho 8.9% 0.3% 1.3% 19.9% 5.1% District of Columbia 8.5% 0.1% 1.8% -5.1% 7.0% Washington 8.5% 1.3% -0.7% 7.8% 12.8% Kansas 8.1% 0.5% -2.8% 9.4% 6.9% Hawaii 7.9% 0.2% 5.2% -1.2% 9.8% Massachusetts 7.7% 1.2% -5.8% 5.5% 13.1% Nebraska 6.9% 0.3% -0.5% 10.6% 7.0% Georgia 6.8% 1.4% -2.3% 15.8% 21.4% 4 It is quite Wyoming startling that given the fiscal stimulus, 6.7% low inflation, 0.1% and low interest 1.8% rates over 8.0% this period, job -9.8% growth has been much lower than in previous recoveries during which the stimulus was much smaller. 4

10 Table 1: 2004 Hispanic Population and Growth in Total Nonfarm Employment by State and Recovery (Continued) Hispanic Population Distribution Actual Change in Employment (Percent) State Within Across Oklahoma 6.3% 0.5% -2.9% 8.6% -7.6% North Carolina 6.1% 1.3% -2.2% 12.6% 16.9% Delaware 5.8% 0.1% 0.1% 7.2% 17.0% Virginia 5.7% 1.0% 1.8% 8.5% 19.2% Maryland 5.4% 0.7% 0.7% 3.9% 16.5% Alaska 4.9% 0.1% -0.3% 7.9% 10.1% Arkansas 4.4% 0.3% -0.2% 14.2% 13.0% Indiana 4.3% 0.7% -0.7% 11.1% 9.6% Wisconsin 4.3% 0.6% -2.1% 11.1% 8.4% Pennsylvania 3.8% 1.2% -2.0% 3.3% 4.6% Michigan 3.7% 0.9% -5.7% 9.8% 14.5% Iowa 3.5% 0.3% -2.3% 9.7% 3.1% Minnesota 3.5% 0.4% -1.9% 11.3% 10.8% South Carolina 3.1% 0.3% -1.8% 8.8% 15.1% Tennessee 2.8% 0.4% -0.6% 14.4% 13.3% Louisiana 2.8% 0.3% -0.3% 9.9% -5.5% Missouri 2.6% 0.4% -2.8% 9.2% 11.4% Montana 2.4% 0.1% 3.3% 15.5% 0.6% Ohio 2.2% 0.6% -4.1% 8.4% 8.4% Alabama 2.2% 0.2% -0.2% 9.8% 11.5% New Hampshire 2.1% 0.1% -0.3% 12.0% 24.3% South Dakota 2.0% 0.0% -1.0% 15.9% 9.4% Kentucky 1.9% 0.2% -1.2% 11.4% 9.8% Mississippi 1.7% 0.1% -0.7% 14.6% 7.3% North Dakota 1.5% 0.0% 0.8% 11.5% 0.1% Vermont 1.0% 0.0% 0.6% 8.5% 15.5% Maine 0.9% 0.0% -2.3% 4.8% 14.9% West Virginia 0.8% 0.0% -1.1% 9.4% -1.7% U.S. Average 100.0% -0.3% 10.0% 10.0% Notes: Authors tabulations from published BLS data: Hispanic population data come from Table 4: Annual Estimates of the Population by Race Alone and Hispanic or Latino Origin for the United States and States: July 1, 2004 (SC-EST ). Source: Population Division, U.S. Census Bureau. To better understand how the geographic location of Hispanics affected their economic position in the recovery, we calculated an average state employment growth, weighted by each state s Hispanic population. This shows that the employment in geographic areas populated by Hispanics fell by 0.16 percent from 2001 through the first half of Weighting each state s growth by its white and black populations generates declines of 0.96 and 1.06 percent, suggesting that blacks and whites are more adversely impacted by the slow job growth. 1 Hispanics did better in employment because the states with larger Hispanic populations are in the growing parts of the country. In states with at least 10 percent Hispanics, employment grew 5

11 by 0.15 percent. In states with at least 20 percent Hispanics, average employment increased by 0.28 percent. Growth in states such as Nevada, Arizona, and Florida help to offset stagnation and losses in California, Texas, New York and Illinois. The Labor Market Status of Hispanics: Economic Vulnerability Hispanics, particularly Mexican Americans, have a very high employment-population ratio and a high rate of working full-year, which shows that they are doing well in one dimension of the labor market. But they also have the lowest wages, lowest health insurance coverage rates, and lowest pension coverage rates. The main reason for this is that they have fewer years of schooling than other Americans (table 2). Hispanic men (women) have completed an average of 10.4 (10.6) years of schooling, compared to 13.2 (13.2) for non-hispanic men (women). Although the education gap is narrower among new entrants to the labor market, Hispanics still have completed fewer years of schooling than other Americans. The lower educational attainment of Mexican-American men explains a large portion of the gap. Mexican Americans comprise over 60 percent of the Hispanic population. Another key difference between Hispanics and non-hispanics is that approximately 90 percent of non-hispanics are U.S.-born, compared to about 50 percent of Hispanics. Many of the foreign-born are from Mexico. While the media often focuses on unemployment rates, most labor economists concentrate on employment-population rates. This is because employment rates are more clearly measured: employment and population are sharp concepts whereas unemployment depends on labor participation decisions, which vary with job opportunities, as workers become discouraged by poor opportunities or encouraged by good ones. Published BLS data indicate that Hispanic men have employment-population ratios that are 24 points higher than Hispanic women (76.1 versus 52.1) and higher than the ratios of both white and black men. Hispanic, white and black women all have similar employment-population ratios, ranging from 52 to 58 percent. Looking at the Hispanic population in even more detail, we see that Mexican Americans have the highest employment-population ratio followed by Cubans and Puerto Ricans. Among teenagers, Hispanic women and blacks have the lowest employment-population ratios. These low ratios potentially reflect school enrollment decisions, as people who are still going to college are counted in the relevant population. When we limit ourselves to young high school dropouts and high school graduates who are not enrolled in college, Hispanic employment looks much better. High school dropouts of any race and ethnicity are significantly disadvantaged in the labor market, especially black high school dropouts. Employment-population ratios range from 12.3 percent for black youth, to 37.7 percent for white youth, and 39.3 percent for Hispanic youth. Finally, looking at the proportion of the workers who work full-year as another measure of the 6

12 Table 2: 2004 Summary Statistics Panel A: Men All Non-Hispanics Hispanics Mexican Puerto Rican Cuban Years of Schooling Potential Experience % with no more than High School Degree Age MSA s Hispanic Population (%) Mexican American 63.0 Puerto Rican 9.0 Cuban 4.0 South American 18.0 Other Spanish st Generation nd Generation rd or Higher Generation Reside in MSA with Large Undocumented Population State s Hispanic Population (%) New Entrants Years of Schooling Potential Experience % with no more than High School Diploma Age MSA s Hispanic Population (%) Mexican American 65.0 Puerto Rican 9.0 Cuban 2.0 South American 18.0 Other Spanish st Generation nd Generation rd or Higher Generation Reside in MSA with Large Undocumented Population State s Hispanic Population (%) Notes: Authors tabulations from the 2004 March Annual Demographic files of the Current Population Survey. The detailed Hispanic categories are based on the BLS definitions. The Non-Hispanic sample is limited to whites and blacks. strength of employment, we see in table 3 that Hispanics have a high rate of full-year employment. Among Hispanics, Mexican Americans have the highest percentage who work full-year. But the flip side of the good employment record is low weekly wages and benefits. Hispanics fall markedly below whites and less markedly below blacks in the proportion of the employed with private health insurance and pension coverage. They have the lowest wages, lowest health insurance rates, and lowest pension coverage (table 3). Compared to whites and 7

13 Table 2 cont.: Summary Statistics in 2004 Panel B: Women All Non-Hispanics Hispanics Mexican Puerto Rican Cuban Years of Schooling Potential Experience % with no more than High School Diploma Age MSA s Hispanic Population (%) Mexican American 60.0 Puerto Rican 10.0 Cuban 4.0 South American 19.0 Other Spanish 7.0 1st Generation nd Generation rd or Higher Generation Reside in MSA with Large Undocumented Population State s Hispanic Population (%) New Entrants Years of Schooling Potential Experience % with no more than High School Diploma Age MSA s Hispanic Population (%) Mexican American 63.0 Puerto Rican 10.0 Cuban 3.0 South American 18.0 Other Spanish 6.0 1st Generation nd Generation rd or Higher Generation Reside in MSA with Large Undocumented Population State s Hispanic Population (%) Notes: Authors tabulations from the 2004 March Annual Demographic files of the Current Population Survey. We use the current BLS terminology to identify individuals by race and ethnicity. The Non-Hispanic sample is limited to whites and blacks. blacks, Mexican-American outcomes are worse. One exception is that Mexican-American usage of food stamps is lower than that of blacks. The only Hispanic group that has food stamp usage rates similar to blacks is Puerto Ricans. We speculate that this may be due to Puerto Ricans not facing the same barriers to entering the program as other Hispanic groups. For all of these outcomes, excluding foreign-born individuals does little to close the differences between Hispanics and non-hispanics. 8

14 Panel A: Men All Table 3: 2004 Labor Market Outcomes by Gender and Detailed Race and Ethnicity Work Full-Year Average Weekly Wages Private Health Insurance Have Pension Food Stamps White 67.7% $ % 47.9% 3.3% Black 54.8% $ % 44.1% 10.2% Hispanic 69.0% $ % 25.8% 7.6% New Entrants Mexican American 70.7% $ % 23.3% 8.0% Puerto Rican 57.2% $ % 38.1% 14.0% Cuban 63.2% $ % 30.9% 8.0% White 54.8% $ % 27.5% 4.7% Black 39.8% $ % 26.3% 14.2% Hispanic 54.5% $ % 18.7% 9.2% Mexican American 57.4% $ % 16.8% 8.7% Puerto Rican 41.3% $ % 27.1% 22.9% Cuban 48.3% $ % 32.4% 8.3% Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. Both part-time and full-time workers are in the sample. Weekly wage is the ratio of income from wages and salary in calendar year and weeks worked. A new entrant is an individual with no more than 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. Table 3 cont.: 2004 Labor Market Outcomes by Gender and Detailed Race and Ethnicity Panel B: U.S.-Born Men (Generations 2 and 3+) Average Private Health All Work Full-Year Weekly Wages Have Pension Food Stamps Insurance White 67.7% $ % 48.1% 3.3% Black 53.3% $ % 45.2% 10.7% Hispanic 60.4% $ % 36.5% 8.5% Mexican American 61.2% $ % 35.6% 8.1% Puerto Rican 56.8% $ % 38.5% 14.0% Cuban 71.4% $ % 38.8% 1.0% New Entrants White 54.6% $ % 27.4% 4.8% Black 38.1% $ % 26.5% 15.0% Hispanic 47.4% $ % 23.7% 10.4% Mexican American 49.4% $ % 22.6% 8.9% Puerto Rican 40.1% $ % 27.2% 23.3% Cuban 44.7% $ % 34.8% 0.0% Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. Both part-time and full-time workers are in the sample. Weekly wage is the ratio of income from wages and salary in calendar year and weeks worked. A new entrant is an individual with no more than 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. 9

15 Table 3 cont.: 2004 Labor Market Outcomes by Gender and Detailed Race and Ethnicity Panel C: Women All Work Full-Year Average Private Health Weekly Wages Insurance Have Pension Food Stamps White 54.0% $ % 43.4% 5.0% Black 49.8% $ % 43.2% 16.5% Hispanic 46.5% $ % 28.4% 11.8% New Entrants Mexican American 44.2% $ % 27.2% 12.0% Puerto Rican 49.4% $ % 34.8% 20.2% Cuban 45.1% $ % 37.0% 11.3% White 50.1% $ % 25.9% 6.8% Black 42.5% $ % 26.9% 20.9% Hispanic 40.3% $ % 18.0% 11.8% Mexican American 39.1% $ % 16.8% 11.6% Puerto Rican 44.8% $ % 16.8% 19.7% Cuban 53.8% $ % 34.4% 6.6% Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. Both part-time and full-time workers are in the sample. Weekly wage is the ratio of income from wages and salary in calendar year and weeks worked. A new entrant is an individual with no more than 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. Table 3 cont.: 2004 Labor Market Outcomes by Gender and Detailed Race and Ethnicity Panel D: U.S.-Born Women (Generations 2 and 3+) All Work Full- Year Average Weekly Wages Private Health Insurance Have Pension Food Stamps White 54.4% $ % 43.5% 5.0% Black 49.5% $ % 43.6% 17.1% Hispanic 49.5% $ % 34.6% 13.2% Mexican American 49.5% $ % 33.9% 12.5% Puerto Rican 48.9% $ % 34.2% 20.3% Cuban 59.6% $ % 45.0% 4.4% New Entrants White 50.3% $ % 25.9% 6.8% Black 42.2% $ % 27.0% 21.9% Hispanic 41.6% $ % 20.0% 13.1% Mexican American 42.0% $ % 19.7% 12.9% Puerto Rican 44.2% $ % 15.8% 19.9% Cuban 50.0% $ % 38.1% 6.5% Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. Both part-time and full-time workers are in the sample. Weekly wage is the ratio of income from wages and salary in calendar year and weeks worked. A new entrant is an individual with no more than 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. 10

16 Linking Industry and Demographic Change The changing distribution of employment among industries potentially affects Hispanic Americans differently than other Americans, because Hispanics are concentrated in a different set of industries. Table 4 shows the distribution of employment by industry in 2001 by race, ethnicity and gender. Men are relatively more likely than women to be employed in manufacturing, an industry that lost jobs from 2001 to They are also concentrated in trade, an industry that has shown little if any growth. Mexican-American and white men are overrepresented in construction, a growing industry. Black men are disproportionately employed in services, education, health and government. Women are concentrated in slow growing wholesale and retail trade, but also in the growing education and health services sector. We translate these patterns into expected shifts in demand for a demographic group s employment by computing a fixed weight index of the potential shift in employment for a group. To do this, we multiply each group s 2001 industry employment share by its industry employment growth from 2001 to 2004, and sum the products to obtain a weighted average growth of employment. Table 5 reports these estimated changes. For men, the shift due to changes in employment in the industries in which they work ranges from -1.7 percent for white and black men to -2.5 percent for Mexican-American men. The main reason for the negative expected employment growth for men overall and for the differences among the groups is their greater concentrations in the manufacturing sector. The expected declines for out of school white, black and Mexican-American men are larger than for other men because proportionately fewer of these men are employed in education and health services, where employment expanded most in the recovery. Larger shares of these young men are employed in trade and professional and business services, major sectors that have contracted. In contrast with the shifts in employment against men, the fixed weight industry growth calculations suggest employment increases for white women, black women, and some Hispanic women. The estimates indicate that Mexican-American women s employment stagnates. For all white and black women, the predicted increases in employment are 1.5 and 1.6 percent. The increases are 0.6 and 0.9 percent for Cuban and Puerto Rican women. The estimates for new entrants suggest modest growth and a decline for out of school youth. These expected increases are driven by women s presence in the education and health services sector. Larger shares of older women are in this sector, while younger and less-skilled women tend to be employed in trade and professional and business services. 11

17 Table 4: 2001 SIC Industry Distributions by Race, Ethnicity, Gender and Experience Panel A: All Men Women Puerto Rican Cuban Mexican American Puerto Rican Cuban White Black Hispanic Mexican American Industry White Black Hispanic Mining 1% 0% 1% 1% 0% 0% 0% 0% 0% 0% 0% 0% Construction 12% 9% 16% 18% 7% 13% 2% 1% 1% 1% 1% 0% Manufacturing 17% 16% 18% 19% 19% 12% 9% 9% 14% 16% 12% 12% Transportation & Public Utilities 9% 14% 7% 6% 12% 11% 4% 6% 4% 3% 6% 5% Trade 20% 19% 24% 23% 23% 25% 22% 17% 24% 25% 19% 21% FIRE 5% 4% 3% 2% 5% 5% 8% 7% 6% 5% 8% 11% Professional & Business Services 12% 14% 14% 12% 16% 16% 12% 13% 18% 17% 14% 15% Education and Health Services 14% 15% 7% 6% 12% 12% 37% 40% 27% 26% 36% 31% Public Administration 5% 6% 3% 2% 4% 4% 4% 7% 3% 4% 4% 3% Panel B: New Entrants Men Women Puerto Rican Cuban Mexican American Puerto Rican Cuban White Black Hispanic Mexican American Industry White Black Hispanic 24 Mining 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Construction 11% 5% 15% 17% 8% 10% 1% 1% 1% 1% 0% 1% Manufacturing 13% 12% 14% 15% 10% 5% 6% 6% 9% 10% 7% 8% Transportation & Public Utilities 6% 10% 6% 5% 11% 11% 3% 6% 4% 3% 6% 5% Trade 30% 31% 32% 31% 31% 32% 31% 28% 34% 35% 30% 32% FIRE 4% 4% 3% 2% 5% 6% 7% 8% 7% 7% 7% 8% Professional. & Business Services 15% 17% 15% 13% 21% 16% 14% 15% 15% 15% 15% 12% Education and Health Services 13% 15% 8% 7% 10% 17% 32% 33% 26% 25% 31% 31% Public Administration 3% 4% 2% 2% 2% 3% 3% 4% 2% 3% 3% 2% Notes: Authors calculations from the 2001 ORG CPS file. The columns are the share of a particular group in each industry. All corresponds to all men at least 18 years of age who work in the either the public or private sector. New entrants refer to respondents with no more than 10 years of potential experience. Out of school youth refers to 16-to-24-year-olds who were not enrolled in school at the time of the survey. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise.

18 Table 4 cont.: 2001 Industry Distribution by Race, Ethnicity, Gender and Experience Panel C: Out of School Youth Men Women Industry White Black Hispanic Mexican American White Black Hispanic Mining 1% 0% 0% 0% 0% 0% 0% 0% Construction 19% 8% 20% 22% 1% 0% 1% 2% Manufacturing 15% 13% 16% 17% 8% 6% 13% 14% Transportation and Public Utilities 5% 9% 4% 4% 3% 4% 3% 3% Trade 34% 36% 31% 30% 45% 39% 41% 41% FIRE 1% 2% 1% 1% 6% 6% 6% 5% Professional & Business Services 14% 20% 14% 13% 18% 18% 19% 19% Education and Health Services 4% 8% 2% 2% 17% 23% 14% 12% Public Administration 1% 1% 1% 0% 1% 2% 1% 1% Mexican American Notes: Authors calculations from the 2001 ORG CPS file. The columns are the share of a particular group in each industry. All corresponds to all men at least 18 years of age who work in the either the public or private sector. New entrants refer to respondents with no more than 10 years of potential experience. Out of school youth refers to 16-to-24- year-olds who were not enrolled in school at the time of the survey. We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise.

19 Table 5: Expected 2001 to 2004 Change in Employment (Assuming 2001 Industry Shares and Actual CES Employment Change) Men White Black Hispanic Mexican American Puerto Rican Cuban All -1.7% -1.8% -2.3% -2.5% -2.3% -1.4% New Entrants -1.2% -1.4% -1.8% -2.0% -1.5% -0.3% Out of School Youth -2.3% -2.1% -2.4% -2.5% Women All 1.5% 1.6% -0.1% -0.3% 0.9% 0.6% New Entrants 1.3% 1.3% 0.4% 0.1% 0.9% 0.9% Out of School Youth -0.5% 0.3% -1.2% -1.5% Notes: Entries are constructed by multiplying a demographic group s 2001 industry employment shares (Table 4) by the industry s percentage employment growth from 2001 to 2004, and summing the products to obtain a weighted average growth of employment in the industries that employed the group. Industry employment growth is the difference from 2001 to In 2003 the industry codes changed. To link 2001 with 2004, we had to make several assumptions. The following list the 2001 SIC (2003 SIC) codes. If an industry shown in Table 8 is not listed below, a direct match could be made: Transportation (Transportation and Warehousing), Communication and Public Utilities (Information), Utility and Sanitary Services (Utilities), Finance, Insurance and Real Estate (Financial Activities), Entertainment and Recreation (Leisure and Hospitality), Professional and Business Services (Personal services including private households, business, auto and repair services, Personal services excluding private households), Education and Health Services (Hospitals, Medical services, except hospitals, educational services, Social services), Other Professional Services (Other services). We use the current BLS terminology to identify individuals by race and ethnicity. Results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise. Undocumented Immigration Our analysis only captures the experiences of Hispanics who show up in the formal government data and is likely to miss the experiences of the undocumented immigrants who have received a great deal of public attention. Most analysts believe that the 2000 Census and ensuing Current Population Surveys capture a large portion of the undocumented immigrants, so that our results are likely to apply to the bulk of the Hispanic population, including the majority of undocumented migrants. One way to see if our results are sensitive to possible mis-measurement due to undocumented workers not covered in the data is to see whether the results vary between states with the largest undocumented populations, where the undercount will be most serious, and other states. In 2000, 15 states contained 87 percent of the undocumented population. 2 Table 6 reports these results for the current recovery and previous two recoveries. Although standard errors are not reported, our narrative reflects whether the estimates are measured with precision. 3 The estimates are not different from those derived from the full sample. Even new entrants in those states that have the largest undocumented immigrant populations do not appear to be differentially impacted. Still, we would expect the undocumented who are missed in the CPS to have a somewhat different work experience than those who are counted. They probably have higher employment rates (which would strengthen our finding that Hispanics have such a high employment rate) but are in worse jobs than others (which would strengthen our finding that Hispanics are in low wage jobs with few benefits). 14

20 Table 6: Change in Employment-Population Ratios by Race, Ethnicity, Gender and Experience from End of Recession through Third Year of Recovery (Percentage Point Change) Panel A: All States Men Women All White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Panel B: States with Large Undocumented Immigrant Populations Male Female All White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Author s calculations from the micro data Outgoing Rotation Group Files of the Current Population Survey. The columns correspond to the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. States categorized as having the largest undocumented immigrant populations are as follows: California, Texas, New York, Arizona, New Jersey, North Carolina, Oregon, Massachusetts, Illinois, Nevada, Virginia, Washington, Colorado, Georgia, and Florida. In 2000, these states contained 87 percent of the undocumented population. In 1990, the share was 90 percent. The estimates were obtained from the Pew Hispanic Center. and We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. The Sluggish Employment Recovery and Growing Economic Insecurity The slow pace of job growth in the 2001 recovery led to greater Hispanic economic vulnerability, particularly for Mexican Americans. From 2001 to 2004, the employed shares of Hispanic men and women their employment-population ratios fell by 1.8 and 1.4 percentage 15

21 Panel A: Men Table 7: Change in Labor Market Outcomes from 2001 to 2004 by Gender, Race, Ethnicity and Experience (Percentage Point Change) All Work Full-Year Weekly Wage Health Insurance Coverage Included in Pension Plan at Their Firms Food Stamp Usage White -1.8 $ Black -1.1 $ Hispanic 0.0 $ Mexican American 0.0 $ Puerto Rican -0.9 $ Cuban 5.5 $ New Entrants White -0.5 $ Black -1.5 $ Hispanic -3.2 $ Mexican American -2.2 $ Puerto Rican 4.1 $ Cuban -0.1 $ Panel B: Women All White -2.4 $ Black -2.6 $ Hispanic -1.4 $ Mexican American -1.8 $ Puerto Rican -5.5 $ Cuban -5.1 $ New Entrants White -1.9 $ Black -2.9 $ Hispanic -4.5 $ Mexican American -5.1 $ Puerto Rican 2.5 $ Cuban $ Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. A new entrant is an individual with no more than 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. points (table 6). This deterioration is concentrated among Mexican Americans. Their ratio fell by 2.4 percentage points. Hispanic new entrant men experienced the largest erosions in their employment-population ratios, falling by 5.5 percentage points. The ratio of new entrant Hispanic women dropped by 2.8 percentage points. Considering the five previously discussed indicators of economic security suggests increased Hispanic labor market insecurity (table 7). Full-year work trended downward for 16

22 Panel A: Men Table 8: Changes in U.S.-Born Labor Market Outcomes by Gender and Race (Percentage Point Change) Work Full- Year Weekly Wages Health Insurance Coverage Included in Pension Plan at Their Firms Food Stamps White Black Hispanic Panel B: Women White Black Hispanic Notes: Authors tabulations from the March Annual Demographic files of the Current Population Survey. To be included in the sample, an individual had to be at least 16 years of age. To be included in the wage sample, the individual also had to work in the public or private sector. A new entrant is an individual with no more than 10 years of potential experience. U.S.-born individuals are second generation, third and higher generations. Standard errors are available upon request. most Hispanics, especially for women. Weekly wages show a mixed pattern. Health insurance coverage either remained the same or fell. For example, coverage among Mexican-American men and women fell 1.7 and 2.0 points. The decline in coverage was even larger for new entrant Mexican Americans. Other measures also indicate rising insecurity for some groups of Hispanics. During the recovery, Mexican Americans experienced a decline in the share of workers that actually received a pension. Food stamp usage, another measure of economic security, rose during the recovery for Mexican-American women by 2.6 percent. These findings remain when we limit our analysis to U.S.-born individuals. Even nonimmigrant Hispanics have been adversely impacted by the slow pace of job growth. Table 8 reports our five indicators for non-immigrant whites, blacks and Hispanics. Most striking is the decline in Hispanic health insurance coverage and the increase in food stamp usage. Is Growing Economic Insecurity a New Feature of Recoveries? We now place the 2001 to 2004 developments for Hispanics in a broader context. We compare changes in our list of outcomes during the current business cycle to previous business cycles. Table 6 reports this analysis for employment-population ratios. 4,5 Hispanics gained employment during the 1980s recovery, followed by stagnation during the 1990s recovery, and losses during the current recovery. This pattern holds for most Hispanic groups, with the losses during the current recovery larger for Hispanic new entrants and out of 17

23 school youth. They are more sensitive to fluctuations in the overall economy than the general population. The employed share of the Hispanic population increased by points from 1982 to 1985 (table 6). During the 1990s recovery, Hispanic ratios remained unchanged. From 2001 to 2004, their ratios fell by 1.8 and 1.4 percentage points. The movement of Mexican- American employment-population ratios in each recovery dominates the overall changes. Although not measured with a high level of precision, the estimated 2001 to 2004 changes for Cubans and Puerto Ricans show either stagnation or deterioration. Do changes in employment-population ratios differ between areas with larger and smaller proportions of Hispanics? To get a first cut at this issue, we examined whether a systematic relationship exists between changes in employment-population ratios during the recovery and a metropolitan area s share of Hispanics (table 9). In fact, an area s Hispanic population is associated with smaller reductions in Mexican-American employment-population ratios. The reverse is the case for blacks. Blacks and whites in metropolitan areas with larger Hispanic populations experienced larger drops in their employment-population ratios. The pattern holds for new entrants as well. This effect on black and white new entrants in the current recovery is larger than during the 1990s recovery. We also obtain qualitatively similar findings when we limit our samples to U.S.-born individuals. Table 9: Changes in Employment-Population Ratios by Percent of Hispanic Metropolitan Area Population Panel A: Men <3% 3 to 9% 9 to 30% 30% or more All White Black Hispanic Mexican American New Entrants White Black Hispanic Mexican American Notes: Author s calculations from the micro data Outgoing Rotation Group Files of the Current Population Survey. The columns correspond to the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise. 18

24 Table 9 cont.: Changes in Employment-Population Ratios by Hispanic Metropolitan Area Population Panel B: Women <3% 3 to 9% 9 to 30% 30% or more All White Black Hispanic Mexican American New Entrants White Black Hispanic Mexican American Notes: Author s calculations from the micro data Outgoing Rotation Group Files of the Current Population Survey. The columns correspond to the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise. Another measure of economic security is the share of people working all year (table 10). During the recovery, full-year employment increased for all Hispanics. This pattern of gains begins to break down from , with losses emerging in the most recent business cycle. Most Hispanics experience a reduction in full-year work from 2001 to The shift from gains to losses is quite prevalent among new entrants. Their losses exceed the losses in the general population. Table 10: Change in Percent Working Full-Year by Recovery and Expansion (Percentage Point Change) Recovery Expansion Men Women Men Women All White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Author s calculations from the micro data of the March Annual Demographic Files of the Current Population Survey. The columns correspond to the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise. 19

25 Table 11: Change in Hourly Wages by Race, Ethnicity, Gender and Experience from End of Recession through Third Year of Recovery Men Women All White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Author s calculations from the micro data Outgoing Rotation Group Files of the Current Population Survey. The columns correspond to the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. Out of school youth are non-enrolled 16-to-24-year-olds. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. All figures in percent, except where noted otherwise. Further, Hispanic earnings data reveal a mixed pattern of change. Changes in real earnings during the recoveries are dominated by the trend decline in real earnings that occurred from the mid-1970s to the mid-1990s, broken only by the increases that occurred during the second half of the 1990s. As a result, real earnings during the 1980s and 1990s recoveries fell. Real earnings began to increase after 1995, breaking the two-decade decline. The extremely low unemployment rates of the 1990s boom that lasted for several years helped to generate the real earnings growth. A return to stagnation and decline occurred even for Hispanics during the current recovery, further dispelling public perceptions that Hispanics have received major benefits (table 11). Other Indications of Growing Economic Insecurity Recent patterns of job growth are associated with trends in Hispanic private health insurance coverage, pension coverage and food stamp usage. Table 12 suggests that the absence of strong and sustained job growth and increased expenses of these forms of compensation have reduced the competitive pressures for employers to provide benefits. For the lowest paid workers, it has increased the need for social support. For Hispanic women and many other groups, employer-provided health insurance coverage increased from 1991 to Hispanic men were the exception. Their coverage fell. From 2001 to 2004, Hispanic employer-provided health insurance coverage decreased. The drop in coverage occurs among Mexican-American men and 20

26 Table 12: Change in Health Insurance, Pensions and Food Stamp Usage by Expansion and Recovery (Percentage Point Change) Panel A: Health Insurance Coverage Recovery Expansion Men Women Men Women Group White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Authors calculations from the micro data March Annual Demographic Files of the Current Population Survey. The columns correspond to the 1982 to 1989 and 1991 to 2000 expansions, and the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. women, and is similar to the drop in coverage of whites and blacks. New labor market entrants of all racial and ethnic backgrounds appear to have lower likelihoods of working in firms that offer health insurance. Economic insecurity for Hispanics also increased due to low and falling pension benefits. With respect to pensions, the share of Hispanics with a pension increased during the 1990s recovery, but fell during the most recent one. Changes among Hispanics are similar to those of whites and blacks, but Hispanics are more insecure, because they have much lower private health insurance and pension coverage. Mexican Americans drive the Hispanic trends. Finally, increased use of food stamps by Hispanics since 2001 further reveals the growth in economic insecurity. 6 As for blacks, this growth in insecurity is not new for Hispanics. The food stamp usage of both groups grew from 1991 to For most groups, the growth in food stamp usage is similar across recoveries (table 12). The table shows the ability of strong job growth to reduce usage. From 1991 to 2000, Hispanic men and women s usage fell by 7.9 and 8.4 percent, respectively. 21

27 Table 12 cont.: Change in Health Insurance, Pensions and Food Stamp Usage by Expansion and Recovery (Percentage Point Change) Panel B: Workers included in Pension Plan Recovery Expansion Men Women Men Women Group White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Authors calculations from the micro data March Annual Demographic Files of the Current Population Survey. The columns correspond to the 1982 to 1989 and 1991 to 2000 expansions, and the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. Table 12 cont.: Change in Health Insurance, Pensions and Food Stamp Usage by Expansion and Recovery (Percentage Point Change) Panel C: Food Stamp Usage Recovery Expansion Men Women Group Men Women White Black Hispanic Mexican American Puerto Rican Cuban New Entrants White Black Hispanic Mexican American Puerto Rican Cuban Notes: Authors calculations from the micro data March Annual Demographic Files of the Current Population Survey. The columns correspond to the 1982 to 1989 and 1991 to 2000 expansions, and the current and two previous recoveries: 1982 to 1985, 1991 to 1994, and 2001 to All respondents are men and women who are at least 16 years of age. New entrants have 0 to 10 years of potential experience. We use the current BLS terminology to identify individuals by race and ethnicity. Standard errors and results for South Americans and Other Spanish respondents are available upon request. 22