Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Earnings differences among Black, White and Hispanic males and females: the impact of overeducation, undereducation and discrimination
(USC Thesis Other)
Earnings differences among Black, White and Hispanic males and females: the impact of overeducation, undereducation and discrimination
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EARNINGS DIFFERENCES AMONG BLACK, WHITE AND HISPANIC MALES AND FEMALES: THE IMPACT OF OVEREDUCATION, UNDEREDUCATION AND DISCRIMINATION by Naomi Turner Verdugo A Dissertation Presented to the I FACULTY OF THE GRADUATE SCHOOL I UNIVERSITY OF SOUTHERN CALIFORNIA I I In Partial Fulfillment of the I Requirements for the Degree j DOCTOR OF PHILOSOPHY (Sociology) December 198 5 UMI Number; DP31850 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. Dissêrtation ftibi sbng UMI DP31850 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106- 1346 UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CAUFORNIA 90089 This dissertation, written by Naomi Turner Verdugo under the direction of h Æ Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILO SO PHY Ph.D. So Vf87 V DeanJ)f Grai if Graduate Studies Date S e p t^ b e r 6, 1985 DISSERTATION COMMITTEE airverson : ABSTRACT This dissertation examines earnings differences i between blacks, whites and Hispanics, both males and ; females, in order to estimate the effects of overeducation, undereducation and discrimination on ' earnings inequalities. While much sociological and economic research has been devoted to examining inter-ethnic earnings differences, none have included i I overeducation (Sullivan, 1978; Clogg, 1979) or its I corollary, undereducation, as independent variables in an ' earnings model. This dissertation helps to fill this ; void. It also has other features not regularly found in : research of this type. For example, due to the : heterogeneity of the Hispanic population, this I dissertation analyzed Cubans, Mexicans, Puerto Ricans and ! other Hispanics separately. Females are included in the 1 1 sample and, as with males, are analyzed by race/ethnic ! group. This dissertation also included structural and I j Labor Utilization Framework (Hauser, 1974, 1977) variables j in addition to standard human capital items, i Data are from the 1980 Census "A" Sample of the I Public-Use Microdata. Because of the difficulties associated with analyzing such a large data base, a I stratified random sample of these data were initially I selected, and cases were further limited to black, white I and Hispanic civilian noninstitutionalized wage and salary ' workers between the ages of 25 and 64 who worked at least 1,365 hours in 1979. Twelve ethnic-sex groups were I analyzed— white, black, Cuban, Mexican, Puerto Rican and other Latin males and females. The resulting subsample included 31,827 males (of whom 54 percent were nonwhite) and 19,832 females (of whom 58 percent were nonwhite). The earnings function included the following independent variables: years of education; years of 1 experience; sector of employment (government or private); I state-level unemployment rates; occupational status scores; marital status; weeks unemployed during the year (1979); hours worked during the year; English language proficiency; country of citizenship; and several dummy variables-— region of residence, period of immigration to the U.S. and overeducation/undereducation. The dependent variable is the natural log of annual earnings for 1979. A second earnings function, identical to that just ; discussed except that it excluded years of education is : also tested. I Among the major findings are these: (1) Undereducation ' was statistically significant for five of the six female I groups but only one of the male groups. As expected, j undereducation tended to lower earnings, and the earnings i of females are more dramatically affected than the : earnings of males. However, results for this variable i ; differ a great deal according to the earnings model used. 1 ! For example, when a second model is examined which omitted j I ■ the variable years of education, the results for 1 1 ' ! I undereducation are more likely to be significant. : i j (2) Regression decomposition reveals that the earnings of i I minorities and females suffer due to discrimination. ! Minority males had their earnings reduced by a minimum of i I I I $2,02 5 (for Cuban males) to a maximum of $3,917 (for ; I Puerto Rican males), relative to white males with similar j characteristics. The earnings of females are also reduced , I due to discrimination, with losses ranging from $6,560 I (for blacks) to $7,091 (for whites), Hence, if minorities and females received the same returns to education, I i i I experience, occupation, etc. as those received by white | i ! j males, a very large portion of the male-female and | ■ minority-majority earnings gap would be closed. I (3) Finally, being female has far more negative impact on ' ; I j earnings than being a racial minority. Indeed, among the : 1 i , females, gender appears to swamp the ethnic differentials, ! i I ' though ethnicity appeared to make a significant difference j I in the earnings of males. ! L _ .. I iv ACKNOWliEDG EMEMTS This dissertation would never have been completed but for the help and support of many people and organizations. I am grateful to the Graduate School of I the University of Southern California and the Haynes Foundation for awarding me a generous fellowship and to the Business and Professional Women's Foundation for the | Sally Butler International Scholarship. I also wish to | I thank the National Council of La Raza as well as Dr. Abdin i Noboa, formerly with the Latino Institute. Both : I organizations supported my earlier work in labor market ; I analysis. Of course the views expressed in this | I , dissertation do not necessarily represent the views of any ; organizations or individuals who provided me with support I and assistance. Thanks are also owed to Dr. Havens Tipps who read a I draft of this dissertation. Both he and Caroline Davis I I jGleiter provided liberal doses of encouragement and i prodded me when I needed it. ! I would like to thank the members of my dissertation I committee. Professors Audrey James Schwartz, H. Edward Ransford and David M. Heer. They all provided me with helpful suggestions. Professor Heer, who served as the V committee's chair, was particularly helpful and generous with his time--carefully reading many drafts of this dissertation. His interest and enthusiasm for the project encouraged me when my own was flagging. I To my parents, Robert and Beverly Turner, I extend my I gratitude for their continued interest in and support of ! 1 my graduate studies. Tremendous thanks are also owed to I i my husband, Richard Ruiz Verdugo, who read drafts of the dissertation, helped me to organize my thoughts and even generated computer graphics. I am forever grateful for ; his support throughout this project. And finally, thanks ! to my son, David Ari, who provided me with an overriding incentive for completing this work. L_„ VI TABLE OF CONTENTS Page Acknowledgements V Chapter 1 Chapter 2 Chapter 3 Chapter 4 ; Chapter 5 Introduction A. The Research Topic B. Why this Topic is of Concern C. How this Study Differs from Previous Research D. Organization of the Dissertation Notes to Chapter 1 Research Methods A. The Data B. Criteria for Selecting Subsamples for Analysis C. Specifying the Model D. Control Variables E. Human Capital Variables F. Structural Variables Notes to Chapter 2 Review of the Literature A. Theories of Inequality B. Inter-Ethnic Earnings Differences C. Male-Female Earnings Differences D. Overeducation E. Summary Notes to Chapter 3 Research Findings A. Within Group Results B. Between Group Results C. Discrimination D. Summary Notes to Chapter 4 Summary and Conclusions A. Summary B. Major Findings C. Discussion D. Conclusions E. Suggestions for Further Research Notes to Chapter 5 14 28 68 144 VI - J Page Bibliography 167 Appendix A. Sample Data Tabulations 177 Appendix B. 1980 Census Long-Form Questionnaire 181 Occupational Status Scores 192 States within each Geographic Region 195 Appendix E. Correlation Matrices 197 Appendix F. Regression Decomposition Tables 222 Appendix C Appendix D vxii CHAPTER 1 INTRODUCTION A. The Research Topic Earnings differences between racial minorities and majorities and between males and females has been the topic of numerous studies. These studies typically attempt to uncover the causes of disparities by using one of several well-known theoretical models designed to analyze earnings. Such models include the human capital framework (Becker, 1964; Mincer, 1970), structural models (Stolzenberg, 1978) and the Labor Utilization Framework (Sullivan, 1978; Clogg, 19 79), among others. Though each of these models has contributed to our knowledge of earnings differences, there is still much to be learned; especially why pay gaps continue to exist between minorities and majorities, and between males and females. This dissertation reviews what economists and sociologists have learned about earnings differentials, and adds to this literature by testing several new hypotheses. I This research used multiple regression to examine j earnings determinants for employed persons in each of I several groups. The groups studied were white, black and I Hispanic^ males and females. Hispanics were analyzed by I I specific national origin. That is, persons of Mexican I origin or descent (referred to as Mexicans throughout this paper) were contrasted with Puerto Ricans, Cubans and 2 "other" Latins . Variables tested as possible influences on annual ' earnings for 1979, the dependent variable, included: ! standard human capital variables (e.g., years of education; and work experience); structural items (e.g., percentage of a state's labor force which experienced at least some unemployment in 1979 and sector of employment— government | versus private); and demographic or control variables (e.g., age and marital status). In addition to the more usual variables included in the earnings model, two non-standard human capital-type variables were included asj independent variables; overeducation and undereducation. j This enabled me to estimate the effects of overeducation j and undereducation on earnings. Overeducated or "mismatched" workers, as defined by Hauser (1974, 1977), and Sullivan (1978), are those individuals whose education exceeds that of the average worker in the same occupation. Undereducation is the converse of this: those workers whose level of schooling is below the average for their occupation . (A complete description of how these variables were operationalized is provided in Chapter 2.) Following the assumptions of the human capital framework (discussed in Chapter 3), I expected that, other things being equal, overeducated people would earn more than others working at the same occupation, and undereducated people would earn less. This dissertation tested this assumption. It should be emphasized that the main interest and contribution of this research was to assess the : ! differential impact of overeducation, undereducation and 1 discrimination (or group-specific factors) on black, whitej and Hispanic males and females. Some believe that I minorities and females are often overqualified for their jobs, but fail to advance to positions for which they are more suited because of racial and sexual discrimination. Analysis of the effects of over- and undereducation on earnings helped test this perception. Estimating separate I but identical models for each group enabled me to study the ways in which race/Hispanic ethnicity and sex interacted to affect earnings. i Group-specific factors refer to variables which differ I among the groups after controlling for the variables in I j the model. Some researchers involved in the decompositionj of such differences with respect to earnings refer to this| as a measure of the "cost" of being discriminated against in the labor market, though others have criticized this interpretation arguing that earnings differences not accounted for by the model may be due to factors besides discrimination, such as the quality of education received or family socialization. Hence, it has been argued that j this technique may overestimate the effects of | discrimination on earnings. Other researchers note, however, that this same technique may also be seen to underestimate discrimination as it excludes any discrimination faced in obtaining an education and j entering the labor force (Cain, 1984). For the purposes j of this study, that portion of the earnings decomposition i i not explained by the model, the residual, has been referred to as the cost of discrimination since this is the best method to date for estimating the impact of discrimination on earnings (Williams, 1982:54). For more on this technique, see Chapter 4. The main questions addressed by this study were: (1) What were the magnitudes of the earnings differences among white, Hispanic and black males and females in 1979?; (2) How have overeducation and undereducation affected the I earnings of black, white and Hispanic males and females?; | (3) What is the effect of discrimination on the earnings j of white females, minority females and minority males?; | I and (4) What is the impact on earnings of being both | I I minority and female, as opposed to being either minority or female? B. Why this Topic is of Concern Income refers to all monies received in a year, both earned income (e.g., wages and salaries) and unearned income (such as unemployment compensation. Social Security, pension distributions, public assistance payments, interest, dividends and the like). Earnings, those monies received in the form of wages, salaries and net income from self-employment^^ , are by far the largest component of income for most people. Current Population [ Survey data for 1979 indicate that 48.8 million families j i received wages or salaries and that this form of payment i j j comprised 84 percent of the average total income for thesej I families (U.S. Bureau of the Census, 1981:7). Earnings is I ; 1 of interest to researchers because it is a major | I determinant of life style. In addition to money required | for the purchase of necessities such as food and shelter, ; it is also used to buy education and job training, j I I I childcare, leisure, and other goods and services. ! ! Earnings also has a great deal to do with how one is | j I j regarded in the community and thus is used to enhance j one’s prestige. Further, earnings not only affects the j ' life style and prestige of the wage earner, but that of the entire family. The status attained by one's offspring is influenced by the earnings of the parents. For example, it is often the parents who make decisions and pay bills associated with children's education. In this way, the earnings of parents may influence the quality and quantity of education received by their offspring. : Earnings is, thus, a major determinant of the way in which; we and our children live. The tremendous disparities that existed between whites| and blacks in terms of earnings and education was a major impetus for the civil rights legislation of three decades ago. Now, some 30 years after Brown v. Board of Education, and 20 years after passage of the Civil Rights Act, disparities still remain between whites and minorities. Indeed, there are some researchers who claim that, after adjusting for inflation, there has been no real reduction in the black-white earnings gap. As noted by Squires, while the nonwhite/white median family income ratio has increased from .54 in 1947 to .62 in 1974, ...the absolute income difference between these two groups [based on 1971 constant dollars] has been getting larger....Non-white families, therefore have been falling further behind whites since 1947. In addition, since, a greater proportion of wives in non-white families work, income inequality is further suppressed by focusing on family income. When the income of unrelated individuals is the unit of analysis, the nonwhite/white ratio has 6 i . ..J declined. This ratio dropped from .71 in 1947 to .58 in 1960, then rose to .73 in 1965 and declined again to .67 in 1971. Much has been made in recent years of the rising black middle class, but the facts do not support prevailing beliefs about the supposed equalization of the races, at least in terms of income (Squires, 1979 : 162-163). I Others have also pointed out that since 197 5 the median | i income ratio of black families to white families has ! I declined (Felder, 1984) or at least has not increased | (Cain, 1984). Cain notes that this lack of progress sincej the mid-1960s may be due to increases in black ! I female-headed families and high unemployment rates since | 1975. "Whatever the reason, progress regarding... family income differences...has been painfully slow" (Cain, 1984:4). Similar results have been found for Hispanics (Long, 1977). Blacks and Hispanics often differ tremendously from whites in terms of the jobs at which ; they work, the level of education attained, prevalence of home ownership and numerous other conditions (U.S. Commission on Civil Rights, 1978). Much of this appears to be related to earnings differences. This analysis of earnings differences should help to identify the determinants of earnings, reveal the magnitude of existing inequalities between groups, and evaluate that portion of the earnings difference due to discrimination and other factors not included in the model. c . How this Study Differs from Previous Research Studies of the labor force are anything but rare. Indeed, current research in sociology is rife with studies of the labor force, occupations and labor markets. How i then can yet another study of the American labor force be | justified? There are five ways in which this study j j differed from previous research. First, the most ; I important distinction between the topic presented here and other research is that in this study overeducation and I undereducation have been used as independent variables in | the regression model. Though studies have been conducted | t which focus on the incidence and distribution of I overeducation among various populations, none have used ! overeducation as an independent variable in an earnings | 5 ! model . Consequently, the degree to which various | 1 measures of underemployment (such as overeducation) affect earnings is not known. A major contribution of this research has been the examination of the impact of overeducation and undereducation on earnings. These variables will also serve to test the assumption that females, particularly minority females, tend to be I overqualified for their jobs, and that they are confined I to largely low-paying jobs because of discrimination. Second, the data base used was the 1980 Census r Public-Use Microdata. Because these data have only recently become available, few analyses using 1980 Census data have been conducted to date. Though more recent data have been analyzed, namely the Current Population Surveys, they have much smaller sample sizes than the 1980 Census ! Microdata, often not large enough to stratify Hispanics by: national origin (e.g., Mexicans, Puerto Ricans, Cubans and! other Latins). Third, analysis of Hispanics by four specific ethnic ! groups, is another key feature of this study. Much ! i research of this type compares whites and blacks, but only, I I i rarely are Hispanics included in such studies. (However, | I I as the size and influence of the Hispanic population ; : I j increases, a growing number of surveys are collecting data' j I I on Hispanic origin or descent.) Even when Hispanics are ! . included, they are often analyzed as if they were a homogeneous group when, in fact, Hispanics comprise a veryi heterogeneous population. Also, as of 1980 Mexicans I comprised about 60 percent of the Hispanic population j (U.S. Bureau of the Census, 1983b : 40) so characteristics I of Hispanics, when analyzed as a single group, would tend ' I ' 1 to reflect Mexicans. Generalizing from Hispanics to j specific nationality groups, such as Cubans, can thus I I ' I prove very misleading. For these reasons it is preferable to study specific Hispanic ethnic groups separately rather than combining them into a single group. In this study the following Hispanic groups were analyzed: Mexicans, Puerto Ricans, Cubans, and other Latins. Fourth, much labor force research includes only ; males. Given the large percentage of women in the labor force^, and the trend toward more consistent female | employment throughout the life cycle (Niemi and Lloyd, 1981:74), it is important that females be included in economic research. In this study, in addition to considering black, white and Hispanic males, I have considered females by race and Hispanic origin group. Fifth, the inclusion of structural and Labor Utilization Framework variables, as well as the more standard human capital items, distinguishes this research from most other labor market studies. Structural variables included in this study were sector of I ' employment, region of residence, and percent unemployment ' by state. Variables such as these permit study of the i 1 I impact of specific conditions, beyond an individual’s own I characteristics, which may be affecting earnings. Labor I t Utilization Framework variables are overeducation and its corollary, undereducation. Hence, elements of several theoretical models— human capital, structural and the 10! Labor Utilization Framework— have been combined in the earnings model used in this research. D. Organization of the Dissertation This dissertation contains five chapters, plus a bibliography and appendices. Discussion of the research methods. Chapter 2, follows the introduction. The data sources are described as are the sample selection criteria. The regression model is also presented in Chapter 2, and each variable is operationalized and discussed. The literature review is presented in Chapter 3, Within this chapter the literature on theoretical models of inequality is briefly reviewed as is the more applied research on earnings differentials (both inter-ethnic/racial and male-female earnings differentials, as well as research which estimates discrimination) and overeducation. Hypotheses relating toi i the effects of overeducation, undereducation and the ' i impact of labor market discrimination on ; inter-ethnic/racial and male-female earnings differentials ; are also presented in Chapter 3. ' I Results of the analysis are presented in Chapter 4. ; Descriptive statistics of the variables are presented. j Regression results and statistical analyses of ' Hi discrimination (i.e., the "cost" of being minority and/or female) are also presented. Chapter 5 summarizes the research and draws conclusions. The bibliography and appendices follow. Items to be included in the appendices are: (A) tabulations showing distributions of the sample data by age, marital status, hours worked, etc. across race/ethnic and sex groups; (B) the 1980 Census long-form questionnaire; (C) information pertaining to the occupational status scores used in this analysis; (D) a table listing the states within each region of the country; (E) correlation matrices for each of the twelve population subgroups; and (F) a series of 24 tables showing results of the regression decomposition using, in turn, each of twelve ethnic-sex groups as the standard population and data from the two regression models analyzed. 121 J Notes for Chapter 1 1. Throughout this dissertation the terms "whites" and "blacks" will refer to non-Hispanic whites and non-Hispanic blacks, respectively. Hispanics may be persons of any race. 2. "Other" Latins refers to those persons from Spanish-speaking countries in the Caribbean (with the exception of Cuba and Puerto Rico), Central and South America and Spain. 3. I am indebted to Professors Philip Garcia and David M. Heer for developing the concept of undereducation and suggesting it be analyzed in this study. 4. Persons classified as self-employed have been excluded from this study. 5. Conversations with Professors Clifford Clogg and Teresa Sullivan, pioneers in the Labor Utilization Framework field, have assured me that this is indeed a unique topic. 6. Figures from the March 1984 Current Population Survey indicate that 53.2 percent (49,210,000) of women age 16 and over were in the labor force, though 7.7 percent of them were unemployed. Of the 45,414,0 00 employed women, 72 percent worked full-time (35 or more hours per week), while 28 percent worked part-time (Hayghe, 1984:32). 13 CHAPTER 2 RESEARCH METHODS This chapter provides information pertaining to the research methods used in this study. The data are described as are the sample selection criteria. Data are : presented which describe the characteristics of persons included in the subsamples, such as distributions by race,, sex and Hispanic ethnicity. Variables used in the regression model are presented and operationalized, and ! I statistical methods are discussed. A. The Data i i I I The primary source of data for this study is a | I subsample of cases from the 1980 Census "A" Sample of the ! I Public-Use Microdata. Data in the "A" Sample were drawn ' i from the long-form questionnaire which was sent to 19.4 i percent of all households. "A" Sample includes over I : one-quarter of the households that received the long-form I questionnaire. Fully 5 percent of all those enumerated are included, comprising over 11 million persons and 4 million housing units. (See column 1, Table 2.1.) B. Criteria for Selecting Subsamples for Analysis To minimize the problems associated with analyzing such a large data base, a stratified random sample of these data were initially selected for analysis in this 14 Table 2.1 Unweighted Tally of Persons Included in "A" Sample of the 1980 Census, and Selected Subsamples Used in this Study Initial Percent Incl. Research Race/Origin Group "A" Sample Size (1) Subsample Selected (2) in Initial Subsample (3) Subsample® Males Females (4) White 9,038,697 89,849 1% 14 , 601 8,337 Black 1,313,902 26,523 2% 3,046 2,721 Mexican 438,590 26,296 6% 3,570 1 ,632 Puerto Rican 101,751 25,977 25% 3,009 1,578 1 Cuban 41,106 23,983 58% 3,889 3,120 Other Latin 157,307 26,696 17% 3,714 2,444 TOTAL 11,091,353 219,324 -- 31,827 ' 19 , 832 i ^ The number of cases in the regressions and other analyses may be less than those presented here depending on the number of cases deleted due to missing data, NOTE: The "A" Sample includes persons residing in group quarters. The "Initial" and "Research" Subsamples exclude such persons. The Research Subsample was also limited to noninstitutionalized civilian wage and salary workers employed at least 1,365 hours in 1979 and having earnings in excess of $0. Also, only blacks, whites and Hispanics between the ages of 25 and 64 were included. study. Groups were selected from "A" Sample as follows 1 percent of whites, 2 percent of blacks, 6 percent of Mexicans, 25 percent of Puerto Ricans, 58 percent of 151 I Cubans, and 17 percent of other Latins, (See columns 2 ! and 3, Table 2.1.) Because this study concerns the analysis of earnings differentials, efforts were made to analyze comparable ! I workers. For example, only year-round full-time workers | were selected for analysis. (Persons who worked at least ' 1,365 hours in 1979 are defined as year-round full-time ■ I T I , workers for the purposes of this study .) Consequently, j I unemployed persons, discouraged workers and others who I I ! ' were not employed for at least 1,365 hours in 1979 have I been excluded from the analysis. ! ; In addition to persons working less than 1,365 hours, j I unpaid family workers, military personnel and persons j 1 living in institutions or group quarters were also | I excluded. Further, persons who were not black, white or j Hispanic have been excluded, as were persons under age 25 ■ i I 1 (because many in the 16-24 age group are still in school) I I or over age 64. This research was also limited to personsj ; having positive earnings. Finally, self-employed persons ■ were excluded for two reasons. First, factors which lead ' j to financial success in self-employment are likely to be i very different from factors which influence the earnings i I of wage and salary workers. Therefore, human capital ! models would be less effective predictors of earnings for 16- - . J the self-employed as compared to wage and salary workers. Second, the number of self-employed minorities by sex are too few to permit a comparative analysis of these persons. The subsample of cases which remain and are analyzed in this study are thus comprised of black, white and ! Hispanic civilian, noninstitutionalized wage and salary workers between the ages of 25 and 64 who worked at least i 1,365 hours in 1979 and earned at least $1. Column 4 in Table 2,1 presents the number of persons selected for this! I analysis by race, Hispanic ethnicity and sex. (Descriptive data for this subsample are presented in Appendix A.) C. Specifying the Model In order that the determinants of earnings can be identified and compared, it is necessary to develop a regression model which is the same for each ethnic-sex group. The following model will be used in this dissertation. ! lnY=f(EDUC, EXPER, SOVEREDt, SECTOR, UNEMP, SREGION^, OCCUP, MARITAL, WEEKSUN, HOURS, LANG, CITIZEN, ZIMMIGi) where : InY = natural log of annual earnings for 1979; EDUC = years of education; i . Z J EXPER = years of work experience; ZOVEREDî = a series of dummy codes indicating whether an individual is overeducated, undereducated or average ; SECTOR = sector of employment (government versus private); UNEMP = state unemployment rate; ZREGIONi = a series of dummy codes for region of residence ; OCCUP = occupational status scores for 3-digit Census occupation codes; MARITAL = marital status; WEEKSUN = weeks unemployed in 19 79; HOURS = hours worked in 1979 ; I I LANG = English proficiency; i , CITIZEN = country of citizenship; I I ZIMMIGi = a series of dummy codes for date of immigration I to U.S. and a code for native born Americans. j Variables were selected for inclusion in the model if ! they were: (1) common to research of this type (e.g., I years of education); (2) necessary for testing specific I hypotheses (e.g., overeducation); or (3) potentially I helpful in analyzing earnings differences among the ethnic j I groups considered here (e.g., English proficiency). ; : I I Though many econometric models use hourly or weekly j earnings as the dependent variable, this can sometimes ! prove misleading, particularly for seasonal workers. (For L_____________ _ . .A5 example, analysis of hourly wages in the case of a farmworker making $10 per hour but working only 9 months of the year would not accurately illustrate the income position of that worker relative to another who earns $10 i I per hour and works the year round.) Consequently, the , dependent variable for this study is annual earnings in ' 1979. Analysis of the earnings variable revealed that it : was not normally distributed, but was skewed toward higher! I earnings levels. The distribution of the natural log of ] annual earnings was then analyzed. This distribution was less skewed and more closely resembled the normal 2 distribution . (See Table 2.2) Therefore, the natural log of annual earnings in 1979 will be the dependent variable. Table 2.2. Summary Statistics for Actual Earnings and the Natural Log of Earnings from the Weighted Subsample of Cases Selected for Use in this Study Actual Statistic Earnings Log Earnings Mean 16052.919 9.490 Standard Error 6.115 .000 Standard Deviation 10133.283 .689 Median 14005.000 9.547 Mode 12005.000 9.393 Skewness 2.060 -1.656 Kurtosis 7.516 10.036 19] It should be noted that earnings refers to all wages, salary, commissions, bonuses or tips received from all jobs before payments for personal income taxes and deductions for Social Security, union dues. Medicare, etc. (See Item 32a in the Census questionnaire. Appendix | I B. ) I The independent variables in the earnings model can be defined as one of three types. These are: control or demographic variables; human capital variables ; and i structural variables. Rationales for selecting each , variable are presented in the following paragraphs. , D. Control Variables } English proficiency (LANG), date of immigration ; (IMMIGi) and country of citizenship (CITIZEN) are potentially useful in explaining earnings differences between Hispanics and other groups. Certainly one's ability to speak English is an asset in finding work and negotiating wages. It is possible that country of citizenship and date of immigration could be proxy j measures for knowledge of American society and labor markets. Marital status (MARITAL) is commonly included in ; i studies of this type as a control item. Previous research ' 1 has found that married men tend to earn more than 201 unmarried men (Verdugo and Verdugo, 1984), It is assumed i that this is the case because married men work more hours, are employed more consistently throughout the year and are more advanced in their careers due to their greater age. However, it is not the case that married women earn more than unmarried women. Hours worked in 1979 (HOURS) is a combination of both hours per week and weeks per year worked and, therefore, , takes into account periods of unemployment. This item hasj been included in the model since we know that the number j I of hours and weeks worked have positive effects on I I earnings. The number of weeks unemployed (WEEKSUN) would, | like HOURS, have a direct bearing on earnings. It is expected that weeks unemployed is inversely related to j earnings. j Occupation (OCCUP) is used as a control item since one's earnings are, in part, determined by occupation. However, merely dummy coding broad occupational categories could prove misleading due to the tremendous range of occupations and salaries within a single broad category. ' Consequently, occupation has been coded according to the occupational status scores developed by Ford and Gehret using data for the total experienced labor force from the 1980 Census (specifically, the one percent Public Use 21 Sample). Status scores were based on an average of median earnings and median education for each 3-digit occupation. However, due to the small number of persons in some of the detailed occupations, I have combined some occupations and computed weighted average occupational i I i status scores. (For more on this see Appendix C.) Status; I 3 I , scores for occupations range from zero to 99 . : E. Human Capital Variables j Perhaps the most commonly used human capital item is I years of education attained (EDUC). Education represents | i training, increased knowledge, and enhanced skills. It is j expected that these items would increase one’s earnings. j Years of work experience in the labor force (EXPER) isi included in the model as a determinant of earnings. It is expected that, other things being equal, earnings rise as labor force experience increases. While the 1980 Census does not provide data on actual years of work experience, an estimate of job experience has been computed based on the age of the worker and the years of education attained. (For details on how this variable has been operationalized, see Table 2,3.) Overeducation (OVERED) is not commonly included in human capital earnings functions. Indeed, as mentioned in Chapter 1, this is the first such study to include 221 - ^ ! overeducation as an independent variable in a model of this type. Still, overeducation can be seen as a human capital variable since it reflects investment in. education. As defined in the Labor Utilization Framework, overeducated persons are those whose education exceeds (byi ! over one standard deviation) that of the average worker in the same occupation. The concept of undereducation (UNDERED) is original to* i this research. Workers are defined as undereducated if j their education falls below the average for workers at j j their same occupation by more than one standard | deviation. Note that, as with the occupation variable (OCCUP), analyses done to determine whether persons were overeducated, undereducated or average, were based on a collapsed set of 3-digit occupations to ensure that there were at least 50 persons in each occupation. Based on human capital theory, which holds that education is a proxy measure for worker productivity, we would expect that overeducation would enhance one's earnings while undereducation would diminish earnings. OVEREDi identifies three educational categories (overeducated, undereducated and average) using two dummy variables. 23; 1 F. Structural Variables Including structural variables in the earnings model is an attempt to measure the impact of, not specific jobs, but larger characteristics of the labor market. Sector of employment (SECTOR), defined as government versus private ! sector employment, is one such structural variable. Given' the supposedly close surveillance of equal opportunity | practices of government, it may be less easy to discriminate against minorities and women in that sector. The unemployment rate for each state in 1979 (UNEMP) is an attempt to measure the relative strengths of labor markets at the state level. While comparing unemployment rates for smaller units of analysis, such as SMSAs, would be preferable (Szymanski, 1976:408), this subsample of Census data are not suited to analysis below the state level. In any case, even state labor markets can be seen 4 to be highly diverse . Region of residence (REGIONt) was included since there is evidence that wage levels vary between regions. This can be particularly important in analyzing minorities, such as blacks and Hispanics, who tend to be concentrated in particular areas of the country (Featherman and Hauser, 19 78). 24! . . . J The model described above will be estimated separately for each of the twelve race/ethnicity/sex groups considered in this study. A complete listing and description of the variables is presented in Table 2.3. i I Table 2.3. Description of Variables Used in the Analysis 1 Variable Operationalization InY Natural log of wages and salary in 1979. EDUC Highest grade completed (as of 1980). EXPER Maximum potential years of work experience. EXPER = (AGE - EDUC) - 5 OVEREDi A series of two dummy variables measuring three educational classifications. OVERED = overeducated. A worker is overeducated if his/her education is more than one standard deviation above the average for his/her 3-digit Census occupation code, (Occupation refers to 1980.) UNDERED = undereducated. A worker is undereducated if his/her education is more than one standard deviation below the average for his/her 3-digit Census occupation code. (Occupation refers to 1980.) Omitted = average (neither overeducated nor unde r educ at ed) NOTE: In tables OVERED is listed as OVEREDi while UNDERED is noted as OVERED2 to indicate that they are dummy variables. SECTOR Sector of employment (as of 1980). 0 = private sector 1 = government sector UNEMP State unemployment rates for 1979, 251 REGION I OCCUP MARITAL I WEEKSUN A series of three dummy variables measuring the four geographic regions of residence (as of 1980) . NORTHEAST SOUTH WEST Omitted = North Central (Appendix D lists the states included in each region.) NOTE: In tables NORTHEAST is listed as REGIONi, SOUTH as REGION2 and WEST as REGION3 to indicate that they are dummy variables. Occupational status scores (ranging from 0 to 99) for the total experienced civilian labor force. Occupations were collapsed to ensure that each detailed occupation contained at least! 50 persons. (For more on this see Appendix C.) i Weighted averages of the occupational status scores, as developed by Ford and Gehret, were computed to account for combining some of the I occupations. | Marital status (as of 1980). 0 = married (including separated) j 1 = unmarried (including widowed and divorced) Number of weeks unemployed in 1979. (Zero for | those who experienced no unemployment.) | HOURS LANG CITIZEN IMMIGi Number of hours worked in 1979. HOURS = (hours worked per week worked in 1979) in 19 79) (weeks English language proficiency (as of 1980). 0 = speak English well, very well or speak English only 1 = do not speak English well or not English speaking Country 0 = not 1 = U.S of citizenship a U.S. citizen citizen (as of 1980) A series of two dummy variables measuring three periods of immigration to the U.S. IMMIGi = 1970-1980 IMMIG2 = before 1970 Omitted = born in the U.S. 26 Notes for Chapter 2 1 . This figure of 1,365 hours was arrived at by figuring year-round full-time as persons working at least 39 weeks per year and 35 or more hours per week. (39 x 35 = 1,365.) While this definition is somewhat less than that used by the Census Bureau (a minimum threshold of 1,365 hours per year as compared to 1,750 hours), it has been defined so as to include persons who work 9 months out of the year, such as teachers. Though 9 months of work is less than year-round, such persons would seem to have a strong attachment to the labor force. While some other labor force studies include anyone who worked one week or more, a somewhat high threshold has been set for this study in order to exclude students working summers and others only temporarily attached to the labor force. 2. In addition, Pareto (1897) found that income was distributed lognormally. 3. It should be noted that 1980 Census data pertaining to occupation and certain other variables refer to the reference week rather than the previous year, 1979. (The reference week is that week in 1980 which immediately preceded the week in which the long-form questionnaire was completed.) However, earnings, weeks of unemployment and various other items refer to 1979. Hence, variables which refer to 1979 are mixed with variables that refer to the reference week in 1980. There is thus some error introduced into the model. One such error, for example, is for those persons who changed from one 3-digit occupation code in 1979 to another occupation in 1980. This is not a serious problem. Such mixing of variables is unavoidable and frequently done by researchers using Census data. John Priebe, an analyst at the Bureau of the Census, has indicated that only a small fraction of individuals change occupations within a 1 2-month period, so the error introduced by using occupation for 1980 in a model of earnings for 1979 should be negligible. 4. Data for this variable are from U.S. Bureau of the Census, Statistical Abstract of the United States: 1984 (104th ed.), Washington, D.C. : 1983, p. 424. 2^ CHAPTER 3 REVIEW OF THE LITERATURE There has been much research to date on the issues of earnings differences between various ethnic and racial groups. However, some topics are frequently omitted from this literature. This dissertation examines several of these oft-omitted topics. First, though previous research has examined between-group earnings differences, the typical focus is on the earnings of white and black males only. Hence, little is known about inter-minority differences, such as those between blacks and Puerto Ricans. This study, however, examines earnings differences between various Hispanic ethnic groups as well j as whites and blacks. Second, given the high levels of I female labor force participation, it is surprising how ! much of the current research continues to focus on males I I exclusively. Hence, this study examines differences ! between males and females for each of these groups. I I ■ Third, though overeducation has been studied, it is still : a fairly new research topic and has never been included as an independent variable in an earnings model. This I i research does, however, examine the impact of ! ; overeducation and undereducation on earnings. ! 1 Consequently, this research is an attempt to fill some of 28 these gaps in the research. This research focuses on three major areas: (1) earnings differences between Hispanics, blacks and whites; (2 ) earnings differences between the sexes for each of these ethnic/racial groups; and (3) the impact of overeducation and undereducation on earnings. Hence, this review of the literature addresses each of these areas as well as others. In Part A various sociological and economic theories dealing with earnings inequality are reviewed. In Part B selected literature pertaining to inter-ethnic earnings differences is reviewed. Though research on earnings differences between blacks and whites is reviewed, a special effort has been made to review the literature on Hispanic-black, Hispanic-white and intra-Hispanic earnings differences. Part B also includes a review of literature estimating the impact of racial/ethnic discrimination on earnings. In Part C, the literature pertaining to male-female earnings differences is discussed as is literature on the effect of sex i I discrimination on earnings. Part D reviews the literature on overeducation.^ Finally, Part E presents a brief summary of the chapter. 29 A. Theories of Inequality The disciplines of sociology and economics have evolved several paradigms for the study of inequality. Much sociological research utilizes the status attainment I framework (Blau and Duncan, 1967) to explain socioeconomic* outcomes, while economists often rely on the human capital' framework. Status attainment research emphasizes ; socialization experiences, such as family origins and educational attainment as predictors of occupational and earnings outcomes. Human capital research emphasizes the j role of investment in human capital as a means of ! I increasing earnings. (Human capital is broadly defined to j I include education, work experience, training, improved ! health, migration and job information.) (See, for i I example, Becker, 1964 and Mincer, 1974.) The rationale ofj ! human capital proponents is as follows: "Those who invest to increase the quantity or quality of human capital can expect greater incomes because those investments presumably cause increases in marginal productivity" (Parcel and Mueller, 1983:6-7). Only recently has there been any inter-disciplinary research, and this has primarily come from sociologists testing or applying aspects of human capital theory to research on income or occupational inequality. Many data bases are more suited 30 to human capital applications than to the status attainment framework, since the latter may require time-series data, or at least information on parents' educational and occupational backgrounds. Consequently, | 1 this dissertation emphasizes the human capital framework | over status attainment theory due to the lack of parental ' ; information to be found in Census data. The human capital - I and status attainment frameworks are probably the most | I commonly used sociological and economic models for I ; studying inequality. ! * Granovetter (1981) criticizes the human capital and ; status attainment models for focusing exclusively on ’ supply side factors in analyzing income differences. He I ( I argues that demand side factors must be considered, since ! I "...whether one's 'investment' pays off depends on whether j ' there is demand for what one's acquired skills can ; produce..." and whether the individual has been matched with a job commensurate with his or her skills (Granovetter, 1981:18). While the education, experience and overeducation : variables included in the earnings model used in this I research can be seen to reflect investments in human I capital, the occupational status variable can be seen as I ! measuring the "value" of specific occupations to society 31 (Horan, 1978). In that sense, demand side factors are, if only indirectly, incorporated into the model used in this : research. In addition to the human capital variables some ^ structural variables, which measure characteristics of labor markets as opposed to individual characteristics, ! have been included in the earnings model. The importance of such variables has been noted by Parcel and Mueller 1 (1983:2-3): I Theoretical analyses of institutional discrimination suggest that the handicaps j economic minority groups face are external to the ; human capital or background characteristics they . embody and instead are a function of institutional arrangements that operate to their | disadvantage. ; Variables such as region of residence and state unemployment rate may be seen as structural variables. | Region, for example, is a variable consistent with dual or segmented labor market theory. The dual labor market perspective argues that jobs are located in one of two sectors, primary or secondary. Jobs in the primary labor market offer opportunities for advancement and training, and provide benefits to the workers, often including health insurance and retirement pensions. Jobs in the secondary labor market offer no upward mobility, no opportunities for acquiring new skills, and little or no 32 I I benefits (Piore, 1975). Work in the secondary labor market may often be intermittent or seasonal. It is • argued that women, the urban poor and black teenagers comprise the majority of employees in the secondary labor market. To the extent that minorities and women are concentrated in these labor markets and that the impact of j these labor markets differ by geographic region, the j regional variables may prove to be important predictors of earnings. Zucker and Rosenstein note, however, that human I I I capital variables have "...been found to be more important| : i in determining income in the primary, as compared to the j secondary labor market" (1981:870). Others have also I ^ noted that education and age are only modest determinants of earnings for those employed in secondary labor markets (Freeman, 1981b). Given the selection criteria of the sample, particularly the stipulation that workers be 1 employed at least 1,3 65 hours during the year and be at ' least 25 years old, I suspect the vast majority of persons I ' selected are employed in the primary labor market. Hence, i ; it is expected that, in this analysis, human capital variables will prove to be more important than the structural variables. 33 B. Inter-Ethnic Earnings Differences Recall that the dependent variable in this analysis is ; the log of annual individual earnings for 1979. While family income might be a more realistic unit of analysis for analyzing the actual disposable income available to | ■ I family members, there are limitations associated with its I , i [ use. These limitations stem from differences in the | number of persons and number of workers across family units (Wohlstetter and Coleman, 1972:3). This leads to problems in the interpretation of earnings differences, : particularly inter-ethnic differences. As Michelson (1968:86) explains, i (i)t is important in people as well as commodities to separate price and quantity in ; ' determining total cost of income. For example, | ; if two families earn the same income, but there I ! are three full-time workers in one family, one in| ! the other, then the process should not be judged 1 as equal, though the outcomes may be. i ; I , Indeed, others have also argued that the use of family , j income to assess earnings inequality or discrimination can g j be misleading. "Since family income results from a ^ plethora of combinations of varying numbers of persons working some full and some part time at different levels, ; it is a poor (and notoriously misleading) measure of economic discrimination" (Villemez, 1978:772). For these 1 reasons individual earnings has been selected as the 34 r I dependent variable in this analysis of inter-ethnic earnings differences. One last point with regard to the dependent variable, earned income, is this: Black families are more heavily dependent on earned income than are white families; white ; families derive a significant proportion of their' income from interests, dividends, and rent. This' means that black family income is more heavily I affected by rising unemployment rates than white ; family income (Felder, 1984:55), I This fact also suggests that estimates of income inequality based on all sources of income would reveal a i larger black-white income gap than estimates based on f earned income alone. Hence, this dissertation probably i 1 underestimates black-white disparities relative to ! estimates based on disposable family income from all | sources. One of the goals of this research is to assess the i magnitude of inter-group earnings differences. Another is to assess the degree or "cost" of discrimination on the earnings of minorities and women. Both these topics will be reviewed in this section. Concern with reducing inter-group inequality stems I ; largely from federal legislation, namely the Civil Rights i ! Act of 1964 of which Title VII prohibits discrimination in : employment on the basis of race, color, religion and ' national origin. (The law was expanded in 1967 through 35 Executive Order 11246 to cover sexual discrimination as well.) Basically, the law requires that firms refrain from "'disparate treatment'" and that they apply neutral standards which do not result in "disparate impact." | There is agreement among researchers across the political spectrum that blacks earn less than whites (Felder, 1984; Sowell, 1981; Squires, 1979; Stolzenberg, | 1975; Williams, 1982) and that Hispanics fall somewhere in; between (Briggs, Fogel and Schmidt, 1979). However, there, is much disagreement about why this is the case. Williams; (1982), for example, in noting that black-white income ; ratios for female professionals are far narrower than for I their male counterparts (1982:55), speculates that i ...a good part of the income differential between, white and black male professionals may lie in occupational distribution differences between the two populations. This would imply that even if white and black males were paid identical incomes i within an occupation, significant differences 1 would occur when measuring median professional ! income (Williams, 1982:64). ! j Like Williams, Stolzenberg (1975) also sees the ' differential distribution of black and white men across occupations as a major factor in black-white earnings I I differentials. Indeed, Stolzenberg found that, within i detailed occupations, differential returns to education ’ between the two groups do not account for the black male-white male earnings gap. He argues that 36 . . . the best way to assure black men a fair return on their educational qualifications is probably to assure them fair access to employment in all occupations. The problem does not appear to be that black men are less able than white males who work in the same occupation to convert their schooling into earnings. Rather, it seems to be that the blacks are less successful than whites in converting their schooling ; into employment in better-paying occupations ■ (Stolzenberg, 1975:314). | Though Stolzenberg approaches this research from a human ; capital perspective by focusing on education, he found that differences in educational quantity and quality j cannot explain black-white earnings differences among ' I males in the same occupation (Stolzenberg, 1975:316). ' ' I I Hence, it appears that blacks are not only more heavily ! I concentrated in lower-paying occupations than are whites, ! but that when employed in the same occupation, black males i tend to be paid less than white males. Squires, in attempting to explain "... what little progress has periodically occurred..." with respect to the ! white-nonwhite earnings gap, states that improvements "can be accounted for by a combination of factors. And education is a minor one at best" (Squires, 1979:167). Among other factors cited by Squires are the movement of blacks from the South to the North where wages are higher, and civil rights legislation which has enabled blacks to enter white-collar and skilled jobs. Yet, he notes, "(o)ne thing which is clear is that much more progress has. 3 7; been made in equalizing educational attainment than in equalizing income..." (Squires, 1979:167). Further, Squires argues that those nonwhites who have made it and have penetrated the high paying jobs "...have needed more education than whites in order to do so....While it is | difficult to pinpoint the effect of formal education on j racial inequality, with all things considered, it appears ] ! that education has been given far more credit than it ‘ deserves in terms of reducing that inequality" (Squires, 1979 : 169-170) . Similar results have been found for Chicanos. Garcia, citing research by Poston, Alvirez and Tienda, notes that "Chicanos with equal education levels are less likely to | hold equal occupational status and earn less than their | Anglo counterparts" (Garcia, 1979:4). j Freeman, however, believes that "deterrents to j economic progress" differ for blacks and Hispanics. Increases in earnings among blacks "...have been achieved by black women (who have attained virtual economic parity with white women in earnings) and more educated and skilled blacks" (Freeman, 1981b:137). But he notes that there has developed a widening gap between those blacks employed in the "mainstream economy" and those outside the mainstream. That is, while "... an increasing proportion 38 of employed black men hold better jobs, an increasing proportion are also apparently out of the mainstream economy" (Freeman, 1981b:140). Further, "(t)raditional equal employment activities do not seem to offer a route ! out of economic distress for many less skilled black | workers" (Freeman, 1981b:107). In essence. Freeman's ! ! argument is that civil rights legislation and affirmative j action measures have helped to advance middle-class ; 1 blacks, while those most in need remain unaffected. With | regard to Hispanics, he states that the "...lack of , education and lack of skills appear to be the most j important deterrents to economic progress" (Freeman, i 1981b:107). Just as there are a variety of explanations for inter-group earnings differences, researchers are not in agreement on the impact of discrimination on earnings. Cain identifies two definitions of discrimination. "First, economic discrimination may be defined as | long-lasting inequality in economic well-being among individuals based on their color, gender or ethnic ties. I Second, economic discrimination is also defined as I I differences in pay or wage rates for equally productive groups" (Cain, 1984:2). Based on the first definition, economic inequality, Felder argues that "...racial 39 ; discrimination continues to impede the economic progress 1 of black Americans" since blacks working full-time earn less than whites working full-time, even after controlling for educational level and occupational category (Felder, 1984:55). i Following Cain's (1984) second definition, there is | ' much research which compares "equally productive groups" through the use of multivariate techniques (Ragin, Mayer I j and Drass, 1984:221). The discussion which follows I I concerning the "costs" of discrimination is based on : research utilizing multivariate methods. Typically, black males are found to experience more i ; intense forms of discrimination than do Hispanic males - I (Freeman, 1981b:142). That is, black male workers are I ! ! ; subject to higher costs due to discrimination than are | Hispanic male workers or, more specifically, Mexican malesj j (Reimers, 1982; Schmidt, 1976; Verdugo and Verdugo, j I 1984). (Seldom have researchers attempted to estimate the j ! I cost of minority status among females.) Freeman, for | j example, notes that among Hispanics, unlike blacks, | ; "(d)ifferences in pre-market resources rather than j I I I unexplained 'residual' discrimination appear to be the I I ! j prime cause of economic disadvantage..." (Freeman, : I I ' 1981b:142). Indeed, using data from the March 1979 ; I I ! I I I_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Current Population Survey, Freeman found that ..nearly all of the white-Hispanic family income gap is attributable to education attainment. Earnings for individuals, however, yield the more moderate conclusion that about 40 percent of the gap is due to education" j (Freeman, 1981b:144). He also cited several studies by | 1 Cordelia Reimers in which nearly all of the white-Hispanici I differences were due to "background factors," though results varied for each of the Hispanic origin groups. In another study, Reimers (1982) found that race has no I 1 ; significant impact on the wages of Hispanics,.." (Reimers,| I 1982:11). That is, the earnings of black Hispanics and I white Hispanics are not I 2 I statistically different. She also found that ' (i)n states where Hispanics constitute larger I fractions of the population, Mexican, Puerto i Rican, and 'Other Hispanic' men have lower wages I than elsewhere. However, white and Cuban men I earn at least as much in states with high I concentrations of Hispanics as they earn I elsewhere. The negative effect is significant I for Mexicans and 'Other Hispanics." This may be 1 evidence that discrimination affects Hispanics I more when they are a large proportion of the labor force, as in the Southwest (Reimers, I 1982 ; 18) . Further, Reimers found that discrimination exercises I significant negative effects on the earnings of Puerto : Rican, black and "Other Hispanic" males, but appears to reduce the earnings of Mexicans by only 6 percent. "Low 41 levels of education are apparently a much more serious problem than discrimination for Mexicans. The Cuban-Anglo wage differential can be completely explained by differences in observable characteristics, especially recency of arrival in the U.S. and language handicaps" (Reimers, 1982:27). Others have also found that blacks are more discriminated against in the labor market than are Mexican Americans (e.g., Carliner, 1976; Gwartney and Long, 1978; Verdugo and Verdugo, 1984). Schmidt (1976) also found that blacks seem to face more intense forms of cultural prejudice than do Mexican Americans. C. Male-Female Earnings Differences The male-female earnings gap has been the subject of numerous books and articles in both the scholarly and popular press. It has been well documented that median earnings for females are between 60 and 65 percent of that for males and have been for the past several decades (Shack-Marquez, 1984:15). Reasons proffered for this disparity are also numerous and include females' lack of skills, education and training, lack of consistent work experience, tendencies toward part-time work, selection of jobs based on the ease with which they can be re-entered (e.g., after periods of child-bearing and rearing). 42j competing responsibilities (e.g., for the home and family), lack of drive and motivation to excel in the workplace, and discrimination. Sowell, for example, states "...that marriage and motherhood are the major | factors in the occupational status of women relative to ; men" (Sowell, 1984:101). While there is disagreement over! the factors which cause the male-female earnings gap, it , is agreed that, causes aside, females do earn less than ! males, considerably less. Indeed, Freeman states that, 1 I "(1 )ow-earning workers have certain distinct characteristics. For the most part, they are black, poorly educated, relatively unskilled, female, and located; in certain industries" (1981b:106, emphasis added). Further, having defined "permanently disadvantaged" as being "... in the lowest decile of the male earnings distribution for 70 percent or more of the time over a j decade..." Freeman found that about 5 percent of employed j i I male heads of household are "permanently disadvantaged," ! I ' while "... 60 percent of women household heads who are in j I i the labor force in any given year are in the lowest I I j earnings decile for men" (1981b:106). Two other findings ' j pertaining to the male-female earnings differential from I I Freeman's research on "troubled workers" follow: (1 ) ^ While economic developments of the 1970s helped to improve 431 the positions of black and Hispanic workers, "(t)he situation for women heads of households, however, shows little evidence of change, and sluggish economic growth has meant that persons at the bottom have hardly improved their absolute earnings" (Freeman, 1981b:107). In fact, "...the income of female-headed homes has not risen over time. Indeed, the income of female-headed homes was higher relative to that of male-headed homes in 1969 than in 1978. This, of course, reflects the fact that there has been little, if any, rise in the female-male wage ratio in recent decades" (Freeman, 1981b:133). (2) Along these same lines it was found that (p)erhaps the group with the most serious labor market problem is working women who are heads of households. Their annual earnings place them in the bottom decile of the earnings distribution for men to a greater extent than any other defined group. Unlike blacks, whose median wage and salary earnings have risen rapidly in the past two decades, women have not fared well, although nonwhite women have closed the gap between themselves and white women (Freeman, 1981b:107). Others have also noted that white and nonwhite females have closed the earnings gap, though this has not been true for males to the same degree, or between males and females at all. For example, in 1959, the earnings of black females were 61 percent of white female earnings. By 1972 this gap had narrowed to 91 percent, though the I_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 4 4 , male-female wage gap was not at all reduced during this period (Squires, 1979:174). Also, Williams (1982) found that the correlation coefficient measuring the similarity of earnings among black and white female professionals was .85, far more similar than the measure for males which came to .6 8 . Indeed, Squires writes that "... it is safe to conclude that women have not made the progress relative to men that nonwhites in general have made relative to whites. In fact, it appears that if the position of ' women, in terms of their occupational status relative to men, has changed at all, it has changed for the worse" (Squires, 1979:174). Regardless of race or ethnic background, it seems to be a very consistent finding that females earn less than males. Previously we presented data which showed that females, as a group, earn less than males, but this also holds for females within specific race and ethnic groups. ' For example, female Chicanas have been found to be far I more subject to poverty level earnings than were Chicano I men (Garcia, 1979). Still, both minorities and females I earn less than white males. As noted by Squires, "(o)ne . factor which partially explains the economic position of I minorities is that they are more dependent on their female j ■ members for their income than are whites. More minority | 45 ' families are headed by women, and more minority wives are ^ forced to work in order to make ends meet" (Squires, ' 1979:170). Even considering all women together, not just I ! minority women, it has been found that "(a)t least 70 ! percent of the working women earn money which provides the ' basic necessities of life for themselves and their i I families" (Squires, 1979:177). Hence, the notion that j females are not committed to the labor force and are I merely working for pin money is antiquated if, indeed, it ^ was ever true. Squires also notes that while women today ' are earning more, are working in higher status jobs and ! have thus experienced improvement over time, relative to , males little, if any, progress has been made. ; Other researchers have also found that males and j females tend to experience high degrees of occupational segregation. That is, few occupations employ large numbers of both males and females (Fuchs, 1971). It has also been found that within the same detailed occupation in the same local labor market, men and women tend to be employed in different firms— men in the high-wage firms and women in the low-wage firms (Blau, 1979). Rytina and Blanchi (1984) found that under the occupational coding scheme used in the 1970 Census, 59 percent of all occupations were male-intensive, 2 1 percent were 46 female-intensive and 20 percent were neutral. Even using the revised 1980 occupational coding scheme, figures remained much the same, though the proportion of female-dominated occupations was slightly reduced while the proportion of neutral occupations increased (Rytina and Bianchi, 1984 : 14-15). This suggests that females tend| to be concentrated in fewer occupations than are.males. Some researchers have also explored the impact of j male-female occupational differences on the earnings gap. | I An early study (Sanborn, 1964) found that most of the j male-female earnings difference is due to occupational 1 differences. Chiswick et al. (1974) attributed 28 percent] of the male-female earnings difference to occupational factors. This is less than that found by Sanborn, but Chiswick's analysis did not involve the use of detailed occupations. Though few today would argue that women are working for pin money, some still believe that female commitment to the labor force is transitory at best. Recent research suggests that this, however, is not the case. Niemi and Lloyd note that "(a)11 the available evidence implies that recent increases in female labor force participation, particularly among women with young children, are the result of a trend toward long-term career commitment 47 rather than an increase in marginal workers with high turnover rates" (1981:74). They also found that the inflation of the 1970s, which has served to end real wage growth or at least reduce real wages over the decades, has contributed to increases in female labor force participation. "It appears that family labor supply * behavior is influenced, not only by current wages and prices, but by the expectation that significant inflation | will continue into the future, which may well be one ! factor contributing to the rise of the two-earner family" i (Niemi and Lloyd, 1981:74-75). ‘ I Another study which focused on Puerto Rican, Mexican j I and Cuban females, found labor force participation rates to be highest for Cuban females, closely followed by ■ i Mexicans, with Puerto Rican participation rates lagging far behind (Ortiz and Cooney). Other studies (Cooney, 1975; Cooney and Warren, 1979) have shown that the presence of preschool children, number of children, family income (aside from the female's) and education "...significantly influence the participation of Mexican and Puerto Rican women" (Ortiz and Cooney, p. 309). But, the authors warn that the demand for low-skilled labor, which varies considerably across regions, has much to do with the diversity of labor force participation rates _48J ' " 1 I among Hispanic groups. The geographic location may I explain some of the variation in labor force participation ; rates. For example, Puerto Ricans are concentrated in the I Northeast where the demand for low-skilled workers has been declining, while Mexicans and Cubans are concentrated' I in the Southeast and Southwest where labor markets have , been expanding. ' j j Research into status inconsistency will now be briefly: I reviewed as much of it focuses on females in the labor j force. Status inconsistency refers to, for example, j i minority doctors or female executives. As noted by ‘ I I Epstein, "...few in the professions find that good can ! I come from being born of the wrong sex, race, religion, or * ! ethnic group" (1973:912). Epstein studied 31 black | I I I females in New York deemed to be occupationally successful. I in fields such as law, medicine, dentistry, university I I I teaching, journalism and public relations, in order to identify the characteristics which might explain why these females were successful in non-traditional careers. Giveni I the cumulative disadvantage (or "double negative") of ■ i black females, it would be expected that they would be at | the bottom of the "occupational pyramid." She found that ; these women were smarter, more talented and more specialized than their white male colleagues. "Thus they 49 paid more for the same benefits (or 'goods'), if they were permitted to acquire them at all" (Epstein, 1973:912). She found several factors which might explain the success of these black females. First, black females have easier access to white society than black males because they are | not perceived as so threatening or so powerful precisely because they are females. Second, the fact of being blackj cancelled the negative impact of being female for the | black women working in a white professional environment. j That is, "...a black woman is viewed as lacking the ; 'womanly' occupational deficiencies of white women— for | example, seeking a husband— and the black woman's sex status is given a higher evaluation" (Epstein, 1973:914). Third, being a black woman professional creates a relatively unique status for which there may be no established "price." Hence, the black female lawyer, say, may be in a better position to negotiate her own salary. Fourth, being unique, or, as Epstein phrased it, being a "stranger," she may be freer to choose or may be forced into an alternate lifestyle. "This choice was made by many black women forced to enter the occupational world because of economic need, and, in turn, it created selective barriers which insulated the women from diversions from occupational success and from ghetto 50 culture, thus strengthening ambitiôn and motivation" (Epstein, 1973:914). These women appeared to be slightly ; less likely to marry and far less likely than other women to see marriage and a husband as a path out of the labor force. Indeed, Epstein writes ; (a)1though the white college-educated woman is ' strongly deterred from focus on a career when she' marries (though she may work), the black woman who marries a black college-educated man cannot ■ consider withdrawing from the marketplace. She knows that her husband's education is no guarantee of his financial success. It has been clearly established that the discrepancy in ' income between white and black male college , graduates is wider than the gap between incomes ! of those who are less well educated.... I It seems probable, too, that black women view I careers differently than white women who expect to combine marriage and career. White women like j to view their work as supplemental to the ! husband's. They tend not to think of their work ' as a career growing out of their own life aims. ; Black women tend less to view their work as a I 'hanger-on' activity. One gets the feeling in ! interviews with them that the quality of their t lives is determined by their own endeavor and is 1 less a response to their husband's occupation ' situation. Perhaps this is a function of their I relatively high self-confidence" (Epstein, I 1973:925). I Hence, the high labor force participation rates of black I females may, in part, stem from the relatively low wages I I of black males. Indeed, the married black women I ' interviewed by Epstein were less likely to consider withdrawing from the labor force if their occupational success surpassed that of their husbands' since the 51 , female’s income contributed substantially to the family's ■ i standard of living. Another reason for this finding may ■ be the higher probability of marital breakup among black I ■ women as compared to white women, and that they are less ' likely to remarry after divorce than are white women. The; results of one time-series analysis confirmed this I I finding. O'Neill (1981) found that married women's labor ; I force participation rates do seem to correspond to | ; husbands' wage rates. i While many claim that marriage and child rearing : ; i account for a large part of the male-female earnings gap, ! the successful black women interviewed by Epstein espoused j : I ! views on marriage and child rearing which were different | ! I from the norm. For example, with regard to motherhood l ...(t)he black mothers interviewed seemed far j less anxious about their children than whites. | They did not insist that is was their sole | responsibility to care for their children, nor | did they fear that their absence from home during| the children's early years would be harmful to their psychic and physical growth (Epstein, 1973 : 928) . ' Further, these women seemed to have access to family , members who cared for their children. Epstein also notes j I I that upper-class blacks have a lower fertility rate than | : I : other groups. Other factors include the following; these' i i \ women tended to be raised in families where the mothers | i I I worked, often as professionals or quasi-professionals; the i j I j I I 52I families stressed middle—class values; many were West Indians— a group noted for their economic success (Sowell, 1984); and these women received much encouragement from their families in gaining an education. One of the black ' women interviewed, a West Indian, said "'Girls or boys— whoever had the brains to get education was the one pushed to do it and encouraged" (Epstein, 1973:922). Epstein contrasted this attitude with the often less than ; i enthusiastic response of white families to their 1 daughter's education, particularly education beyond the i bachelor's degree, which is sometimes seen as detrimental to marriage. Still, even these highly successful black women found that getting suitable jobs was harder than getting into the right schools. For example, the medical doctors often ended up working in clinics, in government employment or had to confine their medical practice to the black community. Finally, Epstein concludes on a pessimistic note. ...(G)iven the limits imposed by the current social structure, only the most extraordinary black women, those who are intellectually gifted and personally attractive, can make it. The fact that some do indicates that an enormous amount of energy in the social system must be directed to keeping others out (Epstein, 1973:932). I I I I 531 While the case of occupationally successful minority women suggests there are differences between such women and other minority women, as well as the successful minority women and successful white women, it appears that there are areas in which all women, regardless of race, j ! experience obstacles to varying degrees. However, j Epstein’s conclusions are based on interviews with only 3l| black women, all living in New York, so findings should bej I considered tentative. It is also quite possible that, j over the 13 years since the data were collected, ( I I j differences between successful career women of all races j ‘ have narrowed. I I One finding common to research tracing earnings over j the life cycle is that the earnings of males and females ! peak at different ages, Mellor (1984) found that median I weekly earnings for males peak at the 35-44 year age I group, while those of females peak at 25-34 years of age. He also found that educational differences and differences ( in the age distribution of males and females accounted for ; little of the male-female earnings difference (only 1 percentage point or $4 per week). However, it was found, ] as with other research, that occupational differences explain much of the median weekly earnings gap, particularly as ever finer measures of occupation were 54 ; used. For example, the male-female median earnings gap narrowed by only 1.6 percentage points if females were : distributed identically to males across 11 broad (1-digit) : occupational categories. However, analysis of 40 2-digit occupational categories revealed that the earnings gap ! would narrow by 5.1 percentage points (such that women I would make 70.1 percent of what men make) if females were I distributed across occupations similarly to males (Mellor, I 1984:18). Mellor further states that "(t)he ratios would I 1 undoubtedly rise even higher if the very detailed Î three-digit occupations were redistributed" (1984:19). j ; (This, however, was not done due to the small number of : observations in many occupations, particularly after I splitting males and females.) However, he cautions, that ! I ' even at the three-digit level there is still a wide i I ! variety of jobs within an occupation. (As an example : Mellor cites physicians, one of the three-digit . occupational categories, which in fact comprises some 85 I specialties.) Others have argued that job experience plays a role in explaining the male-female earnings gap and it does appear that there are sizable differences between males and females in average length of employment with a given employer. In Mellor's study it was found that males had 55 an average of 5.1 years with their current employer as compared to 3.3 years for females. Further, "(m)en in each 10-year age group 35 years and over also had more seniority with their employer than did women. Up to about the mid-30's, job tenure [length of time with a given employer] does not differ significantly by sex" (Mellor, 1984:24). Occupational tenure [length of time within the same occupation] has also been found to differ between males and females. Further, in one study this difference accounted for 4 percent of the male-female earnings gap. However, in combination with potential work experience,^ marital status, part- versus full-time employment, place of residence and major occupation and industry group, 25 percent of the male-female earnings gap was explained (Rytina, 1982). Yet another study found a high degree of convergence in male-female earnings after accounting for differences 4 in skill levels (Sieling, 1984). Data for this study were from the National Survey of Professional, Administrative, Technical and Clerical Pay (PATC), a database collected by the Bureau of Labor Statistics in March 1981. First the earnings of males and females within the same specific occupations were compared without considering skill level. This yielded female-male earnings ratios ranging from 74 to 101 percent with all but two occupations having a ratio of under 90 percent. However, after including skill level in the equation, the ; male-female pay gap narrowed. Indeed, 43 of 48 female-male pay ratios reached 90 percent or more. Sieling thus concludes that the male-female earnings gap "...largely reflects an uneven distribution of men and | I women among the [skill] levels of the occupation— that is,I I different staffing patterns" (Sieling, 1984:32). Also, women in the professions tend to be concentrated in entry level jobs where male-female pay differences are less due to the lack of seniority distinctions. (It should be noted that the PATC data do not indicate length of time at a given skill level, that is, how long it takes to get j I promoted.) : Aside from male-female differences in seniority and l skill level, another common belief is that females earn { less than males because they do not work as many hours or { are less committed to the labor force. Yet one study . found that male-female differences in occupational j distribution, not hours, is the actual cause of the | earnings gap (Mellor, 1984). "The numbers suggest that | the effect on women's earnings as a result of their ! 57 working fewer hours than men is brought about more because women are less likely to hold higher-paying jobs which demand long workweeks than the fact that they are less likely to work overtime and receive premium pay" (Mellor, 1984:25). However, others have found that even among males and females working year-round in the same job, often substantial pay gaps continue to exist. Indeed, Suter and Miller (1973) acknowledge that the intermittent work experience of females contributes to their lower earnings. (They cite the ages of 20-40 years as a time when males are working full-time and females either leave the labor force entirely or work part-time. Either way they accumulate less work experience than males in the same age group and so are less likely to move into higher I paying jobs.) In order to avoid the problem of i differential years of work experience, the authors ' I selected a data set in which females were heavily j committed to the labor force. Using the National Longitudinal Survey (Parnes) data, 5,000 women between the ages of 30-44 were selected who had worked at least six months per year since leaving school. These data were collected in 1967 and were compared with a sample of males ages 30-44 from the 1967 Current Population Survey. The results of this study suggest that women who have 58' consistent work experience are better educated and have higher status occupations than other women. However, despite these characteristics, (c)areer women, who receive higher incomes than other women, have difficulty in breaking away from average-paying jobs.... Although the professional and technical women who have been working for most of their adult lives have significantly higher earnings than other women, they earned only 66% of the male average in these occupations. A similar comparison made for clerical workers showed that career women earned 79% as much as men, and among operatives and service workers women earned only about half as much as men. The figures cited suggest in a rather impressionistic way that when occupation, length of work experience, and age are taken into ! account, about two-thirds of the difference j between the earnings of men and women can be i explained, leaving about 33% attributable to all other factors (Suter and Miller, 1973 : 967-968). j Despite the fact that Suter and Miller's study focused on career women, those women with lengthy and consistent commitments to the labor force, there were still tremendous differences between the regression coefficients of these women and men. For example, the females' returns to education were 76 percent as great as that for males, and the coefficient for occupational status (as measured by the Duncan Socioeconomic Index) is 60 percent as great. The authors conclude that "(t)he inability of women to convert occupational status into income to the same extent as men suggests that much of the remaining 59] unexplained difference in male/female earnings could be attributable to discrimination in payment for jobs with equal status" (Suter and Miller, 1973:971). It is, however, intersting to note that the career women had regression coefficients quite similar to those of black males. This leads to a controversy concerning females' investment in human capital. While Mincer and Polachek (1974) argue that females, in large part, earn less than men because they tend to invest less in their human capital as they expect to interrupt their careers. Further, Polachek (1981) found that individuals expecting to interrupt their careers will choose to enter those occupations in which the penalty for leaving and re-entering is least. However, Corcoran et al. (1982) found that though females, upon their return to work earned lower wages than prior to leaving the labor force, they experienced a rapid growth in wages such that the netj loss due to leaving the labor force is small in the I long-run. Mellor also found that, after analyzing 112 i occupations with 50,000 or more females each, for every 10 : percent increase in percentage female employed in the i occupation median usual weekly earnings in 1982 would fall' by $13. The proportion of females in an occupation 60 accounted for about 19 percent of the male-female earnings gap among these 112 occupations. While it appears fairly clear then that high percentages of females in a given occupation will tend to depress wages, others have found that the sexual segregation of occupations is declining | (Rytina and Bianchi, 1984). I D. Overeducation , Very little research has been conducted on j i overeducation and even less on how overeducation affects i earnings. Overeducation refers to the condition of having! I more education than that required to do a job j satisfactorily. This condition has been operationalized I { I I and labeled in various ways— "mismatch" (Sullivan, 1978; I ' ; Clogg, 1979), "credentialism" (Berg, 1970), and i "overeducation" (Freeman, 1975; Rumberger, 1981). j However, all these concepts speak to the same issue ; * namely that the "...tremendous expension of formal i I education [has exceeded] the upgrading of skill requirements..." (Squires, 1979:103). Overeducation is, according to the Labor Utilization Framework (Hauser, 1974 and 1977), a form of "underemployment." Though education has often been viewed as the most important institution in promoting equality between the races and sexes, it has not, alas, been terribly effective in this regard. Though 61 differences in years of education attained between males and females and whites and nonwhites have narrowed, this narrowing has not resulted in a reduction of the earnings , gap (Squires, 1979). Indeed, since the 1970s, there has i been a dramatic increase in the number and proportion of i college educated workers in the labor force. Though the supply of college educated workers has increased, the . demand has not (Freeman, 1980). As a result. Freeman i finds that the relative earnings returns to highly | educated workers may have declined. This finding has, j however, been disputed by Smith and Welch (1978). They i argue that earnings declined only for the baby boom I generation and that this was due to their large cohort | size (which served to increase the competition for jobs) and their entry into the labor force during a period of economic decline. For older workers Smith and Welch found that earnings remained constant over time. Others have defined overeducation not in terms of earnings so much as skill levels required to do a job. One such interpretation is that, as educational attainment levels have increased, employers have raised minimum educational requirements for jobs beyond that which is really necessary (Berg, 1970). This is known as "credentialism" and can be illustrated through an example 62 of queuing theory. Queuing theory suggests that all workers apply for better paying jobs in which the employer selects who he thinks is best, leaving the non-selected ones to lower paying, undesirable jobs. If an over-supply of applicants exists, then credentials are utilized to screen out the labor pool (Garcia, 1979:4). The interpretation of Clogg (1979) and Sullivan (1978), with respect to overeducation, is that some workers have educational levels far beyond that of others working at the same occupation, and thus the skills of these persons are being inadequately utilized. Research into the distribution of overeducation among various subgroups of the population have yielded conflicting results. Garcia (1979) found that 13 percent of the Chicanos in his sample were overeducated. Most likely to be overeducated were those at either extreme of the occupational status scale— professionaIs/managers and agricultural workers. Women are less likely to be overeducated than are men, and this seems to be a very consistent finding. However, this is not a particularly heartening finding for women. "This pattern...is not to be interpreted to mean that women are better off than men in this respect. On the contrary, it reflects the fact that women...have less education than men and, therefore, are less eligible to 63 . . J experience this form of underemployment" (Hauser, 1977:13). Because, contrary to Garcia (1979), educated segments of the population are most likely to experience mismatch, the highly educated younger generation is | I particularly susceptible (Sullivan, 1978). Indeed, due to increases in educational attainment, it has been found ; that the prevalence of overeducation has increased (Clogg ; and Shockey, 1984). . Clogg (1979) reports that blacks and nonblacks tend toi I be equally as likely to experience mismatch. "Mismatch j j ' I shows some trend toward 'parity' between blacks and j nonblacks, indicating greater educational attainment of I I blacks and perhaps also greater difficulties for educated I I I I blacks in receiving occupational rewards commensurate with! ; I ; their educational attainment" (Clogg, 1979:133). , : Because research into overeducation has been limited ! I and findings have been mixed, it is difficult to develop j hypotheses concerning the distribution of overeducation among the population. It does, however, seem clear that the highly educated are most prone to overeducation. Even j less research has been conducted on the impact of ^ overeducation on earnings. Only Stolzenberg (1975) makes ' brief mention of the possible relationship. Stolzenberg | argues that, within a given job, persons with more 64; education will generally earn more than those with less education. E . Summary Though the research findings reviewed in this chapter have not always been in agreement, some common findings j suggest possible outcomes of this research effort. For I example, it is expected that white males will be found to | earn more than females. Some research has also suggested ^ that being female reduces earnings to a far greater extent: than does minority status (Almquist, 1975). Previous research on males suggests that blacks experience higher costs due to labor market discrimination than do j Mexicans. Research into whether or not race and sex j discrimination interact to depress the earnings of j minority females below those of white females and minorityi males is inconclusive. Though it is commonly assumed that j "double jeopardy" does indeed serve to reduce the earnings| I of minority females relative to white females, research byI Epstein (1973) suggests this may not always be the case. | I I Perhaps minority females work harder due to greater economic need or cultural factors, and so equal or exceed the earnings of white females. However, given the tiny sample upon which Epstein's results are based, assumptions in this regard are difficult to make. Finally, though no 65 1 earlier research has been done which examines the impact of overeducation on earnings, it is assumed that within a given occupation, earnings are higher for those with more education. L ................................. ........... Notes for Chapter 3 1. Because the concept of undereducation is unique to this study, there exists no previous research on the topic, to review. 2. These results, however, should be interpreted with caution since the sample analyzed by Reimers contained I very few black Hispanics. : 3. Potential work experience was operationalized as 1 follows: (Age - Years of Schooling - 6). ; 4. Skill level refers to a numeric ranking: lower : numbers for entry level; middle numbers for experienced ; workers; and high numbers for supervisory or managerial I positions. Occupations considered include accountant, I auditor, attorney, chemist, personnel director, job analyst, buyer, engineering technician, drafter, computer ' operator, photographer, accounting clerk, messenger and purchasing assistant. 67; _ _ J CHAPTER 4 I RESEARCH FINDINGS j There are several sections in this chapter. Part A I reviews the findings for each ethnic-sex group 1 separately. Comparisons between groups are not drawn, ' though means, statistical significance and the direction I j of association of variables in the regression model are discussed. Part A concludes with a discussion of the standardized regression coefficients so that those variables having the greatest impact on earnings, the I dependent variable, can be identified. Part B compares j findings between groups by examining the raw score I regression coefficients and amounts of variance explained I by the model. In Parts A and B results are presented for ! human capital variables first, followed by the structural variables, and lastly the control variables. Part C I presents results of the regression standardization (or decomposition) in order to determine the estimated impact I I of discrimination on the earnings of each ethnic-sex I group. Part D presents the summary and conclusions for ^ this chapter. I A. Within Group Results 'A.I. Human Capital Variables ! The earnings model considered in this study includes I four human capital variables, years of education completed I_______________________ 68 (EDUC), years of work experience (EXPER), and a dummy variable having two educational classifications. (These classifications are OVERED and UNDERED which represent overeducation and undereducation, respectively. Those , workers with average levels of education for their 1 ' occupations comprise the omitted category.) For each of I these variables except experience, there are striking I differences among the twelve ethnic-sex groups considered. (See Table 4.1, for means and standard I deviations of all the variables.) A .1.a. Means Probably the most commonly used human capital variable I is years of education completed (EDUC). Average I educational attainment is highest for white males and , females— 13 years. The next highest education level is 12 years for black females and Cuban and other Latin males and females. Black males average 11 years of education as do Puerto Rican females. Puerto Rican males and Mexican ; females have 10 years of education, and Mexican males have I the lowest average with 9 years. It is interesting to I note that among blacks, Mexicans and Puerto Ricans, the i : females have higher average education levels than their male counterparts. For whites, Cubans and other Latins, 'males and females have about the same levels of 6 9i o OO CO ON * -i r- * —4ro ao tA < 7 N ro o O Nt ro M" t4 C N I NO O CM r~ * ON «A ON r- O o r- t4 LA CM CO r- a o r- r- n4 - «N I nJ - ro r- • N t CM CM *4 St MJ o CO r4 r4 t4 VO m ro th M" (M <4 N -M T —1 CM CO oa CO o# m r — CO o O » - f«NI r4 ON CM t4 St CM NO r- LA CD CD +4 « NI sr CO LO CO r- nO t4 C 7 N CO Nf O t4 CM >t OD *4 ro t f l u c ro ro o o ( N I lA f- O O CD CM r- ro t4 ON O CD H < S | fO • • # # # • • • # • # • • « • # U» 3# Q> c r > o ro NO ro C T N >» 04 t-l CM NT r— 1-4 o ID CM C <X ro <7> ' r > CM CM CO CM lA ON ON Nf ft ro CO ON >t r- 4 O CM CM r- m O O r- ON ro r— >t St (M «A ON O o o r- M" ON CM ro O N» M- r- ro <x> CM Nt Nt ro Nt CO • • • # • • • • M" r4 *4 ro CM Nt q f4 CM o c mi St Oi r4 tn XI NO OO <30CM n4 f - t#4 t4 lA r4 - 4 - tA O r* ~ ro o NO t4 E < 0 1 >4“ r— w - $ i r i CO lA ro «4 tA ro ON NO Nt o st t4 ON o « 24 C ro (NI s4 o CO f# N» O CO tA t4 «4 CO LA CM r- »4 CM ■H f O # • # « # • # # • # • • # # • • H o ON ( J N ro tA CM tA 2: <M St r- f4 ro f4 <t u> O tH r- r4 nO CM CM O ON NO St ro CM tA ON t- Nt 4- o lA m r- nO CO vO O o r- CD CO ON CM st tA st CO O LO# o r — vf CO <M CO O n4 tA CM LA Nt o St O »4 i4 t4 ml CO • • • • • • • # • u> -J ro CM t4 ro CO ro c rH CM St o 3Q 44 t4 m CM -M mi fO r4| «N I r— o sr M- lA r- CO Nf tA CO r4 r* CM CM st CM CO •H OOI NO m t r * OO u> lA ON O t4 CD « J N tA CD ro O r — CM »4 > c ro ro r * - o r4 CM OO CM tA O o CM CO CM o ON O O m • • • a o m r4 ro lA tA «4 £ rH CM vt o -o r4 e CM fO •o CM CO N» o O' O CO CM ro #4 ON ON r* CO CM CM tA CO c • ro nO cm «H ON 00 CO CM NO f- O h- St CM NO •4 < X > tA (0 Q nO r - t CM ro CM ro t4 n4 - Nf ro t4 ro t- GO O *4 O t4 •H CO • • • # CO CO CM »4 CO CM tA r4 CM CM -O < 0 1 Nt c ♦4 fO •r4 q r- < T* nO OO ON lO ro CM o tA O st ON o st t- I S . O U1 31 vT ro CM o ON r- ON CO r4 NO ON t- O t- O CO O CM c c r — o o rH o .4 r— CM ro r4 CM r4 f- >t O ON o CD ru m • # • # 0» <0 O' ro ro tA O o 3E ac tH cm M> CM CM CM (A # 9 •H *4 C N J •4 CM CO «4 Z z JQ O Q oc Z z Z < ZD t i l t4 O ro 0£ UJ lU o Q. o o o CL » — to LA |s jLD o 1 —1 • < 4 l _ ) UJ QÊ 0£ Z 14 ' 14 14 3 X Q£ LD 14 L. >- Z) CL UJ Ui o UJ o o o O OC LU ZD Z > — X X fO (0 c Q X > > UJ Z UJ UJ LU o «* LU O <X ►4 X X h — > r4 UJ UJ a a CO o oc ac <X o X DC JZ _J O 14 1 4 tH cm 70 v t 4 in >s r-i f B c <x Of JZ c 4 i n c o Q 4 > 9 O X> «. f O T3 C < 9 - t - » to T3 C m lA c ( 0 9 -o « t - » c o o to# <o 23 tu| UJ LAI 41 XI Q to z x 3 UJ r a i -4 nO( C 9 X CD to 9# c r o 9 X a •HIQ 41 to < 0 1 -» I U 9# q •HI C 0 | (0 9 X Q to a rq q 31 CJl c ro 9 X %0 vO CM f s - ^ CO fM «4 r- OO LA CM • • • • <M »4 ON UN ^ O m r- r— LA O O C J N o • • • • ON A4 4 «4 A4 A4 nO vO |s- A - NO ON A 4 L A %0 « 4 A 4 A4 A4 A4 A4 O st CO O OO VA .4 CO A4 O O' A4 ro «4 A4 in OO o A4 vo NO r- A4 n O ON A4 ro • • • • ro »4 GO OO st ON Nt nO OO «4 LA «4 00 0 4 • • • i4 A4 r4 A4 4 A4 ON A4 A- m st ro vO A4 in ro • • • • st A4 A- vO A4 vO O ro ro A4 m CO nO 4 o A4 ro m ‘ A OO A4 ON CO o ON f s . ro A4 4 st OO m OO NO r * » O' s j - O CO ( N | L I N U L > A4 OO O' ro GO 4 • • • m A4 U> O' A4 A4 4 Nf 4 Vt A4 in t— ro no A4 A4 A4 ON in Nf OO A4 r — 4 no ro O' Nt CO r- on O O CD 4 4 ro nO OO ro ro 4 ro ro m vo -t st st CD 4 m in 4 4 <M • • • o CD CD in CD A4 m O' NO St in A4 ro ON ON ro OO O' 00 st st 4 nO ro so A4 sj r- CD OO ro sO r — ON St CD A4 st C D OO A4 sO OO C D 4 A4 ON % —4 00 ro St ON OO A4 CD Nt in ro m A4CD A4 CD ro st NO NO 4 CD CD OO in in st nO OO A4 nO st OO on T-* O A4 30 tn A- A- O O' St st 4 A4 A4O' A4 4 ro ro st st o A4 ro A- A- NO ro in NO V —4 St in ON ON 4 in 4 A> A- CD CD ro A4 st st A- A4 • • m A4 nO A4 4 St o A4CD 4 NO CD A- nO ro A4A* CD • ON 4 St st T - 4 A4 ro ON A- A- o A- ro ro st m -st GO • m A4 CD A4 o st ro ro 00 NO NO A- m m A- 4 lA A- 4 ON in A- 4 A4 st st A- 0 4 4 f O " O < T > O r- 4 C D O' C D sO CD 4 CD m A- so CD 00 O CD on CD in vO OO r- A- ro ro Nt st 4 O' ro CO o m m 4 O A4 OO A4 O n O O' O' st st ro st ON A- A4 CO CO ro m 4 CD NO sO A4 CD ON A * ro ro CD st st in nO A4 st st nO ro st 9 XI < 0 JU 1 9 4 A4 4 A4 ro _J Z Z Q Q Ô Î Z Z Z «* 3 UJ Qc LU LU 0 0 - 0 o o e x . 4 — m tn a40 o U I oc or A- X A# 4 4 ZD 41 oc t o 4 *4 >- ZD Q- LU LU CD t u O O CD O OC LU ZD Z 4 — X C O X >• > LU Z LU lU LU U < lU O l-l X 4 LU LU O O to ZD OC QC OC Q X Z* X —4 CD 4 A4 O A# X X 71 < x >o tn O ON ro 00 sO CO ro o o s t A 4 O' OO A- s t in m o s t N O N O ON 4 m 4 ON st CD A 4 A J ON A- r o U N ■ri Q O A- «M A4 r o ro O N st S t s t A4 s t OO 1 — 4ro st s t S t in a OO • • • • # • # • • • • ■ • • • • • • >s r# ro 4 in A4 in r-i HI 4 A4 AI ro rq ro c -Jl <t 1 m m ON fO m 4 s t s t in CO ON OO A 4 O N CO CO A- 4 w s r O N CO A - so O N s t ON A4 1 — 4 O N m A4 st O ' st in ON 9 91 c o tn 4 O 4 1 —1 4 A4 A 4 S t 4 ro As A - 4 sO AJ A4 JC q r o • • • • • # # • • • • • • • • • • • H +1 9 ON T - lA i NO A4 in r r # X * - #A J s f O N C ON • H 1 — 4 i n E ON 4 m OO sO < 3 0A- CO ro 0 0 C O O 1 — 4m ro o OO 9 r- o r- A4 A- A4 OO A4 A* N O ON in N O A- C O ON o ro H o in o Al A » ro m m st st A 4 s t s t s t vO st st s t s t 1 —« OO a s t A 4 st A 4 m r — I q 4 A 4 ro r - i q ro <( 3| m OO CO O A4 C M 4 A4 O OO in s t O N C O A - A- A4 o C N 4 4- o s Q A4 A 4 A- A 4 in s t st A - A 4 O N i n O A - OO o s t O c o O A- 4 4 4 4 CM n O O ro A4 s A A - ro in A 4 As r o • $ • * • • • in 9 ON 4 s O s O A 4 ro c X 4 A4 s t 1 — 4 o O - H I / o rvj H U4 r o -0 • H O O CO A- ON ro A4 ro s t o 4 O st i n 1 — 4As i n m As > ap q to 4 s O A- o O A - in 1 — 4 N O s t A - ON 4 m A4 ro C O 9 u4 noo O n O OO 4 s r s t A- s t ro A4 O s t ro O N ro 4 1 — 4 4 Q U - l m OO • • • • • $ « • • • • • • • # • • • en T - 4 ro A4 NO TJ 09 # — 4 A4 o L . 1 ro r o o i x> *4 s r 4 OOro ro ro ro ON A* s t ON O ro in o st CO o c q A| m i nro o O C O o O As ON s t s t A- in C O 4 ro r o 91 C o O O sO O I M A4 i n A- 4 O in ro NO sO 1 — 4O N O o H q 1 0 • • • OO 04 9 ON o A4 n O 4 o X e —4 A4 s t NO T3 O N C 4 r o ON s O # — 4ro s O CO in A- 4 ON O N o o AI O N A4 A4 NO i n s t sO # — 4A- s t o ON O O O ON NO n O 4 O As 1 — 4in o c Q S O ON i n 4 s t st ON 4 s t S t ro s t A- rO ro St ro ro < 0 OO • • • 9 q m # — 4 ro A4 O X q 4 A4 O m ro / - N > 9 0 3 q ON # — 4A- 4 s t 4 A4 A 4 s t As ON ro 4 ( 3 0s t ro in ON • q s f s r t n ro A- 1 — 4s t 1 — 4 sO ro N O o i n ONA- C O s t m H c ON 4 A4 o A4 A4 S t O ro m < 3 0ro A- ro 1 — 4A- 4 1 — 4 C r o • • • $ • • « o 9 CO o A4 in f s - o o X 4 A4 ro CA ON s t 4 4 AJ 4 A 4 ro —Iz z ta O Q QC z z z q LU 1—4 9 r o OC L U UJ a a . o o o CL A* CO LO rs| O o r —1 ■ri l _ >UJ Ûf oc Â- X 4 4 4 q 4 X OC L3 4 4 4 JQ L >- Z3 O l L U LU o UJ o o e> o QC L U q Z A - X X r o r o c O X r > > LU z lU LU l U o <X L U o c 4 X X h- > r — 4LU L U a u C O Z 5 o c OC oc o X 3: X q O 4 4 A I 72 education. In no case do females lag behind males in average educational attainment. ; Years of work experience (EXPER) range from a low of I I 22 years to a high of 28 years. It should be noted, ■ however, that this variable is likely to be the maximum I potential years of experience. Census data do not provide I information about actual number of years employed. This variable was operationalized such that each year following an individual's completion of schooling is counted as a year of experience (see Table 2.3). However, this may be an overestimate of the years of actual work experience, particularly for women who tend to enter and leave the labor force more frequently for childbearing and other reasons (Felmlee, 1984:181). Except for Cubans, all the jgroups have 22-24 years of work experience. Cuban males ! have 28 years of experience and Cuban females"27 years. I jThis difference appears to be due to their greater age. I(Cuban males have an average age of 44, while Cuban jfemales average 43 years of age. Whites, on the other I ■hand, average 41 years for both males and females. See I Table A.l in Appendix A.) [ Differences in OVERED and UNDERED (overeducation and iundereducation) are most striking of all. In spite of females having roughly the same or more education than 73^ 1 ' males for each ethnic/race group, proportionally fewer are overeducated. (Perhaps the relatively high levels of education for females and their relatively narrow ; distribution among occupations account for this finding. Recall that, by definition, overeducation is dependent on the average educational level for each occupation as well : as the standard deviation. Thus it would seem that, other things being equal, occupations in which females are concentrated will tend to have relatively high average i education levels. This may reduce the percentage of I persons defined as overeducated in female-dominated I ! occupations.) Among males, Cubans are the most likely to i I be overeducated, with 13 percent so defined, followed by I 12 percent of other Latins and 11 percent of whites. ! Black, Mexican and Puerto Rican males are far less likely to be overeducated with 8, 5 and 5 percent overeducated, respectively. The percentage of overeducated females I tends to be lower. Cuban females are most likely to be jovereducated with 12 percent so defined, followed by 7 ! percent of other Latins, and 5 percent for white and black 'females. As with males, Mexican and Puerto Rican females are the least likely to be overeducated with only 3 percent for each group. Though Mexican and Puerto Rican males and females were 1 least likely to be overeducated, they are most likely to be undereducated. Among males, 33 percent of Mexicans are undereducated followed by 23 percent of Puerto Ricans, 22 i percent of Cubans, 17 percent of other Latins and 15 j ■ percent of blacks. For both males and females whites were ! the least likely to be undereducated with 10 percent and 9 ; percent, respectively. Among females, the most likely to : be undereducated are Mexicans with 27 percent, followed by ■ 20 percent of Puerto Ricans, 17 percent of Cubans, 16 ' I , I percent of other Latins and 12 percent of blacks. In | ! I summary we find that whites have the most education and jare less likely to be undereducated than other groups. i I Cuban and other Latin males and females are most likely to : be overeducated given their occupations. : A.1.b. Other Statistics j According to the human capital framework, we would ! expect, other things being equal, that persons with more i ! education and experience would earn more than those with less education and experience, and indeed this is the ; case. Regression coefficients for both variables are I positive for each ethnic-sex group. (Table 4.2 presents I standardized regression coefficients for all independent I variables in the model. Table 4.3 presents the same for a i 1 ! I I I _________________________ 75j Table 4.2. Standardized Regression Coefficients of All Independent Variables in the Analysis (First Model: With EDUC) Independent Variables EDUC EXPER OVERED 1 OVERED 2 SECTOR UNEMP REGION 1 REGION 2 REGION 3 OCCUR MARITAL WEEKSUN HOURS LANG CITIZEN IMMIG 1 IMMIG 2 iOiifi Siack baalGao Puerto kiaaa. Lubao Other Laiio 170«« ,158«« .015 .104»« .097»» .013 .220»» .080«« -.025 .183»» .103»» .010 .189»» .019 -.024 .149»» .120»» .018 047»* — .040 .001 -.005 .040 -.028 ,077»« 043«« 047«* .003 .026 -.123»* -.024 -.026 -.024 .062»» .008 -.109»» -.042»» -.068»» . 029 -.050»» .021 —.06 6» 085«« -.222»» -.239»» -.115»» -.104»» -.072»» 006 -.052»» -.115»» -.039 -.010 -.059» 268«« 120»» 08d«« 088«« 013 001 020» .269»* -.074»» -.067»» .072»» -.007 -.015 -.029 .185»» -.095»» -.051»» .103»» -.062»» .053» -.046 .185»» -.077»» -.090»» .066»* -.043» -.023 -.036 ’ .213»» -. 102»» -.088»» .098»» -.059»» .069»» -.001 .297»» -.096»» -.071»» .085»» -.067»» .013 -.094»» 013 -.008 .017 -.033 .031 .006 «Significant at the .05 level (using the two-tailed t-test). ««Significant at the .01 level (using the two-tailed t-test). i Table 4.2. (Cont'd.) Standardized Regression Coefficients of All Independent Variables in the Analysis (First Model: With EDUC) I Independent Variables EOUC EXPER OVERED 1 OVERED 2 SECTOR UNEMP REGION 1 REGION 2 REGION 3 OCCUP MARITAL WEEKSUN HOURS LANG CITIZEN IMMIG 1 IMMIG 2 âlâ£k Puerto kl£âD. Other ialio .140«* .088«« .013 .111»» .039 -.002 .069 .002 - -.029 .144»* .081»» .037 .073# .030 -.019 .052 .031 .040 .048«« -.058»» -.081» -.056 -.045» -.049» .029«« .061«« .007 .036 .049» -.039 .005 -.001 — .003 -.022 -.008 - « 067 .042» .020 -.021 -.007 . 043 -.021 .052»* -.135»» -.174»» -.081»» -.115»» -.088»» .025» -.036 -.059 -.038 .006 -.059 .328»» . 047»» .068»» .172»» .016 .013 .007 .321»» -.001 -.062»» .070»» -.002 -.028 -.003 .281»» — . 0 06 -.077»» .07 2»» -.096»» .078 .068 .268»» .014 -.097»» . 052« -.055» . 041 .001 .277»» -.004 -.130#» .101»» -.105## .060»» — .040 .344»» .033 -.106»» .112»» -.076»» . 066» -.043 .018 .009 .109»» -.007 .008 . 028 «Significant at the .05 level (using the two-tailed t-‘test). ««Significant at the .01 level (using the two-tailed t-test). 77j Table 4.3 Standardized Regression Coefficients of All Independent Variables in the Analysis (Second Model: Without EDUC) Independent Variables hhlls bmklgao Puerto Blsao- other kalio EXPER OVERED -.121#* .075** .069*# .042# .030 .021 • 065## .05 3## -.003 .041## .097## .065## OVERED -.109## -.089#* -.109## -.097## -.051*# -.101## SECTOR UNEMP REGION -.065#* .043** -.046## .010 .026 -.124## -.009 -.028 -.021 .074## .010 .104## -.033# -.073## .028 -.043## .026 —.069# REGION -.091## -.230## -.248## -.105## -.110## —.071## REGION 0 -.051#* —.114## -.031 -.011 -.062# OCCUP MARITAL WEEKSUN HOURS LANG CITIZEN IMMIG .354## -.119## -.092## .087## -.015# -.001 .021# .309## — .07 5«# -.065## .073*# -.007 -.014 -.027 .251*# — .09 2## — -.059## - .100## -.080## .060# —.05 7* — .252## .076## .096## .067## .057## .022 .034 .284## -.103## -.088## .100*# -.0 68## .080## .010 .360## -.094## -.074## .083## -.078## .014 -.091*# IMMIG 2 .015 -.005 .015 .033 .043 .012 «Significant at the .05 level (using the two-tailed t-test) . ««Significant at the .01 level (using the two-ta iled t-te st) . 78 Table 4,3. (Cont’d.) Standardized Regression Coefficients of All Independent Variables in the Analysis (Second Model: Without EDUC) Indep endent Variables EXPER QVEREO 1 OVERED 2 SECTOR UNEMP REGION 1 REGION 2 REGION 3 OCCUP MARITAL WEEKSUN HOURS LANG CITIZEN IMMIG 1 IMMIG 2 âlâ£k Puerto bfl2i£â0 &l£dO. £ubao Other LâliO .061## .032»# .015 .030 -.012 .051 -.016 .064## .020 .009 .022 .056## -.097## -.103## -.115# -.126## -.077## -.072## .048## .062## .006 .047## .049# -.039 .011 -.012 -.002 -.002 -.002 -.072 .047## .020 -.023 -.003 .045# -.025 .056## -.135## -.176## -.074#* -.117## -.089## .028# -.037 -.060 -.035 .007 T . 062 .399## .051## .070## .167## .010 .013 .006 .379## -.001 -.064## .065## -.006 -.027 -.003 ,306## .323## -.005 .017 -.079## -.101## .071## .049# -.103## -.069## .082# .038 .065 -.001 .304## -.004 -.130## .099## -.111## .065## -.039 .366## .033 -.107## .112## -.081## .068## -.043 .016 .011 .109#* -.009 .010 .029 st the .05 level (using the two-tailed t-test). at the .01 level (using the two-tailed t-test). 79 I second regression model in which the variable years of ' education completed was omitted.) Pursuing this same ; argument, we would expect the overeducated to earn more I than others and the undereducated to earn less. This, I however, does not appear to be the case for every ethnic-sex group. When the education variable (EDUC) is j left in the model, overeducation is negatively related to learnings for half of the twelve groups. That is, for I white and black females, and Mexican and Cuban males and females, the overeducated earn less than those with less education. The findings for these six groups seem to lend some support to Freeman's (19 76) view that there is no iincome benefit derived from education in excess of that I necessary for a given occupation. However, for white and ; black males, and Puerto Rican and other Latin males and ! females, overeducation is positively associated with ,earnings. However, overeducation is not statistically t : significant for any of the twelve groups, and relative to i ■many of the standardized regression coefficients for other Variables, overeducation does not appear to be a major [determinant of earnings. When the variable EDUC is omitted I from the regression model results for OVERED change. For !example. Table 4.3 indicates that overeducation enhances earnings for eleven of the twelve ethnic-sex groups, and is statistically significant for eight groups. The regression results for undereducation are clearer ; and not unexpected. Undereducation is negatively related ; to earnings for ten of the twelve ethnic-sex groups. Only I for Mexican and Cuban males is undereducation positively j related to earnings. Results for UNDERED are even more ! consistent in the second model. Here undereducation is negatively associated with earnings and statistically significant for all twelve groups. This suggests that, consistent with the human capital framework, persons ! having less education tend to earn less than their I counterparts at the same occupation. However, given that I ! for six of the twelve groups overeducation was also ! negatively related to earnings, it appears that, in many cases, there is an educational threshold beyond which I earnings returns decline. j Of the four human capital variables, none are statistically significant for all twelve ethnic-sex groups I I I according to the main regression equation (which includes ! the variable years of education completed). However, jeducation is significant for ten of the groups; only for I Mexican and other Latin females is it not significant. : Experience is shown to be significant for all the males 1 i except Cubans. Among females experience proved to be 811 significant for whites and Puerto Ricans only, i Overeducation was not statistically significant for any of : the groups in the main model, though undereducation was ’ significant for six groups (i.e., white males and all i I females except Puerto Ricans). For all groups, both ; J education and experience were positively associated with earnings. Overeducation was negatively associated with earnings for half the groups, two male groups (Mexicans ^ and Cubans) and four female groups (whites, blacks, j Mexicans and Cubans). Undereducation proved to be I ’ I negatively associated with earnings for ten of the twelve . ! ethnic-sex groups. ] I Results for overeducation and undereducation were more | I i consistent and more likely to be significant in the second : jmodel. Undereducation served to reduce earnings for all ! I jtwelve groups and was statistically significant for all. I Overeducation was positively associated with earnings for 1 eleven groups and significant for eight groups, i The relative importance of each variable in determining earnings is shown by the magnitude of the standardized regression coefficients. (See Table 4.2) Education proved to be the most important determinant of earnings among the human capital variables for eleven groups (all but Mexican females for whom undereducation ^3 was the greatest determinant of earnings, followed by education). Among the four human capital variables I experience was the second most important determinant of I earnings for seven groups, and undereducation for four ! groups. In considering all the variables, education was I among the five most influential determinants of earnings for all groups except Mexican and other Latin females. In summary, it appears that education is an important determinant of earnings for ten of the twelve ethnic-sex groups, and that, as expected, education and experience I I enhance earnings. Overeducation was not statistically ! significant for any of the twelve groups in the first 1 ; model. Further, it had mixed effects on earnings. For i six of the groups overeducation had a positive effect on I earnings. This would be expected based on the assumptions j of the human capital framework. However, for the six i remaining groups it appears that overeducated workers are I in no way rewarded financially for their additional jeducation as compared to non-overeducated workers. That : is, overeducation was negatively associated with ^ earnings. Results for undereducation were more consistent and less surprising. First, undereducation proved to be I statistically significant for half of the ethnic-sex ! groups. Second, as expected, undereducation had a 83; --J negative impact on earnings for ten of the twelve groups. ; Finally, undereducation was negatively associated with I earnings for each of the six groups in which j undereducation was statistically significant. These results suggest that having more than the average I education for a job (i.e., being overeducated) is not financially beneficial for some groups (i.e., Mexican and Cuban males and females, and white and black females). For these groups education beyond the average for a specific occupation may not be worth the cost of additional training. However, for the remaining six groups (i.e., white and black males, and Puerto Rican and other Latin males and females) overeducation does appear I to enhance earnings to some degree. On the other hand, I I j for ten of the twelve groups workers with less education ! than the average for their jobs (i.e., the undereducated) I appear to suffer financially from this lack of education. This suggests that optimal returns are obtained by those persons having average educational levels for their I specific occupations. Finally, it appears that for most i I groups education and experience are more important I determinants of earnings than are overeducation or I Iundereducation. It was, however, discovered that results for undereducation and overeducation differ dramatically 84 depending on whether or not the variable years of education completed is included in the model. When ! education is omitted from the model, the overeducation and I undereducation variables act as predicted by the human . capital framework. That is, overeducation is positively j associated with earnings for eleven groups, while undereducation is negatively associated with earnings for all twelve groups. A. 2. Structural Variables « -------------------------- ! The earnings model considered in this study includes ! five variables defined as structural components. These j i are : sector of employment (SECTOR)— government versus j private; state unemployment rate (UNEMP); and three ; regional variables, the Northeast, South and West. The I North Central region is the omitted category. ] I A .2.a. Means I ; Sector of employment (SECTOR) reveals that Cuban males I and females are least likely to be government employees '(probably due, in part, to not being U.S. citizens), while ; I I I black males and females are most likely to be government ' I employees. Also, a greater percentage of women than men I are employed by federal, state or local government. For I men, the percentage in government employment ranges from a i I low of 11 to a high of 26. For women, 12 to 32 percent 85 are employed in the public sector. The average state unemployment rate (UNEMP) ranges from 5.4 to 6.6 percent. Puerto Ricans appear to be at the greatest disadvantage, with the average state unemployment rate for both the males and females at 6.6. No doubt, this is due to Puerto Ricans being heavily concentrated in the Northeast, the region having the highest average unemployment rate in 1979. (About 71 percent of all Puerto Ricans in this sample are located in that region.) Mexicans tend to be located in states that I had relatively low levels of unemployment in 1979. The ; average state unemployment rate for Mexican males and I ' females was 5.4 percent. This is below the U.S. average I of 5.8 percent for 1979. (Note that Mexicans are heavily 'concentrated in the western and southern portions of the country, areas not as plagued by unemployment as the j industrial Northeast.) Slightly over half of the blacks in the sample are located in the South, with roughly 9 percent in the West, 20 percent located in the Northeast and another 20 percent in the North Central region. Whites were far more evenly distributed among the regions than were the minority groups. A. 2.b . Other Statistics With regard to the statistical significance of the structural variables, there was one item that was 86 significant for all of the twelve groups considered— the ' southern region. The Northeast was also significant for ■ white, black, Puerto Rican and other Latin males, while i the West was statistically significant for black, Mexican I ■ and other Latin males, and white females. These three ! regions, when significant, were usually negatively associated with earnings relative to the omitted category. This means that, among those groups for which the Northeast, South and West were statistically ; significant, earnings were usually less than in the North : Central region. The only exception to this is for white I females who earned more in the South and West than in the ]Northeast or North Central regions. This means that I though whites females do better economically in the South, I I followed by the West, than in other regions, the other I ‘groups do better in one of the regions outside the South, land earn most in the North Central region. Sector of employment was statistically significant for I white and Cuban males and females, as well as Puerto Rican ,and other Latin males. Interestingly enough, it appears that black and Puerto Rican males, and white, black, Mexican and Cuban females earn more through government employment than in the private sector. However, the earnings of white males (as well as Mexican, Cuban and 87 other Latin males, and Puerto Rican and other Latin : females) were lower in government than in private sector j employment. This effect, government having beneficial impacts on earnings for some minorities and most females, but a negative impact on the earnings of white males, may ' be due to more discriminatory wage scales experienced by I women and black and Puerto Rican males in the private sector. The state unemployment rate is significant for white males and females, Cuban males and black females. Among I ■ the four groups for which this variable is statistically ; significant, it is negatively associated with earnings only for Cuban males. This means that in states with I I higher unemployment rates the earnings for employed Cuban imales are depressed. This is as expected. Curiously : enough, unemployment is positively associated with earnings for white males and females and black females. I This is contrary to what one would expect, since it I appears that for these groups, the higher the unemployment j I rate the greater are their earnings. One possible ! explanation is that some of those states which had i I I relatively high levels of unemployment also had relatively jhigh wage rates. Indeed, Figure 4.1 bears this out. The ,figure plots the relationship between state unemployment 88 Figure 4 .1 State-Level Unemployment Rates and Mean Household Earnings, 1979 10 7- U N E 6 -j U P L 0 Y U E N T 4- 3-! 2 - j 7 i4 0- 10000 6000 9000 11000 12000 13000 UEAM HOUSEHOLD EARNINGS 14000 15000 SOURCE: Unemployment data are from U.S. Bureau of the Census, 1983:424. Earnings data are from the 1980 1 decennial census. L 89 I rate and mean household earnings for 1979. These items were correlated at .436 which was statistically ; significant at the .05 level. This confirms the roughly ' linear relationship which is illustrated in Figure 4.1. I The standardized regression coefficients show that j among the five structural variables the South is the I j single most important determinant of earnings for all jgroups except white females. (For white females this I variable was second to state unemployment rate.) The i I South is among the top four most important of all the I variables for black, Mexican, Puerto Rican and Cuban males and females, and other Latin females. In summary, the structural variables revealed some : interesting findings. First, the earnings of black and I ; Puerto Rican males, and most of the female ethnic groups I ! were higher in government employment relative to employment in the private sector, while the opposite was true for white males and some other groups. This suggests jthat women, black men and Puerto Rican men may face I I greater wage discrimination in the private sector than in I government. Second, the regional distributions of i I minorities appear to be highly skewed. Over half the j black population and over 60 percent of all Cubans live in the South. Also, over half of the Mexican population is 90; concentrated in the West, while over 70 percent of all Puerto Ricans are living in the Northeast. The South was I statistically significant for every group. Further, at I ; least one regional location appears to be among the more : : i ■ important determinants of earnings for each of the twelve | groups. Third, UNEMP, the variable measuring the state | unemployment rate, led to curious results. It seems that , for white males and females and black females earnings are , higher as the state's unemployment rate increases. For | I iCuban males, however, the findings for UNEMP were as ! expected. For this group it was found that the higher the • j unemployment rate the lower the earnings. One would | I ■ ! expect that high unemployment rates would serve to reduce ^ earnings. A possible explanation is that unemployment was , I high in some of those states which have high wage rates. i 1 I (Perhaps heavily unionized industries within particular ! I states prevented wage rates from declining in spite of ! high unemployment rates.) The graph shown in Figure 4.1 I points to a somewhat linear relationship between state i iunemployment rates and mean household earnings. Finally, ' only for white and black females did UNEMP appear among the top eight determinants of earnings. : A .3. Control Variables There are eight control (or demographic) variables. These variables are: English language proficiency (LANG) two variables measuring period of immigration (IMMIG), marital status (MARITAL), hours worked in 1979 (HOURS), number of weeks unemployed in 1979 (WEEKSUN), a measure of occupational status (OCCUP), and country of citizenship (CITIZEN). ! A .3.a. Means i I Examination of the mean occupational status scores I across groups reveals some striking differences. The I occupational status (or prestige) score has been estimated | ' using education and income rankings of detailed occupations. Scores range from zero to 99.^ Mean ; status scores are highest for white males with 60, I followed by Cuban males and white females (each with 52), 'Other Latin males with 49, black males with 45, Puerto I Rican males with 44, black females with 43, Mexican males I and Puerto Rican, Cuban and other Latin females with 42 each, and Mexican females with 38. I The vast majority of persons in this sample are I married. This is not surprising given that the sample was I limited to persons ages 25 through 64. However, it was I surprising to find that females in the sample were less likely to be married and significantly more likely to be ; unmarried. Perhaps the relatively high threshold of hours : worked (1,365 hours for the year) accounts for these 92j ' results. I Among each of the twelve groups, the number of weeks unemployed averages less than one. Indeed, WEEKSUN ranges I i from a low of 0.6 to a high of 0.9. Black males had the ! highest average number of weeks unemployed (0.9). ! The average number of hours worked in 1979 was, not surprisingly, highest for white males with 2,221 hours. (Note that working 40 hour weeks for 52 weeks comes to 2,080 hours in one year.) Among males, blacks and Puerto Ricans had the lowest averages, 2,101 hours and 2,079 j hours, respectively. Females appear to work a lower average number of hours. Cuban females had the highest 1 average with 2,014 hours, and Puerto Rican females had the ] lowest with 1,961 hours. Several of the control variables were included ’ specifically to aid in analyzing the earnings of Hispanics. These variables are English language : proficiency (LANG), country of citizenship (CITIZEN) and ! I jperiod of immigration (IMMIGi and IMMIG2 ). Comparing ; the means across groups for these variables yields I interesting results. With respect to LANG we see that 1 virtually all the white and black males and females speak ! English well, very well or are native English speakers. Among the Hispanic ethnic groups this is not the case. 93 Twenty-four percent of Mexican men and 17 percent of ' [ Mexican women reported that they either do not speak English well or do not speak it at all. Similarly, 17 and I : 20 percent of other Latin males and females, respectively, did not speak English well. Cubans appear to have the largest percentage least able to speak English well, with j32 and 38 percent so reporting for males and females respectively. The percentage of non-English speakers (or persons who speak English with difficulty) is lowest among I the Hispanics for Puerto Ricans. Fifteen percent of j Puerto Rican males and females reported difficulties with i English. I U.S. citizenship was also, not surprisingly, far more I common among whites and blacks than among Hispanics. The ipercentage of persons who were hot U.S. citizens ranged I 'from 1 to 3 percent for white and black males and jfemales. Similar percentages were found for Puerto jRicans. However, for the remaining three Hispanic ethnic I groups far higher percentages were not U.S. citizens. I About 29 pecent of Mexican males and 22 percent of Mexican jfemales were not U.S. citizens. Among other Latins this figure was 35 pecent for both males and females. Cubans jhad the largest percentage of non-U.S. citizens with 41 ! and 42 percent for males and females, respectively. 94, i The variable IMMIG i identifies those persons who I immigrated to the U.S. between 1970 and 1980. (Recall I these Census data were collected in 1980 so IMMIGi I ' represents persons who have been in the U.S. ten years or ■ less.) IMMIGz measures earlier arrivals, those coming I to the U.S. before 1970. Only between 1 and 3 percent of white and black males and females immigrated to the U.S. during either of these periods. The remainder are native born Americans. Among Puerto Ricans, similarly low ; percentages immigrated during these periods, ranging from ; 2 to 4 percent. This is due to the fact that Puerto Rico is a territory of the U.S. and so moving from Puerto Rico to the U.S. mainland does not be constitute immigration. (The term immigration refers to movement across I international boundaries. Thus, when Puerto Ricans come I , to the U.S. mainland they are not considered immigrants.) ; For Mexicans and other Latins roughly the same percentage I immigrated during each of the two periods. (Among Mexican ! males 20 percent came during each period, for Mexican I females 15 and 16 percent came, other Latin males were split 26 and 27 percent, and other Latin females, 26 and 29 percent.) Cuban percentages were highly skewed with 74 percent of both males and females reporting that they came to the U.S. before 1970. Nineteen percent of the Cuban 95 males and 20 percent of the Cuban females arrived between 1970 and 1980. A.3.b. Other Statistics Among all 17 variables in the model, occupational status has proven to be the single most important ; determinant of earnings for all groups except Mexican males. (For Mexican males it is the third most important determinant following Southern residence and years of education.) For each of the twelve ethnic-sex groups, OCCUP is statistically significant and positively 1 I I associated with earnings. This means that for each i increase in occupational status, earnings increase. Marital status is statistically significant for all I j the males but, among the females, it is only significant I for whites. Married men in each of the six ethnic/race I groups earn more than their unmarried counterparts. For j women results are not so clear. For two groups married I females have a very slight earnings advantage over their j unmarried counterparts. However, for four other groups, j the unmarried females hold an earnings advantage. For whites, the only group among females for which marital status was statistically significant, unmarried females I I earned more than their married counterparts. For each of the male ethnic/race groups marital status appeared among 96 ' the top eight determinants of earnings. In no case did it ! appear among the top eight for females. WEEKSUN and HOURS are statistically significant for ! all groups. Further, as expected, weeks of unemployment i ' is negatively associated with earnings, while hours worked ; is positively associated with earnings. I ' English proficiency proved to be statistically I I significant for each of the Hispanic ethnic groups. As I expected, not speaking English was found to be negatively I associated with earnings. (This was true for all the ' ethnic-sex groups except white females for whom this jvariable was not statistically significant.) Country of citizenship (CITIZEN) was statistically significant for 'Mexican males, Cuban males and females and other Latin I females. For all of these groups U.S. citizenship was positively associated with earnings. IMMIGi ; (immigrating to the U.S. during the period 1970-1980) was ' statistically significant for white and other Latin males jonly. For the latter group immigrating to the U.S. during I this period was found to be negatively associated with earnings. Again, this is as expected, since it is assumed that recent immigrants would have less knowledge of the U.S. labor market and would be less able to negotiate wages than earlier arrivals or native born Americans. For 97 white males, however, immigrating during this period was positively associated with earnings. This discrepancy may be due to the specific occupational background of recent immigrants. For example, Russian Jewish immigrants and I refugees (who are classified as white) would tend to be highly educated and skilled workers (Kats, 1982:664) whereas some of the Hispanic immigrants are less likely to i be so educated and skilled. These results may also , reflect the greater ease of assimilation into American | ! society of white immigrants such as Russian Jews or i i I Poles. Their respective ethnic and religious communities 1 I I I have, in some cases, provided services to these groups i * I I beyond that available to other immigrant groups. j ! i IMMIG2 f coming to the U.S. before 1970, was ; I i statistically significant only for Mexican females. This I variable was found to be positively associated with I earnings. This means that Mexican females arriving in the U.S. before 1970 tend to earn more than native born females of Mexican descent. This is probably due to the positive selection of immigrants (Cornelius, 1978:20). i : Cornelius, for example, found that migrants were less fatalistic, less submissive to authority, exhibited a J "stronger need for achievement" and had a stronger I propensity to plan for the future relative to I 98| _ ■ non-migrants. Similarly, other studies have found that I ; international migrants have a high degree of 1 self-confidence and self-reliance, and are also highly ; individualistic (United Nations, 1973:237). In summary, examination of the control variables I highlighted several interesting findings. The average I ' occupational status scores varied considerably among the ethnic-sex groups and was highest for white males. For eleven of the twelve ethnic-sex groups, occupational ; status was the single most important determinant of i earnings. The vast majority of persons in the sample were ' married, with males more likely to be married than I ' females. For each ethnic group, the married males earned ! more than their unmarried counterparts. For females I results were mixed, though in four of the six ethnic groups unmarried females had an earnings advantage over I married; females. Marital status proved to be statistically significant for each of the male ethnic groups, but, among the females, it was only significant I for whites. Marital status was among the major determinants of earnings for each male ethnic group, but never appeared among the eight most influential variables for females. 99, The number of weeks unemployed and hours worked were both statistically significant for every group. Also, as . expected, periods of unemployment were negatively associated with earnings, while hours worked was , positively associated with earnings. i One or more of the variables added specifically to aid | ! in the analysis of earnings for Hispanics, LANG, CITIZEN, IMMIG1 and IMMIG2 , proved to be major determinants of . earnings for each of the Hispanic ethnic groups except 1 ; Puerto Rican males. However, none of these items appeared ■ I among the top eight most influential variables for whites I or blacks. In only one instance was one of these ; ■ i variables even significant for a non-Hispanic group. (This was IMMIGi for white males.) English proficiency ! proved to be statistically significant for each of the j ! Hispanic ethnic groups. As expected, difficulty speaking | ; English was negatively associated with earnings for all : Hispanics. Further, U.S. citizenship was found to enhance I I earnings in those cases where it was significant. The i variables IMMIG % and IMMIG2 had mixed results, j IMMIG1 was significant for white and other Latin males I while IMMIG2 was significant only for Mexican females. 1 ; These immigration variables were positively associated ' with earnings for the white males and Mexican females, but 1001 I were negatively associated with earnings for other Latin 1 I males. This means that white males immigrating between ! 1970-1980 and Mexican females who immigrated before 1970 ; 1 I earned more than their native born counterparts. This may ; 1 reflect the positive selection of immigrants, and, in the I case of white males, the assistance provided by their ethnic and religious communities in easing their transition to a new country. For other Latin males who immigrated since 1970, results were as expected, their earnings were lower than their native born counterparts and those immigrant counterparts who came prior to 1970. Standardized regression coefficients for IMMIGi and I jIMMIG2 suggest that, for many groups, earnings will I increase as years of residence in the U.S. increases. A.4. Major Determinants of Earnings I In order to identify those variables which had the greatest impact on earnings, the standard scores (as shown jin Table 4.2) were examined. For each of the ethnic-sex Igroups the top eight (out of a total of 17) standard scores were identified and ranked. OCCUP, a measure of ■occupational status, and the Southern region were the only ■variables which appeared among the top eight rankings for 1 jeach of the twelve ethnic-sex groups. Additionally, I ! {Occupational status was the single most influential , 101 determinant of earnings for eleven groups, and was the i third most important variable for Mexican males. Several I variables appeared among the eight most influential for i ■ eleven of the twelve ethnic-sex groups. These were: ( I ; years of education (for all except Mexican females); hours : ! worked (for all except Puerto Rican females); and weeks of ' unemployment (for all except Mexican males). Marital | status was found among the top eight variables for each of j the male ethnic/race groups, but not for a single female group. The undereducation variable ranked in the top I eight for each of the female ethnic-sex groups except 'Other Latin females. Among the males experience was an I i important determinant of earnings appearing among the top ! eight determinants for all groups except Cubans. The control (or demographic) variables are proportionally found most often among the eight most j influential, closely followed by the human capital variables. Structural variables appear among the top I eight least often. I |B. Between Group Results I Comparing regression results between groups involves i ; use of the metric or raw score regression coefficients jpresented in Table 4,4. Because the dependent variable, annual earnings in 1979, has been logged, these 102 j coefficients are interpreted as percentages. For example, ■ leach additional year of schooling added 3.4 percent to the i earnings of white males according to results presented in t I Table 4.4. Also, white males who did not speak English I ' Swell earned 13.6 (or roughly 14) percent less than white ! I males who spoke English well or fluently. i IB.1. Human Capital Variables I : As reported in Part A of this chapter, educational j attainment (EDUC) was statistically significant for ten of | the twelve ethnic-sex groups. It also appears to be substantively important for most of the groups. After controlling for all other variables each one-year I increment in education increases earnings by a minimum of I 10.9 percent (for other Latin females) and a maximum of 3.7 i I I ^ percent (for Mexican males). It is surprising to note the | I I jslightly greater returns to education received by Mexican ! and Puerto Rican males over white males. All other i groups, both male and female, have lower returns to education than white males. Among males, blacks receive i I the lowest returns to education at 2.1 percent. Among ; females, whites receive the highest returns (3.0 percent ifor each additional year of education) while other Latin females receive the lowest. Among all groups, the four lowest returns to education are received by black males 103 J Table 4.4. Metric Regression Coefficients and Standard Errors of All Independent Variables in the Analysis (First Model: With EDUC) Independent Variables U011& Glacis da&icao Puerto Eicaa- Cuban Other Lalin EDUC . 034** . 021** .037** .035** .030** .025** (.003)g/ ( .007) (.006) (.007) ( .005) (.005) EXPER .008** . 005** .005** .006** .001 .007** ( 0) (.001) (.001) (.001) (.001) (.001) OVERED .031 .033 -.084 . 033 -.048 .037 ( .018) ( . 049) (.058) (.063) (.037) (.036) OVERED -.099** -.078 .002 -.009 .065 -.049 (.019) ( .047) (.041) ( . 044) ( .036) (.037) SECTOR -.129** .004 — .050 .118** -.089** -.089** (.013) (.028) (.034) (.033) (.031) (.027) UNEMP .0 24** .017 -.019 .007 -.072** .015 (.005) ( .012) (.021) (.020) (.021) (.013) REGION -.070** -.215** -.169 -.173** .04 5 -.095* ( .014) (.040) (.117) ( .047) (.054) (.043) REGION -.116** -.314** -.371** -.279** -. 144** -.119** (.012) (.032) (.048) (.055) (.045) (.042) REGION 3 -.010 -.131** -.171** -.102 -.023 -.079* ( .015) (.049) ( . 041) (.058) (.055) (.039) OCCUP .007** .0 0 8** .006** .00 6 ** .006** .008** ( 0) ( . 001 ) (.001) (.001) (.001) ( .001) MARITAL -.2 01** -.121** -.190** -.13 8** -.131** -.159** (.013) ( .028) (.032) (.031) (.026) ( .024) WEEKSUN -.020** -.015** -.013** -.024** -.024** -.017** ( .002) ( .004) ( . 004 ) (.005) (.004) (. 004 ) HOURS 0** 0** 0** 0** 0 * * 0** ( 0) ( 0) ( 0) ( 0) ( 0) ( 0) LANG -.136 — .116 -.107** -.086* -.084** - .119** ( .078) (.270) ( .035) (.037) (.027) (.031) CITIZEN -.005 — .068 .087* -.132 . 094** .018 (.061) (.129) ( .043) (.128) (.023) (.031) IMMIG .146* -.139 -.085 -.191 -.001 -.142** IMMIG b Int Adj. 1 2 c (.075) .053 (.034) 8.381 .208 (.130) -.043 ( . 108) 8.580 .161 (.050) .031 (.041 ) 8.592 .192 (.114) -.137 ( .075) 8.568 .133 (.045) .047 ( .038) 8.991 .205 (.036) .009 (.029) 8.462 .244 a Standard errors in parentheses. t > Int = intercept, e 2 1 Adjusted R = 1—(1-R ) (n-l)/(n —p), u; her e p = number of parameters. * Significant at the .05 level (using the two-tailed t-test). ** Significant at the .01 level (using the two-tailed t-test). 104 Table 4.4. (ContVd.) Metric Regression Coefficients and Standard Errors of All Independent Variables in the Analysis (First Model: With EDUC) Independent Variables EDUC EXPER OVERED 3 OVERED SECTOR UNEMP REGION REGION > 2 REGION OCCUP MARITAL WEEKSUN HOURS LANG CITIZEN IMMIG 1 IMMIG Int 2 c hhXia Adj. R .231 .030** .028** .011 (.004) (.008) (.008) .004** .002 0 (.001) (.001) (.002) -.032 - . 0 06 -.111 (.028) (.065) (.096) -.097** - . 13 0** -.117* (.023) ( .051) (.054) .040** .056 .008 (.014) (.029) (.040) .031** .033** -.001 (.005) (.013) (.027) -.009 -.070 -.019 (.017) (.042) (.152) -.064** -.193** -.235** Puerto ElfifiO. .026** (.009) .0 0 5** (.002) .135 (.093) -.092 (.057) -.036 (.040) -. 007 ( .026) -.097 ( .061) -.171** Other IfillD .011* .009 (.005) (.006) .001 .002 (.001) (.001) -.034 .099 (.037) (.052) -.07 0* - .085* (.034) (.043) .075* -.012 ( .030) (.031) .020 .030 (.021) (.015) -.029 -.029 ( .058) -.140** (.054) -.134** ( .014) (.033) (.065) (.066) (.050) ( .051) .038* -.103 -.077 -.096 .014 -.076 (.017) (.053) (.057) (.076) (.059) (.050) .00 8** .009** .008** .008** .007** .009** ( 0) (.001) (.001) (.001) (.001) (.001) . 0 57** -.002 — .008 .019 -.005 .043 (.012) (.025) (.033) (.033) (.020) (.024) -.016** -.017** -.018** -.027** —.030** -.024** ( .002) (.005) (.006) (.007) (.004) (.004) 0** 0** 0** 0* 0** 0** ( 0) ( 0) ( 0) ( 0) ( 0) ( 0) .131 -.022 -. 164** -.102* -.126** -.121** ( .084) (.237) (.054) (.047) (.025) (.036) .066 -.129 .124 .216 .070** . 088* ( .068) (.132) (.066) (.150) (.022) (.035) .052 -.018 .126 .003 -.059 -.062 ( .095) (.141) (.076) (.132) (.046) (.042) .057 .039 .193** -.024 .011 .039 (.035) ( .092) (.055) (.088) (.040) (.033) 7.425 7.997 8.288 8.079 8 .233 7.967 .195 .182 .158 .239 .246 Int * Intercept, c 2 2 Adjusted R = 1-(1—R ) ( n-1 )/ ( n-p ) * inhere p = number of parameters. * Significant at the .05 level (using the two-tailed t-test). ** Significant at the ,01 level (using the two-tailed t-test). L. 105 I ! and three Hispanic female groups (black males— 2.1 i percent, Mexican and Cuban females— 1.1 percent, and other ! Latins— 0.9 percent). It is interesting to note that I within each ethnic-sex group, with the exception of i ; I blacks, females have lower returns to education than their ; male counterparts. ' Each additional year of work experience (EXPER) served i to augment earnings for each ethnic-sex group except Mexican females for whom this coefficient was zero. For white males, experience increased earnings more than for other groups (adding 0.8 percent for each year of experience). Other Latin males followed with returns of 10.7 percent. Mexican females received no returns to I experience, and Cuban males and females received returns I of only 0.1 percent each. Again, the returns to I experience for males tended to exceed the returns for ! females. I The results also indicate that, for six ethnic-sex I I groups, overeducated persons tend to earn less than ,persons who are not overeducated. (These groups are [Mexican and Cuban males, and white, black, Mexican and Cuban females.) Among these groups, Mexican males and females realize the greatest earnings disadvantage by being overeducated. The earnings of overeducated Mexican 106 j females are about 11 percent less than their counterparts I who are not overeducated. For Mexican males the ! overeducated earn about 8 percent less. Overeducated I workers in the remaining six ethnic-sex groups realize an I earnings advantage over those who are not overeducated, ; This advantage ranges from 3.1 percent (for white males) to 13.5 percent (for Puerto Rican females). Results differ when years of education completed is omitted from the model. Table 4.5 presents these results. As shown, overeducation serves to enhance I earnings for all groups except Mexican females for whom ! results were not statistically significant. Further, I jovereducation seems to have a large impact on earnings for many of the twelve groups. Earnings were increased from 1.6 percent for Cuban females to 24 percent for Puerto Rican females. 1 Now returning to the model which includes educational I I attainment, I find that for ten of the twelve ethnic-sex Î I groups (all but Mexican and Cuban males) undereducation ,was negatively associated with earnings. The I undereducated suffered earnings reductions ranging from 0.9 percent (for Puerto Rican males) to 13 percent (for black females). The earnings of undereducated females tend to suffer more than do the earnings of undereducated 107 r Table 4.5. Metric Regression Coefficients and Standard Errors of All Independent Variables in the Analysis (Second Model: Without EDUC) Independent Variables WbilA aifisE Puerto &l££Q. Cwb£0 Other EXPER .006** .004** .002* .006** 0 .006** ( 0)1/ (.001) (.001) (.001) ( 0) (.001) OVERED .152 .108* .069 .175** .083** .134** (.016) (.043) (.053) (.058) ( .030) (.031) OVERED 2 -.230** -.173** -.172** -.166** -.082** -.176** (.017) (.037) (.031) (.033) (.027) (.028) SECTOR -.108** .016 -.020 .141** -.069* -.076** (.013) (.028) (.034) (.033) (.031) (.027) UNEMP .024** .017 -.021 .010 -.078** .019 (.005) (.013) (.021) (.020) (.021) (.013) REGION -.069** -.216** -.146 — .16 5** .04 3 -.010* (.014) (.040) (.118) (.047) (.054) (.043) REGION -.125** -.325** -.385** -.253** -.152** -.119** (.012) (.032) (.048) (.055) ( .045) (.042) REGION 3 -.001 -.128** -.169** -.080 -.026 -.084* (.015) (.049) (.042) (.058) (.055) (.039) OCCUP .010** .009** .008** .008** .008** .009** ! ( 0) (.001) (.001) (.001) ( 0) ( 0) 1 MARITAL -.198** -.121#* -.184** -.136** -.182** -.15 6** (.013) (.028) (.032) (.031) (.026) (.024) WEEKSUN -.0 21** -.015** -.015** -.026** -.024** -.018** (.002) (.004) (.004) (.005) ( .004) (.004) 1 HOURS 0** 0** 0** 0** 0** 0** i ( 0) ( 0) ( 0) ( 0) ( 0) ( 0) 1 LANG -.154* -.111 -.138** -.117** -.097** -.138** 1 (.078) (.270) (.035) (.037) (.027) (.031) CITIZEN -. 004 - . 064 .099* -.126 .109** .019 (.061) (.129) (.044) (.129) (.023) (.031) IMMIG 1 1 .156* -.130 -.107* -.183 .018 -.138** 1 1 (.076) (.130) (.050) (.115) (.045) (.036) ! IMMIG . 060 -.026 .027 -.138 .065 .018 1 ^ ' b (.034) (.108) ( .041) (.076) (.038) (.029) Int 8.726 8.807 8.998 8.886 9.306 8.697 Adj. R .198 .158 .183 .125 .197 .238 Standard errors in parentheses. b Int = intercept, c 2 2 Adjusted R = l-Cl-R ) (n-1)/(n-p), where p = number of parameters * Significant at the .05 level (using the two-tailed t-test). ** Significant at the .01 level (using the two-tailed t-test). 108 Table 4.5. (Cont'd.) Metric Regression Coefficients and Standard Errors of All Independent Variables in the Analysis (Second Model: Without EDUC) Independent Variables Int = intercept, c £ 2 Adjusted R = 1-Cl-R ) (n-1)/(n-p), where p = number of parameters. * Significant at the .05 level (using the two-tailed t-test). *» Significant at the .01 level (using the two-tailed t-test). Ublls filfiCE Puerto fii£aa. Cuban Other Lalin EXPER .003## .001 -.001 .002 .001 .001 ( 0) (.001) (.002) (.002) (.001) (.001) OVERED .082## .099* -.062 .237** .016 .138** (.025) • (.057) (.088) (.087) (.028) (.045) OVERED -.194#* -.232** -.167** -.206** -.118*# -.124** (.020) (.042) (.039) (.043) (.026) (.033) SECTOR .066** .072** .017 -.019 .084** -.005 (.014) (.028) (.039) (.039) (.030) (.031) UNEMP .032** .032* -.001 -.002 .019 .031* (.005) (.013) (.027) (.026) (.021) (.015) REGION J — .008 -.070 -.014 -.105 -.031 -.035 (.017) (.042) ( .152) (.061) (.058) (.054) REGION -.069#* -.193** —.237** -.158* -.142** —.136** (.014) (.033) (.065) (.066) (.050) (.051) REGION 3 .042* -.101 -.078 -.087 .015 -.080 (.017) (.053) (.057) (.076) (.059) (.050) OCCUP .010** .011** .008** .009** .007** .009** ( 0) (.001) (.001) (.001) (.001) (.001) MARITAL .062** -.002 -.007 .023 -.005 .043 (.012) (.025) (.033) (.033) (.020) (.024) WEEKSUN -.017** -.018** -.019** -.028** -.031** -.024** (.002) ( .005) (.006) (.007) (.004) (.004) HOURS 0** 0** 0** 0* 0** 0** ( 0) ( 0) ( 0) ( 0) ( 0) ( 0) LANG .085 -.078 -.177** -.128** -.132** -.130** (.085) (.237) (.054) (.047) (.024) (.035) CITIZEN .066 -.125 .129* .202 .077** .091** (.069) (.133) (.066) (.150) (.022) (.035) IMMIG .047 -.016 .119 -.004 -.057 -.062 (.095) (.142) (.076) (.132) (.046) (.042) IMMIG .058 .051 .193** -.330 .013 .040 b (.035) (.092) (.05 5) (.088) (.040) (.033) Int 7.7 59 8.313 8.413 8.325 8.350 8.054 Adj. R .225 .192 .182 .153 .238 .245 109 males within the same ethnic/racial groups. Undereducation is statistically significant for five of I the female groups (i.e., all but Puerto Ricans) and one I j male group (white males). However, results based on Table I j 4.5, the model which omits education, show a consistently I negative impact on earnings. Being undereducated served to reduce earnings by a minimum of 8.2 percent for Cuban males and a maximum of 23 percent for white males and black females. Undereducation was statistically significant for all groups in this model. ! In summary, Mexican and Puerto Rican males received ithe greatest returns to education followed by white ■males. Black males, and Mexican, Cuban and other Latin ! females received the lowest returns to education. White I I jmales received the highest returns to experience while jreturns were lowest for females, particularly Mexican I I females. Overeducation appeared to have a negative effect I on earnings for six ethnic-sex groups, and a positive I effect on the remaining six groups when education was included in the model. IFor the same model undereducation ; seemed to exercise a negative effect on earnings for ten I I of the twelve ethnic-sex groups. Further, females tended I to have their earnings lowered more by undereducation than did males from the same ethnic groups. However, when 11Q ~! education was omitted from the model results for over- and j undereducation tended to be as expected. Overeducation j increased earnings significantly for eight groups, and : I undereducation significantly reduced earnings for every I I I ' ethnic-sex group. ! ! I 1 B.2. Structural Variables | Sector of employment (SECTOR) proved to be | I statistically significant for only half of the ethnic-sex I jgroups. Among these six groups, three earned more in j 'government than in the private sector. (These groups were | I Puerto Rican males and white and Cuban females.) For i I I ; white, Cuban and other Latin males, however, earnings were | ! I ■ lower in government than in private industry (13, 9 and 9 i I percent, respectively). Puerto Rican males realized the jgreatest benefit from government employment, earning 12 I percent more in government. Cuban females earn 8 percent I more in government, and white females earn 4 percent I more. (Results were not significant for the remaining ; groups.) I State-level unemployment rates were significant predictors of earnings for only four of the twelve groups. With each one percent increase in unemployment, earnings for Cuban males decline by about 7 percent. Curiously, for black and white females and white males 111 ■ earnings rose as unemployment rose. The earnings of white I and black females increased by 3 percent, while those of I white males increased by 2 percent. This may indicate j that whites, and black females, tend to be in i i recession-proof occupations, while Cuban males are not. i An alternative explanation is that earnings tend to be higher in those states having higher unemployment rates. This may, in part, reflect the effect of collective bargaining in those states which tend to be heavily unionized. Collective bargaining may serve to maintain I high wage rates in the face of relatively high ! ^ unemployment rates. Certainly, it would reduce the I elasticity of wages. I I When statistically significant, earnings in the Northeast, were lower than in the North Central region I I(the omitted category). Black males are at the greatest I I disadvantage in the Northeast, with their earnings lowered I by almost 22 percent. The earnings of Puerto Rican males I are 17 percent lower in the Northeast, followed by other Latin males with 10 percent lower earnings and white males |with earnings 7 percent lower. Females, too, had lower jearnings in the Northeast relative to the North Central j region, but to a lesser degree than males, and it was not {significant for any of the females. 112 Of all four regions, earnings were lowest in the South . for every group. (This variable is also statistically significant for every group.) Mexican males are worse off ! 1 i than others earning 36 percent less in the South than in j ! the North Central region. Black males earn 31 percent j I 1 ! less in the South, Puerto Rican males earn 28 percent | less, Mexican females 24 percent less and black females 19 ; j percent less. Of all groups, white females are least I penalized by working in the South, though even their ; : earnings are 6 percent lower than in the North Central ; i region. Earnings also tend to be lower in the West than in the j I I : North Central region for all groups except white and Cuban females. However, living in the West does not have as [economically depressing an effect on earnings as living in ! the South. Of all the groups Mexican males receive the greatest negative effect earning 17 percent less than in I the North Central region. Black males follow earning 13 percent less, while Puerto Rican males and black females ! earn 10 percent less. In summary, the South was significant for every I .group. Nine of the twelve groups earned most in the North Central region and all earned least in the South. In comparing across groups, white males and females tended to ; 113 , have the greatest returns (or lowest negative impacts) in I each region. Unfortunately, blacks and Mexicans do worse I I I in those regions where they are concentrated (i.e., blacks I ; in the South and Mexicans in the South and West). Sector ! ! : of employment was significant for six ethnic-sex groups. I Among these groups, Puerto Rican males and white and Cuban females earned more in government than in the private sector. The reverse was true for white, Cuban and other Latin males. Surprisingly, the state unemployment rate had a positive effect on earnings for three of the four j groups for which it was statistically significant. It ; does appear that there is some relationship between I I state-level unemployment rates and earnings. Hence, in ! some states having relatively high unemployment rates, r I wage rates are also high. jB.3. Control Variables I Three of the eight control variables were statistically significant for every group, HOURS, WEEKSUN and OCCUP. Though statistically significant, none of the significant digits are shown for hours worked in Table 4.4 and so will not be discussed. The number of weeks I unemployed is, however, substantively interesting. As expected, each week of unemployment serves to reduce earnings. This negative effect ranges from a high of 3 114 percent for Puerto Rican and Cuban females to a low of 1 I percent for Mexican males. For all groups except whites, : the earnings of females suffer more from periods of I I unemployment than do those of males. Marital status (MARITAL) shows that unmarried males earn less than married males. Among males, unmarried whites are most negatively affected earning 20 percent less than their married counterparts. The unmarried/married differential is narrowest among black males with singles earning 12 percent less than married I I blacks. Though marital status is a statistically I significant determinant of earnings for all males, it is ! only significant among females for whites. Unmarried I ' white women earn about 6 percent more than married white : women. Though results are not significant for other f I females, among the six ethnic/racial groups, in half the junmarried have an earnings disadvantage, and in the other j three groups unmarried workers yield higher earnings than 1 their married counterparts. Differences, however, tend to I ; be slight. ! The results for OCCUP, a measure of occupational jstatus, show that for every increment in occupational I status, earnings increase. While the raw score regression j coefficients show this earnings rise to be slight, 1 115 I percent or less, changes in occupational status amounting I I to differences in status scores of 10 units or more can ( result in substantial earnings increases. It is also well I to keep in mind the impact on earnings of the wide . inter-ethnic differences in mean occupational status ! scores. (Recall, these data were presented in Table i 4.1.) For example, the mean occupational status score of black males was 15 points less than that for white males, while the difference between that for Mexican and white males was 18 points. It is also interesting to note that I not only did females within each ethnic group have lower I ■ mean occupational status scores than males from the same ! ethnic group, but female returns to occupational status I jwere also slightly lower than for males. : The variables included specifically for analyzing :Hispanics (i.e., LANG, IMMIGi, IMMIGz and CITIZEN), ^ were, with one exception, only statistically significant I for Hispanics. When significant, the inability to speak I English well lowered earnings for all groups. Not speaking English well had the greatest negative impact on Mexican females (lowering their earnings by 16 percent), followed by white males (14 percent), Cuban females (13 I percent), other Latin males and females and black males !(12 percent), Mexican males (11 percent) and Puerto Rican 116 . J ; females, Puerto Rican males, Cuban males and black females : ! (10, 9, 8 and 2 percent, respectively). Only for white ! females was the coefficient positive. For this group I ! I earnings were 13 percent higher for those women having ; difficulty speaking English. However, this variable was j i only statistically significant for the eight Hispanic | i ! j ethnic-sex groups. This variable exerted a significant \ impact on earnings for all of the Hispanics except Puerto : 1 Rican females. ' U.S. citizenship (CITIZEN) was positively associated : I ! 'with earnings for those groups in which it was I I statistically significant. Mexican and Cuban males and | ! I other Latin females who were U.S. citizens earned 9 I ,percent more than their counterparts who were not | ,citizens, while Cuban females who were U.S. citizens I I I earned 7 percent more. j Immigrating during the period 1970-80 (IMMIGi) I proved to be significant for white and other Latin males ; only. IMMIGi was negatively associated with earnings I for the other Latin males (lowering earnings by 14 [percent), though for white males immigrating during this jperiod earnings were 15 percent higher than those of [native born whites, the omitted category! Perhaps this can be explained by the occupational backgrounds of the 1171 [■ immigrants. White immigrants, for example, are more I likely to have experience in professional or technical occupations relative to Hispanic immigrants from largely Third World countries. Also, white immigrants may have an ; easier time establishing themselves in American society I relative to minority immigrants due, in part, to I i assistance provided by the active emigre' and religious communities here which serve them. Immigrating to the U.S. before 1970, IMMIG2 / proved to be significant for Mexican females only. Surprisingly, this group earned 19 percent more than women of Mexican descent born in the U.S. (the omitted category). Perhaps this can be explained by the positive selection of immigrants as posited by some researchers (Cornelius, 1978:20; United Nations, 1973:237) and their lengthier stay in the U.S. relative to those who immigrated since 11970. Although this variable was not significant for I other groups, it was positively associated with earnings ; for nine of the twelve groups. In summary, the number of weeks unemployed was statistically significant for all groups and, as expected, periods of unemployment reduced annual earnings. Hours worked was also found to be statistically significant, as has been found in many earnings models (Verdugo and 118J j Verdugo, 1984), but significant digits do not appear in ! the raw score coefficients. This is probably the case I j because WEEKSUN and HOURS express largely the same concept I (that is, weeks worked during the year) though they were I ; not highly intercorrelated (see correlation matrices in ' Appendix E). Also, as found in other research (Verdugo I and Verdugo, 1984) married males earn more than unmarried males. This relationship is not nearly so strong or consistently positive for females. Figures in Appendix A indicate that married males work slightly more hours than I unmarried males— 2,218 to 2,146 hours— (see Table A.2), I and appear to have far more work experience than unmarried males, 24 years as compared to 18 years for the unmarried I I (see Table A.3). Among females the results are mixed. I I I f , j With respect to hours worked, married females work an | : average of 1,979 hours to the 2,021 worked by the : 1 unmarried. Both married and unmarried females, on the I I ' I other hand, appear to have the same amount of work j jexperience, an average of 23 years. | I Results showed that earnings increase for each I j ' increment in occupational status. Since measures of ! ; I ; occupational status are, in part, based on income, it is | not surprising to find this relationship. While earnings rose only 1 percent or less for each increment in 119 occupational status, it is well to keep in mind the often large differences in mean status scores between white males and minority males and females. Gaps average 15 to 20 points and so differences in occupational status can I I amount to sizable earnings differences. i As expected, not speaking English well tends to lower I I earnings. This variable was a significant determinant of I earnings for all eight Hispanic ethnic-sex groups, j Citizenship and period of immigration did not ; significantly affect as many Hispanic groups. It was i found, however, that being a recent immigrant (arriving I within the last 10 years) and not having U.S. citizenship ! I I often lowered earnings. | ! B.4, Variance Explained by the Model | ' Finally, between-group differences in the amount of | : I ! variance explained by the model (R^) should be j I ! j addressed. R^ ranged from 13 percent for Puerto Rican | I males to 25 percent for other Latin females. (See Tables , 4.4 and 4.5) For every ethnic/race group except Mexicans, j j ; 'the R^ for females was higher than for males in the same ' ! i group. Thus the model seems to be better at explaining the earnings of females than males. (Running the I i identical model without logging the dependent variable, that is, taking actual earnings, tended to yield even 120 I higher R^s, probably due to the greater amount of I ! variance to be explained. Taking unlogged R^s yielded I results ranging from 22 percent for black males and Cuban I j females, to 29 percent for other Latin males.) I IC. Discrimination I In this section an estimate of the extent to which minorities and women suffer from discrimination (i.e., the "cost" of being minority and/or female) is made. This procedure is called "regression standardization" or "decomposition" and has been used for similar purposes by others (Duncan, 1968; Duncan, Featherman and Duncan, 1972; Althauser and Wigler, 1972 ; Masters, 1975). I The most commonly used regression standardization 1 ’technique involves estimating an equation for white males, deriving the metric coefficients and the intercept, and i I substituting the minority means into the equation to I ascertain the expected mean earnings for each of the : remaining eleven ethnic-sex groups. The earnings estimated for minorities and white females using this procedure show the mean earnings these groups would 'receive if they had returns to education, occupation and the other independent variables equal to white males. The Idifference between mean earnings for white males and the [estimated minority and white female earnings shows that portion of the earnings gap which is explained by 121 ivariables in the model. The remaining gap represents the j combined effects of discrimination and other determinants I of earnings not included in the model. ; While the method described above is the most commonly 'used, a modification of this method is included here. ! First, the standardization technique described in the 1 I previous paragraph was used with minority/female means I substituted for the white male means and applied to the I white male regression coefficients and intercept. Second, the regression equations for each of the remaining ethnic-sex groups were taken in turn and the means of the other eleven groups were substituted into the equations, jFor example, means from the remaining eleven groups were ; plugged into the regression equation for white females, 'then black males, black females, Mexican males, etc., 'Until decompositions had been run using regression i : coefficients from each of the ethnic-sex groups in turn. I This was done for two regression models: one which ! included educational attainment and one which omitted this i I ‘variable. Results from these 24 decompositions indicated I that estimates of the cost of discrimination may vary jwidely depending on which group is selected as the benchmark. (Results for each of the 24 decompositions are presented in Appendix F.) In order to arrive at a single 122! : best estimate of the cost of minority/female I 2 I status, the following steps were taken: (1) Using I results from the 24 regression decompositions, differences I in the cost of minority/female status between white males - ■ and each of the remaining ethnic-sex groups were calculated; and (2) from this an average cost was computed for each of the minority/female groups. It should be emphasized that these results are not absolutely significant and that they apply to this model only and the specific sample under analysis. It should also be noted ! that, in order to more easily interpret the results, the ! I decomposition analysis was derived from the identical learnings model as the regression results presented ! earlier, except that here the dependent variable was not ; logged. This enables results to be interpreted in terms I .of the estimated dollar cost, rather than the percentage I of logged earnings which is a less intuitive statistic. The accuracy of estimates made using this technique have been debated. Critics of decomposition argue that other variables aside from discrimination could account j for the earnings gap and, indeed, this may be true. I I Therefore, the component of the earnings difference ! attributed to discrimination may also be due to other variables not included in the model (e.g., family jbackground). Williams, for example, in discussing the I _________________________ 123_ regression decomposition method writes, "(s)uch an ; approach, though none better is known, is fraught with ^ many technical difficulties." He adds that variables I known to influence productivity but difficult to account ; for are : chance, genetic endowment and household or j cultural values (Williams, 1982:54). However, I would I argue that factors such as chance (i.e., luck) and genetic endowment are likely to be normally distributed across race and sex groups. Critics of this method would also I argue that the estimated cost of being minority I exaggerates or at least incorrectly estimates the impact 1 ; of discrimination (Wohlstetter and Coleman, 1972:45; I Sowell, 1981:24). However, it can also be argued that I this same technique underestimates discrimination. For jexample, decomposition of an earnings function does not I consider the discrimination minorities and women face in jobtaining education and other productive characteristics I when these items are held constant (Cain, 1984:7). Cain, I i for example, in his analysis of labor market I 1 discrimination does not hold education constant. He notes : Perhaps less education among minorities reflects societal discrimination - not labor market discrimination, but pre-labor-market discrimination. On the other hand, blacks and women may perceive that higher levels of schooling yield smaller earnings for them than 12 I for white men. If this were true, then these ' groups may have curtailed their schooling. In ' which case educational attainment would reflect 1 labor market discrimination (Cain, 1984:7). ! Further, even if job assignments were made strictly on the I ! basis of an individual's productive characteristics, the I acquisition of such characteristics is influenced by the I economic success of the individual's parents. "Thus, the deleterious consequences of past discrimination for the racial minority are reflected in the model by the fact that minority young people have less successful parents, on average, and thus less favorable parental influences on their skill acquistion processes" (Loury, 1981:125). The : regression standardization technique, as operationalized i here, is unable to include the impact of discrimination experienced in obtaining education and other productive characteristics and so may be underestimating the total I I cost of discrimination. Also, since this sample includes I j only employed persons, the estimate of discrimination does j not include the difficulties minorities and women face in I I gaining employment. Therefore, some critics argue that this procedure underestimates discrimination since it considers only the costs of labor market discrimination. Use of the regression decomposition technique requires that several assumptions be made. First, it must be assumed that all major determinants of earnings have been L 125 j included in the model. This is because the residual (or estimated cost of discrimination) encompasses the effects I of variables not included in the model in addition to I j discrimination. If a major determinant of earnings has been omitted from the model the cost of discrimination I could be significantly overestimated. A second but I related assumption is that variables omitted from this model would have little or no effect on the regression results. Variables which are considered by some to have a strong impact on earnings include genetic endowment, ability and chance (Williams, 1982). However, as mentioned earlier, such items may be assumed to be ■normally distributed throughout the population, and would ! not seem to be determined by race or sex. Hence, such I items would have no effect on the regression results. The third assumption is that workers' earnings are achieved by their own efforts. Therefore, it could be argued that family background, which is not considered in the model used here, would have little impact on determining earnings. Indeed, research has shown that family jbackground has an effect on the early career only (i.e., I I the first job); it has no significant impact later (Blau and Duncan, 1967; Featherman and Hauser, 1978). Because cases selected for analysis in this dissertation are 126 f I limited to persons 24 through 64 years of age, we may ! assume that the majority of these persons are not in the i ! earliest stage of their careers, but are fairly well I ensconced in the labor force. Having reviewed the limitations and assumptions of regression decomposition I I will now review the findings. ! As shown in Table 4.6, white males enjoy the highest mean earnings of the twelve ethnic-sex groups, earning an average of $19,966. Cuban and other Latin males follow earning $16,082 and $15,782, respectively. The mean I 'earnings of black, Mexican and Puerto Rican males range I from $13,962 to $13,275. White females have the highest ! mean earnings among the female ethnic groups, $11,162. j This figure, however, is lower than that of the lowest learning males, Puerto Ricans. Black females follow 'earning an average of $10,134, followed by other Latin, 'Puerto Rican and Cuban females with $9,980, $9,675 and I ^$9,421, respectively. Mexican females have the lowest mean earnings with $8,976. Table 4.7 presents average costs of minority/female ; status derived from the regression decompositions which used, in turn, each of the twelve ethnic-sex groups as j standards with data from two earnings functions. An [examination of Table 4.7 shows that, among males, Puerto 127 /J Table 4.6. Mean Earnings by Mean White Male Earnings (1): Group Mean Earnings (2) Race/Ethnicity and Sex 19,966 Difference Between White Male and Minority/Female Earnings (l)-(2) (3) MALES Black $13,962 $6,004 Mexican 13,830 6,136 Puerto Rican 13,275 6 , 691 Cuban 16,082 3 ,884 Other Latin 15,782 4 , 184 FEMALES White 11,162 8,804 Black 10,134 9,832 Mexican 8,976 10,990 Puerto Rican 9,675 10,291 Cuban 9,421 10,545 Other Latin 9,980 9,986 Ricans experience the greatest cost due to racial discrimination (i.e., residual difference, that difference I ! which is not accounted for by the model) averaging j$3,917, This means that Puerto Rican males earn about I $3,917 less than white males having the same ; characteristics (i.e., education, hours worked, ^ occupational status). Black males followed having an I average cost, relative to white males, of $3,379. Mexicans had the third highest cost of discrimination 128 Table 4.7. Average Cost (or White Males and Average Cost of Minority/Female Group Status Relative to White Males Residual) Difference Minorities/Females Minimum Maximum between Rank ' Within ! Sex ; Group ! 1 MALES Black 3,379 1,544 4 ,362 1 2 : Mexican 2,578 291 3 , 748 3 Puerto Rican 3,917 1,236 5,611 1 1 Cuban 2, 025 822 3 ,362 5 ; Other Latin 2,240 986 3,072 4 : FEMALES White 7,091 5,667 7,815 1 1 ! Black 6,560 3,881 7,949 6 Mexican 6 , 664 4,037 8,267 5 ! Puerto Rican 6 ,968 3,995 8,819 3 ! Cuban 7 ,072 5, 728 8 , 608 2 Other Latin 6 , 678 4,911 7,973 4 I among the males, having an average residual of $2,578. ; This was followed by other Latin males with $2,240 and jCubans with a cost of $2,025. These results were not ! surprising. Previous research (Verdugo and Verdugo, i j1984) has found that black males bear higher costs due to discrimination than Mexican males, and the average costs ranging from $2,025 to $3,917 seem reasonable based on other research. Among the females, however, results were more surprising. (This may, in part, be due to the limited 129 ; use of regression decomposition to estimate the costs of ; discrimination against females in previous research.) Estimates of the costs borne by females due to their j racial and/or female status were quite high, often more 'than double that faced by males of the same ethnic/race I I group, ranging from $6,560 to $7,091. Also surprising was the ranking of female ethnic groups based on the average cost of discrimination. White females appear to bear the highest average cost of discrimination earning $7,091 less than white males with similar 'characteristics. They were closely followed by Cuban I females having a cost of $7,072 and Puerto Ricans I earning, on average, $6,968 less than white males. Other ’Latin females ranked fourth with an average cost of I$6,678, followed by Mexican females with $6,664, and, [finally, blacks bearing a cost of $6,560. These results iwere surprising because it was expected that discrimination based on race and discrimination based on sex would combine such that minority females would appear to face greater costs due to discrimination than white [females. Yet with white females bearing the greatest I [average cost of discrimination and black females the least, this does not appear to be the case. Curious though these results are, several possible explanations 130j I come to mind. First, the range of averaged costs for 'females, from minimum to maximum, is less than $550. ! (For males this range is almost $1,900.) Second, the I I earnings of females married to minority males may be more I jcrucial to their family's well-being due to the lower jearnings of minority men relative to white men. Hence, minority females may be more likely than white females to evaluate a job’s desirability in terms of its salary or wage rate as opposed to other factors (e.g., ! satisfaction). Consequently, females whose earnings I : contribute substantially to family income may have a i I greater incentive to take jobs which pay more than would 1 I females whose earnings contribute relatively little to [family income. Similarly, unmarried females with i I dependent children (that is, divorced, widowed or never [married mothers) may be under greater pressure to seek out and accept higher paying jobs than would unmarried i I females who are not responsible for supporting dependents. Minority females would be more likely than white females to be unmarried heads of households with [children. In short, minority females may make greater i iefforts to obtain higher paying jobs, hence accounting for the surprising decomposition results which found that minority females experience slightly lower costs of 131 I discrimination than white females. However, this I : explanation is purely speculative and more research into I 'the costs of minority and female status and the factors I I which contribute to inter-ethnic cost differences among j females is necessary. \ : Most interesting is the finding that the costs of i I : disrimination borne by females swamp that of minority [ I males. Though racial/ethnic discrimination exacts a heavy | price on the males, sexual discrimination appears to j reduce earnings to a far greater extent. While it would [ seem then that the combination of being both minority and female would subject a worker to even greater earnings j discrimination than either sex or race alone, this does ; I not appear to be the case. White females as well as ,minority females bear much the same costs of I discrimination, and far more than the costs borne by I [minority males. Clearly then being female, more than Ibeing minority, serves to reduce earnings. Other research has resulted in similar findings (Almquist, 1975; Palmore and Manton, 1973). Still, minority males are themselves subject to costly amounts of discrimination. 132 f ! I I I D. Summary I As expected, increases in education and experience led I to increases in earnings. Education is statistically I ' significant for ten of the twelve ethnic-sex groups, and ' j I among the eight most important determinants of earnings j i for eleven of the twelve groups. Returns to education are I ! highest for Mexican males, followed by Puerto Rican and white males. Returns are lowest for black males and Mexican, Cuban and other Latin females. Minorities (with ! the exception of Mexican and Puerto Rican males) and white ' I ] 'females are not compensated for education to the same ! I degree as white males. With regard to experience, white j I males receive returns higher than minorities and females, | < and males in general receive far higher returns than do | females. This could indicate two things: (1) The | earnings of females are not influenced by the human ' 1 capital they bring to the labor market to the degree that ; I male wages are so influenced. This suggests that females, ] i who tend to be far more heavily concentrated in specific | (often female—dominated) occupations than males, labor in jobs with relatively flat wage rates. Hence, increases in years of work experience do little to increase earnings. j(2) It is also possible that the method for estimating [years of work experience is highly inaccurate for females L._ .... ......... ... I such that returns to work experience for females must be i I largely ignored. Census data are not designed to analyze ! years of work experience. Research using other data bases 'which collect information on actual years worked would help shed light on methods to improve estimates of female work experience. i Results for the overeducation and undereducation jvariables were sometimes surprising, running contrary to ] human capital assumptions. With regard to the model which included educational attainment, overeducation was not I I statistically significant for any of the twelve ethnic-sex I groups while undereducation was significant for five of ithe six female ethnic groups as well as for white males. : For half of the twelve groups overeducation was positively I associated with earnings. This was as expected since it jwas assumed that, all other things being equal, workers Iwith more education would earn more than their counterparts at the same job with less education. However, for the remaining six groups (Mexican and Cuban [males and white, black, Mexican and Cuban females) results ran contrary to expectations. For them it seems that overeducated workers earn less than their average educated counterparts working at the same job. The impact of undereducation on earnings were less surprising and more " i l l i consistent than for overeducation. Undereducation I negatively affected earnings for ten of the twelve j ethnic-sex groups, and all six of the groups for which it I was significant. These results suggest that for some I ethnic groups, particularly among females, having more than the minimal education for an occupation is not financially beneficial. This may also mean that work experience is more important than education in getting and keeping a job and in negotiating wages. However, for most groups, being undereducated resulted in an earnings disadvantage. Perhaps this is a result of "credentialism," the need for a level of education which is beyond that necessary to adequately perform a given I job, but is nonetheless required to meet arbitrary I requirements set by an employer. In short, employers may i be using education as a method of screening applicants for a job (Berg, 1970). Hence, it is possible that workers iwith less than the average education for a given i joccupation, if not denied entry altogether, may be denied I advancement opportunities, promoted at a slower pace, or may be paid less than the usual wage. In any case, it appears that, in general, optimal earnings returns are ! obtained by workers who have the average level of education for their occupations. Further, just as females 135 ! seemed to receive lower returns to education and i experience as compared to males, they also tended to be ! more disadvantaged with respect to overeducation and jundereducation. The earnings of undereducated minority I females suffered to a greater extent than did the returns o f males within the same ethnic groups. Likewise, overeducated white and black females experienced a negative impact on earnings, while white and black males benefitted from overeducation. Results were contrary to these and were more i predictable when educational attainment was omitted from the model, Undereducation, for example, had a ; consistently negative impact on earnings and was Î j statistically significant for all twelve ethnic-sex I igroups. Overeducation enhanced earnings for eleven of the [twelve groups and was significant for eight groups. Given that years of education completed was omitted from this model, it is likely that OVERED and UNDERED are acting as j proxies for education in this instance. However, the dramatic differences in results among the two models suggests that further research on over- and undereducation is necessary to determine the actual effect of these variables on earnings. 136 ‘ .U White, black, Mexican and Cuban females, and black and , I I Puerto Rican males earned more in government employment I than in the private sector. However, other groups did I ! I not. This suggests that women and some minorities face a ; i j j greater degree of wage discrimination in the private | ! sector. This is not surprising given that government, : until recently, tended to monitor the hiring and promotion ' of minorities and females more closely than private j industry. j I The state unemployment rate led to unexpected results ' I I for black females as well as white males and females. For | I these groups earnings rose as the state unemployment rate | 'increased. Among Cuban males (the other group for which | I this variable was statistically significant) earnings i I ' I declined as unemployment rose. This finding may indicate i I I I that white and black females and white males are employed i jin recession-proof industries and occupations, or ihigh-paying occupations (possibly heavily unionized) in jthose states having relatively high unemployment rates, iState unemployment rate was among the eight most I I influential determinants of earnings for three groups, I Cuban males, and white and black females. i Examination of the distributions of the race/ethnic •groups across geographic regions revealed that black. 137J I Mexican, Puerto Rican and Cuban males and females are [highly concentrated in specific regions, whereas whites are more evenly distributed. For blacks and Mexicans, I ^these distributions had a particularly negative effect on earnings. Earnings were lowest for blacks (and for all I other groups) in the South, a region in which over half of all blacks reside. For Mexicans earnings were lowest in the South followed by the West. Yet almost 90 percent of the Mexicans in this sample live in these two regions. ' For eleven of the twelve ethnic-sex groups, occupational status was the single most influential : determinant of earnings. With every increase in ioccupational status, earnings rose one percent or less. I The mean occupational status score was highest for white males, often 15 to 20 points higher than for other I groups. This means the earnings of white males are often ! 15 to 20 percent higher than for minorities due to 'differences in occupation. j As expected, married males earned more than unmarried [males. This was not the case for females, however, for Iwhom the results were inconsistent and, in all but one case, not statistically significant. Among males, marital ! status was significant for all groups and also proved to be a major determinant of earnings for all. Further 138 I analysis revealed that the sample contained a greater I proportion of unmarried females than males, and that ! married males tend to work more hours, are older and have I I more years of work experience than unmarried males. , The number of weeks unemployed was statistically I I I significant for all groups and was a major determinant of learnings for most. As expected, earnings declined with ! each additional week of unemployment. Hours worked, on the other hand, served to increase earnings, and was among the most influential determinants of earnings for eleven jof the twelve ethnic-sex groups. The metric regression ■coefficient for hours worked rounded to zero for all I ^groups. Hence, hours worked in the year did not : substantively contribute to the model. This probably occurred for two reasons. First, in selecting the sample only persons who worked at least 1,365 hours in 1979 were included. This served to reduce the amount of variance I jthat would have existed if a lower threshold had been I j selected. Second, though not highly intercorrelated, jweeks unemployed and hours worked express much the same [concept— weeks worked during the year. j ; Those variables added specifically to aid in analyzing |the earnings of Hispanics (English proficiency, ^citizenship and period of immigration) yielded interesting 13 results. They tended only to be significant for IHispanics. English proficiency was a major determinant of • earnings for five of the eight Hispanic ethnic-sex [groups. As expected, difficulty in speaking English I served to reduce earnings. Cuban, Mexican and other Latin males and females were far less likely to be U.S. citizens I than were members of the other ethnic groups. Citizenship was statistically significant for half of the Hispanic groups and, for these groups, it was found that citizens ■earned more than non-citizens. Period of immigration ; yielded inconsistent results. Coming to the U.S. between I 1970-80 proved to be significant for white and other Latin 'males only. For the latter group, arrival during this I period was negatively associated with earnings. This was ! expected since it was assumed that recent arrivals would i [have less knowledge about the U.S. labor market and so ; i Iwould be less effective in finding suitable work and in 'negotiating wages. For white males, however, those who arrived during this period earned more than native born whites. Surprising though this is, it likely reflects the jhigh socioeconomic status and occupational backgrounds of [white male immigrants during this period. It may also indicate that white immigrants face fewer obstacles in assimilating than do minority immigrants. 140 I Another surprising result was that Mexican females ' arriving before 1970 had higher earnings than American i 1 ' born females of Mexican descent. Again, this probably ; reflects the positive selection of immigrants during this ' period. i j Analysis of the earnings model also revealed that more ' 2 I variance was explained for females than males, R for i 2 I females ranged from 16 to 25 percent. For males R I ranged from 13 to 24 percent. Among both males and I ! females the model was least effective in explaining the earnings of blacks, Mexicans and Puerto Ricans, 1 Use of the regression decomposition (or standardi- ! ! zation) method to estimate discrimination has been j criticized by some researchers. However, it remains ! the only quantitative method available for analyzing the I economic impact of labor market discrimination and results I from this method are frequently used as evidence in legal ; cases. Using a modified regression decomposition ' 3 technique highlighted several interesting findings. First, white males earn far more than the other ethnic-sex groups. The narrowest gap was between white and Cuban males, yet even here whites earned $3,884 more than Cubans, The white male-Mexican female gap was widest with white males earning $10,990 more than Mexican women. 141J ( i The highest average earnings among females was for whites ; I($11,162). This was lower than for the lowest earning 'males, Puerto Ricans ($13,275). Second, every group \ experienced discrimination relative to white males and I 'this discrimination served to reduce earnings by averages j ranging from $2,084 to $7,091. Third, females experienced | far more earnings discrimination than minority males. The ' costs of discrimination for females ranged from $6,560 to i $7,091. For males the cost was less (from $2,025 to $3,917) and the difference from one ethnic/race group to j : the next was wider than for females. Finally, it appears ' I that being both minority and female does not seem to have j ! the "double jeopardy" effect expected by some ; researchers. While ethnicity makes a tremendous idifference among the males with respect to the costs of ; discrimination, it does not have this affect for females, jIndeed, the range of costs of discrimination were far jwider among males than among females. Females all seem to {face much the same cost of discrimination with ethnicity I I playing a minor role if any at all. Further research into the costs of discrimination, particularly for females, is required to more conclusively explain these interesting results. 142 i Notes for Chapter 4 ' 1. For more on the occupational status scores see Ford I and Gehret; Nam and Powers, 1965; and Powers and HoImberg, : 1978. 1 2. This modification to the standard regression ! decomposition technique was developed by Professor David M . Heer. 3. This involved taking an average of the cost of ; minority/female status as compared to white males after [running a series of decompositions on two earnings I functions using, in turn, each of the ethnic-sex groups as I the standard population. ; CHAPTER 5 i I SUMMARY AND CONCLUSIONS I ! This research has estimated an earnings function among I black, white and Hispanic ethnic males and females. In I addition, using a modified regression decomposition I technique, an estimate of the impact of discrimination on I learnings was made. The results of these analyses will serve to answer the research questions addressed earlier: (1) What was the magnitude of earnings differences between I white, Hispanic and black males and females in 1979?; (2) How do overeducation and undereducation affect the earnings of blacks, whites and Hispanics?; (3) What is the effect of discrimination on the earnings of white females, minority females and minority males?; and (4) What is the impact on earnings of being both minority and female, as lopposed to being either minority op female? ! In this section the four questions listed above will ibe discussed and answered based on the results of this Iresearch. Other findings from the research will also be [presented and conclusions will be drawn. Prior to this a brief summary of the data used, the sample selection icriteria and the earnings model will be presented. ;A, Summary This research is based on an analysis of data from the 1980 Census "A" Sample of the Public—Use Microdata. 144 r ; Through the selection of a stratified random sample the 11 I million plus cases in "A" Sample were reduced to 219,324. I jOf this number, only those persons who were black, white 'or Hispanic civilian, noninstitutionalized wage and salary workers between the ages of 25 and 64, who worked at least I 1,365 hours in 1979 and had non-zero earnings for the year, were selected for analysis. Hence, only those blacks, whites and Hispanics who worked year-round I full-time and were in their prime earnings years were jselected. After these selections were made, the sample I contained over 31,000 males and over 19,000 females. Each i I of the ethnic—sex groups contained a minimum of 1,500 I I cases. ' The earnings model regresses the natural log of annual j earnings for 1979 on a number of independent variables. ; The independent variables can be divided into three I types : human capital, structural and control (or 1 demographic) variables. Human capital variables selected ! include years of education, years of work experience, I overeducation and undereducation. Structural variables I jinclude the state level unemployment rate, sector of ; employment (private or public sector), and three dummy j coded variables representing region of residence. The 'control variables include marital status, hours worked in 145 I 1979, number of weeks unemployed, occupational status, iEnglish language proficiency, period of immigration (two i ; dummy coded variables) and U.S. citizenship. (For more ; specific information on the variables see Table 2.3.) :B. Major Findings i Findings from the regression analysis revealed the ; following : # Education— As expected, education was positively associated with earnings for all groups and proved to be a major determinant of earnings for eleven of the twelve ethnic-sex groups. Black, Cuban and other Latin males and all female ethnic/racial groups are not compensated for education to the same extent as white males. Of all groups, Mexican, Cuban and other Latin females receive the lowest returns to education. And though the educational attainments of females equaled or exceeded their male counterparts within each ethnic—sex group, females tended to receive far lower returns to education than males of the same ethnic group. # Experience— As expected, experience was positively associated with earnings. It appears that males receive higher returns for each year of work experience than females. However, the method of estimating work experience is less than ideal. Consequently, these findings may indicate that estimates of work experience, particularly for females who tend to be less consistently employed than males, are subject to considerable error. An alternative explanation is that females, who tend to be far more highly concentrated in specific occupations (i.e., more narrowly distributed) than males, tend to work in jobs with relatively flat wage rates such that increases in work experience do not result in higher earnings. # Overeducation— Based on the model which Included educational attainment, results for six of the twelve groups ran contrary to expectations. It 146] appears that, for half of the ethnic-sex groups the overeducated worker earns less than the non-overeducated worker, after controlling for all other variables. However, for the other half, the overeducated worker realized an earnings advantage over his or her average educated counterpart. Finally, for each ethnic group males are more likely to be overeducated than females, and unlike previous research which found white males most likely to be overeducated, here Cuban males and females, followed by other Latin males and white males were most likely to be overeducated. (Figures for these groups were 12.6, 12.2, 11.8 and 10.8 percent, respectively.) These results might reflect the high socioeconomic status of many Cuban and other Latin immigrants upon arrival in the U.S., and their difficulty in converting education attained abroad into employment in the U.S. Undereducation— Based on the model which included educational attainment, results for undereducation were less surprising than for overeducation. As expected, for ten of the twelve ethnic-sex groups undereducation seemed to decrease one's earnings relative to other workers. White males and females were less likely than minorities to be undereducated, and females in each ethnic-sex group were less likely than their male counterparts to be undereducated. Further, those females who were undereducated tended to suffer more economically than did their undereducated male counterparts. Over- and Undereducation, Model Two— When educational attainment was omitted from the earnings model results for over- and undereducation were dramatically different although the amounts of variance explained by the two models were much the same. In the second model, overeducation tended to enhance earnings and was more likely to be significant. Also, undereducation had a consistently negative effect on earnings and was statistically significant for every ethnic-sex group. Sector of Employment— As expected, white females and most minorities had higher earnings in government employment than in the private sector. White males, on the other hand, earned more in the 147J private sector. Further, it is interesting to note the high percentage of blacks in government employment; 26 percent of black males and 32 percent of black females, as compared to 18 percent of white males and 23 percent of white females, the next highest group. State Unemployment Rate— This was a major determinant of earnings for three of the twelve ethnic-sex groups and was statistically significant for four. Among Cuban males unemployment operated as expected; as a state's unemployment rate increased, earnings were lowered. However, for black females and white males and females the reverse was true; increases in the unemployment rate were associated with increased earnings. It appears that, for some states, high wage rates persist in the face of relatively high unemployment rates. Perhaps additional research can further explain these findings. Region of Residence— At least one of the three variables identifying region of residence appeared among the most important determinants of earnings for every group. For all groups earnings were lowest in the South. As expected, minorities tend to be more heavily concentrated in specific regions of the country than are whites. Unfortunately, blacks and Mexicans are concentrated in those regions in which they have the lowest earnings. Occupational Status— Occupational status scores (which range from zero to 99) were, on average, far higher for white males than for other groups. While an increase of one unit in this variable increases earnings for each group by about 1 percent, the 15 to 20 point advantage in average occupational status which white males have over most minorities, means they earn 15 to 20 percent more than others due to employment in higher status occupations. Marital Status— As expected, married males earned more than unmarried males. This variable was a significant and important determinant of earnings for males. Further analysis of these results showed married males tended to be quite a bit older, had more work experience, and worked 148 # slightly more hours annually than unmarried males. : Among females, marital status appeared to have an inconsistent and statistically insignificant effect. Married and unmarried females tended to be , the same age and have the same amount of work experience, but married females worked slightly fewer hours annually than their unmarried counterparts. I i Weeks Unemployed— As expected, earnings declined j for each additional week in which the respondent was unemployed. This variable was a major determinant of earnings for all the females and most of the males, and was statistically > significant for all groups. I Hours Worked Annually— Surprisingly, though this ! variable has been found to be a major determinant | of earnings in previous research, as it was here, | the metric regression coefficients were so small ; they rounded to zero. The standardized regression ; coefficients revealed that, nevertheless, hours was among the top eight determinants of earnings for eleven of the ethnic-sex groups. Yet, because the j metric coefficient rounded to zero this variable was not substantively interesting^-. j English Language Proficiency— This was a major determinant of earnings for five of the eight Hispanic ethnic-sex groups. As expected, difficulty speaking English reduced earnings. U.S. Citizenship— Cubans, Mexicans and other Latins were far less likely to be U.S. citizens than were members of the other ethnic/racial groups. When statistically significant, U.S. citizenship was associated with increased earnings relative to non-citizens. This variable proved to be significant for half of the Hispanic groups. Period of immigration— Period of immigration was coded into three categories : immigrating between 1970-80; immigrating before 1970 ; and being U.S. born (the omitted category). Earnings tended to be lower for those Hispanic groups coming to the U.S. during the 1970-80 period. It was expected that the more recent arrivals would have lower earnings than the native born. However, among 149 white males (a group for whom this variable was significant) earnings of the recent immigrants were higher than for native born white males. It is likely that this finding reflects the advantaged occupational backgrounds of recent white immigrants. Similarly, Mexican female immigrants arriving in the U.S. before 1970 had higher earnings than American born females of Mexican descent. This may reflect the positive selection of Mexican immigrants during this period. The earnings model was slightly better able to explain the earnings of females than males, and among both males and females, was best able to explain the earnings of whites, Cubans and other Latins. White males earn considerably more than females and minority males. The white male-minority gap was narrowest for Cuban males (with white males earning $3,884 more than Cuban males) and widest for Mexican females (with white males, on average, earning $10,990 more than Mexican females). Puerto Rican and black males appear to face considerably more labor market discrimination than Mexican, Cuban or other Latin males. For Puerto Rican males the cost of being a minority was $3,917, followed by $3,379 for black males. For Mexican males the cost of discrimination comes to $2,578, followed by $2,240 for other Latin males and $2,025 for Cuban males. Relative to minority males, females experience considerably more labor market discrimination. Also, their average earnings are far lower than those for minority males, not to mention white males. Indeed, the female group having the highest average earnings (white females) earn some $2,000 less than the lowest earning male group (Puerto Rican males). White females appear to experience the greatest cost due to labor market discrimination ($7,091), closely followed by Cuban females ($7,072). Hence, each of these groups earned over $7,000 less than white males having the same amount of education, occupational status, etc. It is surprising to note that white females appear to bear slightly greater costs due to discrimination than other groups. However, for all 150J the female groups, the range of average costs of discrimination was quite narrow. Costs for the remaining groups are as follows: Puerto Rican females, $6,968; other Latin females, $6,678; Mexican females, $6,664; and black females, $6,560. # Results of the decomposition analysis show that females experience considerably higher costs of discrimination than that for minority males although minority males experience substantial costs due to labor market discrimination. Still, the effect on earnings of being female far outweighs the effect of being a racial or ethnic minority. :. Discussion Results of the regression decomposition showed : tremendous costs due to labor market discrimination, I ,particularly for females. These estimates may be jartificially high for several reasons. First, it is I likely that estimates of years of work experience are significantly inflated for females who tend to leave the labor force more frequently than males, often for lengthy I jperiods of time. If this is the case, and years of work experience is overestimated, then womens' returns to work ; experience are not as low as they may at first appear. I j jSecond, the need to combine many of the detailed I 1 occupations in order to have at least 50 cases in every ! ,occupation served to distort the impact of overeducation, Iundereducation and occupational status scores to some I junknown degree. Also, the occupational status scores used !(see Ford and Gehret) were derived from an analysis of a 1 151 r j percent: Public Use Sample of the 1980 Census. Use of the I larger 5 percent sample would serve to increase the number ! of cases in each detailed occupation, and so would lead to I I I more stable estimates. Third, the high estimated costs of I discrimination for Cubans and other Latins, may in part be I due to their inability to convert education and experience ! received abroad into an appropriate job in the U.S. This seems to be a problem common to those who receive professional training outside the U.S. It has also been found that foreign work experience yields low returns in ^the U.S., and, for Mexicans, Puerto Ricans, Central and ! 'South Americans and black immigrants, foreign work 'experience adds "virtually nothing" to wages (Reimers, 1 j 1982 : 16) . |D. Conclusions I Let us now turn our attention to the four main 'research questions addressed earlier in the dissertation. I First, what was the magnitude of earnings differences jbetween white, Hispanic and black males and females in i I 1979? As shown in Table 4.6, the average earnings for I civilian wage and salary workers aged 25-64 who worked at i least 1,365 hours during the year ranged from a high of $19,966 for white males to a low of $8,976 for Mexican females. Not surprisingly, males earned more than _ 152j ; females. Even the highest earning female group (white , females) earned some $2,100 less than the lowest earning I males (Puerto Rican males). After white males, Cuban and I other Latin males have the highest earnings. Among ’ females, the Hispanics earned less than whites and ! blacks, The male-female earnings difference ranged from $3,600 for Puerto Ricans to slightly over $8,800 for whites. This means white women, on average, earn about $8,800 less than white males, while Puerto Rican women earn some $3,600 less than Puerto Rican men. The white ! male—minority/female mean earnings difference is quite I large considering we are comparing year-round full-time I workers. For example, black males earn $6,004 less than ! white males and Puerto Rican males earn $6,691 less than ; white males. Black females earn $9,832 less than white j males, while Mexican females earn nearly $11,000 less than I white males. The wide variation in average earnings among i I the Hispanic ethnic groups, particularly the males, points to the importance of analyzing Hispanic ethnic groups separately. The second question addressed earlier was: How do overeducation and undereducation affect the earnings of blacks, whites and Hispanics? Findings from two separate earnings functions indicate that results for over- and | I . - - " " j I undereducation vary considerably depending on how the : earnings model is specified. For example, in the regression model which included the variable education, it was found that being overeducated tended to lower one's earnings for half the ethnic—sex groups, though it was ; positively associated with earnings for the remaining six jgroups. Overeducated persons in those groups for which earnings were lowered, lost a minimum of 1 percent (for I black females) and a maximum of 11 percent (for Mexican i .females). Overeducation was more likely to reduce the earnings of females than males. Where overeducation was positively associated with earnings, economic benefits ranged from 3 percent (for white males) to 14 percent (for Puerto Rican females). However, when the variable education was removed from the earnings model, .overeducation was, with one exception, found to be 1 positively associated with earnings, and was statistically ! significant for eight of twelve groups. Hence, when education did not appear in the model, overeducation I performed more as predicted by the human capital model, j Similarly, the results for undereducation varied ^ according to the model used. When education was included : in the earnings function, undereducation appeared to lower earnings for ten of the twelve ethnic-sex groups, and more I negatively affected returns for females than for males. I Being undereducated lowered earnings anywhere from 1 ! percent (for Puerto Rican males) to 13 percent (for black ; females). Results for undereducation were more consistent when the variable education was omitted from the model. Here undereducation was shown to reduce the earnings of all twelve ethnic-sex groups, and was statistically significant for all groups. The third research question was: What is the effect 'of discrimination on the earnings of white females, minority females and minority males? As shown in Table I 4.7. females experienced considerably more labor market i ^ discrimination than males. Even so, minority males had I their earnings reduced from $2,025 (for Cuban males) to a maximum of $3,917 (for Puerto Rican males) due to racism 'and other factors not included in the model. Females had their earnings reduced from $6,560 (for black females) to $7,091 (for white females) due to discrimination. Labor jmarket discrimination was costliest for white females, !followed by Cuban females, Puerto Rican females, other jLatin females, Mexican females and black females. Among jmales, Puerto Ricans experienced the greatest amount of I labor market discrimination, followed by blacks, Mexicans, other Latins and Cubans. In short, the impact of 155 I discrimination on the earnings of women and minorities is ’considerable. Labor market discrimination accounts for a j large proportion of the white male—minority/female j earnings gap. This discrimination (i.e., the residual I difference not accounted for by the model) has a wide jrange, explaining from 42 percent of the total earnings I gap for Mexican males to 81 percent of the gap for white ! females. Clearly then, if minorities and white females i I received the same returns to education, experience, I occupation, etc. as those received by white males, a very ; large portion of the male-female and white male-minority I male earnings gap would be closed. I The fourth question was: What is the impact on I earnings of being both minority and female, as opposed to 1 ibeing either minority or female? Again, Table 4.7 i presents data which can help to answer this question. It I is clear that being female exerts a far more negative j impact on earnings than being a member of a racial or I ethnic minority group. Indeed, the single most ’discriminated group in this analysis was white females I and, clearly, sexual discrimination alone (not racial [prejudice) accounts for the high cost of discrimination I experienced by white females. However, minority females, I : without exception, experienced greater costs due to 156 discrimination than did minority males. Hence, it appears ; that race and sex are interacting to some extent in I reducing their earnings. Future research should attempt to parcel out the effects of sex discrimination from those jof racial discrimination in examining labor market costs ; of minority and female status. ! Minorities and white females also received lower returns to experience than white males. Females appeared j I to receive lower returns to their human capital, j ! particularly education, than did males. Though some researchers (Mincer and Polachek, 1974) have suggested ; that females do not pursue education and training to the j I same degree as males because they anticipate intermittent [ I attachment to the labor force (due to childbearing and E other reasons), the analysis of this sample, a relatively j ! j old sample due to the selection criteria, shows that | i I females have average educations similar to males, and « I ; j often exceed the educational attainments of males. Given | I the relatively high educational attainments of this sample ; I of women, their low returns to human capital are I particularly striking and may reflect the low occupational levels in which women find themselves. However, occupational status was controlled for, so perhaps low ; returns to human capital result from the concentration of j I 157 I females in fewer occupations (as compared to males), I ' combined with the reduction in earnings associated with I ^ increasing percentages of females in an occupation. That I is, previous research has found that earnings are higher : in male-dominated occupations as opposed to I female-dominated occupations. This was found to be the ! case by Mellor (1984) in his analysis of Current ! I Population Survey data. His regression model involved 112 occupations each having 50,000 or more females and revealed that ..for each increase of 10 percent in the I proportion of women in the occupation, median usual weekly earnings in 1982 would fall by $13. The equation ! accounted for about 19 percent of the variance in women's i earnings among these occupations" (Mellor, 1984:27). I I The impact of overeducation on earnings appears to be influenced by race/ethnicity and sex, though the impact of !undereducation on earnings appear more consistent across jgroups. For six of the twelve ethnic-sex groups (four I female and two male groups) overeducation served to reduce ! earnings, a finding which ran counter to the expectations of the human capital model. For the remaining six Iethnic-sex groups, however, overeducation was associated jwith increased earnings. These mixed results are not 1 easily interpreted and certainly require further study. 158 ' particularly since an alternative earnings function, one j which omitted educational attainment, yielded very I different results. Results from the regression model I I which included the variable years of education completed, suggests that Freeman (1976) may be correct in asserting ! that the increase in the percentage of workers who have ! educations beyond that required for their jobs has resulted in reduced wage rates, at least for some ethnic-sex groups. Further, most of those groups which did benefit from overeducation did not benefit much. An i I I ; ' j alternative explanation for these results may be that, in j I this relatively aged sample, workers are far enough along | , in their careers that the impact of experience in I I determining earnings outweighs the impact of relative j I ! education (i.e. , overeducation). < j j Using data from the second earnings function (this one ; excluded the variable educational attainment) resulted in I overeducation having a more consistently positive impact I I on earnings and was more likely to be statistically significant. For this model, earnings were enhanced for I the overeducated, with increases ranging from 1 percent to I 8 percent. These results tend to cast doubt on Freeman's 1(1976) hypothesis. However, it is interesting to note ! that even for this alternative model, female returns to I i i 1 I overeducation tended to be less than male returns, i This same model showed undereducation to be I statistically significant for all twelve ethnic-sex i groups, and also bore a negative relationship to earnings ; for all twelve groups. The results for undereducation, were more consistent in the first model (which included education) as compared to the results for overeducation from that model. For ten of the twelve ethnic-sex groups, undereducation served to reduce earnings. This finding, in conjunction with the negative returns to overeducation or otherwise slim increase in earnings due to overeducation found in the first model suggest that the I ! most optimal course in terms of earnings advantage is to I I ! obtain the mean education for that occupation one wishes to enter. However, further study of this issue is ; necessary before one can go beyond speculation on this i I ! point. IE. Suggestions for Further Research I j Interpretation of the findings presented here were I often speculative due to certain gaps in the literature on ’earnings and employment. In such areas additional ! research would be helpful. For example, this and other research (Verdugo and Verdugo, 1984) has noted that ! minorities and females tend to earn more in government 160J ; employment while white males earn more in the private , sector. This suggests that women and minorities face i I barriers to promotion as well as other forms of I discrimination in the private sector. While there is some ; evidence that minorities and women are less likely to be in positions of authority and, if in such positions, will not be compensated at the same levels as white males I(Kluegel, 1978; Wolf and Fligstein, 1979), research which specifically compares public and private sector returns would help to explain the findings presented here. I Additional research on the impact of state and local unemployment rates on earnings would also be useful. Recall that in this research it was found that the I earnings of black females and white males and females I I increased as unemployment rose. This was rather curious and unexpected leading me to speculate that these groups I may be employed in recession—proof occupations and ; industries or are located in states with heavily unionized work forces. But again, these interpretations were purely speculative and further research in this area would be helpful. It was noted here that of the four regions, earnings were lowest in the South for all groups. However, this ! variable had a less negative impact on the earnings of 1.61J , whites than on the earnings of other groups. That is, j though all groups earned less in the South than in the I I other regions, whites did better there than others. I Further research into why this is the case could be i illuminating. t The use of occupational status scores in the analysis ! ! of earnings could be improved upon. Due to changes in the 1 j 1980 Census occupational codes it was not possible to use I the Duncan Socioeconomic Index. It would, therefore, be 'helpful to research of this type to replicate the Duncan ■ Socioeconomic Index using the 1980 Census occupation I codes. Alternatively, it would be helpful to re-estimate the Ford and Gehret occupational status scores using a I ! larger sample of 1980 Census data. (Ford and Gehret used a 1 percent sample and so there may be some instability of the status scores.) Finally, given the complexity, not to j mention costs, of processing large Census data samples, in this research effort it was necessary to process a I subsample of Census data. Because many of the 3-digit I occupations had a small number of cases, it was necessary 'to combine occupations. Hence, I often had to compute : weighted averages of occupational status scores. This may have distorted results to some degree, both with respect 2 to occupational status and over- and undereducation. 162 ; Consequently, replicating the earnings model used here I I with the full 5 percent sample of the 1980 Census could ; result in more stable estimates. I I Further, though much research has been done on the I , impact of earnings within specific occupations, more is I I still necessary. One such area to examine is the I distribution of males and females within specific occupations. Though some have argued that women train for and select those occupations which allow them to easily leave and re-enter the labor market (Mincer and Polachek, ! I j1974), it is doubtful that this would be the case for ; younger women to the extent that it may have been for an i : earlier generation. Hence, a cohort analysis, such as a ! comparison of the distributions of young men and women in i I specific occupations and their earnings would be helpful. ■ Since it has been found that there is little variation ! between the races and sexes with respect to entry level I jwages (Loury, 1981:123), analysis of longitudinal data jwould be best. Indeed, such analyses show that though ,entry level wage rates for black and white males are quite close, subsequent wage growth is "...significantly smaller I for the black workers" (Loury, 1981:123). This could also help to address the issue of the rate of promotions for ! minorities and women as compared to white men. Census 163 ! data may not be the best source for such a research I : effort, since it would be highly desirable to have actual ' figures on years of work experience. Finally, with the comparable worth debate raging, I [ expect more research will be done on earnings differences I between male— and female-dominated occupations. It has I been documented that increases in the proportion of females in an occupation result in lower earnings (Mellor, 1984), but why is this the case? Careful examination of ! skill differences, years of work experience and supply and 'demand factors, among other variables, must be i ■considered. Also, the finding in this dissertation that 'double jeopardy (the intersection of race and sex in I I reducing the earnings of minority females) does not appear to have a large impact on the earnings of minority females I was surprising and needs further study. I Though much research has been conducted on the learnings and labor market experiences of minorities and I women, much of these findings may no longer apply. This j is particularly true now since dramatic social and 1 economic changes during the last 15 years have led to increases in the labor force participation of women and their recent entry into predominantly male occupations. ■Also, these social changes have supposedly increased the 164 i opportunities open to minorities. Because the labor force I is so dynamic, it is often not possible to build upon the ' findings of research done even 10 years earlier. Rather, I j we must add to this research in order to see from where we've come and how far still we have to go. 165: Notes for Chapter 5 I I 1. This curious result— being among the most I influential determinants of earnings yet substantively j uninteresting— is due to the highly unstable nature of I standardized regression coefficients. Because of the I large standard deviation of hours for white males (426), I and the relatively small standard deviation of log I earnings (.6), hours had a high standardized regression ' coefficient. I 2. For over- and undereducation it was necessary to ! compute mean education and the standard deviation for each I occupation. Due to the collapsing of many occupations, there is likely some distortion in these figures. 166 ! BIBLIOGRAPHY ; Almquist, E. M. Untangling the effects of race and sex: ' The disadvantaged status of black women. Social i Science Quarterly, 1975, 56, 129—142 I I Althauser, R. P., & Wigler, M . Standardization and I component analysis. Sociological Methods & Research, ! 1972, 1, 97-135. Becker, G. Human capital. New York : Columbia University Press, 1964. Berg, I. Introduction. In I. Berg (Ed.), Sociological perspectives on labor markets. New York: Academic Press, 1981. Berg, I. Education and jobs: The great training robbery. New York : Praeger, 19 70. :Blakemore, A. E. & Low, S, A. Sex differences in i occupational selection: The case of college majors. I The Review of Economics and Statistics, 1984, LXVI, : 157-163. I IBlau, P. M. & Duncan, O. D. The American occupational j structure. New York : Wiley, 1967. !Bluestone, B., Murphy, W. M., & Stevenson, M. Low wages I and the working poor. Ann Arbor, MI: Institute of I Labor and Industrial Relations, 1973. ; Briggs, V. M . , Fogel, W. & Schmidt, F. H. The Chicano ' worker. Austin, TX: University of Texas Press, 1979. j Cain, G. G. The economics of discrimination : Part 1. ! Focus, 1984, 7, 1-11. i ; Car1iner, G. Returns to education for blacks, Anglos, and I five Spanish groups. Journal of Human Resources, 1976, 11, 172-184. Chiswick, B. R. An analysis of earnings among Mexican origin men. Stanford University: Hoover Intitution, 1977 . Clogg, C. C. Measuring underemployment : Demographic indicators for the United States. New York: Academic j _________ 167 Press, 1979. IClogg, C. C, & Shockey, J. W. Mismatch between occupation I and schooling: A prevalence measure, recent trends and I demographic analysis. Demography, 1984, 235-257. I Clogg, C. C. & Sullivan, T. A. Labor force composition and ; underemployment trends, 1969-1980, Social Indicators ! Research, 1983, 12, 117-152. Corcoran, M ., Duncan, G. J., & Ponza, M. Work experience, job segregation, and wages. Paper presented at the National Academy of Science Conference on Job Segregation by Sex, 1982. Cornelius, W. A. Mexican migration to the United States: Causes, consequences, and U.S. responses. Migration and Development Study Group, #C/78-9, MIT, July 1978. Davis, K. & Moore, W. Some principles of stratification. American Sociological Review, 1945, , 242-249. DeFreitas, G. Ethnic differentials in unemployment among Hispanic Americans. Paper presented at the Hispanic Labor Conference, Santa Barbara, CA, February 4-5, 1982 , Draper, N. R. & Smith, H. Applied regression analysis (2nd ed.). New York: Wiley, 1981. Dune an, O.D. Inheritance of poverty or inheritance of race? In D.P. Moynihan (Ed.), On understanding ' poverty. New York : Basic Books, 1968. ; Duncan, O. D . , Featherman, D. L. & Duncan, B. 1 Socioeconomic background and achievement. New York : Seminar Press, 1972 I Epstein, C. F . Positive effects of the multiple negative: I Explaining the success of black professional women, j American Journal of Sociology, 1973, 7^, 912-935. !Featherman, D, L. & Hauser, R. M. Opportunity and change. I New York : Academic Press, 1978. jFelder, H . E. The changing patterns of black family ; income, 1960—1982. Washington, D.C.: Joint Center for ; Political Studies, 1984. 168 •Felmlee, D. H. A dynamic analysis of women’s employment I exits. Demography, 1984, 2 1 ' 171-83. jFord, K. & Gehret, J. Occupational status scores from the ' 1980 Census Public Use Samples. Johns Hopkins ' University, no date. (Mimeographed.) i 1 Fossett, M. A. Inequality and the relative size of minority populations: An examination of Anglo-Hispanic income inequality in the metropolitan Southwest. ! Austin, TX: Texas Population Research Center, 1982. 1 Freeman, R. B. The overeducated American. New York: Academic Press, 1976. Freeman, R. B. Have black labor market gains post-1964 been permanent or transitory? Working Paper No. 751. Cambridge, MA: National Bureau of Economic Research, 1981. (a) I Freeman, R. B. Troubled workers in the labor market. I Working Paper No. 816. Cambridge, MA: National Bureau : of Economic Research, 1981. (b) I Freeman, R . B. & Wise, D. A. NBER summary report : Youth unemployment. Cambridge, MA: National Bureau of j Economic Research, 1979. I Fuchs, V. R. Differences in hourly earnings between men I and women. Monthly Labor Review, 1971, 9-15. I I Garcia, J. A. Underemployment— Dimensions of underutilized labor force : The case of Southwestern Chicanos. I Revision of paper presented at Southwestern Social i Science Association Meeting, Ft. Worth, TX, March j 28-31, 1979. I !Granovetter, M . Toward a sociological theory of income ! differences. In I. Berg (Ed.), Sociological I perspectives on labor markets. New York : Academic Press, 1981. Grenier, G. The effects of language characteristics on the wages of Hispanic-American males. Journal of Human Resources, 1984, XIX, 35-52. Gwartney, J. D . & Long, J. E. The relative earnings of blacks and other minorities. Industrial and Labor 169 Relations Review, 1978, 3^, 336-345. Hanushek, E. Ethnic variations: Magnitudes and I explanations. In T. Sowell (Ed.), American ethnic I groups. Washington, D.C.: The Urban Institute, 1978. I Hauser, P. M. The measurement of labour utilization, j Malayan Economic Review, 1974, 29, reprinted in I East-West Population Institute Reprint No. 85. .Hauser, P. M . The measurement of labour utilization— more I empirical results. Malayan Economic Review, 19 77, 22, reprinted in East-West Population Institute Reprint No. 85. Hayghe, H. Working mothers reach record number in 1984. Monthly Labor Review, 1984, 107, 31-34. 'Hiestand, D. L. Discrimination in employment: An appraisal ' of the research. Ann Arbor, MI: Institute of Labor I and Industrial Relations, 1970. 'Hogan, D, P. & Pazul, M, The occupational and earnings I returns to education among black men in the North, i American Journal of Sociology, 1982, 22' 905-920. ! Horan, P. M. Is status attainment research atheoretical? American Sociological Review, 1978, 22' 534—541. llams, H . M . & Thornton, A. Decomposition of differences: A j cautionary note. Sociological Methods & Research, I 1975, 3, 341-352. I Jeffries, V. & Ransford, H. E. Social stratification: A i multiple hierarchy approach. Boston: Allyn & Bacon, i 1980. !Kats, R. The immigrant woman : Double cost or relative improvement? International Migration Review, 19 82, I 16, 661-677. IKluegel, J. R. The causes and cost of racial exclusion I from job authority. American Sociological Review, I 1978, 285-301. i Knoke, D . S c Burke, P. J. Log-linear models. Beverly I Hills: Sage, 1980. 170 I Leonard, J. S. Antidiscrimination or reverse discrimination: The impact of changing demographics, j Title VII, and affirmative action on productivity, j Journal of Human Resources, 1984, XIX, 145-174. I Lieberson, S. A piece of the pie: Blacks and white 1 immigrants since 1880. Berkeley : University of 1 California Press, 1980. I : Light, S. C. Multicollinearity in a study of regional ! variations in homicide rates : A comment on Smith and I Parker. Social Forces, 1984, 62, 800-803. Long, J.E. Productivity, employment discrimination, and the relative eonomic status of Spanish origin males. Social Science Quarterly, 1977, 58, 357-373. Loury, G. C. Is equal opportunity enough? American Economic Review, 1981, 22' 122-126. i Masters, S. Black-white income differentials : Empirical I studies and policy implications. New York: Academic : Press, 1975. ! ,Meister, R. J. Race and ethnicity in modern America. Lexington, MA: Heath, 1974. 1 :Mellor, E . F . Investigating the differences in weekly 1 earnings of women and men. Monthly Labor Review, I 1984, 107, 17-28. IMichelson, S. On income differentials by race : An analysis ■ and a suggestion. In The conference papers of the Union for Radical Political Economists. Philadelphia, i 1968, 85-121. 'Miller, A. R., Treiman, D, J. & Cain, P. S. Work, jobs, I and occupations : A critical review of the dictionary : of occupational titles. Washington, D.C.: National Academy Press, 1980. I Mincer, J. The distribution of labor incomes : A survey. Journal of Economic Literature, 1970, 8, 1-26. ; Mincer, J. Schooling, experience and earnings. New York : I Columbia University Press, 1974. 171 I Mincer, J. & Polachek, S. Family investments in human ' I capital: Earnings of women. Journal of Political ' I Economy, 1974, 82^/ S76-S108. i i ■ I 1 Nam, C. B. & Powers, M . G. Variations in socioeconomic ■ structure by race, residence, and the life cycle. I American Sociological Review, 1965, 30^, 97—103. tNiemi, B. T. & Lloyd, C. B. Female labor supply in the : ! context of inflation. American Economic Review, 1981, i I 71, 70-75. j I I 'Ni1sen, S. R. Recessionary impacts on the employment of men and women. Monthly Labor Review, 19 84, 107, 21-25.; O'Neill, J. A. A time-series analysis of women’s labor force participation. American Economic Review, 1981, 71, 76-80. Ortiz, V. & Cooney, R. S. Hispanic female participation in ' j the labor force : A comparative analysis of Puerto ' J Ricans, Mexicans, and Cubans, no date. (Mimeographed.) !Palmore, E.B. & Manton, K. Ageism compared to racism and I sexism. Journal of Gerontology, 1973, 22' 363-369. I Parcel, T. L. & Mueller, C. W. Ascription and labor I markets : Differences in earnings. New York: Academic I Press, 1983. i : Pareto, V. Cours d'économie politique. Lausanne: Rouge, I 1897. Cited in G. S. Sahota, Theories of personal income distribution : A survey. Journal of Economic j Literature, 1978, XVI, p. 3. 'Parker, R . N . & Smith, M. D. High correlations or multicollinearity, and what to do about either : Reply to Light. Social Forces, 1984, 6 2 , 804—807. Piore, M. Notes for a theory of labor market stratification. In R. C. Edwards, M. Reich & D . M . , Gordon (Eds.), Labor market segmentation. Lexington, MA: Heath, 1975. 1 Poirier, D. J., Piecewise regression using cubic splines. I Journal of the American Statistical Association, 1973, i 68, 515—524. 172 ■Powers, M. G. & Holmberg, J. J. Occupational status i scores: Changes introduced by the inclusion of women, i Demography, 1978, 22' 183—204. jRagin, C. C., Mayer, S. E. & Drass, K. A. Assessing j discrimination: A Boolean approach. Amer ican I Sociological Review, 1984, 49, 221-234. ■Reimers, C. A comparative analysis of the wages of Hispanic, black and white American men. Paper presented at the Hispanic Labor Conference, Santa j Barbara, CA, February 4-5, 1982. 'Reimers, C. Why are Hispanic Americans' incomes so low? An analysis of the family incomes and earnings of the major Hispanic origin groups in the U.S. Report prepared for Employment and Training Administration, I U.S. Department of Labor, Grant #21-34-78-60. Robey, B. Black votes, black money. American Demographics, 1984, 6, 4 & 6. Rumberger, R . W . Overeducation in the U.S. labor market. New York: Praeger, 1981. iRytina, N. F . Tenure as a factor in the male-female I earnings gap. Monthly Labor Review, 1982, vol., 32-34. Rytina, N. F. & Blanchi, S. M. Occupational reclassification and changes in distribution by gender. Monthly Labor Review, 1984, 107, 11-17. (Sanborn, H. Pay differences between men and women. Industrial and Labor Relations Review, 1964, 17, ! 534-550. 'Schmidt, F. H. Spanish surnamed employment in the ‘ Southwest. Washington, D.C.: U.S. Government Printing I Office, 1976. I j Sewell, W. H. St Hauser, R. M. Education, occupation, and I earnings : Achievement in the early career. New York : I Academic Press, 1975. Shack-Marquez, J. Earnings differences between men and i women: An introductory note. Monthly Labor Review, ' 1984, 10/, 15-16. 173J ’ Sieling, M. S. Staffing patterns prominent in female-male earnings gap. Monthly Labor Review, 1984, 107, 29-33. ! j Sorensen, A. B. & Kalleberg, A. L. An outline of a theory j of the matching of persons to jobs. In I. Berg (Ed.), j Sociological perspectives on labor markets. New York: I Academic Press, 1981. , Sowell, T. Markets and minorities. New York : Basic Books, 1981. I ' Sowell, T. Civil rights: Rhetoric or reality? New York : William Morrow, 1984. Squires, G. D. Education and jobs: The imbalancing of the social machinery. New Brunswick, NJ: Transaction Books, 1979. Stolzenberg, R. M. Education, occupation, and wage differences between white and black men. Amer ican I Journal of Sociology, 1975, 8%, 299-323. IStolzenberg, R. M, Bringing the boss back in: Employer size, employee schooling, and socioeconomic i achievement. American Sociological Review, 19 78, 43, i 813-828. jSullivan, T. A. Marginal workers, marginal jobs: The j underutilization of American workers. Austin : I University of Texas Press, 1978. ISuter, L. E. & Mil1er, H . P. Income differences between I men and career women. American Journal of Sociology, I 1973, 28, 962-974. Szymanski, A. Racial discrimination and white gain. American Sociological Review, 1976, 4J., 403-414. ISzymanski, A. White workers' loss from racial discrimination: Reply to Villemez. American Sociological Review, 1978, 43, 776-782. I ITreiman, D. J. & Terrell, K. Sex and the process of status I attainment: A comparison of working women and men. I American Sociological Review, 1975, 42, 174-200. 1 Tyree, A. & Treas, J. The occupational and marital mobility of women. American Sociological Review, . 17 4j ; 1974, 39, 293-302. I 'United Nations, Department of Economic and Social Affairs. The determinants and conséquences of ; population trends : New summary of findings on i interaction of demographic, economic and social t factors (Vol. I). New York: Population Studies, No. I 50, 1973. U.S. Bureau of the Census, Money income of families and ; persons in the United States : 1979. Series P-60, No. I 129, 1981. ju.S. Bureau of the Census, Census of population and I housing, 1980: Public-use microdata samples technical documentât ion. Washington, D.C., 1983. (a) U.S. Bureau of the Census, Statistical abstract of the United States : 1984 (104th ed.) . Washington, D.C. : U.S. Government Printing Office, 1983. (b) U.S. Bureau of the Census, Wives who earn more than their husbands. Washington, D.C.: U.S. Government Printing , Of f ice, 1983. (c) iu.S. Commission on Civil Rights, Twenty years after Brown ! The shadows of the past. Washington, D.C., 1974. U.S. Commission on Civil Rights, Social indicators of equality for minorities and women. Washington, D.C., I 1978. :U.S. Department of Labor, Manpower Administration, ; Handbook for analyzing jobs. Washington, D.C.: j Government Printing Office, 1972. !Verdugo, N . T. & Verdugo, R. R. Earnings differentials I among Mexican American, black, and white male I workers. Social Science Quarterly, 1984, 22' 417-425 I iVillemez, W. J. Black subordination and white economic well-being. American Sociological Review, 1978, 43, 772-776. ! Williams, W. E. The State against blacks. New York : New j Press, 1982. 175 I ; Wohlstetter, A, & Coleman, S. Race differences in income. In A. H. Pascal (Ed.), Racial discrimination in ■ economic life. Lexington, MA: Heath, 1972. Wolf, W. C. & Fligstein, N. D. Sex and authority in the workplace: The causes of sexual inequality, American Sociological Review, 1979, 4^/ 235—252. Work interruptions and the female-male earnings gap. Monthly Labor Review, 1985, 108, 50—51. ' Zucker, L. G. & Rosenstein, C. Taxonomies of institutional structure: Dual economy reconsidered. Amer ican Sociological Review, 1981, 46, 869-884. 176 1 APPENDIX A. Presentation of Weighted and Unweighted Sample Data Tabulations Table A.I. Average Age by Race/Ethnicity, Sex and Marital Status (unweighted) Average Age Married Unmarried Total White male female 42 41 3 6 41 41 41 Black male female 41 39 37 39 40 39 Mexican male female 38 37 34 37 38 37 Puerto Rican male female 39 39 36 38 38 39 Cuban male f ema1e Other Latin male female MALES (weighted) FEMALES (weighted) 45 44 40 38 42 41 40 43 36 40 36 41 44 43 39 39 178 Table A.2, Average Hours Worked in 1979 by Race/Ethnicity, Sex and Marital Status (unweighted) Average Hours Worked Married Unmarried White male female Black male female Mexican male female ! I ; Puerto Rican * male ; female ; Cuban male ' female Other Latin male female MALES (weighted) FEMALES (weighted) 2233 1981 2109 1971 2153 1972 2087 1949 2193 2008 2146 1978 2218 1979 2159 2023 2079 2011 2103 2031 2051 1981 2118 2027 2126 2027 2146 2021 Table A.3 Average Years of Work Experience by Race/Ethnicity, Sex and Marital Status (unweighted) Average Years of Experience Marr ied Unmarr ied White male female 24 23 18 23 Black male female 25 22 21 21 Mexican male female 24 22 19 22 Puerto Rican male female 24 23 20 22 Cuban male female 29 27 23 26 Other Latin male female 23 22 18 23 MALES (weighted) FEMALES (weighted) 24 23 18 23 18 qJ APPENDIX B. 1980 Census Long-Form Questionnaire 181 Please fill out this official Census Form and mail it back on Census Day. Tuesday, Apnl 1. 1980 1 9 8 0 Census of the United States Your answers are confidential Bv i« t> » itm# Ï 3 Ü S Cooai. cansu* amptovMa arc suBiact to tine ano /cf imprrsonffianT I Of any 0<acloaur* Of yOur answers Only after 72 yaars 30*S vour information MCOme avaalMle 10 oma> government agerfcres or the ouOfiC fhe same law reouires that vou answer trte questions to m# oest of vour k n o w r e o g e Para personas de habla hispana •Por Soantsn-soeaiung oersons) _ SI US TED CES£A UN CUESnONARIO DEL CENSC EN ESPa n OL 'lame a la of<-na oaf censo Ei njmero da tet<fono se encuemra en ei etkcasriiaoo ae < a æaccidn 0 S I orehere marque esta caeSi O D O T correo an ei soera qua sa e mctuve M e m uat #om ome to tume taiw aioefc o* ouraehms aa a gaopta if our W n o n is to meet swccasWwiN the mem» namonaf and local ctialiangaa we face Thta ta the puroeaa of me 1 M O The eaeeneet rseed for a eooulasien cewaua w ahneai 2 0 0 veera age «men our C onam uM n «ma «mnen *a prmaded mr araeta i. m e hrat census was conducted m 1790 end one has Been lacam e«erv 10 years atnca then The lew under «m«eh the census is taken oretects the eonhdermelitv 0< vour answers Por the neat 72 years — or untM Aoril 1. 2 0 9 2 — only sworn census workers h#ra access to the indivfduai records, and no one dee may see mem Tour answers, when combined wtm the answers from other oeoeie. «aii ormnde the stsusticei figures needed by puBHc and prn sss grouos. schools, busmssa and industry. ar»d federal. SietSi and local govem me nts across the coumry These figures «all hale ad sectors a* American society understand now our oooufetion and housing are changing in th # way. we can deal more effectively wdh today s problems and work toward a bettes future for ad of us. The census is a wtsdy im portant rianéfiai activity Please do your part by fdfmg out diis census form securatefy and cempieteh d you mad it back prometiy m the enclosed postage*oaid envelope, it vwii smre the eimerwe and inconvenience of a census taker hmnng to wait you. Thank you for your cooperation ro.Ce Bwf**w rm# C#w# a # 1 8 2 1 How to fill out your Census Form $ # # Th# f ilM - o u t « M m p i* m :h# yalloiw instruction g u 'd * This g old# mkH hoip imth any probtam a vou m ay hava. If you naad m ora haip call tha Cansus O ffica Tha tatapbona nwmbar of ih a local oMica is shonvn at tha bottom of tna addrass boa on tha front covar Uaa a black pancii to ansnwar tha q u a ttio n s Black pancii IS ban ar to usa than ballp oin t or o thar pans Fill circias O com pfatafy iika this # W h an you w fita m an anaiaar p n m or «vnta d e a rly sura that anaiaars sra providad for auaryona Saa p ag a 4 of ttta g u id a if a room ar or sptnaona a t# * in itt# n ou M hO id O OM not w an i to giva you all tha inform ation for tha form tha quastions o n pagas 1 th ro u g h 5 and than starting «with pagas 6 and 7. fill a pair of pagas for aach parson m tfia housahold Cftack your artawwars Than w n ta your nam e, tha d ata and lalap h o n a num par on pag e 2 0 M ad back this form on Tuasday April 1. or as soon aftarw ward as you can U s e tha ancioaad anvaiopa. no stam p is naadad M aaaa start by answwanng Q uastion 1 baioww Question 1 List in Q uastion 1 T . W hat ia •Family mambars iK«ng hara indudmg babias stik m tha mam a c4 a a eh parson a rh o t AprN 1. I M O . o r t a h o i hospital •Aalatsas H w «ng hare •Lodgers or boarders kwmg here •Other oersons Imng hare • Collage students «who stay here «whia atteryjmg cokage e te y ia g o r irW tin g h a rd aw d h a d ma e l # w h a a ia ? •Parsons ««mo u tu aHy k«a hare but are temporarily aww sy (inciudmg chiidran in boarding school baioww tfta college S h a l l • Persons ««âh a home aisai«mars but w w no stay here most of the «week ««rsie wwonor«g Oo Not List in lawastion 1 • Any parson awway from hare in tha Armmi Forces •Any collage student «who stays somawwhare efsa wmiia atiandmg collage •Any parson i«#lx) usually stays sumawwhara afaa most of the ««laak ««ktde w w ortorw g there. • Any parson awway from hare m an maatubon such as a home for the aged or mental hosprtai •Any person staying or vmnrtg hara ««ho has a usual home If u Than hara a ataymg only tamporanly and has a alsawwhare. plaaaa mark this box Q . the quasbons on pages 2 through 5 only, the address of your usual home on page 2 0 183 am 184 MSOAMswmTMeHOusrnieauesmmsoNm^KGes nnm é i ■üiin t * c. Imr a) I • ü a M c. vaa a < 1 • • • • * » I I 2 1 4 S * 7 » « W U U IZJ4SS 1 a tioiiu mmmn wa» «cwa «m» mjmmj I 2 I 4 s « 7 ê » to 11 12 mmml c o r n u s *. ustOÊm.* CMmus A . lag e w o r .SSfr « 185 i ! .ZBKULaiHHi- Tmmân N O W PLEASe A M S W m Q UESTI ONS H 1 -H 1 2 _______ m YOUN H O U SB i O LD_______ I CTH' Sl an gi-TSôü F»' g; uauMaH4jM UAmatiT siTjnantJM MOUoeatiMm t m M 0 a t l4 » . « M I USOjOOO a ttw.w* i WUOBaWMt * te «■ Mt M M M «Mta. M w— a i - I t e — « 0 t l W a t l 4 4 « o a C M t i T o a t m S M a lM t l t e a t i a * S2D»S2> S M O a ttW n O O a U 2 * « • • M ■ *2 2 » a *2 4 4 U O O a tlO » « 4 0 * 2 2 7 4 « i i o a t i w < 2 7 4 **2 0 4 •U M # « U » n a o * t » a « u o a t i a « M * t S * t U 4 e * U 4 » * 4 0 0 * * 4 0 * ( l a e a t t w « 0 0 * I t e * 12: 4*4 74 9 W I I U 1 2 2 4 t • 7 # a , ai! coma A . %T*teawaa, a*— . -M»CZ amae. 2 4 # te tiaa2na«n 2 «##4 tete# 4»amU«te# :i—'«##2, 2a te» t w m t a 186 : 7»u : U*a m m ' m 3 m tm. 187 M a r 0*- ■ «X 188 L 2. MM a. M l VeaanMwe Ma MaoUMi ■omaMoMMi t» rs M tM O tW S M 1M > ino M t*7 « . IM O M 1«M' . ont» « P M * C hMM - SM»M >* in/. ! IS a OM MM « M M * M m M I U 79t f Mn< MM ttn « IMM - IMb Ma v«v M D M* - SMa m l* I mmm 1 i t n H 1 7 . M «MM W 7 t fUmpimmU MM i a Ob M M M M M Ma Mb TM Me L M M M m m m «Me M e M M MmI«7S« • MB f a ,M M f« e * -4 M W l * 7 j ; M UM — a /Ma* IHC-Jmm w, ItSÎI I Mm n f l — i l iM T ; I MM I fBM* W r aeiMMM ;»;« !a a a AM5W? T tm e o u e s n o N S K m 1 2 1 4 s « 7 ■ 9 10 U t2f 2 1 . /TM B i iM il» HM II IM M b » I B I 'M - a N a M m b m b m M M B a u m a e «a ?<MM>/ AiMM, • • f Tl^» I M B B M <b>7 M Mo t « M M M « m m b b m m I « atom (*» M M M e # M i 12M ■ la fc a m M. 189 un 'MM. V. WWi e H » dmnforottnr tangfbnm. 190 >30 Make Sura You Hava Rllad This Form Comolataty If vou h»v« lisi«d mor« ihan 7 parsons <n Quasiion 1. ptaase make Sure that you nawa filled iha form for (he first 7 people Then mail Pack this form A Census Taker w ill call to o btain the information for the other people • Answered Question 1 on page 1 > Answered Questions 2 through 10 for each person you listed at (Tie top of pages 2 and 3 • Ansuwred Questions H 1 through H32 on pages 3 4. and 5 > Filled a pair of pages for each person listed on pages 2 and 3 That IS. pages 6 and 7 should be filled for the Person m column 1. pages 8 and 9 for the Person m column 2 etc ha#»# noMc* mm m##e m eme i to auosiiomi > 7 mrougn 33 lo» wen osrsonbemMlDioAerii t> # 5 n e w mensfimeymw wcim ialoieoiyio me oonicutar ewson For eeameie. mu nwy n ew lorsonsn le Ml ak the ner e m ry circle* on enrk or on income lor • ie * r * s * i yoing lo cchooi. or • renreo oerton To e w a oiir henng to check— m you to mske *w reorine#n*eei pieeeeoe of th e person w h o filled the fo rm the date The form w as com pleted, and the te lep h o n e n um ber on w hich the p e o p le m this househ old can be called 3 T h a w t s É d e w I a p t » I I the way It was sent to you Mail •( back in the enclosed arweiopa The address of the U S Census Office appears on the front cover of this ouesdonnaire Please be sure thet before you seal theenyeiope (he address shows through the window No stamp is required Than* you yery much 191J APPENDIX C. Occupational Status Scores 192 I Many of the 1980 Census three-digit occupational scores had fewer I than 50 observations in each occupation. In order to compute more stable estimates of occupational status and overeducation, similar occupations were combined creating collapsed occupations having at | least 50 observations. After collapsing the occupations weighted I j ! ' I : I I averages of the occupational status scores were computed for each { , * I ! : I newly grouped occupational category. The three-digit occupations noted below were collapsed as indicated. Each listing within parentheses constitutes a single collapsed grouping. (3,5,6) (17,18,19) (24,34,37) (28,29) (35,36) (45,46) I (47,48) (54,58,59) (63,76) (66,67,68) (74,75,76) (79,78) (87,89) (99,103,104,105) (113 through 117) (118 through 126, 134 through 149, 153, 154) (127 through 129,133) (158, 159) (164,165) (167 through 173) (178, 179) (183,184) (187, 193, 194, 198) (204, 205, 208) (214, 215, 216) (233,235) , (258, 259) (264, 265) (266, 267) (277 through 285) (308, 309) (325, 326, 327) (345, 346, 347) (348, 349, 353) (366, 369, 374) 1(384, 389) (404, 406) (413, 414) (415, 425 through 427) |(416, 417) (437, 438) (456,457) (459, 464) (466, 469) ; (475, 476) (477, 485) (483, 484) (486, 487) (488, 489) i ; (494 through 496) (497, 498) (505, 506) (508, 515) ! (509, 519, 549) (516, 317) (527, 529) (525, 533) I (535 through 543, 547) (553 through 558) (563, 564) I ;(565, 566, 573, 584, 589 through 594, 596, 598, 599) (567, 569) (575, 576) (579, 583) (585, 587) (614, 615, 617) (634, 635) 19^ ' (636, 643 through 646, 649, 655) (637, 639) (656 through 659) (669, 673, 674) (675 through 677, 679, 684) (687, 688) j (689, 693) (694, 695, 699) (705, 707, 713, 715, 717) (723, 724, 725) (726, 728, 729, 733) (735, 737) (743, 749) (753, 755, 758, 763 through 765, 768, 773, 777) (786 through 795) (796, 798) (803, 843) (813, 814) (823, 826) (828, 829,833,834) , (845, 848, 853, 855, 859) <863 through 867) (876, 883) 194 APPENDIX D. States within each Geographic Region 195 Table D.l. States within each Geographic Region North Central South 111inois Indiana Iowa Kansas Michigan Minnesota Missouri Nebraska North Dakota Ohio South Dakota Wisconsin Northeast Connecticut Maine Massachusetts New Hampshire New Jersey New York PennsyIvani a Rhode Island Vermont Alabama Arkansas Delaware District of Columbia Flor ida Georgia Kentucky Louisiana Maryland Mississippi North Carolina Oklahoma South Carolina Tennessee Texas Virginia West Virginia West Alaska Arizona California Colorado Hawaii Idaho Montana Nevada New Mexico Oregon Utah Washington Wyoming L 196 1 APPENDIX E. Correlation Matrices £ ( V ) ^ O ' o o C T ' r-l O rs j O C M O O O O O N O O oooo^m*-i«-ifn C à r - « - I C D Ô f O * H » - l 0«Hv-4000<^J0 o o i A n ^ o > t r i o o r ^ o»-*«-irMNj-«-ifnrsi o o o o o o o o C F N rvj vO m C M o O O O O I I I I I o o > u > « n > f » - * o r ^ o r s j c D c3vo>TirirsjaorsjrricDNj*d^ 0cv)rs|00000000 r— csj ^ GO O' M 3 N O O < - 4 C M m o fn <M O O C D o vnooocor«tnof-49«f-4orsi o^<NiOv-4CDmoNO N O C N i i r \ ooooooooov-400 I I I I I ONO(M(MrMNOOinmNOOOrsjiA orNjrsjO^i/\U>«-4CV)rsjOG0rsjf-4 OOOOOOOOOOOCDCDO o%A^NOsf^oDfnfn*MrMNOOOsf CDv-4'M^O«HlOU>OCV)ONO*HrOCV)0 0*^0000000000000 CO oa ' • T rONOoooooNocor^oo 00N0Orsj>T*-4'^m>9’U>orsiv-4«^JN0 *- 4<MOOOoOOOOrslO«-4O 0 lAO>Csj»-4t-loONO*-i*-**-4n^O rsjrsiaooo*0«"t>9’ «hcsJnOnOOO ^ N f * - 4 0 0 0 * - 4 0 0 0 0 0 U > 0 or-trtmüor->^rsi>orno>^»Os^a' OU%NO>TM>0>9’OCOU^*-4«Hrs|NOUl 0 < N J 0 0 0 0 0 0 0 0 0 0 0 « H * - 4 198 i n or f9 3: <b> O <n m 1 . CO O y - 4C s JA- • ac sA o v4 o CM 1 T3 ID rH • • • • % O v> « —4 -H n 1 1 C o o w o r- CD m <a> * •lA o CO K- « - 4y - 4* — 4« - 4« - 4 199 o r vf i n « H ' • T s o < Z > rsj y - 4 rsj G O O o o o o m o O O 4-4 sO 4A 04 CM O nj O * * 4 4 -t 4 - 4 4 - 4 o < z > < z > o < z > r- o O i A s O O O r — NOfHsO ocf'oooromo 04^00004-49 0 < ^ C O O h - f * - O O r M Of-imfM4-4rMaorsjo ooooooooo o n J - 4-4oooor-ooroao o 4 - 4 m 4 - 4 m o < js 4 - 4 fN i 0 m 0 0 « - 4 0 0 0 4 - 4 0 I I I I I I I 0(7*>f^4-4aor-'GOCr*P^<7S'0 o 4 - 4 u " \ in c js s j* o r— c s i 4 - 4 om4-404-44-40oomo ooor— >f^rsjrM\Om>ffncOGO 04-4r-M3000CO«.J-<Z>NO<T'Os| o s^rnooooooooo omooin'Ososor-r^cT'mforo omcMtHrorsjrotHomomso 0 0 0 0 0 0 0 0 0 0 04 0 0 omrsirnr-'.i'rsjrMrOk-iO'Or— m oorsj»n<r*r-oooorn'04-4rM oooooooooooooo ocT*a04-4in»nr-rn«-j-<7\rM04-44-4co Of\#4-44-4rsjo&m4-4^mmorsjr^4-$ 0 4 - 4 0 0 0 0 0 0 0 0 0 0 0 0 0 04-4s^ou>of^ro<<ih-<Mfvjcoco<y'm or^*"4^4-40n4in4"4s0n4^^(7%^0 0 4 - 4 s ± 0 0 0 0 0 0 0 0 4 - # 0 4 - 4 0 0 OGorsjy-44nmM>cT*sorsjr-cornrM4-4CT*in <z>coaoO'#sooosoroom>^'inr— rMu>4-4 OvnrOs04-H004-44-400000 o o Of-#^4-4cnr-">fr~-o>f^rMocn4-4sfsfij'o> o^<M<^u>r-f^4n<y\h«.f\jrii4-4aorM4-400 0 < M 0 0 0 0 0 0 4 - 4 0 0 0 0 0 4 - 4 m 0 4 - 4 in r- oo ' O I_ _ 200 (A <p> /-\ o CO CO OO o X: t4 o L_ LU VD • O LU t4 3 X ’ H z L _ lU o sO f - 4 A* o A| m » - f t4 o o 3: h- sy # # w c o 1 o r# O fO A- ro OL O o A- *-4 ID so o o O 0» L _ > 1 - 1 # • # e L _ > vy w ~ t c . a 1 o o o OO O' LO CO o C T » o o # QC m o O o m X» . Z) • # • # % o ■ * - » z 1 c o o _J w <x O CO GO OO CM # — > # • O CO O o A- # tH o o o o O C s J (X w # # # # # • < r—< LU 1 0» r-4 # # # # # 4A sO A- OO # — •H *H t4 * —1 201 #-# <N J x : ^ o CM ro 4 - 4 *o o a* ro 4 - 1 >o o o o o 4 - 4 m o o ÛQ ( J * * - • m o > i - r s i m < N J o r * ^ O CM o o r s l o 9 o r * N O r — o i n a * m m o 4 - 4 4 - i 9 ' T 4 - ( O o r n m o o r — '4'<r*s^orsi<Nioo otr*<NjcT\rs4CDr— o m 0 0 4 - 4 0 0 0 0 C V J 0 o c 7 * > t ^ > t ^ r o 4 - 4 0 o r o ii % o Qo OO 4- 4 m < 7 * r — 4 - 4 ' . f o r - 0 4 - ( 4 - 4 0 0 0 f s i o I I I I I o < 7 * m - M c o s O ' O C T ' r — ro>.T o r— 4-4^m 4-40om sro 0 0 4 - 4 0 0 0 0 0 0 0 0 o c o s - T c o r — ro o *n > o rN J s ,t’ <7* orsj4-4roa>rM4-4rNjOf— 04-#00r— 04-44"40*-#04-#0 ooom>ooroo>rO'.»“<Njoaoo O N o r o o o i O i O i 0 0 ^ 4 - i ( ^ ) r o o > o 0 0 0 0 O 4 - # 4 - # 0 0 O r M r s j 0 o o r s j ' 0 * n m O ' « T 4 - 4 r — rooocTso orotf>'.j“msOcx>»-4vnf— rsjor—ro O4-4OOOOr0<M4-4OO4-4rMO I I I I I O ' O ' . t ’ p-irOf— vJ'4-4'04-4rslr-4s,t'mC7> o M > o ^4-#rs#rs#r— rs#0>f04"4sf0 0 * - # 4 - # 0 0 0 0 0 0 0 0 0 0 0 0 oroM3(7*'^r— 4-44-4r— Lnfn4-4oo(7\(7*4-i O f— sorocM^tn'-j'sJ'OOsoro^'^'^ 0 ^ > t ^ 0 0 0 0 0 4 - ( 4 - ( C N | 4 - 4 0 < N J O O I I I I I oromO(7scr*<N(Nr— so<TNtnar*u%cr*u%<N OGOlAs^lAf-Hr— rorvir— ou^mr— rM0000'vfrM4-400sfm0 OcyrnsOOO^^rM v^4-t'0rM4-4rr>«0^iA ooo^oocommco4^^0fM(7*w%4-#(y*4-4 Or0OO4-lOOOOO<MrslOO4-4r04-l4-4 4 -4 rsj ro m sO r— O O a* 04-4rs#fn^w>or-oo L 202] w (0 r —i m z c m u •H X z z> z < y >OO X v4 &_ UJ cy o LU DC X r # Z L. LU y\ ■H r- <0 ►4 r4 Z h- cy *4 c CD o M ■H a. ZD nO < t > CD *4 L CD cy i. o O CD y * s <y> yc # QC LA T3 ■ z> r4 » CD vy -H X C O CD vy yc * — K4 *4 m QC cy 9 <* LU Z CD O CD LA CD r4 CD CD t4 1 CD OO CO CD LA O' O CCI CD • -4 • • O ( 7 > C X > O CD AI CD CD * - # CD m CD ^ ^ m CD ^ rf m O O CD O O CD I I J3 r o >*• tA »0 P- OO 203 0 C» 0 0 0 X X: w 1 1 rv*O' O' 4"i 0 ry ro 0 0 0 0 X X 1 1 1 1 0 tn O' 0 z 0 0 0 * - < Q O O r—004-«rON004-l o o o o o o o o OCTNCM-s^f— soo^oomoo OO^sOOOOf— 0 0 0 0 0 0 0 4 -MOO OCT^OO'O'O'OnO^OOCSO 04-«r— r— c r* o m otrv-sTOooooooo Otn«-i<3oroo^cDO^«^mnini OOU^'TlAOOrMrMO'r-lr-l 0'0*0*-<00004-l000 0 0 3 < N | 0 0 * " 4 f— O f — v O rO O 'C M O ' O 4-1 4-1 >r<NiM>4-iors*4-i>rrM*-« 00000000004-400 LU 4"# L/> 1 4 1 1 1 1 X c GOeg sO COegeg m m u eg g- m 0 0 4-f or m 0 0 <x > 0 0 1 1 1 1 1 ♦■* Iw • 3 *o en 00 CO O' UJ CM g> g-sT U QC 0 0 0 0 0 0 0 0 0 *#- 1 1 1 1 X 44 L eg 0 O' 0 r- m s#- m 0 eg 0 0 X c 1 1 1 1 1 1 1 1 1 0 •H >r (0 0 s#- g- 0 eg =3 4-g m 0 0 0 Q L UJ L_ 1 1 4 1 t 0 0 0 0 >T sO g> 0 s#- 0 m 4-i<Nim>#irNNOr-GC><T*04-«rMfOc#^in o r— oo 204 m c f O u •H OC c o 3 tx o Z3 y> OO co ^ *4 UJ ND O o o X M L . U J O CO + y tel A* O •4 m «4 *4 CD O z A- cy • • 14 *4 c O o *4 O *4 ro m o. y^ o C s l ro *4 r j O o O CD Q » CD *4 • • * c. CD cy *4 c o 1 o o o t 4 C v J *4 /A OO y> o A- CD < M • QC ul o O CD ro “ O Z) *4 • • • • » a cy •4 ♦y X 1 c o o cy <* y-\ O AI C A OO *4 H- O >r A4 m Al # »4 r - 4 o o O CD CD OC cy • • • # • * < r4 Ui Z 1 1 1 Q » r —4 J Q • • « • • < 0 sf » A O A- 00 ► — «4 «4 •4 4 •4 205 o e-N #-# m X fH 9 en m m OO 9 9 « H fH O ' < M O O O O f H O o e n O Q o o X X 1 1 1 1 v Oo Q o O n O r — I I I I I oinpnfnom*-i-Tm 9 s O « - t « - t v M « - t O O v M ooooooooo O N O O > » f — 'O '^O O'^ OO o o o ' O m O ' ^ L A t n r - o omoooooooo 9 0 ' P - ' O s ^ C v | r - i r s j m r - C 0 oesjr— OrHOmLOOOsOr-l Of- vMOOOOOOOO o m c o c n r s i s o O r - i m r s j i A r - o m o o o s r o f N i m ^ T o O f H O O O sfO O O O O O O O O O s o r - r - r n o r - r g t n s O f - i > t * * - f o r 0 9 r H f N j r H m m ' 0 s ^ q - 4 f n m OOOOOf-lOOOOfHf-IO O f - « ^ f H v 0 r g N t ’ O f - i m s 0 a 0 f - i 4 0 o r- m o cf m ^ >Tco<N ioo'«H «-g o o o o o o n «-tooo oego I I I I I I o 0% ^ O O o N ^O O m rsJC O rO ’ s O C T 'rM o o ' O ' ^ r r n o m r g r g f - i e g r n g* o* « h 0f-«00000f-«0000000 O N O * - # N O f g f g w f - & A o o ^ M ) f \ i # - « m OiAO'O'O'sOO O O m rH g) \0 O < 7» O Of-HOnOOOOOs^rHOrHO ^ rH O I I I I I o^oofMOrHsfrsi-^egooaorOf-HLno Os.fO'NOaoa'irvvHiAcoaoofeicoesim^ O t O s r > 0 « - H O O O O s ^ r H f - H O O U > m O Of'iegrsirviO'NOrH-^eooou^rnsOmoocOiA OrHcooofn*HfTijnr— r^isfrno'or— vTVfHNf- ornrHO«-HOoooo(NjrHOf-HrHmegrH rHfsimsfinsOf-aoa'OrHegrnsfmsor— OO 206 w o 4 m z c (0 ja z 3 =D /A O L/> < x > vc 4 L. UJ ND O UJ 4 - 3 X 4 Z C . UJ /A • A » Al A» fO 4 4 Z A» CD 4 c CD o 4 •H fO O. / A 4 3 sO <9 CD 4 C . CD CD L. O O CD XA CO DA • ■ o c LA *o ZD *4 » a CD - p y z IZ o CD -I CD <% «A I — s» 4 4 LA QC CD • <C UJ Z o o o o o o H - CD CD O sf O A- O' O AI CD CD CD CD OO m sf s i - CD Al o O' OO fO 4 CD n O CD o Al CD O O O O I I I JD ra st LA sO Pw OO L, 207j 9 O O C N l GO c f O U l r H r s | O ' O O O O O rH O 9 m O O c f cr* cf o in o r— C 7> O' m o rn 9 9 rH m 9 O O ' ^ t A i n o e g * ^ o o m m o ' o m r — 9 e 9 O 9 m H " 9 *-H 9 O r H 9 c f m s f f n s 0 9 o o 9 9 m m a o m 9 ' ^ 9 r H 9rHrHrH090esl0 9 m 9 o o ' O L n r o m r — 'J* OrgcOM><M<MOO>OrH 9 sJ-9099rHrH99 I I I I I Vh o 9<MO'cn»nfno<T'm'Oen 9 m M3 sr rH m m go «o <7* « - h 9 r n L n < M f M < M O O r H < M O orsir— 'OLn'OrHO'rHrNiom 9 0 9 N O m r H ' s f O G O a ' O D O OiA^OrHrHrHOOOrHO 9 r H r H o > a o t n m r H m < M o m * o O c f O ' o t o r M & n G o o r H a ' r H r — 09009rHrH900rH<M9 o o n j i n r H O c s i r - i A s O r H c n m r — 9r— sj'oorsiMj'OoO'OfnnprjrnrH 099009<NJ00090rH0 I I I I I 9 0 0 ( 7 ' 9 L n * 0 r H C 7 ' O U > 0 ' 9 m o n ^ 9'Ofnrn^rsj4ttnrf>orgNrrHrgrH O rH O O O O O O O O O O O O O 9 9 a'rHOr— moor— a'<N irgcr>p— s O n o O O O rs lr H U > O O ^ m > O e s |c 9 ’ % 0 ^ ^ r H r H OrHmOOOOOrHrHrHrHOrHOO 9 0 L A r H r — a O rH < 7 'tn a O *O O O rH > 0 ^ (7 *s ^ 9 c f m e g r — vnN#*r— ' j ' N o e s j r o n 9 0 ' r — M3 ONfcT'OrHOrHOOmrHOOrHNt’rHO o r— moor— <3D<^rH(7'Otnm^ffirHO>rHC7' O O r H M > « ^ M j c n O r H c X > t ’ rvl>#rHr— 9 < r * c ^ OmOOrHOOrHOOCMCJOrHrH'krrHrH rHegmcfinNOr— ooo'OrHcvirocfvnM^r— OO 208 t / l </p></a><p><a> vO CD *M 4 « O 4 # • # L O CD 4 C o 1 O O O CO r o X> / A OO DO CD m 4 CO • OC u> CD 4 CD Csl TJ %) 4 • # • • % O CD 4 ■4 X 1 C O o VD < t DA O GO LA AJ A I 4 H- o 4 r o r o CD • 4 4 o O o CD O nO ÛC CD • • • # • • H 4 LU Z 1 1 1 O « 4 X # • # • ro H- U> nO 4 OO P - 4 4 4 4 4 209 o ni o u > N ü r H O rH rH <M m O . C D C D O O m O I I I I I O sf O O th *o m O o fH rH fH rH O O rH O O O O O m O Z rH j - o o r s i r H O o o c g o OCT'fOinf— rgfnsoco OOOeg^rsir— rHO ooooooooo r H N O m o O iA r H P — e v i e v j m u * v o m i r t f n r H moooooooo O M 3 r H > * ’ LnrHes|tNJ>^'00O CDf— s ^ > fr s i > r r H OmcsJOOCDOOOOO o O O r— -s f m r— u n r— m rn m m o s o r - e g c N i r g s ^ H ^ m m s T o o m eg rHOOOOOOOO O O O i n c O O r H r H t A > ^ r H r H i n OOsIrHeslrHenmOlrHrsi-OenrH OOOOOOOOOrHfMOO OO'O'rHvOsfGOf— rnC^NOrgrHfSI O r H O O m rsimm'OOrHtn g- rH OOOOOOOOOOOrHOO OU^OrHOOr— ÇT'sOC'esj'^r— rsjcn O f - > f O r H r H f n O r H O * O O r H O m ooooooooooooooo OrvjaONf<J*>TrHiOrHCNjvOrH'OOOCnr— 0 * 0 * 0 « n r H ^ O » n ^ r H f - r H r H m O ^ OrHrHOOOOOOOOOOCvjOO omo'foofnesir— o'ommr— ^ ^ r H O ' ^ m e n m o o o o r H O O o o m o o O ^ e g O O O e g f O r H ^ rH f O r — M ) e g O O < 7 * m O o m - v ^ r g r 0 C n > ^ O < 7 ' r H O O C D < 7 ' O O m OrnOOOrHOOrHOOOOOrH cf O rH rHrsjrnH^fcAsor— coo>OrHevjm>fmsor— QO 210 «B < V U . X Z ) 4 CD 3 OO CD 4 CD K. LU CD O LU 4 H - 3 t X • H Z L LU 4 CD 4 py A» 4 o 4 fO 4 4 CD CD z P - CD • • 4 * 4 c O o 4 • p y O m vO ÛL 4 O rsi LA 4 Z ) sO o o CD w O 4 • e O CD *4 L a 1 o w o o O l A 4 LO 4 o m CD LA • oc LA o o CD «M T3 Z ) 4 • » O CD 4 - p y X 1 1 c o o CD < 4 O m LA «M A4 4 s f o sO rO fNI s r • 4 4 o o o CD CD 4 QC CD • • • • • < * # 4 LU Z • 4 J3 • • • • m LA o 4 QO h- 4 4 4 * 4 4 211 o rH CO C X 3 sj- CD o m eg fo o o o o o m o CD e l O rH r— rH M) O eg rH O rH rH rH o o o o o r— o 00'4h0rg0'0co O s j - e O r H r H c g u i r g o r g o o o o r g o o u i o o o o o r g g j e n O r H s f r g r g c j ' C n f O r H O O O O O O O O O o m O ' O G ' r H ' O c n r — cf O c n i - H o o r g c g o f o o o O C g O O r H O O r H r H O I I I I I I o r — rH C O r H U lO O M D U l c fm C D r H s f * n C D m C D r H M > r* l c r o v A r H O r g r g o o o r g o O o o r * l o O ' O e i r g c r * s j ’ o r g s o O r H c n c O o r — m m c f o cD o O c ^ m O O O r H O O O r H O o u i < 7 * ' O f n m < M ' ^ m f — fOoorH o m r g f o o e g g j ' O O s j ' r n r - o O O O O O O O O O O f A O O o r g c g r — r H u i s ^ o c g s O s ^ f - f O r H O rg CD sj- rH rH sf Ul <N # OJ lA fO rH O O O O O O O O O O O r H O O o m s o m o a o N O r g O s O G O a o o G O c f O C O r H O r H r H O r H C g r g r g r H c ^ O r g o o o o o o o o o o o o o o o OrHuirnmO'-^tiiaouioenoD'Ooeg O O c ^ ^ r H W l r H ^ O m ^ ^ r H O r H ^ OrHrgooooooooootnoo o m o o r O ' O r H o o a r ' r — l A u i e g u i i A f O c f m oo'OONOeiO'O'or— sj- rH rH m o eg eg ou irn ghe goo ooo ooo oLA OO oegr— or»o0'0(7''00oooui'0r— egcnencnr— O r H e g u i e n u i O o o r — s ^ r H O f O r H f - O ' e g a ' O r e j r H O O r H r H O r H O O O O O O m O O rHrgrn>^ui'Of— oocnOrHegmghgi g > f — o o 212 (A <V 4 ro £ O u_ DC u 10 z 4 Z ) 4 r o t o ao V I 4 L LU VD O UJ «*- 3S X •H Z L UJ 4 -H AJ 4 10 4 4 Z 4 CD 4 C W o r i py <0 e x . yc 4 Z> X» « W 4 C O CD U O o o VO D> • oc lA X» ZJ 4 % O CD py X c o o —I <% DC 4 H # 4 4 OO OC CD • UJ Z o o o o o o o m x > o o ro o o o o tn 4 O' 0 0 4 0' o o o I N I o X) 4 X) ^ o X> AI o rvj o o o o o « Jâ m ^ lA X> 4 OO 213 o < T * f> j o r s 1 / ^ > * ■ O • - < C 3 < x > m < M o C D o o r r » C D o O S rxj C T * o r— rs j c 3 O O < 5 * — I C3 T-l o o ^ o o o r ~ ' O u s m i r > r ' - < s j C D A . - l v O O O S ' O . H o c n . - » a o r u m c 3 o < M A s * " « T > e s fs i e s e s o —t T - i o e s e s o « N e s o u s O O v f- C T ' fs - ro O O C T ' O O 0--« T \JCT'U S'*'» -*CS .-40 OaOr4r-IC 3CSCSO<MCS CSfMf«“ OOf«»t/S AO.»-CSCS o a o « - # . - « O m c s r u r U ' O m OO.HOOOOOOOO cs«r»T-ifn«-#r— o . - f a o r * - « - i r - « o e s o o < M e s C S 0 0 r U \ 0 ( 7 * C S m T - l ' 0 ^ « - # A O C D > o r o c T ' o o r - T —rNi'O fs-fs-^ C S C S O C S C S O J .~ « O C S O < M .-IC S O T — aorOvror~-oo.-«m.-« , —t <T> <nj 0 . - l t M . - « r - « T r t f M 0 A C M r n m r s 4 f N â o.-icscsoesm<Mcsooors40 ocsaofs-orNjosrnisjT—Lo--«aocsjo Or-»cS'*’m.*’ris<M>rcs<*>o»-«'rr'* 0.-I.-ICSOOOOOOOOOOO OU>r-ieS'-lfs-Ofs->Or-lli%OS'd' cm «f I— o m < 7 > ' N j i - i o . - i c s r - ( M r M . - « - ^ ( M t n r o O r - 4 < « > ^ o o o o i ' - t O T - i o o m o o 3 < N e s c s r - 4 ( n ' O A i n T M < M r ~ r s > > o r r t ' ^ r n r o t n . ~ « C 3 'O 0 O « a (M ra > frri< n (7 s rM rs tu s .-« rs t'O > ^ omcsi'Omoooo-d-ro.-fcsomrrio cs.-HOcO'OCnoocT*C7'CNJr^aOvr*^aO'Ousc> cseS'00«nc3r>fMcs'>o.>trMrgrMOinrnro o o m O S e 1 1 1 1 1 1 # a m s O O ' * o 214 in O r H « o E <» c fO U •H X O Z z> y - N LO 00 vc «H LU CD O o o X •H L lU /A CD CA py 4 CD rH m •H rH CD CD z 4 CD • • •H rH c O 1 o H py O QO x> <0 CL y\ CD AJ CD •H 3 sO CD AJ rH « O w# • * • C O CD «H t. O 1 o o O ro rH Al yc LO y\ O vO Ai CD • QC LA O o CD ro -O - Z3 • H • # • • » O CD rH py Z 1 c o o _J CD «X DC O *H LA sf 4 sr O CA AJ Al CD • •H «H O O CD O CD c r » QC CD • • • • • • «t rH uu Z 1 « r H • • • • • m H" LA X> 4 OO 4 «H rH w# rH w# 215 J CD '«■ «n <3 iM «n o fvi m vO o o o o o m o o r-« CO .n tH o rvj rNj fNi LT\ r— (n O CD CD a CD 'T CD CD UD rn rn CD {j\ no CD o r o < M r o < M ( A O > o CDCDCDCDCDr\jr4CD C D O ' *- * r \j C D U D O r \j m CDCDCDCDCDCDCDOCD oaoinr^roooroor\j\o CDCDCDCDCDOCDCDCDCD CDr-HCDO^ir\vO*-*Cvi>J-rWl/> 0 ' « ' < r o o r . j < M > o o D > r t < r ' M otn-4 - o o o o o o o o CDO'cniD'r\jo^cOrOfNJ<7'r-4rg o < M ‘r t < - * < v i o r o r - ( 7 ' o m « - t C D O r O —ICDCDCDCDCDCDOCD o t O ' ^ - r n o o t r t r M r n i r k f n n i o CDCDCDCDCD*-«CDCDCDCDrMCDCD cDr— • - « m a o o o r — -*-cDr~oouD(Ni<7* o o < ^ u > r - t n t n r o a o r \ j o o j r 4 r 4 CDCDOCDCDCDC n JCDCDCDCDCDCDCD CD rn > o a ' o o < - * 0 ' o o a o > o r ^ ^ c 7 ' i n v o oo*'0<»irnm<Miaf>jrn«-»oooo»-i OC DCDCDOCDCDCDOCDCDCDCDCDCD o r M < - i r n « n a o m < M t n > f o r n < - « < M > - t r \ j C D * - « - i - C D O O C D C D r N | C D T - t C D C D C n C D C D OCT'»-tO'^aDCDm(T>t-t»-t>*r-OO>-lv0 oot>«^mooo«-»a*oooroorn<Mift OIA(0>00JO<-*<-*O>^O<-*OO>AOO CDsf\OCy\r\jr40sfsfaOr\jCDr\jsf.-4.-4COsf oomc7'rn<x><Ni>forr|aomtor\jaO'*''0'>’ Ofv^v-tOf-tOOoOO^-tooooroo^N o o ( 7 ' C D r - i r \ i r r , ' f i f » o r — oo 216 ( A 0> rH m E 0 » u. c ro u •H oe o • * - » <0 3 z Ol _D CO oo c _ • - » o LU y- LU 3 X •H t. Z LU (0 r s * r- z: ►H • - » t — w c *-* o O •H y m « —$ Q. <9 Z) 'O & . l _ > L O w O O O * < / > Xf oc lO % ZD wH * - » O W c z o o w _l <% * h - o » - » i - y œ vy * <% LU z o o o o Ï M O O O O o r - C M o ' O c o o o o o <M 1 - » r a o r - i r - o o o o r o I I o o m t r t r - « o m t o o s f o o o o C3 * * * * * I a > jQ m N» »o o r- co 217 o oo 'O CXI irv O ro ^ oo en eg O O O O «H O O O o < o > o r \ j m flo *-• in fo fo G O o o « H m a o n * - < M o o m r r > ' O r o o>^ 'T o o o m ro > o OrM«-400>^'^0 I I I I I O r\j 00 oo 'O lA >#■ sf- rg O ' ^ o o o r ^ f H f n o 000000000 o o f o # - 4 r g r g p * » O f \^ m moooooooo rn fH f— 'O o o *H CM o oo rsj ro <0 «-400000000 Q O ^ O O O O « - 4 r O m i A > ^ r M O L m O O ^ r 4 0 ^ \ O U l r 4 W ^ ^ O l A O O O O O O O O O or^oooo>#mu>ooaoinmrsio o r M o o o < j ' ' 0 * H o r ~ < 7 > o o m 00000fH0000<N|«-40 I I I I O l A O ' ^ > ^ r O O ' O f l O U % ( N # * H * - « f * o r » r » r r i ^ , H f ^ o r » f M o o u > o oooooocgooooofHO o o o ^ r » o % u > ' O o o O f M ^ v M c n o o o r » r - o * - « f H r r > « - 4 0 f H < n o « - 4 ^ o o 0«-40000 0«-40000000 O L n & o ^ > f o e g * - 4 0 ^ e g u i p * » c n o o r o o o o a o ^ r - ^ f # r M o o r M u i ^ r g ~ f o r o 0 0 C N | « - 4 0 0 0 0 n ^ 0 0 0 0 - ^ < M 0 O O O O t O m ^ f M r H i m ^ ^ O ^ O O ' f f M O O i n > ^ u > « M o o o o > ^ < - 4 f H O O i / > m o omo^r^^«M>^mr»maoao>^^r-oco<7^ o * M * H ^ f H f ^ r M i n o « O f H i n o > ^ < M o u > r » 0mNO^^00^orn^f4O,-#>ffM,-# ^ fM r» o > t’ u ^ 'O r^ o o < 7 ^ o *-« rM ^ o CO I 218 (A </a><a> ZD CO oo c. x: r4 o UJ W y- UJ 3 X • r 4 C z < J J y\ m r v l r- z t-4 r4 Ë- vy c * - 4 o O •H y r a « - 4 o. o Q » «O (. w C o vy O a o • co -O oc in % 3 * " 4 O vy C z o o _ J < % # h - » - *r4 OC v y • < U J X o o o o r v i o >#• vO o en o o <T» r— o <M m o o o CM o m rv4 h- en C3 *Ni r» m *3 o o o o o « SI Q ^ lA 'O oo 219 o C O o O O ro tA C O (M o o o o o o f*« c n O s j c n rn o o T - 4 ro 'O ^ O fO O O CM 4A O oinoocO'^ O OlAC MM>U >>t^OO 0''^'0ooro>j‘0 o c M o o u o r o o o c ^ a o 0 ' ^ c n * n < r i r * * r s j 4 r » m 0#-4^^000i-40 0 I A ^ » A > ^ 0 * H f ^ C F » f M O ' ^ ^ O O O O O O O oooaoo«-^r^'^*-tr^«-^r^ O ' f ^ O 0 0 r 4 € M O * ^ ' 0 r ^ o r o i / > r O i - 4 r M o ^ , - « < M O 00*^>J‘ l A C M C 0 ' ^ I A f ^ i - 4 r ^ uo CM r- r* SO o lA ~o o ^ >*‘> «‘0^0«-<000^0 O rM O O C V f'O rO fM 'O iA O O 'O lA C O o o o « - ^ ^ i o n o ^ o o «^<7^o > o 00»-«00^«-««-«00rs|rM0 O -f sO oOODvOrMCnOOfOOOCOOO 0 > J ‘ € y ' < M i A i A r O ' 0 0 0 > ^ r \ I O O * - « ooo«-^oo<Mooooo«-^o o > j ‘ *^<T**-*t-«co>*‘ rnoor->u^<7'roi-4 O < M < M O O O < M > * * 0 0 m O O ^ 4 A ^ Of^OOOOOOOOOOOOO OrOCOsrfO<T'0>J‘fOvOCOfVJ O fO lA o f ^ o r ^ < 7 ' < M f ^ ' ^ o o o i A r - « c o r o m 0 « - ^ r r | 0 0 « - ^ 0 0 A I « ^ r < 4 0 0 < M 0 0 o o o i r \ ' 0 < T * o o o i / > f o r - o o * ^ i A s f u ^ r - > O i A > # U > C M O t - < < H O > t ’ ^ O O O i n C M O o r v j v O ' « o o > ^ n * » * ^ a o i - 4 f ^ ( y * a o i A r - > i A r n f ^ o c M i - f O O ' f o o l o * - # 'T o o ^ u % r * r \ j f M r * o r r > « - 4 0 o « - i o o o o r M r M o o ^ % r r ^ ^ < •-4<Mro>t'VA'Or^oo<T‘ 0 * - « r M r O > * ‘ i/>’MOr^ao 22 0J (A 9 9 E 9 C •H ro c- 9 sz o 3 y> O CO OO o C - *"4 CD o lU vy • y- LU « - 4 3 X -H L Z -K UJ y> o CD ro r- CO vO X »-# •H o CD 4- yy • • c *-4 o U i •H r o CO r~ LA rH Ol o r- oo 9 o 'O o o CD C - o r4 • # • C - w vy r-» o o i o y> O m A4 • UJ y> O oo <r oo T5 QC O o o A4 % ZD » - • • • # • 4 - * O vy « - 4 c z • o o sy y\ O </a><a>om- CO H- *-4 * - # 22IJ APPENDIX F. Regression Decomposition Tables 22_2J A brief discussion of the figures presented in Table F.l illustrate how to interpret data in the tables found in Appendix F. An examination of Table F.l reveals that white males have actual earnings far higher than other groups (see columns 1 and 2). Note that white males average $19,966 while the next highest earning group was Cuban males with $16,082. Other Latin males earned $15,782, while black, Mexican and Puerto Rican males followed earning $13,962, $13,830 and $13,275, respectively. Average earnings for the highest paid female group (white females) were less than for the lowest earning male group (Puerto Ricans). The white male-minority/female earnings gap (reported in column 4) is narrowest for Cuban males and widest for Mexican females. White males earn, on average, about $3,884 more than Cuban males and a tremendous $10,990 more than Mexican females. Column 3 presents the earnings minorities and white females would receive if they had the same intercept and regression coefficients as white males. Thus, figures in column 3 show the earnings white females and minorities would have if they were compensated at the same levels as white males. Earnings for each group would rise if they were compensated at the same rate as white males. This Table F .1 Decomposition of Difference in Mean Annual Earnings Between White Males and Other Groups (First Model) Mean Actual Earnings for White Males (1): 19,966 Group MALES Mean E a r n in g s A c t u a l E x p e c t e d ( 2 ) ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) C o st o f Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6) B la c k 1 3 ,9 6 2 1 5 ,5 0 6 6 ,0 0 4 4 ,4 5 9 1, 544 M ex ica n 1 3 ,8 3 0 1 4 ,1 2 2 6, 136 5 , 8 4 4 291 P u e r t o R ic a n 1 3 ,2 7 5 14 ,5 12 6, 691 5 ,4 5 4 1 ,2 3 6 Cuban 1 6 ,0 8 2 1 8 , 8 1 2 3, 884 1, 153 2, 730 O ther L a t i n 1 5 ,7 8 2 1 7 ,9 7 1 4, 184 1 ,9 9 4 2, 189 FEMALES W hite 11 , 162' 1 7 ,2 4 0 8, 804 2 , 725 6, 078 B la c k 1 0 ,1 3 4 1 4 , 3 9 7 9 ,8 3 2 5, 568 4 , 2 6 4 M exican 8 , 9 7 6 13 , 013 10 , 990 6, 952 4, 037 P u e r t o R ic a n 9 , 6 7 5 1 3 ,6 7 0 1 0 ,2 9 1 6 , 296 3 , 9 9 5 Cuban 9 , 421 1 6 ,1 5 9 1 0 ,5 4 5 3 , 807 6, 738 O th er L a t i n 9 , 980 1 5 ,7 1 5 9 , 9 8 6 4 ,2 5 0 5, 735 rise is particularly noticeable among females. (Column 6 | shows the amount by which earnings would increase over j actual average earnings.) ! The white male-minority/female earnings difference | presented in column 4 is divided into two components: that! portion of the difference which can be explained by the I model (column 5); and that portion of the earnings j difference which cannot be explained by factors in the j 224 model (column 6). For five of the eleven groups the residual or cost of ■ discrimination exceeds the portion of the earnings difference accounted for by the model. The decomposition results based on Table F.l indicate . that females experience far more discrimination than males. Indeed, Cuban females are most discriminated | against, to the tune of $6,738. This means that Cuban : I f e m a l e s e a r n o v e r $6,700 l e s s t h a n w h i t e m a l e s w i t h I I identical characteristics because of their race and sex (as well as other factors which are not included in the - model). White and other Latin females also experience a | heavy financial disadvantage relative to white males, j $6,078 and $5,735, respectively. Black females are the | fourth most disadvantaged, earning $4,264 less than similar white males. Mexican females earn $4,037 less than identically qualified white males, while Puerto Rican' women earn $3,995 less. j Though minority males also experience discrimination, I earning from $291 to $2,730 less than similar white males, . for them the "cost" of being minority is far less than for women. (Among the males, Cuban and other Latin males are the worst off, earning $2,730 and $2,189 less than white males, respectively, due to discrimination. Black males 225> follow earning $1,544 less than white males due to discrimination. For Puerto Rican males the cost of discrimination comes to $1,236. Mexican males appear to have the lowest costs associated with labor market discrimination of all the minority groups considered. They earn about $291 less than comparable white males.) Note that the title of each table indicates whether the results were from the first or second model. The first model refers to that earnings function which includes years of education completed. The second model refers to that function which omitted years of education Table 4.5 in Chapter 4 is based on data from both models 226 Table F.2. Decomposition of Difference in Mean Annual Earnings Between Black Males and Other Groups (First Model) ?4ean Actual Earnings for Black Males (1): 13,962 C o st o f Group Mean A c t u a l ( 2 ) E a r n in g s E x p e c t e d ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) Minor i t y / F em a le S t a t u s ( 3 ) - ( 2 ) ( 6 ) MALES W hite 1 9 ,9 6 6 17 ,0 12 - 6 , 0 0 4 - 3 , 0 5 0 - 2 , 9 5 4 M exican 13 , 830 13 , 106 132 856 - 7 2 4 P u e r to R ic a n 1 3 ,2 7 5 1 3 ,8 8 0 687 82 604 Cuban 1 6 ,0 8 2 1 4 ,4 9 8 - 2 , 1 2 0 - 5 3 6 - 1 , 5 8 4 Other L a t i n 1 5 ,7 8 2 1 4 ,6 1 9 - 1 , 8 2 0 - 6 5 7 - 1 , 1 6 3 FEMALES White 1 1 , 1 6 2 1 5 ,2 5 1 2 , 800 - 1 , 2 8 9 4 , 089 B la ck 1 0 ,1 3 4 1 3 ,2 9 9 3 ,828 663 3 , 166 M exican 8 ,9 7 6 1 2 ,4 0 0 4 ,9 8 6 1 , 562 3 , 4 2 4 P u e r to R ic a n 9 , 675 13 ,3 0 2 4 ,2 8 7 659 3 , 6 2 8 Cuban 9 ,4 2 1 1 2 ,7 8 5 4 , 541 1 , 177 3 , 3 6 4 Other L a t i n 9 ,9 8 0 1 3 ,1 8 1 3 ,9 8 2 780 3 ,201 227] Table F.3. Decomposition of Difference in Mean Annual Earnings Between Mexican Males and Other Groups (First Model) Mean Actual Earnings for Mexican Males (1): 13,830 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c te d ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) Minor i F em ale S t a t u s ( 3 ) - ( 2 ( 6 ) MALES W hite 1 9 ,9 6 6 18 , 206 — 6 , 1 3 6 - 4 , 3 7 5 - 1 , 7 6 0 B la c k 1 3 ,9 6 2 1 5 ,2 6 5 -1 3 2 - 1 , 4 3 5 1 ,3 0 3 P u e r t o R ica n 1 3 ,2 7 5 1 5 ,3 8 0 555 - 1 , 5 5 0 2 , 105 Cuban 1 6 ,0 8 2 1 5 ,5 6 3 - 2 , 2 5 2 - 1 , 7 3 3 -5 1 9 O ther L a t i n 1 5 ,7 8 2 1 5 ,5 6 6 - 1 , 9 5 2 - 1 , 7 3 6 - 2 1 6 FEMALES W hite 11 , 162 1 6 ,4 6 4 2 , 6 6 8 - 2 , 6 3 4 5 ,3 0 2 B la c k 1 0 ,1 3 4 1 4 ,5 9 2 3 , 6 9 6 - 7 6 2 4 ,45 9 M exican 8 , 9 7 6 1 3 ,1 9 7 4 ,85 4 634 4 , 221 P u e r t o R ic a n 9 , 6 7 5 1 4 ,7 6 4 4 ,1 5 5 - 9 3 4 5 ,0 8 9 Cuban 9 ,4 21 1 3 ,8 0 9 4 ,409 21 4 ,3 8 8 Other L a t i n 9 ,9 8 0 1 4 ,1 1 2 3 , 8 5 0 -2 8 2 4 , 132 22SJ Table F.4 Decomposition of Difference in Mean Annual Earnings Between Puerto Rican Males and Other Groups (First Model) Mean Actual Earnings for Puerto Rican Males (1): 13,275 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c te d ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 6 ,3 7 8 — 6 ,6 9 1 - 3 , 1 0 3 - 3 , 5 8 7 B la c k 1 3 ,9 6 2 13 ,750 -6 8 7 -4 7 5 - 2 1 1 M exican 1 3 ,8 3 0 1 2 ,3 2 8 - 5 5 5 948 - 1 , 5 0 2 Cuban 1 6 ,0 8 2 1 3 ,3 8 9 - 2 , 8 0 7 - 1 1 4 - 2 , 6 9 3 O ther L a t i n 1 5 ,7 8 2 1 3 ,9 4 7 - 2 , 5 0 7 - 6 7 1 - 1 , 8 3 6 FEMALES W hite 1 1 ,1 6 2 14 , 776 2 , 113 - 1 , 5 0 1 3 ,6 1 4 B la c k 1 0 ,1 3 4 13 , 197 3 , 141 79 3 , 063 M exican 8 , 9 7 6 1 1 ,8 1 8 4 ,2 9 9 1 ,4 5 7 2 ,8 4 3 P u e r t o R ic a n 9 , 675 1 2 ,7 7 0 3 , 600 505 3 , 0 9 6 Cuban 9 , 421 1 1 ,7 5 9 3 ,8 54 1 , 516 2 , 3 3 8 O th er L a t i n 9 , 9 8 0 1 2 ,5 6 6 3 , 2 9 5 710 2 , 5 8 6 229 Table F.5. Decomposition of Difference in Mean Annual Earnings Between Cuban Males and Other Groups (First Model) Mean Actual Earnings for Cuban Males (1): 16,082 C o st o f Group Mean A c t u a l ( 2 ) E a r n in g s E x p e c t e d ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) Minor i t y / F em a le S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 8 ,7 2 4 - 3 , 8 8 4 - 2 , 6 4 2 - 1 ,2 4 2 B1 ack 1 3 ,9 6 2 1 5 ,2 9 2 2 , 120 790 1 ,3 3 0 M exican 1 3 ,8 3 0 1 4 ,3 5 0 2 , 2 5 2 1 , 732 520 P u e r to R ic a n 1 3 ,2 7 5 1 5 ,0 3 4 2 ,8 0 7 1 , 048 1 , 759 O ther L a t i n 1 5 ,7 8 2 1 5 ,7 9 3 300 289 11 FEMALES W hite 1 1 ,1 6 2 1 6 ,7 7 2 4 , 920 -6 9 0 5 , 6 1 0 B la c k 1 0 ,1 3 4 1 4 ,6 3 0 5 ,9 4 8 1 ,4 5 2 4 , 4 9 7 M exican 8 , 9 7 6 1 3 ,6 3 5 7 ,1 06 2 , 4 4 7 4 , 659 P u e r to R ic a n 9 , 675 14 ,3 4 5 6 ,4 0 7 1 , 737 4 , 670 Cuban 9 ,4 2 1 1 4 ,0 8 4 6 ,6 6 1 1 ,9 9 8 4 , 663 O ther L a t i n 9 , 980 1 4 ,1 8 1 6 , 102 1 ,9 0 1 4 , 201 23PJ Table F .6 Decomposition of Difference in Mean Annual Earnings Between Other Latin Males and Other Groups (First Model) Mean Actual Earnings for Other Latin Males (1): 15,782 Group Mean E a r n in g s A c t u a l E x p e c te d ( 2 )__________ (3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) MALES D i f f e r e n c e A c c o u n te d f o r by Model (5 ) C o st o f Minor i t y / F em a le S t a t u s ( 3 ) - ( 2 ) (6) W hite 1 9 ,9 6 6 1 8 ,9 8 0 - 4 , 1 8 4 - 3 , 1 9 8 - 9 8 6 B la ck 1 3 ,9 6 2 1 5 ,5 5 4 1 , 820 229 1 , 592 M exican 1 3 ,8 3 0 1 3 ,8 5 5 1 , 952 1 , 928 24 P u e r t o R ic a n 13 ,27 5 14 ,79 4 2, 507 988 1 , 519 Cuban 1 6 ,0 8 2 1 6 ,7 9 4 - 3 0 0 - 1 , 0 1 2 712 FEMALES W hite 1 1 , 1 6 2 1 6 ,7 3 1 4 , 620 -9 4 8 5, 568 B la c k 1 0 ,1 3 4 1 4 ,6 2 4 5, 648 1 ,1 5 8 4 ,491 M exican 8 , 9 7 6 1 2 ,9 4 2 6, 806 2 , 841 3 ,9 6 6 P u e r t o R ic a n 9 , 675 1 4 ,0 3 0 6, 107 1 , 753 4 , 3 5 5 Cuban 9 ,4 2 1 1 4 ,5 8 7 6 ,3 6 1 1 , 195 5 , 167 O ther L a t i n 9 , 980 1 3 ,9 8 8 5 , 802 1, 794 4 , 008 231 Table F.7. Decomposition of Difference in Mean Annual Earnings Between White Females and Other Groups (First Model) Mean Actual Earnings for White Females (1): 11,162 Group Mean E a r n in g s A c t u a l E x p e c te d (2 ) ( 3 ) Di f f e r e n c e in A c t u a l E a r n in g s (l)-(2) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) C o st o f M i n o r i t y / F em a le S t a t u s ( 3 ) - ( 2 ) (6) MALES W h ite 1 9 ,9 6 6 B la c k 1 3 ,9 6 2 M exican 1 3 ,8 3 0 P u e r t o R ic a n 1 3 ,2 7 5 Cuban 1 6 ,0 82 O ther L a t i n 1 5 ,7 8 2 FEMALES B la c k 1 0 ,1 3 4 M ex ic a n 8 , 9 7 6 P u e r t o R ic a n 9 , 6 7 5 Cuban 9 ,4 2 1 O th er L a t i n 9 ,9 8 0 1 2 ,6 7 6 - 8 , 8 0 4 - 1 , 5 1 4 - 7 , 2 8 9 1 0 ,2 3 8 - 2 , 8 0 0 924 - 3 , 724 9 , 501 - 2 , 6 6 8 1 , 661 - 4 , 3 2 9 1 0 ,0 3 3 - 2 , 1 1 3 1, 129 - 3 , 2 4 2 1 2 ,1 5 5 - 4 ,92 0 -9 9 3 - 3 ,9 2 7 1 1 ,4 8 3 - 4 , 6 2 0 -3 2 1 - 4 ,2 99 9 ,9 3 7 1 , 028 1 ,2 2 5 - 1 9 6 8, 938 2, 186 2 , 224 - 3 8 9 , 698 1 , 487 1 , 464 23 10 , 740 1 ,7 4 1 422 1 ,3 1 9 10 ,3 9 1 1 , 182 771 411 232 Table F .8. Decomposition of Difference in Mean Annual Earnings Between Black Females and Other Groups (First Model) Mean Actual Earnings for Black Females (1): 10,134 Group Mean A c t u a l (2 ) E a r n in g s E x p e c te d (3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) C ost o f M i n o r i t y / F em ale S t a t u s ! ( 3 ) - ( 2 ) ^ (6 ) , MALES t W h ite 1 9 ,9 6 6 1 2 ,5 2 0 - 9 , 8 3 2 - 2 , 3 8 7 - 7 , 4 4 5 ! B la c k 1 3 ,9 6 2 1 0 ,3 2 4 - 3 , 8 2 8 -1 9 1 - 3 , 6 3 7 I M exican 13 ,83 0 9 ,2 3 0 — 3 , 6 9 6 904 - 4 , 6 0 0 i j P u e r t o R ic a n 1 3 ,2 7 5 10 , 032 - 3 , 1 4 1 102 - 3 , 2 4 4 > Cuban 1 6 ,0 8 2 1 1 ,2 6 7 - 5 , 9 4 8 - 1 , 134 - 4 , 8 1 5 O th er L a t i n 1 5 ,7 82 1 1 ,3 2 7 - 5 , 6 4 8 - 1 , 1 9 4 - 4 ,45 5 FEMALES I W hite 1 1 ,1 6 2 1 1 ,3 8 1 - 1 , 0 2 8 - 1 , 2 4 8 -2 1 9 1 M exican 8 , 9 7 6 8, 889 1 , 158 1 , 245 - 8 7 1 P u e r t o R ic a n 9 , 675 9 , 846 459 288 171 1 Cuban 9 , 4 2 1 10 , 091 713 43 670 ' O th er L a t i n 9 ,9 8 0 1 0 ,3 8 6 154 -2 5 2 406 !33j Table F.9, Decomposition of Difference in Mean Annual Earnings Between Mexican Females and Other Groups (First Model) Mean Actual Earnings for Mexican Females (1): 8,976 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c t e d ( 3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) Minor i' F em a le S t a t u s ( 3 ) - ( 2 (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,2 3 3 - 1 0 , 9 9 0 - 3 , 2 5 7 - 7 , 7 3 3 B la c k 1 3 ,9 6 2 1 0 ,3 3 2 - 4 , 9 8 6 - 1 , 3 5 6 - 3 , 6 3 0 M exican 1 3 ,8 3 0 9 , 4 2 6 - 4 , 8 5 4 -4 5 0 - 4 , 4 0 4 P u e r t o R ican 1 3 ,2 7 5 1 1 ,0 4 0 - 4 , 2 9 9 - 2 , 0 6 4 - 2 , 2 3 6 Cuban 1 6 ,0 8 2 1 1 ,1 5 1 - 7 ,1 0 6 - 2 , 1 7 5 - 4 ,9 3 1 O th er L a t i n 1 5 ,7 8 2 1 1 ,0 9 8 - 6 ,8 0 6 - 2 , 1 2 2 - 4 , 6 8 5 FEMALES W hite 1 1 ,1 6 2 1 1 ,1 1 3 - 2 , 1 8 6 - 2 , 1 3 7 -4 9 B la c k 1 0 ,1 3 4 1 0 ,0 1 2 - 1 , 1 5 8 - 1 , 037 -1 2 1 P u e r t o R ic a n 9 , 675 1 0 ,7 0 9 -6 9 9 - 1 , 733 1 , 034 Cuban 9 ,4 2 1 1 0 ,0 5 0 -4 4 5 - 1 , 0 7 4 629 O ther L a t i n 9 , 980 1 0 ,2 2 0 - 1 , 004 - 1 , 2 4 4 240 234 Table F.10. Decomposition of Difference in Mean Annual Earnings Between Puerto Rican Females and Other Groups (First Model) Mean Actual Earnings for Puerto Rican Females (1): 9,675 Group Mean E a r n in g s A c t u a l E x p e c t e d (2 )__________ (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model ( 5 ) C o st o f Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6) MALES W h ite 1 9 , 9 6 6 1 2 ,1 9 9 - 1 0 , 2 9 1 - 2 , 5 2 4 — 7 , 7 6 6 B la c k 1 3 ,9 6 2 1 0 ,0 6 5 - 4 , 2 8 7 -3 9 0 - 3 , 8 9 7 M exican 1 3 ,8 3 0 9, 047 - 4 , 1 5 5 628 - 4 , 7 8 3 P u e r t o R ic a n 13 ,2 7 5 9 ,9 7 8 — 3 , 5 0 0 -3 0 3 - 3 , 2 9 8 Cuban 1 6 ,0 8 2 1 0 ,6 1 4 - 6 , 4 0 7 -9 3 9 - 5 , 4 6 8 O th er L a t i n 1 5 ,7 8 2 1 0 ,5 0 3 - 6 , 1 0 7 -8 2 8 - 5 , 2 7 9 FEMALES W hite 1 1 ,1 6 2 1 0 ,9 8 7 - 1 , 4 8 7 - 1 , 3 1 3 - 1 7 5 B la c k 1 0 ,1 3 4 9 ,7 3 4 -4 5 9 -5 9 - 4 0 0 M exican 8 ,9 7 6 8, 686 699 989 - 2 9 0 Cuban 9, 421 9 , 443 254 232 22 O th er L a t i n 9 ,9 8 0 9 , 528 - 3 0 5 146 - 4 5 2 235 Table F.11 Decomposition of Difference in Mean Annual Earnings Between Cuban Females and Other Groups (First Model) Mean Actual Earnings for Cuban Females (1): 9,421 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c t e d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( D - ( 2 ) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 19 ,9 6 6 1 2 ,2 1 9 - 1 0 , 5 4 5 - 2 , 7 9 9 - 7 , 7 4 6 B la c k 1 3 ,9 6 2 1 0 ,5 1 2 - 4 , 541 - 1 , 0 9 1 - 3 , 4 5 0 M exican 1 3 ,8 3 0 9 , 656 - 4 ,4 0 9 - 2 3 5 - 4 , 1 7 4 P u e r t o R ica n 13 ,2 7 5 1 0 ,4 9 9 - 3 , 8 5 4 - 1 , 0 7 8 - 2 , 7 7 7 Cuban 1 6 ,0 8 2 1 0 ,4 7 3 — 6 , 6 6 1 - 1 , 0 5 2 - 5 , 6 0 9 O ther L a t in 1 5 ,7 8 2 1 0 ,8 1 9 - 6 , 3 6 1 - 1 , 3 9 8 - 4 , 9 6 3 FEMALES W hite 1 1 ,1 6 2 1 1 ,2 8 6 - 1 , 7 4 1 - 1 , 8 6 5 - 1 2 4 B la c k 1 0 ,1 3 4 1 0 ,3 3 6 - 7 1 3 - 9 1 5 203 M exican 8 , 9 7 6 9 ,4 30 445 -9 454 P u e r t o R ica n 9 , 675 1 0 ,2 6 5 - 2 5 4 - 8 4 4 590 O ther L a t in 9 , 980 10 , 046 -5 5 9 - 6 2 5 66 L. 236 Table F .12 Decomposition of Difference in Mean Annual Earnings Between Other Latin Females and Other Groups (First Model) Mean Actual Earnings for Other Latin Females (1): 9,980 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c te d (3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n te d f o r by Model (5 ) M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,7 0 3 - 9 , 9 8 6 - 2 , 7 2 3 - 7 , 2 6 2 B la c k 13 ,9 6 2 1 0 ,6 6 6 - 3 ,9 8 2 -6 8 6 - 3 ,296 M exican 1 3 ,8 3 0 9 , 5 8 1 - 3 , 8 5 0 399 - 4 ,24 9 P u e r t o R ic a n 1 3 ,2 7 5 1 0 ,6 3 4 - 3 , 2 9 5 -6 5 4 - 2 , 6 4 2 Cuban 1 6 ,0 8 2 1 1 ,4 2 4 - 6 , 1 0 2 - 1 , 4 4 4 - 4 , 6 5 8 O ther L a t i n 1 5 ,7 8 2 1 1 ,0 3 0 - 5 , 8 0 2 - 1 , 0 5 0 - 4 , 7 5 2 FEMALES W h ite 1 1 , 1 6 2 1 1 ,3 2 7 - 1 , 1 8 2 - 1 , 3 4 7 165 B la c k 1 0 ,1 3 4 1 0 ,2 4 8 -1 5 4 - 2 6 8 114 M exican 8, 976 9 , 049 - 1 , 0 0 4 931 73 P u e r t o R ica n 9 , 6 7 5 1 0 ,2 2 0 305 -2 4 0 545 Cuban 9 , 4 2 1 1 0 ,0 8 1 559 -1 0 1 660 237 Table F.13. Decomposition of Difference Earnings Between White Males (Second Model) in Mean Annual i and Other Groups Mean Actual Earnings for White Males (1): 19,966 C o st o f Group Mean A c t u a l ( 2 ) E a r n in g s E x p e c te d (3 ) Di f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n te d f o r by Model (5 ) Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES B la c k 1 3 .9 6 2 1 5 ,7 4 8 6 , 004 4 ,218 1 , 786 M exican 1 3 ,8 3 0 1 5 ,3 4 5 6, 136 4 , 621 1 ,5 1 5 P u e r t o R ic a n 1 3 ,2 7 5 1 5 ,2 2 9 6 , 691 4 , 737 1 , 954 Cuban 1 6 ,0 8 2 1 8 ,6 7 4 3 ,8 84 1 , 292 2 , 5 9 2 Other L a t i n 1 5 ,7 8 2 1 8 ,3 0 8 4 . 184 1 , 658 2 , 526 FEMALES W hite 1 1 ,1 6 2 1 6 ,8 2 9 8 , 804 3 , 137 5 , 667 B la c k 10 , 134 14 ,0 1 5 9 , 832 5 ,9 5 1 3 , 881 M exican 8 ,9 7 6 1 3 ,4 7 5 1 0 ,9 9 0 6 ,4 9 1 4 , 499 P u e r t o R ic a n 9 , 675 1 3 ,9 1 4 1 0 ,2 9 1 6, 052 4 , 239 Cuban 9 ,4 2 1 1 5 ,7 6 6 1 0 ,5 4 5 4 ,20 0 6 , 3 4 5 Other L a t i n 9 , 980 1 5 ,7 1 4 9 , 986 4 ,2 52 5, 735 238 Table F.14. Decomposition of Difference in Mean Annual Earnings Between Black Males and Other Groups (Second Model) Mean Actual Earnings for Black Males (1): 13,962 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c te d ( 3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Mode 1 (5 ) Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES White 1 9 ,9 6 6 16 ,909 - 6 , 0 0 4 . - 2 , 9 4 7 - 3 , 0 5 7 M exican 1 3 .8 3 0 1 4 ,4 8 9 132 - 5 2 7 659 P u e r t o R ic a n 1 3 ,2 7 5 1 3 ,4 9 3 687 469 218 Cuban 1 6 ,0 8 2 1 4 ,0 8 1 - 2 , 120 -1 1 9 - 2 , 0 0 1 O th er L a t i n 1 5 ,7 8 2 1 4 , 7 2 2 - 1 , 8 2 0 -7 6 0 - 1 , 0 6 0 FEMALES W hite 1 1 ,1 6 2 1 5 ,0 2 8 2, 800 - 1 , 0 6 6 3 , 8 6 6 B la c k 1 0 ,1 3 4 1 3 ,1 0 9 3 , 828 853 2 , 9 7 5 M exican 8 . 976 1 2 , 5 3 6 4 , 986 1 , 426 3 , 560 P u e r t o R ic a n 9 . 675 1 3 ,3 5 5 4 , 287 607 3 , 6 8 0 Cuban 9 . 421 1 2 ,7 4 0 4 , 541 1 , 222 3 , 3 1 9 O ther L a t i n 9 , 980 1 3 ,1 9 9 3 , 982 763 3 , 2 1 9 Table F.15 Decomposition of Difference in Mean Annual Earnings Between Mexican Males and Other Groups (Second Model) Mean Actual Earnings for Mexican Males (1): 13,830 C o st o f Group Mean A c t u a l (2 ) E a r n in g s E x p e c t e d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n ted f o r by Model (5 ) Minor i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES - White 1 9 .9 6 6 1 7 ,8 5 4 — 6 , 1 3 6 - 4 , 0 2 4 - 2 , 1 1 2 B la c k 1 3 .9 6 2 1 5 ,0 7 7 - 1 3 2 - 1 , 2 4 7 1 ,1 1 5 P u e r to R ic a n 1 3 .2 7 5 1 5 ,5 1 4 555 - 1 , 6 8 4 2 , 239 Cuban 1 6 ,0 8 2 1 4 ,8 4 1 - 2 , 2 5 2 - 1 , 0 1 1 - 1 , 2 4 1 Other L a t i n 1 5 .7 8 2 1 5 ,2 4 1 - 1 , 9 5 2 - 1 , 4 1 1 -5 4 1 FEMALES W hite 1 1 ,1 6 2 1 5 ,9 6 8 2 , 668 - 2 , 1 3 8 4 ,8 0 6 B la c k 1 0 ,1 3 4 1 4 ,1 6 3 3 , 696 -3 3 3 4 , 029 M exican 8 ,9 7 6 1 2 ,9 4 1 4 , 854 889 3 ,9 6 5 P u e r t o R ic a n 9 , 675 1 4 ,7 0 4 4 , 1 5 5 - 8 7 4 5 , 029 Cuban 9 ,4 2 1 1 3 ,0 3 7 4 ,4 0 9 793 3 , 6 1 6 Other L a t i n 9 , 9*80 1 3 ,6 5 8 3 , 850 172 3 , 678 240 Table F.16. Decomposition of Difference in Mean Annual Earnings Between Puerto Rican Males and Other Groups (Second Model) Mean Actual Earnings for Puerto Rican Males (1): 13,275 Group Mean A c t u a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e in A c tu a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n te d f o r by Model (5 ) C o st o f M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 19 , 966 1 6 ,1 1 7 - 6 , 6 9 1 - 2 , 8 4 2 - 3 , 8 4 9 B la c k 1 3 ,9 6 2 1 3 ,6 8 3 - 6 8 7 - 4 0 8 -2 7 9 M exican 1 3 ,8 3 0 1 2 ,5 3 1 - 5 5 5 744 - 1 , 2 9 9 Cuban 1 6 ,0 8 2 1 3 ,0 5 5 - 2 , 8 0 7 220 - 3 , 0 2 7 Other L a t i n 1 5 ,7 8 2 1 3 ,8 0 0 - 2 , 5 0 7 -5 2 5 - 1 , 9 8 2 FEMALES W hite 1 1 ,1 6 2 1 4 ,3 7 5 2 ,1 1 3 - 1 , 1 0 0 3 , 2 1 3 B la ck 1 0 ,1 3 4 1 2 ,9 0 0 3 ,1 4 1 375 2 , 766 M exican 8 , 976 1 1 ,7 6 8 4 , 299 1 , 507 2 , 792 P u e r t o R ic a n 9 , 675 1 2 ,5 9 6 3 , 600 679 2 , 9 2 1 Cuban 9 , 4 2 1 1 1 ,3 7 0 3 ,8 5 4 1 ,9 0 5 1 ,9 4 9 Other L a t i n 9 , 9.80 1 2 ,3 0 4 3 , 295 971 2 ,3 2 4 241] Table F.17. Decomposition of Difference in Mean Annual Earnings Between Cuban Males and Other Groups (Second Model) Mean Actual Earnings for Cuban Males (1): 16,082 C ost o f Group Mean A c t u a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n te d f o r by Model (5 ) M i n o r i t y / Fem ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 19 ,96 6 1 8 ,7 1 8 - 3 , 8 8 4 — 2 ,6 3 6 - 1 , 2 4 8 B la c k 1 3 ,9 6 2 1 5 ,4 6 2 2 , 120 620 1 , 500 M exican 13 ,8 30 1 5 ,0 4 2 2 , 2 5 2 1 ,0 4 0 1 ,2 1 2 P u e r t o R ic a n 1 3 ,2 7 5 1 5 ,3 6 8 2 , 807 714 2 ,0 9 3 O th er L a t i n 1 5 ,7 8 2 1 5 ,9 3 4 300 148 152 FEMALES W h ite 1 1 ,1 6 2 1 6 ,4 9 1 4 , 920 - 4 0 9 5 ,3 2 9 B la c k 1 0 ,1 3 4 1 4 ,3 8 4 5 ,9 4 8 1 , 698 4 ,2 5 0 M exican 8 , 976 1 3 ,8 3 9 7, 106 2 ,2 4 3 4 ,8 6 3 P u e r t o R ic a n 9 ,6 7 5 1 4 ,3 6 8 6 ,4 0 7 1 ,7 14 4 , 693 Cuban 9 , 421 1 3 ,9 6 0 6 ,6 6 1 2 ,1 2 2 4 , 539 O th er L a t i n 9 , 980 1 4 ,1 0 6 6, 102 1 ,9 7 6 4 , 126 242 Table F.18 Decomposition of Difference in Mean Annual Earnings Between Other Latin Males and Other Groups (Second Model) Mean Actual Earnings for Other Latin Males (1): 15,782 Group Mean A c tu a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n te d f o r by Mode 1 ( 5 ) C ost o f M in o r i t F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 18 , 890 - 4 , 1 8 4 - 3 , 1 0 8 - 1 , 0 7 6 B la c k 1 3 ,9 6 2 15 , 643 1 , 820 139 1 , 681 M exican 1 3 ,8 3 0 1 4 ,2 2 7 1 ,9 5 2 1 ,5 5 5 397 P u e r t o R ica n 1 3 ,2 7 5 1 5 ,0 2 9 2 , 507 753 1 , 754 Cuban 1 6 ,0 8 2 1 6 ,6 6 1 - 3 0 0 -8 7 9 579 FEMALES W hite 1 1 ,1 6 2 1 6 ,4 2 0 4 , 620 - 6 3 8 5 ,2 5 8 B la c k 1 0 ,1 3 4 1 4 ,3 8 5 5 , 648 1 ,3 9 7 4 ,2 5 1 M exican 8 , 9 7 6 1 2 ,9 3 0 6 ,8 0 6 2 , 852 3 , 9 5 4 P u e r to R ic a n 9 ,6 7 5 1 4 ,0 2 1 6 ,1 0 7 1 , 761 4 ,3 4 6 Cuban 9 ,4 2 1 1 4 ,3 5 6 6 ,3 6 1 1 ,4 2 6 4 ,9 3 5 O th er L a t i n 9, 980 13 , 815 5 , 802 1 , 967 3 , 835 243j Table F.19. Decomposition of Difference in Mean Annual Earnings Between White Females and Other Groups (Second Model) Mean Actual Earnings for White Females (1): 11,162 Group Mean A c t u a l (2 ) E a r n in q s E x p e c te d (3 ) Di f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e . A c c o u n te d f o r by Model (5 ) C ost o f M i n o r i t y / F em ale S t a t u s ( 3 ) - { 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,8 0 0 - 8 , 8 0 4 - 1 ,638 - 7 , 1 6 6 B la c k 1 3 ,9 6 2 1 0 ,5 8 8 - 2 , 8 0 0 574 - 3 , 3 7 4 M exican 1 3 ,8 3 0 1 0 ,1 5 8 - 2 , 6 6 8 1 ,0 0 4 - 3 , 6 7 2 P u e r t o R ic a n 1 3 ,2 7 5 1 0 ,5 4 0 - 2 , 1 1 3 622 - 2 , 7 3 5 Cuban 1 6 ,0 8 2 1 2 ,1 1 6 - 4 , 9 2 0 -9 5 4 — 3 , 9 6 6 O th er L a t i n 1 5 ,7 8 2 1 1 ,6 8 8 - 4 , 6 2 0 -5 2 6 - 4 , 0 9 4 FEMALES B la c k 1 0 ,1 3 4 1 0 ,0 1 5 1, 028 1 , 147 -1 1 9 M exican 8 ,9 7 6 9 ,3 0 9 2 , 186 1 ,8 5 3 333 P u e r t o R ic a n 9 , 6 7 5 1 0 ,0 0 7 1 ,4 8 7 1 ,1 5 5 332 Cuban 9 , 421 1 0 ,6 0 9 1 , 741 553 1 ,1 8 8 Other L a t i n 9 , S80 1 0 ,4 7 7 1 , 182 685 497 244 Table F.20 Decomposition of Difference in Mean Annual Earnings Between Black Females and Other Groups (Second Model) Mean Actual Earnings for Black Females (1): 10,134 Group Mean E a r n in q s A c t u a l E x p e c te d (2 ) (3 ) D i f f e r e n c e i n A c t u a l E a r n in g s (l)-(2) ( 4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) C o st o f M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6) MALES W hite 1 9 ,9 6 6 1 2 ,5 6 3 - 9 , 8 3 2 - 2 , 4 2 9 - 7 , 4 0 3 B la c k 1 3 ,9 6 2 1 0 ,5 0 9 - 3 , 828 -3 7 5 - 3 , 4 53 M exican 1 3 ,8 3 0 9 , 594 — 3 , 6 9 6 540 — 4 , 2 3 6 j P u e r to R ica n 1 3 ,2 7 5 1 0 ,2 8 0 - 3 , 1 4 1 - 1 4 6 - 2 , 9 9 5 : Cuban 1 6 ,0 8 2 1 1 ,2 2 1 - 5 , 9 4 8 - 1 , 0 8 7 - 4 , 8 6 1 ■ Other L a t i n 1 5 ,7 8 2 1 1 ,4 1 5 - 5 , 6 4 8 - 1 ,2 81 - 4 , 3 6 7 FEMALES W hite 1 1 ,1 6 2 1 1 ,3 3 2 - 1 , 0 2 8 - 1 , 1 9 8 170 M exican 8 , 9 7 6 9 , 064 1 ,1 5 8 1, 070 88 P u e r to R ican 9 , 675 9 , 958 459 176 283 Cuban 9 ,4 2 1 9 , 979 713 155 558 O ther L a t i n 9 , 9 8 0 1 0 ,3 8 5 154 -2 5 1 405 2.4 5J Table F.21 Decomposition of Difference in Mean Annual Earnings Between Mexican Females and Other Groups (Second Model) Mean Actual Earnings for Mexican Females (1): 8,976 Group Mean A c t u a l (2 ) E a r n in q s E x p e c t e d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) C o st o f M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,2 2 9 - 1 0 , 9 9 0 - 3 , 2 5 3 - 7 , 737 B la ck 1 3 ,9 6 2 1 0 ,3 7 5 -4 ,9 8 6 - 1 , 3 9 9 - 3 , 5 8 7 M exican 1 3 ,8 3 0 9 , 505 - 4 , 8 5 4 -5 2 9 - 4 , 3 2 5 P u e r to R ic a n 1 3 ,2 7 5 11 , 149 - 4 , 2 9 9 - 2 , 1 7 3 — 2 , 1 2 6 Cuban 1 6 ,0 8 2 11 , 042 - 7 , 1 0 6 - 2 , 0 6 6 - 5 , 0 4 0 O ther L a t i n 1 5 ,7 8 2 1 1 ,0 7 3 — 6 ,8 0 6 - 2 , 0 9 7 - 4 , 7 0 9 FEMALES W hite 1 1 ,1 6 2 1 1 ,0 6 1 - 2 , 1 8 6 - 2 , 0 8 5 -1 0 1 B la c k 1 0 ,1 3 4 9 ,9 7 6 - 1 , 1 5 8 - 1 , 0 0 0 - 1 5 8 P u e r t o R ic a n 9 ,6 7 5 1 0 ,7 5 7 -6 9 9 - 1 , 7 8 1 1 , 082 Cuban 9 ,4 21 9 , 9 1 5 - 4 4 5 -9 3 9 494 Other L a t i n 9 , 980 1 0 ,1 5 1 - 1 , 0 0 4 - 1 , 175 171 246 Table F.22 Decomposition of Difference in Mean Annual Earnings Between Puerto Rican Females and Other Groups (Second Model) Mean Actual Earnings for Puerto Rican Females (1): 9,675 C ost o f Group Mean A c t u a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s <l ) - ( 2 ) (4 ) D i f f e r e n c e A cco u n ted f o r by Model (5 ) M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 > (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,1 4 6 - 1 0 , 2 9 1 - 2 , 4 7 1 - 7 , 8 2 0 B la c k 13 ,9 6 2 1 0 ,1 1 6 - 4 , 287 - 4 4 1 - 3 , 846 M exican 1 3 ,8 3 0 9 ,2 2 4 - 4 , 1 5 5 451 — 4 , 6 0 6 P u e r to R ica n 1 3 ,2 7 5 10 ,060 - 3 , 6 0 0 - 3 8 5 - 3 , 2 1 5 Cuban 1 6 ,0 8 2 10 , 533 - 6 , 4 0 7 - 8 5 8 - 5 , 5 4 9 Other L a t in 1 5 ,7 8 2 1 0 ,5 1 6 - 6 , 1 0 7 - 8 4 1 - 5 , 2 6 6 FEMALES W hite 1 1 ,1 6 2 1 0 ,8 7 3 - 1 , 4 8 7 - 1 , 1 9 8 -2 8 9 B la c k 1 0 ,1 3 4 9 , 675 -4 5 9 0 -4 5 9 M exican 8 , 976 8 , 739 699 936 -2 3 7 Cuban 9 , 421 9 , 3 3 7 254 338 84 O ther L a t i n 9 , 980 9 , 490 - 3 0 5 185 -4 9 0 2 47. Table F.23. Decomposition of Difference in Mean Annual Earnings Between Cuban Females and Other Groups (Second Model) Mean Actual Earnings for Cuban Females (1); 9,421 C o st o f Group Mean A c t u a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e i n A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Model (5 ) M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,2 9 2 - 1 0 , 5 4 5 - 2 , 8 7 1 - 7 , 6 7 4 B la c k 1 3 ,9 6 2 1 0 ,6 5 0 - 4 , 5 4 1 - 1 , 2 2 9 - 3 , 3 1 2 M exican 1 3 ,8 3 0 9 , 904 - 4 , 4 0 9 -4 8 3 - 3 , 9 2 6 P u e r t o R ica n 1 3 ,2 7 5 1 0 ,6 7 6 - 3 , 8 5 4 . - 1 , 2 5 5 - 2 , 5 9 9 Cuban 1 6 ,0 8 2 1 0 ,4 9 5 — 6 , 6 61 - 1 , 0 7 4 - 5 , 5 8 7 Other L a t in 1 5 ,7 8 2 1 0 ,9 1 4 - 6 , 3 6 1 - 1 , 4 9 3 - 4 , 8 6 8 FEMALES W hite 1 1 ,1 6 2 1 1 ,3 0 3 - 1 , 7 4 1 - 1 , 8 8 2 141 B la c k 1 0 ,1 3 4 1 0 ,3 7 3 -7 1 3 -9 5 2 239 M exican 8 , 976 9 , 569 445 -1 4 8 593 P u e r to R ica n 9 , 675 1 0 ,3 6 9 -2 5 4 -9 4 8 694 Other L a t i n 9 , 980 1 0 ,0 9 3 - 5 5 9 -6 7 2 113 248 Table F .24 Decomposition of Difference in Mean Annual Earnings Between Other Latin Females and Other Groups (Second Model) Mean Actual Earnings for Other Latin Females (1); 9,980 C o st o f Group Mean A c tu a l (2 ) E a r n in q s E x p e c te d (3 ) D i f f e r e n c e in A c t u a l E a r n in g s ( l ) - ( 2 ) (4 ) D i f f e r e n c e A c c o u n te d f o r by Mode 1 (5 ) M i n o r i t y / F em ale S t a t u s ( 3 ) - ( 2 ) (6 ) MALES W hite 1 9 ,9 6 6 1 2 ,7 6 3 - 9 ,9 8 6 - 2 , 7 8 3 - 7 , 203 B la c k 1 3 ,9 6 2 1 0 ,7 9 3 - 3 , 9 8 2 - 8 1 3 - 3 ,16 9 M exican 1 3 ,8 3 0 9 , 787 - 3 , 8 5 0 193 - 4 , 0 4 3 P u e r to R ic a n 1 3 ,2 7 5 1 0 ,7 9 2 - 3 , 2 9 5 - 8 1 2 - 2 , 4 8 3 Cuban 1 6 ,0 8 2 1 1 ,4 2 5 - 6 , 1 0 2 - 1 , 4 4 5 - 4 , 6 5 7 O ther L a t i n 1 5 ,7 8 2 1 1 ,0 9 1 - 5 , 8 0 2 - 1 , 1 1 1 - 4 , 691 FEMALES W hite 1 1 ,1 6 2 1 1 ,3 1 2 - 1 , 1 8 2 - 1 , 3 3 2 150 B la c k 1 0 ,1 3 4 1 0 ,2 6 5 -1 5 4 - 2 8 5 131 M exican 8 ,9 7 6 9, 133 - 1 , 0 0 4 847 157 P u e r t o R ica n 9 , 675 1 0 ,2 9 5 305 - 3 1 5 620 C uban 9 ,4 2 1 1 0 ,0 4 6 559 6 6 625 2 49.</a></p></b></b>
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Making it to the top: a study of Black partners in major law firms
PDF
A generalized scale for the measurement of attitudes toward international issues
PDF
Differential acceptance of a new role for pharmacists
PDF
A comparison of attitudes of university seniors and sales executives toward outside selling, as revealed by clinical interview and questionnaire methods
PDF
Identity change in a therapeutic community
PDF
Effects of an intervention approach on the medication prescribing behavior of physicians
PDF
The effects of centralization and board membership reform on health occupational licensing policies
PDF
The effects of various factors on the academic achievement of Vietnamese students at Gold Coast College
PDF
Student authoring in the American grain
PDF
Development, validation and testing of a new sensor array for intra-articular pressure measurement: in-vitro human lumbar spine intra-articular facet testing
PDF
The impact of developmental events upon the perception of intergenerational family solidarity
PDF
An experimental study of the effect of "human interest" factors on listenability
PDF
The functional gene expression dynamics of Shewanella oneidensis MR-1 under various external voltage potentials
PDF
Realism in the fiction of Frances Burney
PDF
An experimental study of perceived differences in efficient and inefficient voice production in low-pitched male voices by acoustic spectrography
PDF
Magnesium-based photocathodes for triggering back-lighted thyratrons
PDF
Psychological well-being and location in the social structure
PDF
Convergent morphological evolution in hyenas and dogs: the relationship between form and function
PDF
Social factors involved in the success or failure of consumer cooperatives in the United States
PDF
Bureaucrats and old clients: dependence, stigma, and negative sentiment in the service relationship
Asset Metadata
Creator
Verdugo, Naomi Turner
(author)
Core Title
Earnings differences among Black, White and Hispanic males and females: the impact of overeducation, undereducation and discrimination
School
Graduate School
Degree
Doctor of Philosophy
Degree Program
Sociology
Degree Conferral Date
1985-12
Tag
OAI-PMH Harvest
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC11256202
Unique identifier
UC11256202
Legacy Identifier
DP31850
Document Type
Dissertation