Does a foot in the door matter? White-nonwhite differences in the wage return to tenure and prior workplace experience.

Darity, William, Jr.

1. Overview

There is an extensive literature documenting the lower wages of black and Latino workers relative to white workers in the United States. The explanations for these racial and ethnic wage disparities have ranged from differential accumulation of human capital to labor market discrimination. (1) Farmer and Terrell (1996), and more recently Lewis and Terrell (2001), extend this literature by advancing a hypothesis, which we summarize as the theory of ability misperception. This hypothesis explains how group membership may differentially influence the gains associated with workplace experience, and hence the relative wages of black and Latino workers over the course of their working lifecycle.

Their basic idea is twofold: First, employers believe nonwhite employees acquired fewer skills on prior jobs than white workers. Thus, employers will reward additional experience with prior employers to a greater extent for white workers than for nonwhite workers. We refer to this hypothesis as Proposition 1. This first proposition falls under the umbrella of statistical discrimination, which explains wage inequality as a result of employers offering nonwhite workers lower wages as a result of employer reliance on their perception that the distribution of skills for nonwhite workers is worse than that of white workers.

Second, as employers directly observe nonwhite workers on the job, they come to realize that their perceptions concerning the distribution of skills for nonwhite workers were incorrect and, consequently, they have underrated the ability of nonwhite employees. This learning leads employers to update their assessment of the relative productivity of nonwhite workers. Therefore, racial/ethnic differences in the return to additional tenure or seniority with the current employer will be smaller than the racial/ethnic difference in the return to prior experience when Proposition 1 holds. This prediction constitutes Proposition 2 of the theory of ability misperception.

In addition, we introduce a third proposition (or a strong form of Proposition 2), which suggests that the returns to additional tenure for nonwhite employees will actually exceed that of white employees. Thus, Proposition 2 requires that the difference in returns to additional tenure with the current employer between nonwhite and white workers be smaller than the difference in returns to prior experience, while Proposition 3 requires that nonwhite workers actually receive a higher return for tenure. The purpose of this paper is to advance the existing empirical literature aimed at evaluating the validity of the propositions central to the theory of ability misperception using data drawn from the Multi-City Study of Urban Inequality (MCSUI).

If Farmer and Terrell (1996) and Lewis and Terrell (2001) are correct that employers underestimate the skills of nonwhite workers before directly observing them at the worksite, then many capable nonwhite applicants will be denied work, and, if employed, will suffer undue economic losses. This form of discriminatory treatment, regardless of its source, may have deleterious labor market effects on nonwhite workers. It may harm incentives to work if wages are not commensurate with worker productivity, and it may lead to job lock. Once nonwhite workers get a proverbial "foot in the door," it will be relatively more costly for them to switch to another job because their talents are likely to be initially undervalued.

The theory of ability misperception also has implications for affirmative action policies and stricter enforcement of antidiscrimination laws. Moreover, the theory is at odds with Lazear's (1979) prediction that, in the wake of affirmative action, white worker wage gains for additional seniority will be higher than for nonwhite workers. Lazear assumes that affirmative action leads to the hiring of less-productive nonwhite workers, and that employers adjust for this productivity disparity by offering white workers higher wages with additional years of seniority. Lazear does not address Proposition 1 of the theory of ability misperception (i.e., that employers initially undervalue the prior experience of nonwhite workers).

Although neither this paper nor Lazear's (1979) provides a direct test of whether affirmative action leads to the hiring of less-productive nonwhite workers, this paper does provide an examination of the relationship between differential wage gains associated with prior work experience versus differential wage gains associated with on-the-job experience (seniority). Moreover, if the theory of ability misperception is accurate, it would suggest that affirmative action and antidiscrimination laws could be used to offset the undervaluation of prior experience of nonwhite workers. This is somewhat in contrast to Lazear's (1979) proposition that affirmative action leads to the hiring of less-productive nonwhite workers, which in turn leads employers to offer nonwhite workers lower returns to wages for additional years of seniority.

Ability misperception may arise from somewhat benign reasons, such as genuine ignorance of the productivity of a newly hired nonwhite worker. (2) In addition, some employers may undervalue nonwhite workers for more malevolent reasons, such as a racist inclination to discount evidence of nonwhite competence in order to rationalize not hiring these workers or paying them a lower wage than comparable white workers. (3)

Those employers who fit into the benign characterization should be expected to voluntarily adjust upward their perception of talent possessed by nonwhite workers in light of evidence at odds with their prior experience. Employers who are racist may have an impulse to resist adjusting rewards to nonwhite workers even after these workers prove themselves to be more productive than anticipated. However, there are at least two mechanisms that may induce racist employers to adjust the wages of nonwhite workers once they are employed. First, once employed, it may be easier for workers to observe and prove that they are treated unfairly relative to other workers than it would be for those workers who have not yet been hired. Second, it is theoretically possible that once a worker is hired, workplace norms to offer similar pay for similar work may set in, in part to avoid a hostile work environment, which can harm productivity, leading some employers to adjust the initial undervaluation of nonwhite workers. Therefore, we examine if employers who systematically underestimate the ability of nonwhite workers when they join their firm, regardless of how they come to this position, have a tendency to behave as predicted by Proposition 2 (or Proposition 3) of the theory of ability misperception.

Lewis and Terrell (2001), Bratsberg and Terrell (1998), and Wolpin (1992) provide evidence consistent with the theory's predictions by using data on white and black males drawn from the National Longitudinal Survey of Youth (NLSY). However, since the findings of Lewis and Terrell are limited to black males and are not particular to any occupation, their findings are not robust to occupation type and other workplace settings. Moreover, the relative returns to workplace experience for other subaltern wage-earning groups besides black males have yet to be examined. Thus, the question of whether U.S. data generally support the theory of ability misperception remains unresolved.

A shortcoming of the existing studies is that a wide range of workplace setting characteristics expected to affect a worker's wage, including factors such as the racial/ethnic background of a worker's direct supervisor, number of coworkers, or amount of contact with customers, are not taken into account--controls that we include in our study--because such data are not available in the NLSY. Omitting these workplace descriptors may mar the accuracy of the estimated impact of all the factors influencing a worker's wage--including workplace experience measures--making it difficult to determine whether there are racial or ethnic differences in rewards for work experience.

Moreover, it is plausible that the rewards for additional experience are contingent upon these very same omitted factors. Workplaces may differ in the propensity and extent to which wage setters hold inaccurate perceptions of minority workers' ability and in institutional arrangements that govern wage setting. For instance, white supervisors may be more inclined than nonwhite supervisors to underestimate the talent of nonwhite workers.

In addition, in some workplace settings, institutions are in place that may promote comparable gains to work experience for nonwhite and white workers. For example, larger firms and public sector employers

may be more likely to adopt formal codified rules that govern the rewards for workplace experience. This may reduce managerial discretion in constructing rewards and hence the likelihood of racial and ethnic differences in the returns to prior experience and tenure. Thus, to appropriately access the theory of ability misperception, the validity of Proposition 1 and Proposition 2 (or Proposition 3) should be evaluated for particular worksites. Carrying this exercise out for a host of worksites will reveal if the findings are robust. (4)

Our work is distinguished from earlier empirical work on this topic in four ways. First, the model of wage determination we estimate for white and black male workers controls directly for a wider range of workplace setting descriptors. Second, the analysis is conducted on subsamples based on workplace settings to determine if there is robust support for the theory of ability misperception. Third, we explore whether our findings on racial differences in wage gains associated with tenure and prior experience are robust with respect to gender. Finally, using data on Latinos and whites, we explore whether our findings are robust to subaltern ethnic status.

Our results offer strong support for Proposition 1 and for Propositions 2 and 3 for males along with clear-cut evidence consistent with these propositions for a subsample composed of white females and black females. The findings are less conclusive when the returns for white females and Latina workers are compared.

The remainder of this paper is organized as follows. In section 2, the data and methodology are described. In section 3, we present our main findings examining the theory of ability misperception for black and white workers. In section 4, we present our findings for Latino and Latina workers, in relation to white male and female workers, respectively. Finally, in section 5, we revisit Lazear's proposition with regards to affirmative action and offer some concluding remarks.

2. Data and Methodology

Data

The MCSUI is an interview-based survey of 8916 persons administered in the cities of Los Angeles, Boston, Atlanta, and Detroit beginning in 1992. To acquire more accurate responses, attempts were made to match respondents and interviewers by assigning interviewers of a certain race or ethnicity to respondents with that same racial or ethnic heritage.

Data on a rich array of socioeconomic and demographic factors is provided in the MCSUI, including information on a person's seniority with their employer, the age at which they completed their formal schooling, and workplace characteristics at their current job. MCSUI respondents were asked the number of years they had been with their current employer; we use their responses to this question to gauge an individual's level of tenure or seniority with their current employer. Respondents were not asked about the amount of work experience accumulated in previous positions. In this situation, the standard practice (see Bratsberg and Terrell, 1998, pp. 669) is to use prior potential experience (age minus total years of schooling minus 6 [the typical age of grade school entrance] minus years of tenure) as an instrument for actual workplace experience with previous employers.

Assuming tenure is measured accurately, potential prior experience as a proxy for actual experience with previous employers may suffer from three sources of inaccuracy or measurement error. First, not all children enter first grade at age six. Second, a person does not necessarily complete each grade of schooling in a single year. Third, the proxy implicitly assumes that all persons work continuously and at the same intensity (hours per week and weeks per year) once they complete school. This does not occur for a variety of reasons, including being laid off or taking time off for fertility or other family reasons. (5)

In theory, nonwhite workers are more likely to start school "late," to take additional time to complete a level of schooling, or to work less continuously once they finish their formal schooling. As a result, nonwhite workers will have accumulated less workplace experience than whites with the same measured amount of potential experience. (6) If employers believe such discrepancies exist--even if the reward for an additional unit of actual experience with previous employers is the same for white and nonwhite workers--white workers will receive a greater wage return for a one-unit increase in potential experience. Thus it is possible that studies examining the theory of ability misperception, including the studies by Lewis and Terrell (2001) and Bratsberg and Terrell (1998), might suffer from measurement error. (7)

Fortunately, MCSUI provides information on the age at which a respondent completed formal schooling. Using this information, we are able to construct an alternative measure of potential experience with prior employers: age minus age at which formal schooling was completed minus tenure. This measure is likely to be better than the standard measure of potential experience, because it avoids the first two types of measurement error (inaccuracy in the age at which a person enters the school system and inaccuracy in the assumption that a person completes a year of schooling during each calendar year throughout the schooling years). (8)

We characterize an individual's workplace setting by the individuals with whom they work, the tasks the individual regularly performs, and employer characteristics. The MCSUI survey allows us to identify the race/ethnicity of a worker's direct supervisor and the race/ ethnicity for the majority of their coworkers. (9) In addition, respondents indicated whether they have regular contact with customers, are a supervisor, are employed in a large firm (more than 100 employees), are a union member, are a full-time worker, and are a public sector employee.

We partition the data into separate subsamples of women and men. We restrict the analysis to persons aged 16-65 who were working and who were not self-employed. Initially, we further restrict the subsamples to individuals who are identified as either black or white by the surveyor. Later, we analyze subsamples composed of whites and Latinos, again partitioned by gender (see Appendix A).

In constructing the hourly wage variable, MCSUI's designers excluded workers whose wage translates to less than $2 or more than $100 per hour. In addition, MCSUI's designers did not include wage earners in Atlanta who earned over $100,000 annually; to make our analysis comparable across cities, we excluded all workers earning more than $100,000. Furthermore, we do not use observations from Detroit because information on key workplace setting variables, including the race/ethnicity of coworkers and supervisors, was not collected in this city. Persons also were excluded from our subsamples if they did not report information on the full set of variables used in the wage equation we estimate. Appendix B provides a detailed catalogue of observations lost for each restriction imposed in constructing the white and black subsamples. The subsample of males we analyze (given the restrictions we impose) contains 912 observations, 419 blacks and 493 whites, while the subsample of females we examine includes 1359 observations, of whom 805 are black and 554 are white.

Methodology

We begin our analysis by estimating a conventional wage equation augmented with a set of workplace setting descriptors for white and black males. We seek to determine if there are racial differences for males in the wage gains associated with the accumulation of additional prior experience and current employer experience (which we refer to as either "tenure" or "seniority") once we control for workplace settings. The models we estimate are specified as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

and

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

where ln w is the log of the wage a worker receives on their job and i is an individual index. The subscript k identifies a person's gender; k = male (m) or female (f). Superscripts indicate if a person is white (w) or nonwhite (nw). Tenure on the job is denoted by TEN and PEXP represents potential total accumulated labor market experience with prior employers. In addition, squared terms for potential experience and tenure on the job are included to control for nonlinear effects that may result from diminishing returns.

The vector WPSET contains a set of indicator variables that describe a person's work environment. This vector includes measures that capture the supervisor's race/ethnicity, the typical coworker's race/ethnicity, whether the worker supervises other employees, and whether the worker has regular contact with customers. In addition, this vector also contains a measure of the worker's firm size, union status, whether the job is full time, and whether the position is in the public sector. The vector X contains socioeconomic factors found in traditional wage equations, such as education level, health standing, marital status, family composition, location of residence, and immigrant status, along with a person's occupation. (10)

Motivation and attitude, which influence academic achievement, are possible sources of unobserved individual specific heterogeneity that are likely to be correlated with tenure and prior workplace experience. (11) To purge our OLS estimates of this potential source of bias, an indicator of high school performance, high achievement--measured by an average grade of B or better in high school based on a person's recollection--is included in X. (12)

The expected gain in wages due to an additional year of potential prior experience (PEXP) for the average nonwhite worker and the typical white worker are respectively

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Proposition 1 of the theory of ability misperception asserts that the average white worker will realize greater returns to prior experience then a typical nonwhite worker; therefore the theory predicts

{[partial derivative][w.sup.nw.sub.k]/[partial derivative]([PEXP.sup.nw.sub.k]) - [partial derivative][w.sup.w.sub.k]/[partial derivative]([PEXP.sup.w.sub.k])} = {A} < 0.

The difference in the estimated return to greater tenure (TEN) between typical white and nonwhite employees is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Proposition 2 of the theory of ability misperception posits that, relative to white workers, nonwhite workers will fare better in the wage returns to tenure than in the wage returns to potential prior experience--that is, {A} - {B} < 0. Employers are presumed to learn through direct observation of performance that they have underestimated the ability of nonwhite workers, and they revise their earlier assessment of nonwhite talent. (13)

Evidence that both Proposition 1 and Proposition 2 hold provides support for the theory of ability misperception. We extend the literature to include a third proposition, which states that the return to tenure for nonwhite workers actually exceeds that realized by white workers; it is a stronger version of Proposition 2. Proposition 3 is met when the returns to prior potential experience are higher for white worker ({A} < 0)--as is the case of Proposition 1--while the returns to tenure are higher for nonwhite workers ({B} > 0). Thus, even stronger support for the theory of ability misperception arises if Proposition 3 holds. Of course, if Proposition 3 holds, then Proposition 2 necessarily holds as well.

3. Empirical Results: Race and the Returns to Experience

Male Findings

White and Black Male Summary Statistics (14)

Appendix C reports summary statistics for the full subsamples of white male workers and black male employees. These subsamples are separated into high achievers and low achievers, measured by self-reported high school grades for all of the variables used in our analysis. Variable definitions are presented in Appendix D. (15)

The average black male worker earns $12.76 an hour, has been with their current employer for 5.61 years, and has 10.82 years of potential prior experience. Thirty percent of the black males recalled earning at least a B average in high school. Just over half of the black males ended their formal education with the completion of high school. Twenty-seven percent attended college; with 10% holding a college degree. A quarter of the black male workers were living overseas at 16 years of age, 52% were married, and 59% lived in Los Angeles.

In our subsample of black males, 30% had predominantly black coworkers. Thirty-seven percent reported working "mostly" with whites, while 60% indicated they have a white supervisor and 22% report having a black supervisor. A third of the black males supervised other workers, almost one-third were union members, and 19% were employed part-time. Among the black males in our analysis, 18% were employed in managerial or professional work, one quarter worked in a service occupation, and 18% were public sector employees.

White male workers share many of these characteristics. However, white men reported higher wages--an average of $15.95 per hour--had almost an additional year of potential prior experience, and had been with their current employer 10-12% longer (measured in years). Moreover, relative to black workers, white workers were twice as likely to have attended college, less likely to have lived outside of the United States at 16 years of age (only 5%), half as likely to be part-time workers, much more likely to hold a managerial or professional position, and less likely to be employed in a service job. White workers were much more likely to be in a white working environment, with 79% having mostly white coworkers and virtually the same percentage reported having a white supervisor. (16)

Race and the Returns to Tenure and Experience." Males

Table 1 is a summary table that presents our estimates of the impact of prior workplace experience and tenure on wages for the typical black male worker

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

and for the typical white male worker. The racial difference in the estimated returns to prior experience and tenure, and the percentage difference in our estimates of the interracial returns--with whites serving as the reference group--are presented as well.

Coefficient estimates for all of the variables included in Equations 1 and 2 are presented in Appendix E for a model that pools black and white workers and for the models with separate estimates for black or white employees. Virtually all of the estimated coefficients have the expected sign and are highly significant at conventional levels.

Inspection of Table 1 reveals that an additional year of prior experience significantly boosts the wages of white males by about 1%, but has no substantive impact on the wage of black males. This evidence is consistent with Proposition 1. Another year of tenure significantly raises wages of both white and black workers, and the estimated gains are virtually identical at 3.3% for whites and 3.2% for blacks. Thus, we find evidence consistent with Proposition 2, but not for Proposition 3. (17)

Lewis and Terrell (2001) found results comparable to those we report, but only for workers scoring poorly on the Armed Forces Qualification Test (AFQT) (at the 20th percentile or lower). Using a self-reported average high school grade below B as a measure of low achievement, we further stratify the racial subsamples into achievement groups: high- and low-achievement black males, and high- and low-achievement white males. Then, we evaluate Propositions 1-3 by comparing black male high (low) achievers with white male high (low) achievers. The results of this analysis also are presented in Table 1. Among low-achieving males, we find support for Proposition 1 and Proposition 2. These results are in line with those reported by Lewis and Terrell (2001) for low AFQT workers.

We do not find a significant relationship between previous experience and the wage rate for black or white workers who are high achievers (the sample sizes of black and white high achievers are 86 and 137 observations, respectively). However, the estimated coefficient is much larger for white workers than for black workers, as predicted by Proposition 1. Greater tenure significantly raises the wages of both white and black male workers. The evidence we report is consistent with Proposition 3, since black male workers in the high-achievement group receive a significant 5.9% wage increase when they accumulate an additional year of tenure, while the gain for whites is 5.2%.

Male Interracial Differences in the Returns to Tenure and Experience." The Role of Workplace Settings

We turn next to a formal analysis of whether racial differences in returns to prior and current workplace experience are contingent upon workplace settings for black and white males.

In phase II of our inquiry, we investigate whether racial differences in returns to tenure and experience for males are contingent on various job settings to determine whether the theory of ability misperception is applicable across a broad set of work environments. Next, we will describe the empirical strategy we use to conduct this facet of the inquiry. First, we select a particular job setting, such as one where daily customer contact occurs, from those contained in the vector WPSET. Then, we partition the data into two groups according to this variable for each racial group. For example, one partition is between black (white) workers with daily customer contact and black (white) workers without daily customer contact. The next step is to estimate Equations 1 and 2 for each of the subsamples, after eliminating the workplace setting variable used to partition the data while retaining the other workplace setting variables in the model. Finally, we compare the returns to tenure and prior experience for white and black workers in comparable job settings.

Table 2 provides a summary of our estimates of the impact of potential prior experience and tenure on the wages of male workers in each workplace environment and its matched pair in the same form as Table 1. (18) For instance, at the top of Table 2 we report that among workers who report that their coworkers are mostly white, an additional year of tenure or seniority raised white male workers' wages by 3.2%, while a typical black worker realized a 3.9% wage increase. Thus, the percentage wage gain associated with greater tenure was 22% greater for black workers than for white employees. Because so few white employees reported having a black supervisor or having primarily black coworkers, we are unable to determine if racial differences in the returns to workplace experience depend systematically on this aspect of job settings.

Inspection of Table 2 reveals that the estimated wage return to potential prior experience for black male workers is less than 1% and is statistically insignificant,

[partial derivative][w.sub.b.sub.m]/[partial derivative]([PEXP.sup.b.sub.m]) = [[??].sup.b.sub.m] + 2[[??].sup.b.sub.m]([[bar.PEXP].sup.b.sub.m]) [approximately equal to] 0,

in all 15 of the workplace settings evaluated. However, the return to potential prior experience for white workers is always positive and is statistically significant in 10 of the 13 workplace settings evaluated. When it is significant, it ranges in value from 0.6% (Daily Contact with Customers) to 2.5% (Government Worker), with a mean of 1.11%. Table 3 indicates whether each of the propositions is supported by the evidence presented in Table 2 for each workplace setting. Table 3 is a scorecard for the theory of ability misperception in multiple workplace settings. We find evidence consistent with Proposition 1, that the returns to prior experience are higher for white than black workers (i.e.,

{[partial derivative][w.sup.b.sub.m]/[partial derivative]([PEXP.sup.b.sub.m]) - [partial derivative][w.sup.w.sub.m]/[partial derivative]([PEXP.sup.w.sub.m])} = < 0),

in all 13 situations where we are able to estimate returns to potential experience for both groups. (19)

Black male workers obtain a positive and statistically significant wage increase when they accumulate an additional year of seniority or tenure,

[partial derivative][w.sup.b.sub.m]/[partial derivative]([[bar.TEN].sup.b.sub.m]) = {[[??].sup.b.sub.m] + 2[[??].sup.b.sub.m] (/[TEN.sup.b.sub.m])},

in all 15 workplace settings evaluated. The size of the increase, when significant, ranges from 1.7% for black workers who Supervised Others to 4.0% for black workers who were not supervisors and for black workers who had a Black Supervisor. The mean impact of additional tenure across the various workplace settings for black workers is 3.1%.

Additional tenure significantly raises the wages of white workers in all of the 13 workplace settings evaluated. In these settings, the magnitude of the wage gain ranges from 4.5% for white workers at large firms with at least 100 employees to 2.4% for white workers at small firms with less than 100 employees. The impact of an additional year of tenure for white workers across the various workplace settings is 3.3%.

Table 3 reveals that we find support for Proposition 2--that the black-white difference in the rate of return from on-the-job tenure is less than the racial difference from potential experience--in 12 of the 13 workplace settings. Moreover, in four of the workplace settings, we report evidence consistent with Proposition 3: Black workers have a higher rate of return than white workers from additional years on the job, which, along with Proposition 1, is strong evidence in support of the theory of ability misperception. (20)

We view the evidence presented as overwhelmingly consistent with the predictions of the theory of ability misperception across a wide range of workplace settings--a more general finding than the initial report by Lewis and Terrell (2001). The strong lesson provided by Table 3 is that regardless of the workplace setting, black men, relative to white men, receive relatively poor returns to prior workplace experience, but they fare better when the interracial return to tenure is considered.

However, it is interesting to note that the wage return to tenure for black male workers is 54% larger when they have a black supervisor than when they have a white supervisor, controlling for the race/ethnicity of their coworkers. This result suggests that there may be a greater willingness on the part of black supervisors than white supervisors to update erroneous beliefs about the productivity of black workers as they observe performance. However, if black supervisors are more objective than white supervisors in accessing black workers in the first place, then why is the return to prior experience smaller for black workers than white workers even when their supervisor is black? This puzzle can be resolved if a worker's direct supervisor has greater influence in setting or making recommendations that influence pay increases than in establishing their initial wage. (21)

Female Findings: White and Black Female Summary Statistics

Appendix F reports summary statistics for the full subsamples of white female workers and black female employees, as well as these subsamples separated into high achievers and low achievers for all of the variables used in our analysis. White female workers were similar to black female employees on almost all of these descriptive characteristics. However, white female workers earned almost $2 more per hour ($12.92 compared with $11.09) than black women. In addition, relative to black female workers, white women were more likely to work part-time, almost twice as likely to be married, and far less likely to be union members.

Race and the Returns to Tenure and Experience." Females

Panel B of Table 1 presents our estimates of the impact of prior workplace experience and tenure on wages for black female workers and for white female workers. Coefficient estimates for all of the variables are presented in Appendix E for a model that pools females and for the female models that are stratified by race. Virtually all of the estimated coefficients have the expected sign and are highly significant at conventional levels.

Inspection of Table 1 reveals that an additional year of prior experience does not boost the wages of white or black females significantly. However, an additional year of tenure significantly raises the wage of female workers. Moreover, the wage increase is larger for black females (3.4%) than for white females (3.0%). This pattern of similar (and insignificant) returns to prior experience along with significant and relatively high returns to tenure for black females also reemerges when we compare the wage benefit of greater tenure and prior workplace experience both for high-achieving low-achieving females.

It is interesting to note that the size of the wage increase for additional tenure is much larger for the high achievers, regardless of race, than for the low achievers. High achieving black females who gain an additional year of tenure realize a 7.1% wage gain, while the gain is 2.7% for low-achieving black females. For white females, the returns to tenure for high achievers, 4.2%, is not nearly as large as the returns for comparable black females, but it is still much larger than the returns for low achievers, 2.6%.

Table 4 is a summary table that presents our estimates of the impact of potential prior experience and tenure on the wages of black and white female workers for each workplace setting descriptor and its matched pair. (22) The estimated wage return to potential prior experience for black females is statistically insignificant in all 16 workplace settings evaluated. However, the return to another year of prior experience for white female workers is positive and statistically significant in 5 of the 14 workplace settings evaluated, but when significant the return is relatively small, less than 1%. Table 3 indicates whether each of the propositions is supported by the evidence presented in Table 4 for workplace setting. We find evidence consistent with Proposition 1,

{[partial derivative][w.sup.b.sub.f]/[partial derivative]([PEXP.sup.b.sub.f]) - [partial derivative][w.sup.b.sub.f]/ [partial derivative]([PEXP.sup.w.sub.f])} = < 0,

in 9 of the 14 workplace settings where we were able to estimate the return to potential prior experience for both black and white females. (23)

Black female workers obtain a positive and statistically significant wage increase when they accumulate an additional year of seniority or tenure in each of the 16 workplace settings evaluated. The size of the increase ranges from 2.2% (Union Member) to 5.5% (Part-Time Work) and has a mean value of 3.5%. Additional tenure significantly enlarges the wage of white female workers in 12 of the 14 workplace settings evaluated. In these settings, the magnitude of the wage gain ranges from 1.7% (Supervisor) to 4.8% (Part-Time Work), with a mean increase of 3.13%. Interestingly, the return to seniority is 33% larger for black female workers who have a direct supervisor who is black than for those with white supervisors. This result is similar to the situation we detected for black male workers.

Inspection of Table 3, a summary of evidence concerning the theory of ability misperception in the various workplace settings, reveals support for Proposition 2 in all nine of the workplace settings where Proposition 1 holds. Moreover, in eight of the nine settings we find strong support (both Propositions 1 and 3 are satisfied) for the theory of ability misperception. The evidence that we report for females is similar to the findings we report for males, and both sets of findings are consistent with the predictions of the theory of ability misperception across workplace settings. In the next section we probe further to determine whether data from another group of nonwhite workers, Latinos, conforms to the predictions of the theory of ability misperception.

4. Latino Ethnicity and the Returns to Experience

Findings

Latino and Latina Summary Statistics

Appendix G reports summary statistics for full subsamples of Latino male and female workers and, again, for subsamples partitioned between high achievers and low achievers. The subsample of males we analyze (given the restrictions we impose) contains 987 observations (494 Latinos and 493 whites), while the subsample of females we examine includes 1068 observations (514 are Latina and 554 are white). (24) For brevity, we highlight the major differences between these workers and their white counterparts.

The average Latino worker earned $9.08 an hour, which was substantially less than the typical white male employee ([w.sup.w.sub.m = 15.95). Latinos, relative to white males, were three times as likely to have lived abroad at age 16, half as likely to be a manager or professional worker, and far more likely to reside in Los Angeles. The most striking difference lies in the ethnic configuration of the worksite. Over 70% of Latinos had coworkers who were mostly Latino(a), and almost 39% had a direct supervisor who was Latino(a). Only 14% of white male employees reported having mostly Latino(a) coworkers, and a mere 6% have a direct supervisor who is Latino(a).

A similar pattern exists for Latina workers. Seventy percent of them lived outside the U.S. when they were 16 years old. The wage of a typical Latina employee was $7.56, well below the wage of a typical white female worker ([w.sup.w.sub.f] = $12.92). Almost half of the Latina workers had not completed high school. They also tended to work for a direct supervisor who is Latino(a) (40%), and 70% are employed at a worksite where most of the individuals doing similar work are Latino(a).

Latinos: Returns to Tenure and Experience

Panel A of Table 5 presents our estimates of the impact of prior workplace experience and tenure on wages for Latino workers and for white male workers. Coefficient estimates for all of the variables are presented in Appendix H for a model that pools white and Latino workers and for models stratified by ethnicity. Virtually all of the estimated coefficients have the expected sign and are highly significant at conventional levels.

Inspection of Table 5 reveals that greater time on the job with the current employer or with previous employers significantly raises the wage of male workers who are Latino or white. However, examination of Table 6 reveals that for most of the 12 worksite environments evaluated, the wage return to tenure is substantially larger for Latinos than for white males, while white males obtain a larger gain for additional prior experience.

The estimated wage return to prior potential experience for Latino males is statistically significant in only 3 of the 15 workplace settings (too few Latinos were in public sector positions to estimate the return to experience in that setting). However, the wage gain associated with greater tenure is statistically significant in each of the 15 workplace environments examined. In three workplace settings (i.e., Latino(a) Coworkers, a Latino(a) Direct Supervisor, Part-Time Work), there were too few white males to estimate the returns to workplace experience. Thus, we are able to compare inter-ethnic (white vs. Latino) differences in the returns to tenure and prior experience for males in 12 workplace settings. Table 7 summarizes our findings with respect to Propositions 1-3 for white males relative to Latinos. In nine of these 12 settings we find evidence consistent with Proposition 1. Proposition 2 holds for every one of the nine worksite environment where Proposition 1 holds, and for seven of the nine settings Proposition 3 holds. This evidence, like our racial findings for both gender groups, is in line with the predictions of the theory of ability misperception. (25) We turn now to our results for female Hispanics, Latinas, relative to female white workers.

Latinas: Returns to Tenure and Experience

Panel B of Table 5 presents our estimates of the impact of prior workplace experience and tenure on wages for Latina and white female workers while coefficient estimates for all of the variables are presented in Appendix H. Inspection of Table 5 reveals that prior experience is significantly related to the wages for females who are Latina for the full sample of Latino women and for those in the low-achievement subsample, while the wages of white women are not significantly affected by increased prior work experience. Greater seniority with the current employer significantly raises the wage of white female workers, and this is true also for female workers who are Latino. Moreover, wage returns to tenure are substantially larger for the average Latina employee (4.1%) than for the typical white female worker (3.0%).

The estimated wage return to potential prior experience for Latinas is statistically significant in nine of the 16 workplace settings, while the wage gain associated with greater tenure is statistically significant in 13 of the 16 workplace environments (see Table 8). In two job settings (Latino(a) Coworkers and Latino(a) Supervisor), there are too few white females to estimate the returns to experience. Thus, we can compare inter-ethnic (white vs. Latino) differences in returns to tenure and prior experience for females in 14 workplace settings. Inspection of Table 7 reveals that in only five of these 14 settings do we find evidence consistent with Proposition 1. However, whenever Proposition 1 holds, the evidence is also consistent with Proposition 3, which we consider strong support for the theory of ability misperception.

5. Lazear Revisited and Some Concluding Remarks

Lazear Revisited." The Effectiveness of Affirmative Action in Lowering the Black/White Wage Gap

Our findings do not support Lazear's (1979) prediction that the wage gain for additional seniority will be greater for white workers than for nonwhite workers following the introduction of federal affirmative action legislation. Lazear (1979, pp. 556) assumes that future earnings power at a point in time is lower for nonwhite workers than for white workers. However, he is vague on whether the source of this disparity is less human capital accumulation, the acquisition of less productive human capital, or lower ability. According to Lazear, the presumed lower productivity of nonwhite workers leads employers to pay them less than comparable white workers over a period of time.

Moreover, it can be deduced from Lazear that government affirmative action programs help nonwhite workers get a foot in the door and make it more difficult for firms to engage in entry-level wage differences. Thus, he expects firms to respond to affirmative action by paying equivalent wages to entry-level white and nonwhite workers, but also by providing nonwhite employees fewer opportunities to acquire skills on the job. This ultimately re-establishes the racial or ethnic wage gap that existed prior to the legislation (Lazear 1979, pp. 553). Therefore, he predicts the wage growth fostered by greater tenure will be smaller for nonwhite employees than for white workers and offers evidence in line with this viewpoint.

Whenever

{[partial derivative][w.sup.nw.sub.k]/[partial derivative]([TEN].sup.nw.sub.k]) - [partial derivative][w.sup.w.sub.k]/[partial derivative]([TEN].sup.w.sub.k])} = {B} > 0,

our findings are at odds with Lazear's hypothesis. Our comparison of the wage gain associated with additional tenure provides evidence that {B} > 0 for white female workers relative to black females in 12 of the 14 workplace environments examined. (26) We report similar results for Latino males compared to white males: {B} > 0 in 10 of the 12 workplace environments examined. However, when comparing white and black male workers evidence of {B} > 0 only emerges in four of the 13 workplace setting evaluated. Nevertheless, taken as a whole, our findings cast doubt on Lazear's (1979) claim that affirmative action gives rise to relatively low wage gains due to additional time on the current job for nonwhite workers.

Some Concluding Remarks

The theory of ability misperception (Farmer and Terrell 1996; Lewis and Terrell 2001) is a view of the labor market which promotes two ideas: that nonwhite workers face employers who systematically discriminate against them by undervaluing their skills acquired "on the job" with previous employers (Proposition 1) during the initial phase of employment. As nonwhite workers demonstrate their ability to perform while gaining seniority, the application of negative treatment towards them will diminish, leading to less disparity in interracial and inter-ethnic returns to tenure than for prior experience (Proposition 2). In addition, we examine another possibility--that as tenure advances, the wage gain actually will be greater for nonwhite than for white workers (Proposition 3) even though white workers receive a greater return to prior experience.

We evaluated each of these propositions using data on black, Latino, and white workers, separately for women and men, drawn from the MCSUI. The lessons from our empirical exploration are twofold. First, nonwhites, regardless of job setting, receive relatively poor returns to prior workplace experience (the lone exception is Latinas, for whom this is only true in a quarter of the worksites examined). Second, nonwhites typically receive greater wage gains for accumulating additional tenure than whites. Thus, our findings are largely consistent with Proposition 1, Proposition 2, and Proposition 3, and hence for the theory of ability misperception. Our findings are supportive of affirmative action programs that enable nonwhite workers to at least get "a foot in the door," and, hence, afford employers the opportunities to adjust their erroneous perceptions of nonwhite productivity held prior to observing them in the workplace.

Appendix A
Sample Creation: Respondents Coded by Interviewer
Latino, Latina, or White

 Latino and Latina
 (n = 1639)

 Males Females
Source of Data Deletion (n = 655) (n = 984)

Detroit residents 0 0
Older than 64 years of age 21 44
Self-employed 51 40
Retired or out of labor
 force for past five years 40 310
Refused to provide, don't
 know, or missing earnings data 18 23
Earn more than $100,000 per year 0 0
Refused to provide, don't know,
 or missing hours worked per week 0 1
Worked more than
 95 hours per week 1 2
Hourly wage greater than
 $100 or less than $2 5 12
Missing/invalid data on
 any human capital (including
 tenure) variable 10 8
Missing/invalid data on
 any work place
 setting variable 8 21
Missing/invalid data
 on any occupation code 14 4
Missing/invalid data on any
 coworker ethnicity variable 3 5
Total respondents remaining 494 514

Source: MCSUI. Detroit participants were not asked questions about
coworker ethnicity, a control variable in regressions. Persons
earning more than $100,000 in Atlanta were eliminated by the data
collectors.

Appendix B
Sample Creation: Respondents Coded by
Interviewer as Black or White

 Black White
Source of Data Deletion (n = 1084) (n = 1288)

Detroit residents 269 328
Older than 64 years of age 115 1161
Self-employed 54 123
Retired or out of labor force
 for past five years 107 60
Refused to provide, don't know,
 or missing earnings data 81 73
Earn more than $100,000 per year 0 3
Refused to provide, don't know,
 or missing hours
 worked per week 0 2
Worked more than 95 hours per week 0 1
Hourly wage greater than
 $100 or less than $2 10 17
Missing/invalid data on any
 human capital (including
 tenure) variable 6 4
Missing/invalid data on
 any work place
 setting variable 8 4
Missing/invalid data on
 any occupation code 15 12
Missing/invalid data on any
 coworker ethnicity
 variable 0 7
Total respondents remaining 419 493

 Black White
Source of Data Deletion (n = 2092) (n = 1289)

Detroit residents 493 423
Older than 64 years of age 221 259
Self-employed 54 76
Retired or out of labor force
 for past five years 311 158
Refused to provide, don't know,
 or missing earnings data 111 66
Earn more than $100,000 per year 0 0
Refused to provide, don't know,
 or missing hours
 worked per week 4 2
Worked more than 95 hours per week 2 1
Hourly wage greater than
 $100 or less than $2 25 14
Missing/invalid data on any
 human capital (including
 tenure) variable 16 11
Missing/invalid data on
 any work place
 setting variable 30 14
Missing/invalid data on
 any occupation code 14 7
Missing/invalid data on any
 coworker ethnicity
 variable 5 4
Total respondents remaining 805 554

Source: MCSUI. Detroit participants were not asked questions about
coworker ethnicity, a control variable in regressions. Persons
earning more than $100,000 in Atlanta were eliminated by the
data collectors.

Appendix C
Summary Statistics for Variables Used in the Econometric
Analysis: Black and White Males

 Blacks

 High
 Full Sample Achievement
Variable (n = 419) (n = 86)

Hourly wage 12.76 (6.22) 11.42 (5.05)
Potential prior
 experience 10.82 (8.45) 5.19 (4.26)
Tenure 5.61 (6.57) 2.92 (2.46)
High school
 dropout 0.07 (0.25) 0.02 (0.15)
High school
 graduate 0.54 (0.50) 0.59 (0.49)
Community
 college graduate 0.14 (0.34) 0.17 (0.37)
Attended college 0.17 (0.37) 0.18 (0.39)
College graduate 0.10 (0.29) 0.04 (0.19)
High school
 achievement 0.30 (0.46) --
Self-esteem 3.31 (1.26) 3.58 (0.55)
Married 0.52 (0.50) 0.38 0.49
Number of
 dependents 0.84 (1.25) 0.69 (1.13)
Disability 0.14 (0.35) 0.17 (0.38)
Foreign resident at
 16 years of age 0.25 (0.44) 0.38 (0.49)
Atlanta resident 0.28 (0.45) 0.25 (0.44)
Boston resident 0.13 (0.34) 0.12 (0.33)
Los Angeles
 resident 0.59 (0.49) 0.63 (0.49)
Union member 0.30 (0.46) 0.24 (0.43)
Work part-time 0.19 (0.39) 0.45 (0.50)
Firm size 0.54 (1.25) 0.59 (0.98)
Supervise others 0.33 (0.47) 0.17 (0.38)
Customer contact
 daily 0.61 (0.49) 0.73 (0.44)
Manager or
 professional 0.18 (.38) 0.21 (.41)
Craft 0.33 (0.47) 0.46 (0.50)
Services 0.26 (0.44) 0.26 (0.44)
Production 0.07 (0.26) 0.02 (0.16)
Laborers 0.15 (0.36) 0.04 (0.19)
Government 0.18 (0.39) 0.14 (0.35)
White coworkers 0.37 (0.48) 0.29 (0.46)
Black coworkers 0.30 (0.46) 0.19 (0.40)
Hispanic
 coworkers 0.14 (0.34) 0.14 (0.34)
Asian coworkers 0.00 (0.07) 0.00 (0.04)
Coworkers of
 other races 0.19 (0.39) 0.38 (0.49)
Supervisor white 0.60 (0.49) 0.68 (0.49)
Supervisor black 0.22 (0.42) 0.18 (0.39)
Supervisor
Hispanic 0.06 (0.23) 0.03 (0.16)
Supervisor Asian 0.008 (0.09) 0.003 (0.06)
Supervisor of
 other race 0.04 (0.19) 0.10 (0.30)

 Blacks Whites

 Low Full
 Achievement Sample
Variable (n = 333) (n = 493)

Hourly wage 13.34 (6.59) 15.95 (7.79)
Potential prior
 experience 12.68 (9.01) 11.62 (10.01)
Tenure 6.79 (7.41) 6.28 (7.46)
High school
 dropout 0.08 (0.28) 0.03 (0.16)
High school
 graduate 0.51 (0.50) 0.36 (0.48)
Community
 college graduate 0.12 (0.33) 0.15 (0.36)
Attended college 0.16 (0.37) 0.31 (0.46)
College graduate 0.12 (0.33) 0.16 (0.37)
High school
 achievement -- 0.31 (0.46)
Self-esteem 3.19 (1.45) 3.35 (1.35)
Married 0.58 (0.49) 0.61 (0.49)
Number of
 dependents 0.91 (1.29) 0.61 (0.95)
Disability 0.13 (0.34) 0.12 (0.33)
Foreign resident at
16 years of age 0.20 (0.40) 0.05 (0.22)
Atlanta resident 0.29 (0.45) 0.13 (0.34)
Boston resident 0.14 (0.34) 0.38 (0.49)
Los Angeles
 resident 0.57 (0.50) 0.48 (0.50)
Union member 0.33 (0.47) 0.24 (0.42)
Work part-time 0.07 (0.26) 0.09 (0.29)
Firm size 0.52 (1.35) 0.56 (1.49)
Supervise others 0.40 (0.49) 0.35 (0.48)
Customer contact
 daily 0.59 (0.50) 0.53 (0.50)
Manager or
 professional 0.16 (.37) 0.42 (.49)
Craft 0.28 (0.45) 0.25 (0.43)
Services 0.26 (0.44) 0.09 (0.29)
Production 0.09 (0.29) 0.13 (0.33)
Laborers 0.20 (0.40) 0.11 (0.31)
Government 0.20 (0.40) 0.19 (0.39)
White coworkers 0.40 (0.49) 0.79 (0.41)
Black coworkers 0.35 (0.48) 0.04 (0.19)
Hispanic
 coworkers 0.14 (0.34) 0.08 (0.26)
Asian coworkers 0.01 (0.07) 0.01 (0.10)
Coworkers of
 other races 0.11 (0.32) 0.09 (0.28)
Supervisor white 0.57 (0.50) 0.80 (0.40)
Supervisor black 0.24 (0.43) 0.03 (0.18)
Supervisor
 Hispanic 0.07 (0.25) 0.03 (0.17)
Supervisor Asian 0.01 (0.10) 0.03 (0.16)
Supervisor of
 other race 0.01 (0.10) 0.01 (0.10)

 Whites

 High Low
 Achievement Achievement
Variable (n = 137) (n = 356)

Hourly wage 13.81 (5.51) 16.90 (8.45)
Potential prior
 experience 4.61 (4.86) 14.17 (10.17)
Tenure 2.64 (2.78) 7.90 (8.29)
High school
 dropout 0.00 (0.06) 0.04 (0.19)
High school
 graduate 0.24 (0.43) 0.41 (0.49)
Community
 college graduate 0.13 (0.34) 0.16 (0.37)
Attended college 0.48 (0.50) 0.23 (0.42)
College graduate 0.15 (0.36) 0.17 (0.38)
High school
 achievement -- --
Self-esteem 3.75 (0.78) 3.17 (1.51)
Married 0.45 (0.50) 0.69 (0.46)
Number of
 dependents 0.33 (0.78) 0.75 (1.00)
Disability 0.06 (0.24) 0.15 (0.36)
Foreign resident at
16 years of age 0.02 (0.13) 0.06 (0.24)
Atlanta resident 0.13 (0.34) 0.13 (0.34)
Boston resident 0.42 (0.49) 0.37 (0.48)
Los Angeles
 resident 0.45 (0.50) 0.51 (0.50)
Union member 0.12 (0.33) 0.29 (0.45)
Work part-time 0.09 (0.29) 0.09 (0.29)
Firm size 0.24 (0.65) 0.70 (1.72)
Supervise others 0.20 (0.40) 0.43 (0.50)
Customer contact
 daily 0.46 (0.50) 0.56 (0.50)
Manager or
 professional 0.48 (.50) 0.40 (.49)
Craft 0.35 (0.48) 0.20 (.40)
Services 0.06 (0.24) 0.10 (0.31)
Production 0.07 (0.25) 0.15 (0.36)
Laborers 0.05 (0.21) 0.14 (0.34)
Government 0.17 (0.38) 0.19 (0.40)
White coworkers 0.83 (0.38) 0.77 (0.42)
Black coworkers 0.05 (0.22) 0.03 (0.18)
Hispanic
 coworkers 0.07 (0.25) 0.08 (0.27)
Asian coworkers 0.01 (0.12) 0.007 (0.08)
Coworkers of
 other races 0.04 (0.19) 0.11 (0.31)
Supervisor white 0.81 (0.40) 0.80 (0.40)
Supervisor black 0.04 (0.20) 0.03 (0.17)
Supervisor
 Hispanic 0.01 (0.11) 0.04 (0.19)
Supervisor Asian 0.05 (0.22) 0.02 (0.12)
Supervisor of
 other race 0.00 (0.04) 0.01 (0.12)

Source: MCSUI. Weighted means are reported with their standard
errors in parentheses.

Appendix D
Definition of Variables Used in the Econometric Analysis
and Associated Summary Statistics

Variable Variable Definition

Hourly wage Respondents hourly wage at survey date

Tenure Number of years employed by current
 employer at survey date

Potential Age at last year of schooling (out of
 prior experience school at least 15 continuous
 months)-Tenure

High school dropout 1 if respondent failed to complete
 high school, 0 otherwise

High school graduate 1 if respondents highest level of
 schooling is completion of high
 school, 0 otherwise

Community college 1 if respondents highest level of
 graduate schooling is completion of
 community college, 0 otherwise

Attended college 1 if respondents highest level of
 schooling was attended college,
 0 otherwise

College graduate 1 if respondent completed college,
 0 otherwiseVariable

High school achievement 1 if respondent's average grade in
 high school was B or better,
 0 otherwise

Self-esteem Rosenberg self-esteem score; scores
 range in ascending order from
 0 to 4

Married 1 if respondent is married or living
 with a partner, 0 otherwise

Number of dependents Number of dependents in the household

Disability 1 if respondent has a work-limiting
 health condition, 0 otherwise

Foreign resident at 1 if respondent was primarily a
 16 years of age foreign resident before 16 years of
 age, 0 otherwise

Los Angeles resident 1 if respondent resides in Los Angeles,
 0 otherwise

Atlanta resident 1 if respondent resides in Atlanta,
 0 otherwise

Boston resident 1 if respondent resides in Boston,
 0 otherwise

Union member 1 if respondent is a union member,
 0 otherwise

Work part-time 1 if respondent works part-time,
 0 otherwise

Customer contact daily 1 if respondent has customer contact
 daily, 0 otherwise

Firm size Number of workers at respondents firm
 per 1000

Supervise others 1 if respondent supervises other
 employees, 0 otherwise

Manager or 1 if respondent is in a managerial
 professional or professional occupation,
 0 otherwise

Production 1 if respondent is in a precision
 production, craft, or repair
 occupation, 0 otherwise

Services 1 if respondent is in a service
 occupation, 0 otherwise

Craft 1 if respondent is in a craft
 occupation, 0 otherwise

Laborers 1 if respondent is in a laborer
 occupation, 0 otherwise

Government 1 if respondent is in a public
 employee, 0 otherwise

White coworkers 1 if respondents coworkers are mostly
 white, 0 otherwise

Black coworkers 1 if respondents coworkers are mostly
 black, 0 otherwise

Hispanic coworkers 1 if respondents coworkers are mostly
 Hispanic, 0 otherwise

Asian coworkers 1 if respondents coworkers are mostly
 Asian, 0 otherwise

Coworkers of other races 1 if respondents coworkers are mostly
 of other races, 0 otherwise

Supervisor white 1 if respondents supervisor is white,
 0 otherwise

Supervisor black 1 if respondents supervisor is black,
 0 otherwise

Supervisor Hispanic 1 if respondents supervisor is Hispanic,
 0 otherwise

Supervisor Asian 1 if respondents supervisor is Asian,
 0 otherwise

Supervisor of 1 if respondents supervisor is of other
 other race race, 0 otherwise

Black 1 if respondents is black, 0 otherwise

Hispanic 1 if respondents is Hispanic, 0 otherwise

Source: MCSUI.

Appendix E
Wage Determination Estimates Pooled by Race and
Stratified by Race: Males and Females

 Males

Variable Full Sample Blacks
 (n = 912) (n = 419)

Black -0.062 ** (0.031)
Tenure 0.044 *** (0.005) 0.040 *** (0.008)
Tenure squared * 100 -0.087 *** (0.019) -0.076 ** (0.029)
Potential prior
 experience 0.010 ** (0.004) 0.000 (0.006)
Potential prior
 experience
 squared * 100 -0.016 (0.011) 0.013 (0.018)
High school dropout -0.142 *** (0.049) -0.149 ** (0.063)
Community college
 graduate 0.113 *** (0.039) 0.119 ** (0.055)
Attended college 0.207 *** (0.041) 0.233 *** (0.066)
College graduate 0.372 *** (0.054) 0.524 *** (0.101)
High school
 achievement 0.024 (0.035) 0.055 (0.052)
Self-esteem 0.002 (0.012) 0.023 (0.018)
Married 0.127 *** (0.030) 0.101 ** (0.044)
Number of dependents 0.004 (0.013) -0.007 (0.017)
Disability -0.114 *** (0.038) -0.129 ** (0.054)
Foreign resident at
 16 years of age -0.003 (0.043) 0.063 (0.060)
Atlanta resident -0.114 *** (0.036) -0.083 (0.050)
Boston resident -0.177 *** (0.036) -0.189 *** (0.056)
Union member 0.143 *** (0.036) 0.168 *** (0.051)
Work part-time -0.065 (0.045) -0.040 (0.068)
Firm size 0.006 (0.009) 0.006 (0.013)
Supervise others 0.062 * (0.029) 0.029 (0.044)
Daily contact with
 customers -0.065 ** (0.027) -0.060 (0.042)
Craft -0.090 ** (0.039) -0.033 (0.067)
Services -0.222 *** (0.047) -0.220 *** (0.068)
Production -0.038 (0.052) -0.062 (0.085)
Laborers -0.142 *** (0.048) -0.099 (0.072)
Government 0.029 (0.038) 0.115 ** (0.053)
Black coworkers -0.132 *** (0.040) -0.104 ** (0.048)
Hispanic coworkers -0.197 *** (0.051) -0.074 (0.071)
Asian coworkers -0.035 (0.131) 0.063 (0.202)
Coworkers of other
 races -0.074 (0.048) -0.024 (0.074)
Supervisor white -0.050 (0.045) 0.029 (0.064)
Supervisor black -0.102 * (0.056) -0.074 (0.069)
Supervisor Hispanic -0.091 (0.073) -0.025 (0.095)
Supervisor Asian -0.196 ** (0.093) -0.312 ** (0.146)
Supervisor of other
 race -0.146 (0.128) -0.082 (0.171)
Constant 2.331 *** (0.087) 2.130 *** (0.133)
F statistic for the
 equation 21.36 *** [0.000] 10.23 *** [0.000]
Adjusted [r.sup.2] .45 .44

 Males Females

Variable Whites Full Sample
 (n = 493) (n = 1359)

Black -0.003 (0.028)
Tenure 0.045 *** (0.008) 0.044 *** (0.005)
Tenure squared * 100 -0.089 *** (0.025) -0.089 *** (0.018)
Potential prior
 experience 0.016 *** (0.006) 0.001 (0.004)
Potential prior
 experience
 squared * 100 -0.034 ** (0.015) 0.000 (0.001)
High school dropout -0.130 (0.082) -0.234 *** (0.040)
Community college
 graduate 0.100 * (0.057) 0.111 *** (0.030
Attended college 0.185 *** (0.055) 0.279 *** (0.035)
College graduate 0.329 *** (0.067) 0.360 *** (0.052)
High school
 achievement 0.029 (0.050) 0.017 (0.029)
Self-esteem -0.009 (0.016) 0.001 (0.009)
Married 0.140 *** (0.042) 0.011 (0.024)
Number of dependents 0.024 (0.021) -0.003 (0.010)
Disability -0.108 ** (0.054) -0.111 *** (0.031)
Foreign resident at
 16 years of age -0.104 (0.065) -0.047 (0.041)
Atlanta resident -0.150 *** (0.054) -0.142 *** (0.028)
Boston resident -0.192 *** (0.049) -0.046 (0.030)
Union member 0.128 ** (0.053) 0.083 *** (0.031)
Work part-time -0.107 * (0.063) -0.008 (0.032)
Firm size 0.006 (0.013) 0.022 *** (0.007)
Supervise others 0.077 * (0.040) 0.075 *** (0.026)
Daily contact with
 customers -0.070 * (0.038) -0.062 *** (0.024)
Craft -0.123 ** (0.050) -0.168 *** (0.031)
Services -0.216 *** (0.071) -0.268 *** (0.038)
Production -0.034 (0.069) -0.276 *** (0.079)
Laborers -0.188 *** (0.070) -0.203 *** (0.055)
Government -0.033 (0.057) 0.000 (0.030)
Black coworkers -0.106 (0.087) -0.097 *** (0.032)
Hispanic coworkers -0.302 *** (0.075) -0.161 *** (0.041)
Asian coworkers 0.085 (0.179) 0.039 (0.073)
Coworkers of other
 races -0.091 (0.067) 0.012 (0.039)
Supervisor white -0.137 ** (0.066) 0.013 (0.040)
Supervisor black -0.143 (0.112) -0.067 * (0.045)
Supervisor Hispanic -0.128 (0.116) -0.116 ** (0.065)
Supervisor Asian -0.191 (0.126) 0.011 (0.079)
Supervisor of other
 race -0.209 (0.980) 0.145 (0.104)
Constant 2.451 *** (0.118) 2.237 *** (0.076)
F statistic for the
 equation 10.42 *** [0.000] 32.39 *** [0.0001
Adjusted [r.sup.2] .40 .45

 Females

Variable Blacks Whites
 (n = 805) (n = 554)
Black
Tenure 0.046 *** (0.006) 0.042 *** (0.009)
Tenure squared * 100 -0.093 *** (0.021) -0.091 *** (0.035)
Potential prior
 experience -0.003 (0.005) 0.007 (0.006)
Potential prior
 experience
 squared * 100 0.015 (0.013) -0.019 (0.015)
High school dropout -0.207 *** (0.047) -0.252 *** (0.075)
Community college
 graduate 0.077 ** (0.037) 0.158 *** (0.052)
Attended college 0.285 *** (0.049) 0.254 *** (0.053)
College graduate 0.482 *** (0.080) 0.290 *** (0.076)
High school
 achievement 0.010 (0.036) 0.020 (0.049)
Self-esteem 0.007 (0.012) -0.007 (0.015)
Married 0.047 (0.032) -0.029 (0.037)
Number of dependents -0.005 (0.011) 0.000 (0.018)
Disability -0.130 ** (0.038) -0.082 (0.055)
Foreign resident at
 16 years of age 0.014 (0.054) -0.133 * (0.068)
Atlanta resident -0.187 *** (0.036) -0.088 * (0.048)
Boston resident -0.039 (0.041) -0.065 (0.047)
Union member 0.108 *** (0.037) 0.047 (0.058)
Work part-time 0.011 (0.044) -0.007 (0.048)
Firm size 0.021 ** (0.008) 0.023 * (0.012)
Supervise others 0.071 ** (0.034) 0.072 * (0.042)
Daily contact with
 customers -0.032 (0.030) -0.124 *** (0.041)
Craft -0.150 *** (0.042) -0.188 *** (0.048)
Services -0.255 *** (0.049) -0.284 *** (0.078)
Production -0.247 * (0.118) -0.283 ** (0.110)
Laborers -0.164 ** (0.067) -0.254 ** (0.103)
Government 0.018 (0.035) -0.024 (0.055)
Black coworkers -0.092 *** (0.035) 0.028 (0.093)
Hispanic coworkers -0.1039 (0.059) -0.236 *** (0.075)
Asian coworkers 0.026 (0.101) 0.047 (0.108)
Coworkers of other
 races -0.005 (0.058) -0.013 (0.056)
Supervisor white -0.003 (0.054) 0.020' (0.061)
Supervisor black -0.087 (0.055) 0.036 (0.102)
Supervisor Hispanic -0.079 (0.082) -0.220 ** (0.111)
Supervisor Asian 0.027 (0.109) 0.051 (0.119)
Supervisor of other
 race 0.161 (0.119) 0.066 (0.220)
Constant 2.195 *** (0.098) 2.332 *** (0.126)
F statistic for the
 equation 22.46 *** [0.000] 10.79 *** [0.000]
Adjusted [r.sup.2] .48 .38

Source: MCSUI. Coefficients are reported with standard errors
in parentheses. *** Statistically significant at .01 level
for two-tailed test, ** statistically significant at the .05
level for two-tailed test, and * statistically significant
at the .1 level for a two-tailed test. A description of how
each variable is constructed appears in Appendix D.

Appendix F
Summary Statistics for Variables Used in the Econometric Analysis:
White and Black Females

 Blacks

 Full High
 Sample Achievement
Variables (n = 805) (n = 216)

Ln W 11.09 (6.34) 9.55 (4.43)
Potential prior experience 12.61 (9.43) 9.10 (4.08)
Tenure 6.13 (6.86) 2.62 (2.78)
High school dropout 0.06 (0.23) 0.03 (0.18)
High school graduate 0.46 (0.50) 0.41 (0.49)
Community college graduate 0.29 (0.45) 0.33 (0.47)
Attended college 0.14 (0.35) 0.17 (0.38)
College graduate 0.06 (0.24) 0.05 (0.22)
High school achievement 0.34 (0.47) --
Self-esteem 3.26 (1.41) 3.75 (0.71)
Married 0.37 (0.48) 0.28 (0.45)
Number of dependents 0.92 (1.23) 0.95 (1.31)
Disability 0.11 (0.32) 0.07 (0.25)
Foreign resident at 16 years
 of age 0.07 (0.25) 0.06 (0.24)
Atlanta 0.32 (0.47) 0.26 (0.44)
Boston 0.11 (0.32) 0.10 (0.30)
Los Angeles 0.57 (0.50) 0.64 (0.48)
Union member 0.24 (0.43) 0.16 (0.37)
Work part-time 0.14 (0.35) 0.22 (0.42)
Firm size 0.59 (1.61) 0.41 (1.46)
Supervise others 0.26 (0.44) 0.29 (0.45)
Customer contact daily 0.66 (0.47) 0.71 (0.45)
Manager or professional 0.21 (.41) 0.14 (.34)
Craft 0.52 (0.50) 0.63 (0.48)
Services 0.21 (0.41) 0.21 (0.41)
Production 0.01 (0.12) 0.01 (0.08)
Laborers 0.04 (0.20) 0.02 (0.13)
Government 0.25 (0.44) 0.19 (0.40)
White coworkers 0.39 (0.49) 0.45 (0.50)
Black coworkers 0.41 (0.49) 0.39 (0.49)
Latino(a) coworkers 0.07 (0.27) 0.06 (0.23)
Asian coworkers 0.04 (0.21) 0.04 (0.18)
Other coworkers 0.07 (0.26) 0.07 (0.25)
Supervisor white 0.54 (0.50) 0.60 (0.49)
Supervisor black 0.30 (0.46) 0.30 (0.46)
Supervisor Latino(a) 0.05 (0.22) 0.07 (0.26)
Supervisor Asian 0.01 (0.12) 0.003 (0.05)
Supervisor other 0.01 (0.11) 0.01 (0.11)

 Blacks Whites

Variables Low Full
 Achievement Sample
 (n = 589) (n = 554)

Ln W 11.87 (6.99) 12.92 (6.48)
Potential prior experience 14.40 (9.81) 12.98 (10.20)
Tenure 7.91 (7.60) 6.29 (7.02)
High school dropout 0.07 (0.26) 0.05 (0.23)
High school graduate 0.48 (0.50) 0.40 (0.48)
Community college graduate 0.26 (0.44) 0.22 (0.41)
Attended college 0.12 (0.33) 0.27 (0.44)
College graduate 0.06 (0.24) 0.09 (0.29)
High school achievement -- 0.31 (0.46)
Self-esteem 3.01 (1.60) 3.33 (1.36)
Married 0.41 (0.49) 0.68 (0.47)
Number of dependents 0.91 (1.18) 0.83 (1.05)
Disability 0.14 (0.34) 0.13 (0.34)
Foreign resident at 16 years
 of age 0.07 (0.26) 0.05 (0.22)
Atlanta 0.35 (0.48) 0.17 (0.38)
Boston 0.12 (0.32) 0.38 (0.49)
Los Angeles 0.53 (0.50) 0.44 (0.50)
Union member 0.28 (0.45) 0.17 (0.37)
Work part-time 0.10 (0.31) 0.20 (0.40)
Firm size 0.67 (1.68) 0.56 (1.52)
Supervise others 0.25 (0.43) 0.29 (0.45)
Customer contact daily 0.63 (0.48) 0.63 (0.48)
Manager or professional 0.25 (.43) 0.42 (.49)
Craft 0.47 (0.43) 0.44 (0.50)
Services 0.21 (0.41) 0.09 (0.29)
Production 0.02 (0.13) 0.01 (0.12)
Laborers 0.06 (0.23) 0.03 (0.18)
Government 0.28 (0.45) 0.17 (0.38)
White coworkers 0.36 (0.48) 0.80 (0.40)
Black coworkers 0.42 (0.49) 0.03 (0.17)
Latino(a) coworkers 0.09 (0.29) 0.05 (0.22)
Asian coworkers 0.05 (0.22) 0.03 (0.17)
Other coworkers 0.08 (0.27) 0.09 (0.28)
Supervisor white 0.52 (0.50) 0.80 (0.40)
Supervisor black 0.31 (0.46) 0.04 (0.19)
Supervisor Latino(a) 0.04 (0.19) 0.02 (0.15)
Supervisor Asian 0.02 (0.14) 0.04 (0.19)
Supervisor other 0.01 (0.10) 0.01 (0.09)

 Whites

Variables High Low
 Achievement Achievement
 (n = 161) (n = 393)

Ln W 12.00 (4.94) 13.34 (7.03)
Potential prior experience 5.21 (5.94) 16.45 (9.79)
Tenure 3.43 (3.38) 7.56 (7.81)
High school dropout 0.00 (0.03) 0.08 (0.27)
High school graduate 0.34 (0.48) 0.38 (0.49)
Community college graduate 0.21 (0.41) 0.22 (0.42)
Attended college 0.36 (0.48) 0.22 (0.42)
College graduate 0.09 (0.28) 0.09 (0.29)
High school achievement -- --
Self-esteem 3.77 (0.84) 3.13 (1.49)
Married 0.65 (0.48) 0.70 (0.46)
Number of dependents 0.82 (0.99) 0.84 (1.07)
Disability 0.07 (0.26) 0.16 (0.36)
Foreign resident at 16 years
 of age 0.04 (0.20) 0.05 (0.23)
Atlanta 0.17 (0.37) 0.18 (0.38)
Boston 0.46 (0.50) 0.35 (0.48)
Los Angeles 0.37 (0.48) 0.48 (0.50)
Union member 0.17 (0.37) 0.17 (0.37)
Work part-time 0.20 (0.40) 0.20 (0.40)
Firm size 0.67 (1.76) 0.51 (1.40)
Supervise others 0.27 (0.45) 0.30 (0.46)
Customer contact daily 0.52 (0.50) 0.67 (0.47)
Manager or professional 0.43 (.50) 0.41 (.49)
Craft 0.46 (0.50) 0.44 (.50)
Services 0.11 (0.31) 0.09 (0.28)
Production 0.00 (0.05) 0.02 (0.13)
Laborers 0.00 (0.00) 0.05 (0.21)
Government 0.18 (0.38) 0.17 (0.38)
White coworkers 0.87 (0.33) 0.77 (0.42)
Black coworkers 0.01 (0.11) 0.04 (0.18)
Latino(a) coworkers 0.01 (0.09) 0.07 (0.25)
Asian coworkers 0.03 (0.17) 0.03 (0.18)
Other coworkers 0.08 (0.27) 0.09 (0.29)
Supervisor white 0.80 (0.40) 0.80 (0.40)
Supervisor black 0.03 (0.18) 0.04 (0.19)
Supervisor Latino(a) 0.01 (0.08) 0.03 (0.17)
Supervisor Asian 0.08 (0.27) 0.02 (0.14)
Supervisor other 0.01 (0.11) 0.01 (0.08)

Source: MCSUI. Weighted means are reported with their standard
errors in parentheses.

Appendix G
Summary Statistics for Variables Used in the
Econometric Analysis: Latinos and Latinas *

 Latinos

Variables Full High
 Sample Achievement
 (n = 494) (n = 128)

W 9.08 (4.67) 8.30 (3.52)
Potential
 prior experience 12.79 (10.52) 6.91 (5.67)
Tenure 4.49 (4.61) 3.62 (3.68)
High school dropout 0.45 (0.50) 0.33 (0.47)
High school graduate 0.37 (0.48) 0.45 (0.50)
Community college graduate 0.07 (0.25) 0.12 (0.33)
Attended college 0.09 (0.29) 0.10 (0.30)
College graduate 0.02 (0.13) 0.00 (0.00)
High school achievement 0.30 (0.46)
Self-esteem 3.14 (1.36) 3.26 (1.26)
Married 0.69 (0.46) 0.63 (0.48)
Number of dependents 1.37 (1.42) 1.07 (1.20)
Disability 0.08 (0.27) 0.02 (0.16)
Foreign resident
 at 16 years of age 0.70 (0.46) 0.75 (0.43)
Atlanta 0.01 (0.08) 0.00 (0.00)
Boston 0.05 (0.21) 0.06 (0.23)
Los Angeles 0.94 (0.23) 0.94 (0.23)
Union 0.20 (0.40) 0.18 (0.39)
Work part-time 0.15 (0.36) 0.22 (0.42)
Firm size 0.14 (0.42) 0.08 (0.28)
Supervise others 0.23 (0.42) 0.21 (0.41)
Customer contact daily 0.40 (0.49) 0.39 (0.49)
Manager or professional 0.09 (.29) 0.10 (.30)
Craft 0.16 (0.37) 0.18 (0.39)
Services 0.15 (0.36) 0.17 (0.38)
Production 0.18 (0.39) 0.18 (0.38)
Laborers 0.41 (0.49) 0.37 (0.49)
Government 0.06 (0.23) 0.08 (0.27)
White coworkers 0.19 (0.39) 0.19 (0.39)
Black coworkers 0.03 (0.17) 0.03 (0.16)
Latino(a) coworkers 0.72 (0.45) 0.74 (0.44)
Asian coworkers 0.03 (0.17) 0.02 (0.14)
Other coworkers 0.03 (0.16) 0.02 (0.14)
Supervisor white 0.44 (0.50) 0.36 (0.48)
Supervisor black 0.03 (0.16) 0.03 (0.17)
Supervisor Latino(a) 0.39 (0.49) 0.42 (0.49)
Supervisor Asian 0.04 (0.20) 0.04 (0.18)
Supervisor other 0.02 (0.15) 0.03 (0.18)

 Latinos Latinas

Variables Low Full
 Achievement Sample
 (n = 366) (n = 514)

W 9.42 (5.06) 7.56 (3.86)
Potential
 prior experience 15.35 (11.12) 13.70 (10.13)
Tenure 4.87 (4.91) 4.73 (5.60)
High school dropout 0.51 (0.50) 0.48 (0.50)
High school graduate 0.34 (0.47) 0.31 (0.46)
Community college graduate 0.04 (0.20) 0.13 (0.34)
Attended college 0.09 (0.28) 0.06 (0.24)
College graduate 0.03 (0.16) 0.01 (0.12)
High school achievement 0.22 (0.41)
Self-esteem 3.1 (1.40) 3.33 (1.20)
Married 0.71 (0.45) 0.55 (0.50)
Number of dependents 1.51 (1.49) 1.73 (1.37)
Disability 0.11 (0.31) 0.12 (0.32)
Foreign resident
 at 16 years of age 0.67 (0.47) 0.70 (0.46)
Atlanta 0.01 (0.09) 0.01 (0.10)
Boston 0.04 (0.21) 0.05 (0.22)
Los Angeles 0.95 (0.22) 0.94 (0.24)
Union 0.20 (0.40) 0.13 (0.34)
Work part-time 0.12 (0.33) 0.17 (0.37)
Firm size 0.16 (0.46) 0.18 (0.65)
Supervise others 0.24 (0.43) 0.16 (0.37)
Customer contact daily 0.40 (0.49) 0.46 (0.50)
Manager or professional 0.09 (.28) 0.10 (.30)
Craft 0.15 (0.36) 0.29 (0.46)
Services 0.14 (0.35) 0.26 (0.44)
Production 0.19 (0.39) 0.09 (0.29)
Laborers 0.43 (0.50) 0.26 (0.44)
Government 0.05 (0.21) 0.10 (0.30)
White coworkers 0.19 (0.39) 0.22 (0.41)
Black coworkers 0.03 (0.18) 0.01 (0.10)
Latino(a) coworkers 0.71 (0.45) 0.70 (0.46)
Asian coworkers 0.04 (0.19) 0.03 (0.17)
Other coworkers 0.03 (0.17) 0.04 (0.20)
Supervisor white 0.47 (0.50) 0.34 (0.48)
Supervisor black 0.02 (0.15) 0.03 (0.18)
Supervisor Latino(a) 0.38 (0.49) 0.40 (0.49)
Supervisor Asian 0.05 (0.21) 0.05 (0.21)
Supervisor other 0.02 (0.13) 0.04 (0.20)

 Latinas

Variables High Low
 Achievement Achievement
 (n = 150) (n = 364)

W 8.21 (4.85) 7.34 (3.55)
Potential
 prior experience 6.42 (5.67) 15.72 (10.18)
Tenure 2.85 (2.56) 5.25 (6.09)
High school dropout 0.24 (0.43) 0.55 (0.50)
High school graduate 0.42 (0.49) 0.28 (0.45)
Community college graduate 0.18 (0.39) 0.12 (0.32)
Attended college 0.14 (0.35) 0.04 (0.19)
College graduate 0.01 (0.09) 0.02 (0.12)
High school achievement
Self-esteem 3.77 (0.46) 3.20 (1.31)
Married 0.50 (0.50) 0.56 (0.50)
Number of dependents 1.61 (1.23) 1.77 (1.41)
Disability 0.14 (0.35) 0.11 (0.32)
Foreign resident
 at 16 years of age 0.69 (0.46) 0.70 (0.46)
Atlanta 0.02 (0.14) 0.01 (0.09)
Boston 0.05 (0.22) 0.05 (0.22)
Los Angeles 0.93 (0.26) 0.94 (0.24)
Union 0.13 (0.33) 0.13 (0.34)
Work part-time 0.16 (0.37) 0.17 (0.37)
Firm size 0.14 (0.37) 0.19 (0.71)
Supervise others 0.23 (0.42) 0.14 (0.34)
Customer contact daily 0.49 (0.50) 0.45 (0.50)
Manager or professional 0.19 (.39) 0.07 (.26)
Craft 0.29 (0.46) 0.30 (0.46)
Services 0.24 (0.43) 0.27 (0.44)
Production 0.08 (0.27) 0.09 (0.29)
Laborers 0.21 (0.41) 0.27 (0.45)
Government 0.11 (0.32) 0.10 (0.30)
White coworkers 0.27 (0.45) 0.20 (0.40)
Black coworkers 0.02 (0.14) 0.01 (0.09)
Latino(a) coworkers 0.63 (0.48) 0.72 (0.45)
Asian coworkers 0.01 (0.11) 0.03 (0.18)
Other coworkers 0.06 (0.24) 0.04 (0.19)
Supervisor white 0.34 (0.47) 0.35 (0.48)
Supervisor black 0.03 (0.18) 0.03 (0.18)
Supervisor Latino(a) 0.31 (0.46) 0.42 (0.49)
Supervisor Asian 0.11 (0.31) 0.03 (0.17)
Supervisor other 0.07 (0.26) 0.03 (0.18)

Source: MCSUI. Weighted means are reported with their standard
errors in parentheses.

Appendix H
Wage Determination Estimates for a Pooled Sample of White Males and
Latinos or White Females and Latinas, and for Subsamples of Latinos
or Latinas

 Males

 Full Sample Latinos
Variables (n = 987) (n = 494)

Latino(a) -0.100 ** (0.040)
Tenure 0.052 *** (0.005) 0.075 *** (0.010)
Tenure 2x 100 -0.120 *** (0.020) -0.232 *** (0.058)
Potential prior
 experience 0.011 *** (0.004) 0.008 (0.005)
Potential
 prior experience
 x 100 -0.021 ** (0.009) -0.014 (0.011)
High school dropout -0.155 *** (0.035) -0.142 *** (0.039)
Community college
 graduate 0.108 ** (0.044) 0.116 (0.078)
Attended college 0.130 *** (0.042) 0.013 (0.072)
College graduate 0.302 *** (0.055) 0.294 * (0.153)
High school achievement 0.009 (0.032) 0.013 (0.040)
Self-esteem -0.005 (0.011) 0.012 (0.015)
Married 0.120 *** (0.028) 0.097 ** (0.038)
Number of dependents 0.008 (0.011) 0.006 (0.012)
Disability -0.059 * (0.036) -0.020 (0.048)
Foreign resident
 at 16 years
 of age -0.120 *** (0.037) -0.144 *** (0.045)
Atlanta -0.082 ** (0.044) 0.107 (0.159)
Boston -0.061 ** (0.031) 0.049 (0.040)
Union 0.158 *** (0.032) 0.173 *** (0.041)
Work part-time -0.096 ** (0.038) -0.071 (0.048)
Firm size 0.011 (0.010) 0.020 (0.019)
Supervise others 0.066 ** (0.028) 0.023 (0.039)
Customer
 contact daily -0.033 (0.025) 0.029 (0.033)
Craft -0.144 *** (0.040) -0.209 *** (0.075)
Services -0.262 *** (0.049) -0.355 *** (0.077)
Production -0.105 ** (0.048) -0.207 *** (0.077)
Laborers -0.185 *** (0.046) -0.244 *** (0.073)
Government -0.013 (0.043) 0.028 (0.074)
Black Coworkers -0.133 ** (0.064) -0.139 (0.104)
Latino(a) Coworkers -0.206 *** (0.039) -0.100 ** (0.047)
Asian Coworkers -0.019 (0.100) 0.041 (0.117)
Other Coworkers -0.118 ** (0.052) -0.087 (0.087)
Supervisor white 0.007 (0.044) 0.164 *** (0.060)
Supervisor black 0.029 (0.074) 0.171 * (0.098)
Supervisor Latino(a) -0.049 (0.051) 0.051 (0.060)
Supervisor Asian -0.113 (0.074) -0.045 (0.089)
Supervisor other -0.083 (0.101) -0.023 (0.112)
Constant 2.295 *** (0.082) 1.965 *** (0.118)
F statistic for
 the equation 32.85 *** [0.000] 12.33 *** [0.000]
Adjusted [R.sup.2] .54 .45

 Females

 Full Sample Latinas
Variables (n = 1068) (n = 514)

Latino(a) 0.029 (0.039)
Tenure 0.048 *** (0.006) 0.052 *** (0.008)
Tenure 2x 100 -0.113 *** (0.024) -0.115 *** (0.035)
Potential prior
 experience 0.010 ** (0.004) 0.012 ** (0.005)
Potential
 prior experience
 x 100 -0.024 ** (0.010) -0.029 ** (0.012)
High school dropout -0.149 *** (0.037) -0.117 *** (0.049)
Community college
 graduate 0.113 *** (0.037) 0.001 (0.055)
Attended college 0.254 *** (0.043) 0.213 ** (0.090)
College graduate 0.264 *** (0.061) 0.186 (0.131)
High school achievement 0.003 (0.032) -0.011 (0.043)
Self-esteem 0.002 (0.011) 0.022 (0.016)
Married 0.015 (0.025) 0.064 * (0.033)
Number of dependents -0.009 (0.011) -0.014 (0.013)
Disability -0.045 (0.035) -0.026 (0.046)
Foreign resident
 at 16 years
 of age -0.111 *** (0.037) -0.120 *** (0.044)
Atlanta -0.027 (0.039) 0.266 (0.163)
Boston 0.013 (0.028) 0.088 ** (0.036)
Union 0.066 * (0.036) 0.083 * (0.046)
Work part-time -0.004 (0.034) 0.008 (0.049)
Firm size 0.021 ** (0.009) 0.005 (0.018)
Supervise others 0.091 ** (0.031) 0.097 * (0.049)
Customer
 contact daily -0.092 *** (0.028) -0.033 (0.041)
Craft -0.191 *** (0.037) -0.180 (0.070)
Services -0.296 *** (0.045) -0.274 *** (0.074)
Production -0.291 *** (0.068) -0.279 *** (0.097)
Laborers -0.361 *** (0.054) -0.357 *** (0.081)
Government -0.023 (0.038) 0.034 (0.056)
Black Coworkers -0.039 (0.062) -0.096 (0.086)
Latino(a) Coworkers -0.156 * (0.039) -0.131 *** (0.048)
Asian Coworkers 0.056 (0.073) 0.090 (0.102)
Other Coworkers -0.038 (0.043) -0.048 (0.073)
Supervisor white 0.072 * (0.040) 0.108 ** (0.054)
Supervisor black 0.069 (0.068) 0.087 (0.090)
Supervisor Latino(a) -0.006 (0.048) 0.062 (0.054)
Supervisor Asian -0.051 (0.069) -0.015 (0.083)
Supervisor other -0.097 (0.095) -0.105 (0.101)
Constant 2.141*** (0.086) 1.923 *** (0.128)
F statistic for
 the equation 31.12*** [0.000] 11.43 *** [0.000]
Adjusted [R.sup.2] .50 .42

Source: MCSUI. Coefficients are reported with standard
errors in parentheses.
*** Statistically significant at .01 level for two-tailed test,
** statistically significant at the .05 level for two-tailed
test, and * statistically significant at the .1 level for a
two-tailed test. A description of how each variable is
constructed appears in Appendix D.

This work is supported by grant #0213838 from the National Science Foundation. We are grateful for suggestions provided by Barry Hirsch and seminar participants at Bond University, Gold Coast, Australia. In addition, we are grateful for comments from Julie Hotchkiss, Co-Editor for this journal, and two anonymous referees. This paper was presented at the 2004 Southern Economic Association Meetings.

Received March 2005; accepted March 2006.

References

Altonji, Joseph G., and Rebecca Blank. 1999. Race and gender in the labor market. In Handbook of labor economics, edited by Orley Ashenfelter and David Card. Amsterdam: North Holland, pp. 3143-60.

Altonji, Joseph G., and Robert A. Shakotko. 1987. Do wages rise with job seniority? Review of Economic Studies 54:437-59.

Bergmann, Barbara. 1971. The effect on white incomes of discrimination in employment. Journal of Political Economy 79:294-313.

Bratsberg, Bernt, and Dek Terrell. 1998. Experience, tenure, and wage growth of young black and white men. Journal of Human Resources 33:658-82.

Cain, Glen D. 1986. The economic analysis of labor market discrimination: A survey. In Handbook of labor economics, edited by Orley Ashenfelter and Richard Layard. Amsterdam: North Holland, pp. 709-30.

Darity, William A., and Patrick L. Mason. 1998. Evidence on discrimination in employment: Codes of color, codes of gender. Journal of Economic Perspectives 12:63-90.

Farmer, Amy, and Dek Terrell. 1996. Discrimination, bayesian updating of employer beliefs, and human capital accumulation. Economic Inquiry 34:204-19.

Fiske, Susan T., Amy Cuddy, Peter Glick, and Jun Xu. 2002. A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality & Social Psychology 82:878-902.

Fiske, Susan T., and Janet B. Ruscher. 1993. Negative interdependence and prejudice: Whence the affect? In Affect, cognition, and stereotyping: Interactive processes in group perception, edited by Diane M. Mackie and David Lewis Hamilton. San Diego, CA: Academic Press, pp. 239-68.

Goldsmith, Arthur H., Jonathan R. Veum, and William Darity, Jr. 2000. Motivation and labor market outcomes. Research in Labor Economics 19:109-46.

Lazear, Edward. 1979. The narrowing of black-white wage differentials is illusory. American Economic Review 69:553-64.

Lewis, Danielle, and Dek Terrell. 2001. Experience, tenure, and the perceptions of employers. Southern Economic Journal 67:578-97.

Topel, Robert. 1991. Specific capital, mobility, and wages: Wages rise with job seniority. Journal of Political Economy 99:145-76.

Trejo, Stephen J. 1997. Why do Mexican Americans earn low wages? Journal of Political Economy 105:1235-68.

Wolpin, Kenneth. 1992. The determinants of black white differences in early employment careers: Search, layoffs, quits, and endogenous wage growth. Journal of Political Economy 100:535-60.

(1) Cain (1986) and Altonji and Blank (1999) provide excellent surveys of the mechanisms through which race may affect wages. Darity and Mason (1998) summarize theories of discrimination and existing evidence on discrimination against black workers. For recent evidence on wage differences between Latino and white workers in the United States and on the importance of immigrant generation, see Trejo (1997).

(2) See Fiske and Ruscher (1993) for a review of the literature in social psychology on negative stereotyping and minority group status. Fiske et al. (2002) provide empirical evidence that persons tend to hold negative stereotypes based on race and ethnicity.

(3) There is perhaps a third alternative for employers to undervalue the prior experience of nonwhite workers. Bergmann (1971) asserts that the return to skills obtained by black workers may be underrated as a result of "occupational crowding." She argues that blacks are crowded into low-paying jobs and restricted from high-paying or attractive jobs. However, she suggests that those blacks who are able to ascend to desirable positions are willing to accept a lower wage, given the difficulties for them in finding comparable positions with other firms.

(4) However, there is some evidence to suggest that even large and public sector firms systematically discriminate against workers based on race. For example, Darity and Mason (1998) point out several recent class-action employment discrimination lawsuits of large firms in which settlements have reached as high as $176 million.

(5) Using data drawn from the National Longitudinal Survey of Youth (NLSY), Bratsberg and Terrell (1998) find that even potential experience overstates actual experience for male workers.

(6) Bratsberg and Terrell (1998) find that an additional year of potential experience corresponds to 81% of an additional year of actual experience for white workers and 73% of an additional year for black employees.

(7) We are grateful to Barry Hirsch for pointing out that measurement error associated with using potential prior experience to gauge the extent of actual prior experience may lead to differential returns to prior experience for whites and nonwhites.

(8) Altonji and Blank (1999, pp. 3147), using data from the Current Population Survey for March 1996, report that employed nonwhites work fewer weeks per year and fewer hours per week than their white counterparts. Thus, even our measure of prior potential experience may overstate the amount of experience nonwhite workers acquired for each year of work on previous jobs relative to white workers.

(9) The coworker variable was constructed from the participants' response to the following question: "What (is/was) the race and ethnicity of most of the employees doing the kind of work you (do/did) in this location?"

(10) Altonji and Shakotko (1987) and Topel (1991) suggested that unobserved individual heterogeneity might correlate with wages, tenure, and prior experience, resulting in inconsistent estimates of the parameters if Equations 1 and 2, are estimated by ordinary least squares (OLS). In each study, an instrumental variables estimator was proposed to purge the model of correlation between the error term and both tenure and experience. However, others have found little variation in results across estimator choice. For example, see the research of Bratsberg and Terrell (1998) and Lewis and Terrell (2001).

(11) Goldsmith, Veum, and Darity (2000) offer evidence that motivation is an important determinant of wages using data drawn from the National Longitudinal Survey of Youths.

(12) We recognize that this measure of ability is dependent on the respondent's recollection and the rigor of the classes taken at the respondent's high school. Ideally, a more objective and standardized measure of achievement would have been preferred. Nonetheless, information is gained by using the respondents' recollections of their high school performance as a measure of achievement.

(13) The assumption adopted is that no such revisions occur with prior experience. Workers who switch jobs are again subject to assessment based on employers" perceptions of group ability because their individual performance has not been observed by that employer. Thus, nonwhite workers would face a recurring burden of proving themselves should they decide to switch jobs.

(14) Note that for all summary statistics, weights are used to generate represented characteristics of the population.

(15) Appendix C reveals a somewhat surprising result: A weighted average of 25% of black male respondents in our sample resided abroad at 16 years of age. Many of these respondents were from Boston, which has a substantial Caribbean population and is composed of 38% foreign-born black males. Based on our calculations using the Integrated Public Use Microdata Series of 1990 census data, we find that about 30% of non-Hispanic black males between 16 and 65 years of age who reported positive wages in the Boston metropolitan area were foreign-born. Nonetheless, given the presence of foreign-born blacks in our sample, the reader may be concerned that ability to speak English may be an uncontrolled attribute that influences the racial differences in wage returns resulting from experience and tenure in-out sample. However, in both the 1990 census and in our sample, less than 5% of black respondents were unable to speak English fluently. Furthermore, English speaking ability may be of particular concern for black males who are "high achievers" because an estimated 38% of them were living abroad at 16 years of age (i.e., 13 of 86 high achievers). The MCSUI interviewers rated the English speaking ability above good or better for 12 of the 13 high-achieving respondents who lived abroad at 16 years of age; six of these individuals had "excellent" English speaking ability. Since only one of the high-achieving respondents who lived abroad was rated as having poor English speaking ability, language ability does not appear to explain the racial gap in rewards to workplace experience. Had we simply included an additional control for English ability, the lack of variation in the variable would not have allowed us to estimate all parameters in many of our models stratified by workplace settings.

(16) In our data, a typical high-achieving (i.e., a high school grade point average of B or better) black (white) male earns less than the average low-achieving black (white) male, which may seem surprising. A likely explanation for this is that high achievers have much lower tenure levels, both among blacks (2.92 < 6.79) and among whites (2.64 < 7.90) due to greater job mobility or job shopping. The consequence may be lower average wages. Of note, the coefficient for high achievers was found to be positive (although statistically insignificant) for our pooled regression results for the log of wages and our subsamples of blacks and whites.

(17) Although our findings are consistent with the theory of ability misperception, it is important to recognize that evidence consistent with Proposition 2 can be explained by worker action as well as by employer updating. For example, the poor return to prior experience provides an incentive for black workers to stay on the job, which may lead them to work harder than white workers. Thus, even if employers fail to update erroneous beliefs about racial differences in the productivity gains associated with workplace experience, there will be a smaller racial gap in the wage return to tenure than for additional experience. We thank an anonymous referee for pointing this out.

(18) The full set of coefficient estimates for each equation estimated is available from the authors upon request.

(19) In all three situations where the estimated return to potential experience was statistically insignificant for both white males and black males, the estimated coefficient was larger in magnitude for white workers, which is consistent with Proposition 1.

(20) It is conceivable that part-time workers would receive a lower reward for prior and current experience than full-time workers with similar observable characteristics. However, even after removing part-time workers from the pooled sample (leaving only full-time workers), the rewards for workplace experience and on-the-job tenure were virtually unchanged. It is interesting to note that in workplace settings where management might have greater leeway to assign rewards for additional workplace seniority because of fewer rules (e.g., in small firms or in companies that hire nonunion workers), black males received larger returns to tenure than white workers. Although we are not certain why this pattern emerged, smaller firms and nonunionized firms do provide employers with greater latitude to adjust for previous miscalculations of a worker's talents. The same pattern of interracial differences in the reward to tenure emerges among employees who do not supervise other workers.

(21) Unfortunately, we have too few white employees with black supervisors to investigate if the rewards to workplace experience for white workers is influenced by the race of their direct supervisor. Our findings also reveal that the return to tenure for black males is much smaller when they have black coworkers than when their coworkers are white. It would have been useful if we had information on the skills of the coworkers or degree and quality of worker interaction to explore why this is the case.

(22) The full set of coefficient estimates for each equation estimated is available from the authors upon request. As was the case for males, few white female employees reported having a black supervisor or having primarily black coworkers. Therefore, we are unable to compare the returns to workplace experience that black women obtain in these workplace settings with those of white women.

(23) The workplace settings that were not consistent with Proposition 1 were occupations where workers had daily contact with customers, supervised other employees, were employed in a small firm, a union member, or were engaged in public sector employment. We do not detect an obvious pattern why Proposition 1 is not satisfied in these occupations.

(24) Appendix A provides a detailed catalog of observations lost for each restriction imposed in constructing the (non-Latino) white and Latino(a) subsamples.

(25) Among Latino males, the wage return to tenure is 13% smaller when their direct supervisor is Latino rather than white (which differs from our finding of a substantially larger return to black males when their supervisor is black), and the return is slightly larger when their coworkers are mostly white rather than mostly Latino.

(26) We evaluate whether {B} > 0 in all worksite environments, not solely those where Proposition 1 holds.

Arthur H. Goldsmith, Department of Economics, Washington and Lee University, Lexington, VA 24450, USA; E-mail GoldsmithA@wlu.edu; corresponding author.

Darrick Hamilton, Milano--The New School of Management and Urban Policy, 72 Fifth Avenue, New York, NY 10011, USA; E-mail hamiltod@newschool.edu.

William Darity, Jr., Department of Economics, University of North Carolina at Chapel Hill, Chapel Hill, NC 27514, USA; E-mail darity@unc.edu.

Table 1. The Impact of Tenure and Potential Prior Experience
on Wages for White and Black Males and Females: Summary Table

Panel A: Effect on Wages of Males

 Full Sample

 Black White Black - White
Variable (n = 419) (n = 493) (% Change)

Tenure 0.032 *** 0.033 *** -0.001 (-3.0)
Potential
 prior 0.002 0.009 *** -0.007 (-77.8)
 experience
Adjusted [R.sup.2] .44 .40

 High Achievement

 Black White Black - White
Variable (n = 86) (n = 137) (% Change)

Tenure 0.059 *** 0.052 *** 0.007 (13.5)
Potential
 prior -0.006 0.013 -0.019 (-146.2)
 experience
Adjusted [R.sup.2] .51 .28

 Low Achievement

 Black White Black - White
Variable (n = 333) (n = 356) (% Change)

Tenure 0.028 *** 0.030 *** -0.002 (-6.7)
Potential
 prior 0.004 0.008 *** -0.004 (-50.0)
 experience
Adjusted [R.sup.2] .46 .41

Panel B: Effect on Wages of Females

 Full Sample

 Black White Black - White
 (n = 805) (n = 554) (% Change)

Tenure 0.034 *** 0.030 *** 0.004 (13.3)
Potential 0.001 0.002 -0.001 (-50.0)
 prior
 experience
Adjusted [R.sup.2] .48 .38

 High Achievement

 Black White Black - White
 (n = 216) (n = 161) (% Change)

Tenure 0.071 *** 0.042 ** 0.029 (69.0)
Potential -0.002 0.003 -0.005 (-166.7)
 prior
 experience
Adjusted [R.sup.2] .35 .36

 Low Achievement

 Black White Black - White
 (n = 589) (n = 393) (% Change)

Tenure 0.027 *** 0.026 *** 0.001 (3.8)
Potential 0.002 0.003 -0.001 (-33.3)
 prior
 experience
Adjusted [R.sup.2] .51 .39

Source: MCSUI. Coefficient estimates of the total effect of tenure on
wage {[partial derivative][w.sub.i]/[partial derivative] [TEN.sub.i] =
[[gamma] y + 2[lambda]([bar.TEN].sub.i])} and the total effect of
potential prior experience {[partial derivative][w.sub.i]/[partial
derivative] [PEXP.sub.i] = [beta] + 2 [phi]([[bar.PEXP].sub.i])} on
wage are reported. The p-values from the F-tests of the total effects
are indicated by asterisks (*** < .01, ** < .05, * < .1). The percentage
change is black total effect minus white total effect relative to the
white total effect. If the total effect is zero for any group, the
percentage change is assigned a value of 100. The table is based on
a wage regression that includes potential prior experience and tenure,
along with their squares, as well as controls for schooling level
completed, health status, demographics, firm characteristics, customer
contact, city of residence, occupation of employment, racial/ethnic
composition of coworkers, and racial/ethnic background of supervisor.
A full list of covariates and how they are constructed appears in
Appendix D.

Table 2. The Impact of Tenure and Potential Prior Experience on
ln Wages for White and Black Male Workers: For Subsets of Data
Defined by Workplace Setting Descriptors

 Total Change in Wage

 White Coworkers

Variable Black White Black-White
 (n = 159) (n = 374) (% Change)

Tenure 0.039 *** 0.032 *** 0.007 (21.9)
Potential prior
 experience 0.005 0.009 *** -0.004 (-44.4)
Adjusted [R.sup.2] .37 .40

 Total Change in Wage

 Black Coworkers

Variable Black White Black-White
 (n = 173) (n = 29) (% Change)

Tenure 0.028 *** -- --
Potential prior
 experience 0.002
Adjusted [R.sup.2] .30 -- --

 White Supervisor

 Black White Black-White
 (n = 206) (n = 385) (% Change)

Tenure 0.026 *** 0.034 *** -0.008 (-23.5)
Potential prior 0.002 0.007 ** -0.005 (-71.4)
 experience
Adjusted [R.sup.2] .37 .39

 Black Supervisor

 Black White Black-White
 (n = 123) (n = 23) (% Change)

Tenure 0.040 *** -- --
Potential prior 0.000
 experience
Adjusted [R.sup.2] .45 -- --

 Daily Contact with Customers

 Black White Black-White
 (n = 251) (n = 284) (% Change)

Tenure 0.030 *** 0.034 *** -0.004 (-11.8)
Potential prior 0.002 0.006 * -0.004 (-66.7)
 experience
Adjusted [R.sup.2] .43 .42 --

 No Daily Contact with Customers

 Black White Black-White
 (n = 168) (n = 209) (% Change)

Tenure 0.027 *** 0.032 *** -0.005 (-15.6)
Potential prior 0.000 0.010 ** -0.010 (-100.0)
 experience
Adjusted [R.sup.2] .43 .36 --

 Supervisor

Total change in Black White Black-White
wage with respect to: (n = 126) (n = 194) (% Change)

Tenure 0.017 * 0.029 *** -0.012 (-41.4)
Potential prior 0.007 0.008 -0.001 (-12.5)
 experience
Adjusted [R.sup.2] .43 .49 --

 Nonsupervisor

Total change in Black White Black-White
wage with respect to: (n = 293) (n = 299) (% Change)

Tenure 0.040 *** 0.033 *** 0.007 (21.2)
Potential prior 0.001 0.006 -0.005 (-83.3)
 experience
Adjusted [R.sup.2] .41 .28 --

 Large Firm

Total change in Black White Black-White
wage with respect to: (n = 192) (n = 188) (% Change)

Tenure 0.034 *** 0.045 *** -0.011 (-24.4)
Potential prior 0.007 0.014 *** -0.007 (-50.0)
 experience
Adjusted [R.sup.2] .36 .46 --

 Small Firm

Total change in Black White Black-White
wage with respect to: (n = 227) (n = 305) (% Change)

Tenure 0.027 *** 0.024 *** 0.003 (12.5)
Potential prior 0.000 0.005 -0.005 (-100.0)
 experience
Adjusted [R.sup.2] .38 .32 --

 Union Members

Total change in Black White Black-White
wage with respect to: (n = 111) (n = 98) (% Change)

Tenure 0.032 *** 0.028 *** 0.004 (14.2)
Potential prior -0.004 0.014 ** -0.018 (-128.6)
 experience
Adjusted [R.sup.2] .45 .35 --

 Nonunion Members

Total change in Black White Black-White
wage with respect to: (n = 308) (n = 395) (% Change)

Tenure 0.031 *** 0.031 *** 0.000 (0.0)
Potential prior 0.005 0.009 ** -0.004 (-44.4)
 experience
Adjusted [R.sup.2] .37 .40

 Government Workers

Total change in Black White Black-White
wage with respect to: (n = 93) (n = 82) (% Change)

Tenure 0.034 *** 0.037 *** -0.003 (-1.0)
Potential prior 0.002 0.025 ** -0.023 (-92.0)
 experience
Adjusted [R.sup.2] .40 .51 --

 Non-Government Workers

Total change in Black White Black-White
wage with respect to: (n = 326) (n = 411) (% Change)

Tenure 0.031 *** 0.032 *** -0.001 (-3.1)
Potential prior 0.004 0.008 ** -0.004 (-50.0)
 experience
Adjusted [R.sup.2] .39 .36 --

 Full-Time Work

Total change in Black White Black-White
wage with respect to: (n = 380) (n = 442) (% Change)

Tenure 0.030 *** 0.035 *** -0.005 (-14.3)
Potential prior 0.003 0.009 *** -0.006 (-66.7)
 experience
Adjusted [R.sup.2] .43 .40 --

 Part-Time Work

Total change in Black White Black-White
wage with respect to: (n = 39) (n = 51) (% Change)

Tenure -- -- --
Potential prior -- -- --
 experience
Adjusted [R.sup.2] -- -- --

Source: MCSUI. Coefficient estimates of the total effect of tenure
on wage {[partial derivative][w.sub.1]/[partial derivatives][TEN.sub.i]
= [gamma] + 2[lambda]([[bar.TEN].sub.i])} and the total effect of
potential prior experience [partial derivative][w.sub.i]/[partial
derivative][PEXP.sub.i] = [beta] + 2 [phi]([[bar.PEXP].sub.i])} on wage
are reported. Thep-values from the F-tests of the total effects are
indicated by asterisks (*** < .01, ** < .05, * < .1). The percentage
change is black total effect minus white total effect relative to the
white total effect. If the total effect is zero for any group, the
percentage change is assigned a value of 100. The table is based on a
wage regression that includes potential prior experience and tenure,
along with their squares, as well as controls for schooling level
completed, health status, demographics, firm characteristics, customer
contact, city of residence, occupation of employment, racial/ethnic
composition of coworkers, and racial/ethnic background of supervisor.
A full list of covariates and how they are constructed appears in
Appendix D.

Table 3. Summary of Findings on Proposition 1, Proposition 2,
and Proposition 3 for Alternative Work Place Settings: Blacks
and Whites for Males and Females

 Males

 Status of a Proposition

Variables Proposition 1 Proposition 2 Proposition 3

Full sample [check] [check] [check]
High achievement [check] [check] [check]
Low achievement [check] [check] [check]

 Stratification
 Results

Black coworkers Insufficient data
White coworkers [check] [check] [check]
Supervisor black Insufficient data
Supervisor white [check] [check] [check]
Daily contact with
 customers [check] [check]
No daily contact
 with customers [check] [check]
Supervisor [check] [check]
Not a supervisor [check] [check] [check]
Large firm [check] [check]
Small firm [check] [check] [check]
Union member [check] [check]
Not a union member [check] [check] [check]
Government worker [check] [check]
Not a government
 worker [check] [check]
Full-time worker [check] [check]
Part-time worker Insufficient data
Share of cases when
 a proposition
 holds (yes/cases
 evaluated) 13/13 12/13 4/13

 Females

 Status of a Proposition

Variables Proposition 1 Proposition 2 Proposition 3

Full sample [check] [check] [check]
High achievement [check] [check] [check]
Low achievement [check] [check] [check]

 Stratification
 Results

Black coworkers Insufficient data
White coworkers [check] [check] [check]
Supervisor black Insufficient data
Supervisor white [check] [check] [check]
Daily contact with
 customers
No daily contact
 with customers [check] [check] [check]
Supervisor
Not a supervisor [check] [check] [check]
Large firm [check] [check] [check]
Small firm
Union member
Not a union member [check] [check] [check]
Government worker
Not a government
 worker [check] [check]
Full-time worker [check] [check] [check]
Part-time worker [check] [check] [check]
Share of cases when
 a proposition
 holds (yes/cases
 evaluated) 9/14 9/9 8/9

Source: MCSUI. Proposition 1, [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII] Proposition 2, [MATHEMATICAL EXPRESSION
NOT REPRODUCIBLE IN ASCII] Proposition 3, [MATHEMATICAL EXPRESSION
NOT REPRODUCIBLE IN ASCII] Subscripts indicate gender and
superscripts indicate race or ethnicity. The table is based on a wage
regression that includes a quartic in tenure and prior experience,
and controls for schooling level completed, health status,
demographics, firm characteristics, city of residence, occupation of
employment, racial/ethnic composition of coworkers, and raciallethnic
background of supervisor. A full list of covariates and how they are
constructed appears in Appendix D. A check for Proposition 1 plus a
check for either Proposition 2 or Proposition 3 (even stronger) is
evidence consistent with the predictions of the theory of ability
misperception.

Table 4. The Impact of Tenure and Potential Prior Experience on
ln Wages for White and Black Female Workers: Subsets of Data
Defined by Workplace Setting Descriptors

 Total Change in Wage

 White Coworkers

Variable Black White Black-White
 (n = 262) (n = 406) (% Change)

Tenure 0.035 *** 0.027 *** 0.008 (29.6)
Potential prior
 experience 0.004 0.005 -0.001 (-20.0)
Adjusted [R.sup.2] .48 .33 --

 Total Change in Wage

 Black Coworkers

Variable Black Black Black-White
 (n = 292) (n = 27) (% Change)

Tenure 0.032 *** -- --
Potential prior
 experience -0.003 -- --
Adjusted [R.sup.2] .43 -- --

 White Supervisor

 Black White Black-White
 (n = 376) (n = 426) (% Change)

Tenure 0.031 *** 0.030 *** 0.001 (3.3)
Potential prior 0.002 0.005 * -0.003 (-60.0)
 experience
Adjusted [R.sup.2] .47 .39 --

 Black Supervisor

 Black Black Black-White
 (n = 292) (n = 27) (% Change)

Tenure 0.040 *** -- --
Potential prior 0.000 -- --
 experience
Adjusted [R.sup.2] .48 -- --

 Daily Contact with Customers

 Black White Black-White
 (n = 540) (n = 361) (% Change)

Tenure 0.035 *** 0.030 *** 0.005 (16.7)
Potential prior 0.002 0.001 0.001 (100.0)
 experience
Adjusted [R.sup.2] .45 .37 --

 No Daily Contact with Customers

 Black Black Black-White
 (n = 265) (n = 193) (% Change)

Tenure 0.034 *** 0.029 *** 0.015 (17.2)
Potential prior 0.001 0.002 -0.001 (-50.0)
 experience
Adjusted [R.sup.2] .55 .36 --

 Supervisor

 Black White Black-White
 (n = 170) (n = 167) (% Change)

Tenure 0.025 *** 0.017 * 0.008 (47.1)
Potential prior -0.003 -0.005 0.002 (40.0)
 experience
Adjusted [R.sup.2] .49 .33 --

 Non Supervisor

 Black Black Black-White
 (n = 635) (n = 387) (% Change)

Tenure 0.037 *** 0.034 *** 0.003 (8.8)
Potential prior 0.001 0.002 -0.001 (-50.0)
 experience
Adjusted [R.sup.2] .46 .36 --

 Large Firm

 Black White Black-White
 (n = 389) (n = 228) (% Change)

Tenure 0.024 *** 0.026 *** -0.002 (7.7)
Potential prior -0.002 0.008 ** -0.010 (-125.0)
 experience
Adjusted [R.sup.2] .47 .44 --

 Small Firm

 Black Black Black-White
 (n = 416) (n = 326) (% Change)

Tenure 0.045 *** 0.035 *** 0.010 (28.6)
Potential prior 0.002 0.002 0.000 (0.0)
 experience
Adjusted [R.sup.2] .44 .34 --

 Union Members

 Black White Black-White
 (n = 191) (n = 87) (% Change)

Tenure 0.022 *** 0.010 0.012 (120.0)
Potential prior 0.001 0.001 0.000 (0.0)
 experience
Adjusted [R.sup.2] .42 .52 --

 Nonunion Members

 Black Black Black-White
 (n = 614) (n = 467) (% Change)

Tenure 0.039 *** 0.034 *** 0.005 (14.7)
Potential prior 0.000 0.005 * -0.005 (-100.2)
 experience
Adjusted [R.sup.2] .43 .35 --

 Government Workers

 Black White Black-White
 (n = 262) (n = 406) (% Change)

Tenure 0.032 *** 0.015 0.017 (113.3)
Potential prior 0.002 -0.007 0.009 (128.6)
 experience
Adjusted [R.sup.2] .47 .59 --

 Non-Government Workers

 Black Black Black-White
 (n = 401) (n = 262) (% Change)

Tenure 0.035 *** 0.037 *** -0.002 (5.4)
Potential prior 0.000 0.007 ** -0.007 (-100.0)
 experience
Adjusted [R.sup.2] .46 .34 --

 Full-Time Work

 Black White Black-White
 (n = 262) (n = 406) (% Change)

Tenure 0.031 *** 0.029 *** 0.002 (6.9)
Potential prior 0.002 0.004 * -0.002 (-50.0)
 experience
Adjusted [R.sup.2] .50 .44 --

 Part-Time Work

 Black Black Black-White
 (n = 401) (n = 262) (% Change)

Tenure 0.055 ** 0.048 ** 0.007 (113.3)
Potential prior -0.006 -0.003 -0.003 (-100.0)
 experience
Adjusted [R.sup.2] .35 .11 --

Source: MCSUI. Coefficient estimates of the total effect of tenure
on wage {[partial derivative][w.sup.i]/[partial derivative][TEN.sub.i]
= [gamma] + 2[lambda]([[bar.TEN].sup.i])} and the total effect of
potential prior experience {[partial derivative][w.sub.i]/[partial
derivative][PEXP.sub.i] = [beta] + 2[phi]([[bar.PEXP].sub.i] on wage
are reported. The p-values from the F-tests of the total effects
are indicated by asterisks (*** < .01, ** < .05, * < .1). The change
is black total effect minus white total effect relative to the white
total effect. If the total effect is zero for any group, the percentage
change is assigned a value of 100. The table is based on a wage
regression that includes potential prior experience and tenure, along
with their squares, as well as controls for schooling level completed,
health status, demographics, firm characteristics, customer contact,
city of residence, occupation of employment, racial/ethnic composition
of coworkers, and racial/ethnic background of supervisor. A full list
of covariates and how they are constructed appears in Appendix D.

Table 5. The Impact of Tenure and Potential Prior Experience
on Wages for White, Latino and Latina Workers by
Gender: Summary Table

Panel A: Effect on Wages of Males

 Full Sample

Model Latino White Latino-White
specification (n = 494) (n -493) (% Change)

Tenure 0.054 *** 0.033 *** 0.021 (63.6)
Potential prior
 experience 0.004 * 0.009 *** -0.005 (-55.6)
Adjusted [R.sup.2] .45 .40 --

 High Achievement

Model Latino White Latino-White
specification (n = 128) (n = 137) (% Change)

Tenure 0.046 *** 0.052 *** -0.006 (-11.5)
Potential prior
 experience 0.003 0.013 -0.010 (-76.9)
Adjusted [R.sup.2] .48 .28 --

 Low Achievement

Model Latino White Latino-White
specification (n = 366) (n = 356) (% Change)

Tenure 0.058 *** 0.030 *** 0.028 (93.3)
Potential prior
 experience 0.004 * 0.008 *** -0.004 (-50.0)
Adjusted [R.sup.2] .42 .41 --

Panel B: Effect on Wages of Females

 Full Sample

 Latina White Latina-White
 (n = 514) (n = 554) (% Change)

Tenure 0.041 *** 0.030 *** 0.011 (36.7)
Potential prior
 experience 0.005 ** 0.002 0.003 (150.0)
Adjusted [R.sup.2] .42 .38 --

 High Achievement

 Latina White Latina-White
 (n = 150) (n = 161) (% Change)

Tenure 0.046 ** 0.042 ** 0.004 (9.5)
Potential prior
 experience -0.004 0.003 -0.007 (-233.3)
Adjusted [R.sup.2] .29 .36 --

 Low Achievement

 Latina White Latina-White
 (n = 364) (n = 393) (% Change)

Tenure 0.034 *** 0.026 *** 0.008 (30.8)
Potential prior
 experience 0.006 *** 0.003 0.003 (100.0)
Adjusted [R.sup.2] .48 .39 --

Source: MCSUI. Coefficient estimates of the total effect of
tenure on wage {[partial derivative][w.sub.i]/[partial derivative]
[TEN.sub.i]; = [gamma] + 2[lambda]([[bar.TEN].sub.i])} and the total
effect of potential prior experience {[partial derivative]
[w.sub.i]/[partial derivative][PEXP.sub.i] = [beta] + 2[phi]
([bar.PEXP].sub.i] on ware are reported. The p-values from the
F-test of the total effects are indicated by asterisks (*** < .01,
** .05, * < .1). The percentage change is Latino total effect minus
white total effect relative to the white total effect. If the total
effect is zero for any group, the percentage change is assigned a
value of 100. The table is based on a wage regression that includes
potential prior experience and tenure, along with their squares, as
well as controls for schooling level competed, health status,
demographics, firm characteristics, customer contact, city of
residence, occupation of employment, racial/ethnic composition of
coworkers, and firm characteristics, customer contact, city of
residence, occupation of employment, racial/ethnic composition
of coworkers, and racial/ethnic background of supervisor. A full
list of covariates and how they are constructed appears in
Appendix D.

Table 6. The Impact of Tenure and Potential Prior Experience
on In Wages for White Maleand Latino Workers: For Subsets
of Data Defined by Workplace Setting Descriptors

 Males

 White Coworkers

 Latino White Latino-White
 (n = 104) (n = 374) (% Change)

Tenure 0.049 *** 0.032 *** 0.017 (53.1)
Potential
 prior
 experience 0.011 0.009 *** 0.002 (22.2)
Adjusted [R.sub.2] .26 .40 --

 Males

 Latino/a Coworkers

 Latino White Latino-White
 (n = 343) (n = 41) (% Change)

Tenure 0.057 *** -- --
Potential
 prior
 experience 0.005 * -- --
Adjusted [R.sub.2] .40 -- --

 White Supervisor

 Latino White Latino-White
 (n = 173) (n = 20) (% Change)

Tenure 0.054 *** 0.034 *** 0.020 (58.8)
Potential
 prior
 experience 0.003 0.007 ** -0.004 (-57.1)
Adjusted [R.sub.2] .38 .39

 Latino/a Supervisor

 Latino White Latino-White
 (n = 173) (n = 20) (% Change)

Tenure 0.055 *** -- --
Potential
 prior
 experience 0.002 -- --
Adjusted [R.sub.2] .38 -- --

 Daily Contact with Customers

 Latino White Latino-White
 (n = 192) (n = 284) (% Change)

Tenure 0.058 *** 0.034 *** 0.024
Potential
 prior
 experience 0.011 ** 0.006 * 0.006
Adjusted [R.sub.2] .45 .42 --

 No Daily Contact with Customers

 Latino White Latino White
 (n = 302) (n = 209) (% Change)

Tenure 0.046 *** 0.032 *** 0.014 (43.8)
Potential
 prior
 experience 0.001 0.010 ** -0.009 (-90.0)
Adjusted [R.sub.2] .46 .36 --

 Supervisor

 Latino White Latino-White
 (n = 117) (n = 194) (% Change)

Tenure 0.041 *** 0.029 *** 0.012 (41.4)
Potential
 prior
 experience 0.011 * 0.008 0.003 (37.5)
Adjusted [R.sub.2] .40 .49 --

 Non Supervisor

 Latino White Latino-White
 (n = 377) (n = 299) (% Change)

Tenure 0.049 *** 0.033 *** 0.016 (48.5)
Potential
 prior
 experience 0.003 0.006 -0.003 (-50.0)
Adjusted [R.sub.2] .44 .28 --

 Large Firm

 Latino White Latino-White
 (n = 128) (n = 188) (% Change)

Tenure 0.043 *** 0.045 *** -0.002 (41.4)
Potential
 prior
 experience 0.003 0.014 *** -0.011 (-78.6)
Adjusted [R.sub.2] .41 .46 --

 Small Firm

 Latino White Latino-White
 (n = 366) (n = 305) (% Change)

Tenure 0.057 *** 0.024 *** 0.033 (48.5)
Potential
 prior
 experience 0.002 0.005 -0.003 (-60.0)
Adjusted [R.sub.2] .42 .32 --

 Union Members

 Latino White Latino-White
Tenure (n = 109) (n = 98) (% Change)

 0.024 * 0.028 *** -0.004 (14.3)
 0.005 0.014 ** -0.009 (-64.3)

Adjusted [R.sub.2] .35 .35 --

 Nonunion Members

 Latino White Latino White
Tenure (n = 385) (n = 395) (% Change)

Potential 0.064 *** 0.031 *** 0.033 (106.5)
 prior 0.003 0.009 ** -0.006 (66.7)
 experience
Adjusted [R.sub.2] .41 .40 --

 Government Workers

 Latino White Latino-White
 (n = 26) (n = 82) (% Change)

Tenure -- 0.037 *** --
Potential
 prior
 experience -- 0.025 ** --
Adjusted [R.sub.2] -- .51 --

 Non-Government Workers

 Latino White Latino-White
 (n = 468) (n = 411) (% Change)

Tenure 0.054 *** 0.032 *** 0.022 (68.8)
Potential
 prior
 experience 0.003 0.008 ** -0.005 (-62.5)
Adjusted [R.sub.2] .43 .36 --

 Full-Time Work

 Latino White Latino-White
 (n = 429) (n = 442) (% Change)

Tenure 0.053 *** 0.035 *** 0.018 (51.4)
Potential
 prior
 experience 0.004 0.009 *** -0.005 (-55.6)
Adjusted [R.sub.2] .45 .40 --

 Part-Time Work

 Latino White Latino-White
 (n = 65) (n = 51) (% Change)

Tenure 0.053 * -- --
Potential
 prior
 experience 0.003 -- --
Adjusted [R.sub.2] .18 -- --

Source: MCSUL Coefficient estimates of the total effect of
tenure on wage {[partial derivative][w.sub.i]/[partial derivative]
[TEN.sub.i] = [gamma] + 2[lambda][[bar.TEN].sub.i])} and the total
effect of potential prior experience {[partial derivative] [w.sup.i]/
[partial derivative] [PEXP.sub.i] = [beta] + 2[phi] ([[bar.PEXP].sub.i]
on wage are reported. The p-values from the F-tests of the total
effects are indicated by (*** < .01, ** < .05, * < .1). The
percentage change is Latino Latino total effect minus white total
effect relative to the white total effect. If the total effect is
zero for any group, the percentage change is assigned a value of
100. The table is based on a wage regression that includes potential
prior experience and tenure, along with their squares, as well as
controls for schooling level completed, health status, demographics,
firm characteristics, customer contact, city of residence,
occupation of employment, racial/ethnic composition of coworkers,
and racial/ethnic background of supervisor. A full list of
covariates and how they are constructed appears in Appendix D.

Table 7. Summary of Findings on Proposition 1, Proposition 2,
and Proposition 3 for Alternative Workplace Settings:
Whites, Latinos, and Latinas

 Males

 Status of a Proposition

Variables Proposition 1 Proposition 2 Proposition 3

Full sample [check] [check] [check]
High achievement [check] [check]
Low achievement [check] [check] [check]

 Stratification Results

Coworkers Latino/a Insufficient data
Coworkers white
Supervisor Latino/a Insufficient data
Supervisor white [check] [check] [check]
Daily contact with
 customers
No daily contact with
 customers [check] [check] [check]
Supervisor
Not a supervisor [check] [check] [check]
Large firm [check] [check]
Small firm [check] [check] [check]
Union member [check] [check]
Not a union member [check] [check] [check]
Government worker Insufficient data
Not a
 government worker [check] [check] [check]
Full-time worker [check] [check] [check]
Part-time worker Insufficient data
Yes / No (for 9/12 9/9 7/9
 stratifications)

 Females

 Status of a Proposition

Variables Proposition 1 Proposition 2 Proposition 3

Full sample
High achievement [check] [check] [check]
Low achievement

 Stratification Results

Coworkers Latino/a Insufficient data
Coworkers white [check] [check] [check]
Supervisor Latino/a Insufficient data
Supervisor white [check] [check] [check]
Daily contact with
 customers
No daily contact with
 customers
Supervisor
Not a supervisor
Large firm [check] [check] [check]
Small firm
Union member [check] [check] [check]
Not a union member
Government worker
Not a
 government worker [check] [check] [check]
Full-time worker
Part-time worker
Yes / No (for 5/14 5/5 5/5
 stratifications)

Source: MCSUI. Proposition 1, [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII] Proposition 2, [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII] Proposition 3, [MATHEMATICAL EXPRESSION NOT
REPRODUCIBLE IN ASCII]. Subscripts indicate gender and superscripts
indicate race or ethnicity. The table is based on a wage regression
that includes a quartic in tenure and prior experience, and controls
status, demographics, firm for schooling level completed, health
characteristics, city of residence, occupation and racial/ethnic
background of supervisor. A full list of covariates and how they are
constructed appears in Appendix D. A check for Proposition 1 plus a
check for either Proposition 2 or Proposition 3 with the predictions
of the (even stronger) is evidence consistent theory of ability
misperception.

Table 8. The Impact of Tenure and Potential Prior Experience
on In Wages for White Femaleand Latina Workers: For Subsets
of Data Defined by Workplace Setting Descriptors

 Females

 White Coworkers

 Latina White Latina-White
 (n = 107) (n = 406) (% Change)

Tenure 0.048 *** 0.027 *** 0.021 (77.8)
Potential
 prior
 experience 0.001 0.005 -0.004 (-80.0)
Adjusted [R.sub.2] .49 .33 --

 Females

 Latino/a Coworkers

 Latina White Latina-White
 (n = 333 (n = 40) (% Change)

Tenure 0.035 *** -- --
Potential
 prior
 experience 0.008 *** -- --
Adjusted [R.sub.2] .27 -- --

 White Supervisor

 Latina White Latina-White
 (n = 212) (n = 426) (% Change)

Tenure 0.037 *** 0.030 *** 0.007 (23.3)
Potential
 prior
 experience 0.003 0.005 * -0.002 (-40.0)
Adjusted [R.sub.2] .37 .39 --

 Latino/a Supervisor

 Latina White Latina-White
 (n = 164) (n = 23) (% Change)

Tenure 0.032 *** -- --
Potential
 prior
 experience 0.013 *** -- --
Adjusted [R.sub.2] .36 -- --

 Daily Contact with Customers

 Latina White Latina-White
 (n = 223) (n = 361) (% Change)

Tenure 0.042 *** 0.030 *** 0.012 (40.0)
Potential
 prior
 experience 0.003 0.001 0.002 (200.0)
Adjusted [R.sub.2] .38 .37 --

 No Daily Contact with Customers

 Latina White Latina-White
 (n = 291) (n = 193) (% Change)

Tenure 0.049 *** 0.029 *** 0.02 (69.0)
Potential
 prior
 experience 0.006 ** 0.002 0.004 (200.0)
Adjusted [R.sub.2] .42 .36 --

 Supervisor

 Latina White Latina-White
 (n = 69) (n = 167) (% Change)

Tenure 0.029 0.017 * 0.012 (70.6)
Potential
 prior
 experience 0.011 -0.005 0.016 (320.0)
Adjusted [R.sub.2] .42 .33 --

 Nonsupervisor

 Latina White Latina-White
 (n = 445 (n = 387) (% Change)

Tenure 0.041 *** 0.034 *** 0.007 (20.6)
Potential
 prior
 experience 0.005 * 0.002 0.003 (150.0)
Adjusted [R.sub.2] .41 .36 --

 Large Firm

 Latina White Latina-White
 (n = 177) (n = 228) (% Change)

Tenure 0.039 *** 0.026 *** 0.013 (50.0)
Potential
 prior
 experience 0.003 0.008 ** -0.005 (-62.5)
Adjusted [R.sub.2] .48 .44 --

 Small Firm

 Latina White Latina White
 (n = 337 (n = 326) (% Change)

Tenure 0.044 *** 0.035 *** 0.009 (25.7)
Potential
 prior
 experience 0.007 ** 0.002 0.005 (250.0)
Adjusted [R.sub.2] .36 .34 --

 Union Members

 Latina White Latina-White
 (n = 88) (n = 87) (% Change)

Tenure 0.009 0.01 -0.001 (10.0)
Potential
 prior
 experience 0.018 ** 0.001 0.017 (1700.0)
Adjusted [R.sub.2] .45 .52 --

 Nonunion Members

 Latina White Latina-White
 (n = 426) (n = 467) (% Change)

Tenure 0.049 *** 0.034 *** 0.015 (44.1)
Potential
 prior
 experience 0.002 0.005 * -0.003 (-60.0)
Adjusted [R.sub.2] 0.35 0.35 --

 Government Workers

 Latina White Latina-White

 (n = 63) (n = 98) (% Change)

Tenure 0.053 ** 0.015 0.043 (286.7)
Potential
 prior
 experience 0.019 * -0.007 0.026 (371.4)
Adjusted [R.sub.2] .53 .59 --

 Non-Government Workers

 Latina White Latina-White

 (n = 451) (n = 456) (% Change)

Tenure 0.040*** 0.037*** 0.003 (8.1)
Potential
 prior
 experience 0.005* -- 0.002 (-28.6)
Adjusted [R.sub.2] .35 .34 --

 Full-Time Work

 Latina White Latina-White
 (n = 450) (n = 455) (% Change)

Tenure 0.040 *** 0.029 *** 0.011 (37.9)
Potential
 prior
 experience 0.005 ** 0.004 * 0.001 (25.0)
Adjusted [R.sub.2] .44 .44 --

 Part-Time Work

 Latina White Latina-White
 (n = 64) (n = 99) (% Change)

Tenure 0.035 0.048 ** -0.013 (27.0)
Potential
 prior
 experience 0.005 -0.003 0.008 (266.7)
Adjusted [R.sub.2] 0.09 0.11 --

Source: MCSUI. Coefficient estimates of the total effect of tenure
on wage {[partial derivative][w.sub.i]/[partial derivative][TEN.sub.i]
= [gamma] + 2[lambda]([[bar.TEN].sub.i])} and the total effect of
potential prior experience {[partial derivative][w.sub.i]/[partial
derivative][PEXP.sub.i] = [beta] + 2[phi] ([[bar.PEXP].sub.i])} on
wage are reported. The p-values from the F-tests of the total effects
are indicated by asterisks (*** < .01, ** < .05, * < .1). The
percentage change is Latina total effect minus white total effect
relative to the white total effect. If the total effect is zero for
any group, the percentage change is assigned a value of 100. The table
is based on a wage regression that includes potential prior experience
and tenure, along with their squares, as well as controls for schooling
level completed, health status, demographics, firm characteristics,
customer contact, city of residence, occupation of employment,
racial/ethnic composition of coworkers, and racial/ethnic background
of supervisor. A full list of covariates and how they are constructed
appears in Appendix D.