13
How Unions Affect Minority Representation in Building Trades Apprenticeship Programs CIHAN BILGINSOY* University of Utah, Salt Lake City, UT 84112 I use program-level data to compare the relative representation of blacks and Latinos in the construction industry apprenticeship programs organized with and without trade union participation. Econometric analysis shows that there are significant differences between the black and Latino experiences. The black share is higher in union-management joint programs, but the Latino share appears to be higher in the unilateral employer programs. Although both groups have lower representation in the higher status (electrical and mechanical) and higher paying occupations, the Latino share is more sensitive to earnings. I. Introduction Racial and ethnic minorities ~ were historically under-represented in the crafts labor force, and especially in the higher paying, higher skilled occupations (Marshall, 1965 Marshall and Briggs, 1967; Mills, 1972; Northrup and Foster, 1975; Penn, 1998). Fol- lowing the passage of Title VII of the Civil Rights Act of 1964, the status of minori- ties in skilled crafts and strategies for increasing their participation in these occupations became topics of public discussion. Two issues raised in these debates were the roles of trade unions and of apprenticeship training. Historically, unions have been more prominent in the building trades than elsewhere in the economy, and the crafts labor force continues to be more heavily unionized. Moreover, access to skilled trades requires extensive front-loaded training, and unions exercise substantial control over the organization and administration of apprenticeship, the traditional port of entry. The popular press, policy papers, and academic literature often contend that white male crafts use union power to discriminate against minorities and women (Northrup and Foster, 1975). Indeed, the literature has documented that much of the past century is replete with examples of racially exclusive practices of trade unions. Following this line of reasoning, it may be reckoned that the integration of demographic groups with inadequate representation in the crafts labor force requires either the removal of union control over apprenticeship programs or the replacement of apprenticeship by alterna- tives that are free from union intervention. This view, however, is not uncontested. One argument is that apprenticeship is only one of many routes of entry into the crafts and, therefore, discriminatory practices of the unions cannot be too serious an impedi- JOURNAL OF LABOR RESEARCH Volume XXVI, Number 3 Summer 2005

How unions affect minority representation in building trades apprenticeship programs

Embed Size (px)

Citation preview

Page 1: How unions affect minority representation in building trades apprenticeship programs

How Unions Affect Minority Representation in Building Trades Apprenticeship Programs

CIHAN BILGINSOY*

University of Utah, Salt Lake City, UT 84112

I use program-level data to compare the relative representation of blacks and Latinos in the construction industry apprenticeship programs organized with and without trade union participation. Econometric analysis shows that there are significant differences between the black and Latino experiences. The black share is higher in union-management joint programs, but the Latino share appears to be higher in the unilateral employer programs. Although both groups have lower representation in the higher status (electrical and mechanical) and higher paying occupations, the Latino share is more sensitive to earnings.

I. Introduction

Racial and ethnic minorities ~ were historically under-represented in the crafts labor force, and especially in the higher paying, higher skilled occupations (Marshall, 1965 Marshall and Briggs, 1967; Mills, 1972; Northrup and Foster, 1975; Penn, 1998). Fol- lowing the passage of Title VII of the Civil Rights Act of 1964, the status of minori- ties in skilled crafts and strategies for increasing their participation in these occupations became topics of public discussion. Two issues raised in these debates were the roles of trade unions and of apprenticeship training. Historically, unions have been more prominent in the building trades than elsewhere in the economy, and the crafts labor force continues to be more heavily unionized. Moreover, access to skilled trades requires extensive front-loaded training, and unions exercise substantial control over the organization and administration of apprenticeship, the traditional port of entry.

The popular press, policy papers, and academic literature often contend that white male crafts use union power to discriminate against minorities and women (Northrup and Foster, 1975). Indeed, the literature has documented that much of the past century is replete with examples of racially exclusive practices of trade unions. Following this line of reasoning, it may be reckoned that the integration of demographic groups with inadequate representation in the crafts labor force requires either the removal of union control over apprenticeship programs or the replacement of apprenticeship by alterna- tives that are free from union intervention. This view, however, is not uncontested. One argument is that apprenticeship is only one of many routes of entry into the crafts and, therefore, discriminatory practices of the unions cannot be too serious an impedi-

J O U R N A L OF LABOR R E S E A R C H

Volume XXVI, Number 3 Summer 2005

Page 2: How unions affect minority representation in building trades apprenticeship programs

452 JOURNAL OF LABOR RESEARCH

ment to racial/ethnic integration (Mills, 1972). A stronger counterargument is that, their past performance notwithstanding, trade unions can enhance the opening up of the trades to the nontraditional workers and use collective bargaining institutions to facili- tate outreach and recruitment efforts, to disseminate information, and to establish pre- apprenticeship programs to provide basic skills for the minority youth.

I inquire into the relationship between the patterns of racial/ethnic segregation and union involvement in apprenticeship training. Since the 1960s, minorities made sig- nificant inroads in the skilled construction trades, as well as the apprenticeship pro- grams, and the share of the minority apprentices in the 68 largest occupations in 1990 reached 22.5 percent (Government Accounting Office, 1992; Penn, 1998), exceeding the overall share of minorities in the labor force. The debate over the role of trade unions in the admission of minorities into the trades, however, has endured and, on occasion, escalated against the background of the prevailing wage law debates (Bernstein, 1993).

The scope of my study is limited to the civilian apprenticeship programs registered at the federal or state level which are required to abide by the federal or federally rec- ognized state training standards (29 CFR Part 29.5). These programs are established and administered either jointly by trade unions and employers signatory to a collective bar- gaining agreement or unilaterally by employers. The co-existence of two types of pro- grams subject to the same set of standards and that are, in principle, identical in terms of occupational lines, curricula, wage structure, and time spans, permits the assessment of the role of trade unions in the racial/ethnic integration of the crafts labor force through a comparative analysis of the minority shares under alternative sponsorship schemes. I pursue this question by estimating how program and labor market characteristics affect black and Latino representation in the apprenticeship programs. Utilization of program- level data and the application of multivariate econometric methods distinguish my study methodologically from the earlier studies that relied on aggregate data and whose find- ings are limited to the presentation of bivariate statistics.

Section II describes the main features of apprenticeship training in the building trades and presents background information on the racial/ethnic composition of apprenticeship programs during the 1989-1995 period. Section III presents the empir- ical model of the determination of the shares of blacks and Latinos in apprenticeship training. Empirical results are reported and discussed in Section IV. The concluding section summarizes the major findings.

II. Apprenticeship Programs in Building Trades

The administrative structure of the current National Apprenticeship Program is the product of legislative attempts to expand apprenticeship training in the 1930s. Cur- rently, the Bureau of Apprenticeship and Training (BAT) of the Department of Labor oversees the apprenticeship system with the primary objective of promoting appren- ticeship training. To this end, either the BAT or, in 27 states, the BAT-recognized state apprenticeship councils (SAC) determine the apprenticeable occupations and training standards, enforce compliance with these standards, register programs, and provide

Page 3: How unions affect minority representation in building trades apprenticeship programs

CIHAN BILGINSOY 453

technical assistance to establish and develop apprenticeship programs. Although the registration of programs is not mandatory, program sponsors may prefer registration because it implies that the programs adhere to apprenticeship standards, and the BAT or SACs certify apprentices who finish these programs as skilled craft workers. Another incentive to register is that employers do not pay registered apprentices journey-level wages in federal contracts.

Contemporary apprenticeship provides rigorous formal training on all aspects of a trade by combining on-the-job practical training with theoretical in-class instruction in a structured format. Registered apprenticeship training is time based: Apprentices are required to complete 2,000 hours of on-the-job training and 144 hours of related in- class instruction per year, and the term length of construction trade programs is typi- cally three to four years, depending on the occupation. Upon completion, the worker is certified as a skilled journey-worker.

Apprenticeship programs are sponsored either jointly by union and contractors signatory to a collective bargaining agreement or unilaterally by employers (henceforth "joint" and "non-joint" programs, respectively, following the BAT terminology). Joint programs have been prominent institutions in the union sector of the industry and remain an integral component of the collective bargaining agreements. The oldest joint programs still in operation were established as early as the 1930s. In these programs the apprentice is indentured to the Joint Apprenticeship Training Committee (JATC) com- posed of equal numbers of union and management representatives. The JATC coordi- nator, who is usually a rank-and-file union member, runs the day-to-day program. The JATC determines, within the BAT or SAC guidelines, the entry requirements and the curriculum and monitors the advancement of apprentices. New apprentices are selected from a pool of applicants who meet basic qualifications on the basis of a written aptitude test and interview. Training is financed by cent-per-labor-hour fees paid by the signatory contractors. Although apprenticeship programs are often identi- fied with the union sector of the industry, not all union workers enter the construction industry via this route. Apprenticeship is not necessary for a worker to achieve the journey-level status and qualify for union wages and benefits. While precise numbers are not available, no more than 50 percent of the union construction workers enter the industry via apprenticeship (Bilginsoy, 2003).

Nor is it the case that formal apprenticeship training exists only in the union sec- tor. Although workers in the open-shop sector of the industry are more likely to receive training through informal methods or as helpers, nonunion contractors have established and registered apprenticeship programs in increasing numbers, especially since the 1970s. These programs are organized either by a single employer or by an employer group under the leadership of a contractor organization (single and multi-employer programs, respectively, following the BAT terminology). 2 Typically, local chapters of a contractor association such as the Associated General Contractors (AGC) or the Associated Builders and Contractors (ABC) administer the multi-employer programs, and the participating contractors share the burden of financing (Northrup and Foster, 1975: 240; Business Roundtable, 1980). The majority of the open-shop apprentices are

Page 4: How unions affect minority representation in building trades apprenticeship programs

454 JOURNAL OF LABOR RESEARCH

trained in these programs. Information about the single-employer programs is scarce. There are a vast number of these programs, and a few of these appear to be organized by very large contractors and employ large numbers of apprentices, but the rest are minuscule in terms of program size.

The present analysis is based on the Apprenticeship Information Management Systems (AIMS) database of the BAT of the Department of Labor which provides information on registered apprentices from 36 states who participate in the AIMS. 3 The BAT lists more than 800 occupations as "apprenticeable." For my purposes, I selected the largest ten occupations, which accounted for more than 90 percent of all appren- tices. Women apprentices were removed from the dataset to isolate any confounding effects of gender. 41 also removed apprentices enrolled in programs for which the spon- sor type was not recorded.

A total of 153,680 new, male apprentices were registered in 5,309 apprentice- ship programs over the 1989-1995 period. Joint programs accounted for most of these apprentices (71.7 percent), followed by multi-employer and single-employer non-joint programs (19.0 percent and 9.3 percent, respectively). 5 In terms of the number of new apprentices admitted over the period, the average size of joint and multi-employer non- joint programs was substantially larger (84 and 66 apprentices, respectively) than the single-employer non-joint programs (four apprentices).

Among these apprentices 8.5 percent were black, 7.4 percent were Latino, 81.4 per- cent were white, and the remaining 2.7 percent were other races/ethnicities (Native American, Pacific Islander, Aleutian). Table 1 shows the shares of blacks, Latinos, and whites in the selected occupations across the joint and non-joint programs. Blacks con- stituted a larger fraction of apprentices in the joint programs (9.0 percent vs. 7.3 percent, p << 0.01), whereas the share of Latinos in these programs fell short of their share in the non-joint programs (7.3 percent vs. 7.9 percent, p << 0.0 l). Within joint programs, black representation was highest in the roofing (14.5 percent) and operating engineer (14.1 percent) trades, and Latino share was highest in the roofing trade (10.0 percent). Non-joint programs admitted very few apprentices in the bricklaying, operating engi- neer, painting, and structural steel occupations. Ignoring these occupations, the rela- tive representation of blacks in the non-joint programs was highest among carpenters (11.5 percent) and painters (9.1 percent), whereas Latinos' shares were highest among roofers (33.9 percent) followed by pipefitters (9.6 percent).

Over the 1989-1995 period, the average annual black and Latino labor market shares in the states included in this study were 8.8 percent and 6.4 percent, respectively (CPS, Outgoing Rotation Files, 1989-1995). Since the shares of minorities among incoming apprentices are similar to their shares in the overall state labor force, and even higher in the case of Latinos, these findings do not suggest that the minorities were particularly disadvantaged in gaining entry into apprenticeship training. The sum- mary statistics do not control, however, for the possible effects of other variables on minority shares. The next two sections examine the determinants of black and Latino shares in greater detail in a multivariate context.

Page 5: How unions affect minority representation in building trades apprenticeship programs

CIHAN BILGINSOY 455

Table 1

Program and Ethnic~Racial Distribution of lncoming Apprentices by Occupation, 1989-1995

Joint Programs

%Black %Latino %White

Non-joint Programs

N %Black %Latino %White N

Bricklayer 10.0 7.5 80.9 4,522 I 1.9 5.3 79.5 604 Carpenter 8.7 8.6 79.8 27,207 11.5 8.7 75.5 3,511 Electrician 7.5 6.1 84.3 25,955 7.4 7.0 83.3 23,478 Operating engineer 14. l 4.7 76.2 3,719 4.1 12.1 72.0 346 Painter 10.9 8.4 78.4 5,550 9.2 13.8 72.6 434 Pipefitter 7. I 5.8 84.8 8,144 5.3 9.6 83.3 2,498 Plumber 7.7 6.3 84.2 7,712 5.5 3.9 88.4 7,985 Roofer 14.5 I0.0 73.0 9,757 8.9 33.9 52.8 1,531 Sheetmetal worker 7.7 5.8 84.6 9,188 6.1 8.6 83.0 2,875 Struc. steel worker 9.2 7.6 78.1 8,492 15.1 7.6 76.2 172

All occupations 9.0 7.3 81. l 110,246 7.3 7.9 82.3 43,434

Note." Shares of other racial/ethnic groups are not reported. Source: BAT/AIMS.

III. Empirical Model

I adopt a "program-level" approach to examine the determinants of the racial/ethnic composition apprentices. The unit of analysis is the apprenticeship program in a given occupation and a particular year. There were 14,334 such program-occupat ion-year observations over the 1989-1995 period, during which 149,450 new black, Latino, and white apprentices registered (after deleting other racial/ethnic groups and observations with missing information). The empirical analysis predicts the share of each racial/ ethnic group in each observation as a function of characteristics of the demander of apprentices (i.e., the supplier of training), labor supply factors, and occupational con- trols. In the estimation of the determinants of the shares, I converted the data to the individual-apprentice level by duplicating each program-occupation-year observation by the number of apprentices in that observation. Once the data are unfolded in this way, it is straightforward to employ the multinomial logistic regression method to esti- mate the probability that an apprentice belongs to one of the three ethnic/racial groups. Thus, the empirical model is given by:

Prob(Y = j ) = eO~X/[1 + e~X + e~X], (1)

where Y is the variable that records the race/ethnici ty of the apprentice, j indexes blacks (B) and Latinos (L) (white is the base group), and X is the vector of indepen- dent variables.

Page 6: How unions affect minority representation in building trades apprenticeship programs

456 JOURNAL OF LABOR RESEARCH

The empirical methodology amounts to the estimation of a reduced form equation where the elements of the covariate vector X--observable characteristics of the appren- ticeship program, labor market, the occupation, and other control variables--explain the minority shares in training. In their discussion of the obstacles to the entry of the blacks into the trades, Marshall and Briggs (1967) and Mills (1972) emphasized that their low representation is attributable to discrimination and segregation by the trade unions, unwillingness of blacks to enter apprenticeship, and their lack of qualifications and familiarity with the industry. These considerations call attention to the need to con- trol for supply and demand factors in determining the minority share in training. At the same time, they underscore the difficulties that would be encountered in the interpre- tation of econometric findings due to the interdependence of the supply and demand factors. Discrimination and segregation may lead to both the reluctance of the excluded groups to enter the industry and their lack of access to networks that provide informa- tion about issues ranging from the nature of the trades to vacancies in apprenticeship positions. Conversely, minority youth who are disadvantaged in education would find it more difficult to gain acceptance to apprenticeship because the latter is employ- ment as well as training, and the program sponsor is likely to prefer applicants equipped with superior basic skills over others. Given these interdependencies and the conse- quent difficulties encountered in disentangling the supply and demand factors, caution is warranted in interpreting the estimated impact of the observed variables and draw- ing policy implications. 6

The first set of variables is the program characteristics. The variable of prime interest is the program sponsor type. Joint program is a dichotomous variable that takes the value of one, if the program is sponsored jointly by trade unions and employ- ers, and zero if it is sponsored unilaterally by employers. The expected sign of this vari- able is not known a priori. If there is greater employee discrimination in the union sec- tor, and trade unions engage in exclusionary practices in the admission of apprentices, the likelihood that an apprentice is black or Latino would be lower in these programs in comparison with the unilateral employer programs (and hence a negative sign or an odds ratio less than unity). On the other hand, if collective bargaining institutions are more effective in pooling resources to engage in outreach efforts, active recruit- ment, and pre-apprenticeship training of minority youth, the sign of the estimated coef- ficient would be reversed. 7

The second variable, New apprentices, is the total number of apprentices admit- ted to the program during a calendar year. s The shares of minorities should increase by program size because the National Guidelines for Apprenticeship Standards devel- oped by the BAT require program sponsors to submit for approval an "Affirmative Action Plan and Selection Procedures" in conformity with Federal Affirmative Action Regulations if the program either currently employs or expects to employ in the near future five or more apprentices (29 CFR Part 30). Larger programs may also enjoy economies of scale in organizing recruitment drives and pre-apprenticeship training and be able to attract workers of color in greater numbers. The size of apprenticeship programs measured in terms of total registered apprentices at a point in time is not

Page 7: How unions affect minority representation in building trades apprenticeship programs

CIHAN BILGINSOY 457

available. To the extent that there is a positive correlation between the total number of trainees in the program and the number of new entrants, New apprentices variable serves as a proxy for the program size.

The third program variable, Term length, is measured by the on-the-job (OJT) training-hour requirement of the program. There is some (albeit limited) variation in the OJT requirements across programs that provide training in the same occupation, and the term length is included in order to capture how such variation affects the racial/ethnic distribution of apprenticeships. Given the occupation, apprentices trained in programs with shorter requirements presumably acquire fewer skills. They also work for a shorter period for the apprenticeship wage, which is lower than the qualified journey-level wage. The likelihood that an apprentice is black or Latino would vary inversely with the term length if it is easier for him to enter programs that impart less training or the oppor- tunity cost of training, on the margin, is higher for minority apprentices.

The second set of independent variables relate to the local labor market. The first two, Black share in LF and Latino share in LF, are the annually measured relative shares of blacks and Latinos, respectively, in the state labor market. These demo- graphic composition variables account for the supply side of the labor market, with the expectation that the minority shares in apprenticeship vary directly with their over- all share in the state labor force. This is obviously not a refined method to capture the supply side of the story. As pointed out earlier, inadequate basic training or the lack of information about training programs, occupation, or the industry may reduce minor- ity applications and result in their lower representation in training in spite of a high share in the overall labor force (Marshall and Briggs, 1967; Mills, 1972: 159). Lacking detailed information on the characteristics of the pools from which apprentices are drawn, more precise controls cannot be adopted at this time.

The third labor market variable, Unemployment rate, is the annual state construc- tion industry unemployment rate. As construction jobs become scarce, so do the train- ing jobs. Program sponsors are likely to reduce the admitted number of apprentices during such periods, unless they make strategic planning based on the long-term needs of the labor market. Mills (I972: 162) suggested that the lack of jobs in the highly volatile construction industry could create hurdles for the minority workers to find stable employment which would lower their representation in the trades. The unem- ployment rate is included to assess whether the declining number of training jobs affects different groups equally.

The third set of variables is occupation fixed effects to control for occupational characteristics. As an alternative specification, I also used the weekly median earn- ings in each trade in 1990 as reported by the Bureau of Labor Statistics. Results of this alternative specification are not fully reported in the following section but dis- cussed when they are substantially different from the estimates of the specification with the occupation fixed effects.

Finally, geographic region (North, East, South, and West, following the Census Region Classification) and entry-year dummies are also included as controls.

Page 8: How unions affect minority representation in building trades apprenticeship programs

458 JOURNAL OF LABOR RESEARCH

IV. Estimation Results

Tab le 2 reports the es t ima ted odds ra t ios and c o n f i d e n c e in te rva l s o f the r e g r e s s i o n

model . A c c o r d i n g to the regress ion results , un ion par t ic ipat ion in appren t i cesh ip has

di f ferent results for blacks and Latinos. The es t imated odds rat io o f the Joint program

is 1.08, i.e., the odds o f an apprent ice be ing b lack is 8 percent h igher in the j o in t pro-

g rams than the non- jo in t p rograms . This d i f f e rence is s ta t is t ica l ly h igh ly s ign i f i can t

Tab le 2

Multinomial Logistic Estimates That Apprentice Is Black or Lat ino--

Odds Ratios and Confidence Intervals ( C1)

Blacks Latinos

Odds Ratio 95% CI Odds Ratio 95% CI

Program Variables Joint program a 1.08** 1.03-1.14 0.93* 0.88-0.99 New apprentices admitted b 1.02"* 1.01-1.02 1.02"* 1.02-1.02 Program term c 0.98 0.96--1.00 0.93** 0.90--0.95

Labor Market Variables Black share in LF 1.08"* 1.08-1.08 1.03"* 1.03-1.04 Latino share in LF 1.03"* 1.02-1.03 1.13** 1.13-1.14 Unemployment rate 0.97 0.94-1.00 0.84** 0.81-0.88

Occupation Effects d Bricklayer 1.78** 1.58-2.00 1.98** 1.72-2.27 Carpenter 1.36** 1.28-1.45 1.28** 1.19-1.38 Operating engineer 2.47** 2.19-2.80 1.07 0.90-1.27 Painter 1.97** 1.76-2.21 1.95** 1.71-2.22 Pipefitter 0.91 * 0.83-0.99 0.99 0.89-1.10 Plumber 1.10"* 1.02-1.18 1.04 0.95-1.13 Roofer 3.00" * 2.74-3.28 3.54* * 3.16-3.96 Sheetmetal worker 1.13** 1.04-1.22 1.22** 1.11-1.33 Structural steel worker 1.45** 1.31-1.60 1.13* 1.00-1.27

Region Effects e East 1.06 0.98-1.15 1.39** 1.23-1.57 North 0.98 0.92-1.04 1.45"* 1.31-1.61 West 0.67** 0.61-0.73 3.27** 2.99-3.57

N 149,450

Notes: ** (*) indicate that the odds ratio is statistically significantly different from unity at the 1 percent (5 percent) level. Entry year dummies are included in the regression but not reported, aControl: Non-joint program. ~Measured in units of 10. "Measured in units of 1,000 hours of OJT. aControl: Electrician. ~Control: South. Sources: Black and Latino shares in labor force and the unemployment rate are calculated from the CPS Out- going Rotation Files, 1989-1995; all other variables are from BAT/AIMS.

Page 9: How unions affect minority representation in building trades apprenticeship programs

CIHAN BILGINSOY 459

(p << 0.01). The Latino experience is, however, different. The odds of a joint pro- gram apprentice being Latino is lower by 7 percent (p < 0.02). Thus, in comparison with the non-joint programs, black representation is higher and the Latino representa- tion is lower in joint programs. Comparison of the confidence intervals indicates that the effect of program sponsorship is statistically significantly higher for blacks in com- parison with Latinos. Furthermore, the estimated coefficient for Latinos is not as robust as it is for blacks. Once occupation fixed effects were replaced by weekly median wages, the estimated odds ratio for Latinos is still less than unity (0.96), but with a much lower level of statistical significance (p < 0.16). By comparison, the odds ratio for blacks is 1.19 under the alternative specification (p << 0 .01 ) . 9

In summary, the empirical evidence is mixed concerning the hypothesis that a higher degree of racial/ethnic integration is achieved in the absence of trade union participation in apprenticeship training. This is certainly not the case for the blacks; in fact, black share is higher in joint programs. Although the hypothesis appears to be supported for Latinos, the estimates are not robust. Nonetheless, union-manage- ment sponsorship does not have a favorable impact on the Latino share paralleling that for blacks.

The contrasting findings for blacks and Latinos indicate that there is no simple link between the participation of the trade union in training and minority share in apprenticeship and suggest the need for a comparative historical analysis as a future avenue of research. One possible interpretation is that the exclusionary behavior of trade unions continued, but during the period under consideration Latinos replaced blacks as the target of discrimination. An alternative explanation is that blacks had the advantage of building institutions to integrate with the trades at an earlier date, as rec- ognized by Marshall and Briggs (1967) who emphasized that the removal of dis- criminatory procedures is to little avail if black youth are not better educated, trained, counseled, and actively recruited through the cooperative efforts of the civil rights organizations and trade unions. They commended the work of local organizations, such as the Workers Defense League of New York, the NAACP-Urban League Man- power Advancement Program in Cleveland, and the Chicago apprenticeship pro- gram supported by the union and management as well as efforts at the city, state, and federal government levels in the pursuit of this goal. Thus, by the end of the 1980s, there may have been institutions that enhanced the participation of blacks in training in the union sector of the industry, but this was not the case for the Latinos. A historical analysis would help to uncover changes in trade union attitudes vis-h- vis minority groups, effectiveness of Civil Rights organizations in the recruitment of blacks into the trade unions, and whether comparable supportive institutions existed for Latinos.

Turning to other variables, the size of the program is directly related to the odds that an apprentice is black or Latino. The marginal effects on the probabilities for the two groups are identical. The likelihood that an apprentice is black or Latino rises by about 2 percent when the number of new apprentices enrolled in a program in a given year rises by ten (from the mean value of 10).

Page 10: How unions affect minority representation in building trades apprenticeship programs

460 JOURNAL OF LABOR RESEARCH

Odds ratios for the Program term suggest that Latinos are more likely to be in pro- grams with shorter lengths but black representation is not affected by this variable. These results, however, may be attributable to the contamination by the high levels of correlation between the program term lengths and occupational dummies. Within- occupation variation in program terms is limited. Programs in five occupations (elec- trical, carpentry, pipefitting, plumbing, and sheetmetal) have predominantly 8,000 hours of OJT while the other trades have 6,000 hours. Thus, occupation dummies may pick up the effect of term length variable in Table 2. Indeed, when occupation fixed effects were replaced by weekly earnings, both blacks and Latinos are concentrated in programs with lower OJT requirements. Estimated odds ratios were 0.85 and 0.84, respectively, and both were statistically significant at the 1 percent level.

The supply side variables are also highly significant and show that the represen- tation of minorities in apprenticeship varies directly with their relative size in the labor force. A 10 percentage-point increase in the share of blacks in the labor force raises the likelihood that an apprentice is black by 8 percent. An increase by the same magnitude in the share of Latinos in the labor force raises the odds of an apprentice being Latino by 13 percent.

The business cycle affects the Latino share. A five percentage point increase in the unemployment rate lowers the likelihood of an apprentice being Latino by 16 percent. The estimated odds ratios for blacks are only slightly lower and not statistically signifi- cant. Under the alternative specification with median wages, however, the odds ratio for blacks was 0.96 (p < 0.03). Nonetheless, the impact on the Latino share, greater both in terms of magnitude and statistical significance, indicates another arena where the black and Latino experiences in apprenticeship differ substantially. The Latino representation is highly volatile and fluctuates procyclically as Mills suggested (1972), whereas the black share is not as sensitive to business cycles in magnitude or statistical significance.

According to the estimated odds ratios of occupation effects, the probability that an apprentice is black or Latino is highest in the bricklaying, painting, roofing, and, for blacks, in operating engineer occupations. Conversely, it is lowest in the electrical, mechanical, and, for Latinos, in operating engineer trades. Thus, except for the oper- ating engineers, the occupational distribution of blacks and Latinos are broadly simi- lar. Using weekly earnings instead of the occupation effects also shows that minori- ties are concentrated in the lower paying occupations. A $100 increase in weekly earnings lowered the likelihood of being black or Latino by 24 percent and 33 per- cent, respectively. These findings are consistent with the observations made in the pre- vious studies. Thus, the historical pattern of greater concentration of minorities i n what are commonly viewed as "low-prestige" or lower paying occupations persisted in the first half of the 1990s. Furthermore, this is especially severe for the Latinos as the 10-percentage point difference between them and the blacks in the effect of earnings is substantial and statistically significant.

Another noteworthy finding concerns high representation of blacks training in operating engineer programs. Given the small number of non-joint programs in the

Page 11: How unions affect minority representation in building trades apprenticeship programs

CIHAN BILGINSOY 461

occupation, this finding probably reflects primarily the practices in the joint programs. According to Mangum and Walsh (1994: 142), the share of minorities in the operat- ing engineer unions increased after the 1970s because of successful outreach and recruitment efforts of the union locals following the challenges and lawsuits by civil rights groups. The present results concerning apprenticeship programs are consistent with their observations of rising numbers of blacks among operating engineers. The unanswered question, however, is why the Latino share is not similarly higher.

The geographical distribution of minority apprentices also exhibits substantial differences. The odds that an apprentice is black are the same in the East, North, and West, but significantly lower in the West. For Latinos, however, the odds are the highest in the West and the lowest in the South. In comparison with the South, an apprentice is more likely to be a Latino in the West by a factor of three and in the East or North, almost by a factor of one and a half. These findings indicate substantial geo- graphic differences in the supply of labor by blacks and Latinos to apprenticeship programs (after controlling for their shares in the state labor force) or in the recruitment efforts of the programs. Furthermore, the ranking of odds suggests a roughly inverse association between the probabilities of an apprentice being black or Latino across the four regions. The source of this apparent substitutability between the two minor- ity groups is an interesting question for future research.

VI. Conclusion

The construction industry is distinguished from manufacturing in terms of the craft nature of the work, temporary employment, and the lack of a seniority-based promo- tion system. Mills (1972: 158) emphasized that, under these conditions, segregation and exclusionary practices are more likely in hiring and training. Since collective bargain- ing is an influential factor in the formal training of new recruits, in the eyes of many observers, the union is the usual suspect in obstructing the racial/ethnic integration of the labor force. The abundance of historical examples of exclusion of blacks from trade unions and the crafts labor force further fuel the criticism levied against the unions.

After the Civil Rights movement of the 1960s, blacks and, subsequently, Latinos entered the trades in increasing numbers, although both groups were more heavily represented in the lower skilled trades and relatively scarce in occupations that are con- sidered more prestigious. This study contributes to the literature in two ways. First, it uses apprenticeship program-level data. This level of disaggregation is best suited for the analysis because hiring decisions are made at this level. Second, for the first time, it compares the relative representation of apprentices of color in programs organized with and without trade union participation, controlling for other factors.

Empirical results underscore that merging blacks and Latinos into a single "minori- ties" category in the present context is inappropriate. There are substantial differences between the experiences of these groups. First, over the 1989-1995 period, the relative representation of blacks was higher in union-management joint programs. This finding is robust and statistically highly significant. By contrast, Latinos appear to be more

Page 12: How unions affect minority representation in building trades apprenticeship programs

462 JOURNAL OF LABOR RESEARCH

heavily represented in non-joint programs. While the latter finding is less robust, the dif- ference between the two groups concerning the impact of sponsor type on their rela- tive shares is unequivocal. Second, the Latino share was adversely affected by the unemployment rate whereas the black share was less sensitive to the business cycle. Third, occupational distributions of black and Latino apprentices, with the exception of operating engineers, were similar, with lower concentrat ion of both groups in the higher status electrical and mechanical trades. When the occupational impact is mea- sured by earnings instead, it is observed that both groups are more heavily represented in the lower paying trades, albeit with a statistically significant difference between them. Latino share was more sensitive to median earnings. The divergent experiences of blacks and Latinos draw attention to the importance of understanding the process of integration of nontraditional workers into the craft labor force and call for a historical analysis of differences and similarities between the patterns of integration of different minority groups. In particular, an analysis of the impact of the earlier entry of blacks on the Latino latecomers should reveal a wealth of knowledge and policy lessons regarding the access of all nontraditional workers to the skilled trades.

Northrup and Foster (1975, p. 346) wrote: "the open shop sector is both more hos- pitable as a whole to minority employment and, being without craft restrictions and union rigidities, more capable of dealing with the problem (of expanding minority employment in construction)." I addressed only a piece of the larger question of minority participation in the construction labor market. For the blacks, this suggestion is not supported by the data within the realm of apprenticeship, at least when hospitability is measured by the minority representation. It may be true, however, in the case of Latinos.

NOTES

*1 thank an anonymous referee for comments. An earlier version of this paper was presented at the 54th IRRA Annual Meeting, Atlanta, GA, on January 2002. This work was supported by a University of Utah Faculty Fellow Leave, and completed while I was visiting professor at the College of Administrative Sci- ences and Economics, Koq University, Istanbul.

~Hereinafter, I will refer to racial/ethnic minorities simply as minorities.

2The same distinction also exists in the union sector as well, but joint programs are almost exclusively multi- employer programs.

-~Nonparticipating states are CA, CT, DC, DE, HI, LA, MD, NC, NH, NY, OR, VA, VT, WA, and WI.

4For a discussion of women's share in apprenticeship programs in a larger sample of trades, see Berik and Bilginsoy (2002).

~Only 0.6 percent of apprentices were in single-employer joint programs.

~This problem is commonly encountered in the empirical testing of segregation hypotheses derived from the human capital theory as well. It may be instructive to compare the underpinnings of the present approach with that of the human-capital-based studies. The latter, which dear mostly with gender segregation (Polachek, 1981), typically estimates the probability of employment either in a particular occupation or in an occupa- tion with particular characteristics as a function of the worker' s demographic traits and human capital endow- ments. Thus, these studies emphasize the supply side of the labor market and explain occupational distribu- tion as an outcome of individual tastes and preferences, given human capital and biological endowments.

Page 13: How unions affect minority representation in building trades apprenticeship programs

C I H A N B I L G I N S O Y 463

They are usually silent about the demand side of the market. In effect, the methodological difference between this study and the supply-side studies is the switching the independent and dependent variables: While they predict the likelihood that a worker with given characteristics is in a particular occupation, I predict the like- lihood that a worker in a specific training program belongs to a particular racial/ethnic group.

VThese considerations presume that the joint and non-joint programs choose from the same pool of candi- dates. This may not be true. The minority share in, say, joint programs could be higher if the minority appren- tices happen to apply to joint programs in relatively larger numbers. In view of this possibility, it would have been preferable to estimate the probability that a worker is accepted as an apprentice conditional on having applied in the first place. Due to the lack of data on the applicant pool, this route cannot be pursued.

8Continuous variables are measured as deviations from the mean.

9These regression results are available from the author. It should also be noted that the analysis is incomplete without the examination of the performance of blacks and Latinos in training. Bilginsoy (2003) found that while minorities are more likely to drop out as a group, joint program apprentices are more likely to com- plete apprenticeship programs. That study did not distinguish, however, between blacks and Latinos.

R E F E R E N C E S

Berik, GBnseli and Cihan Bilginsoy. "Unions and W o m e n ' s Training for the Skilled Trades in the U.S." Review of Black Political Economy 29 (Spring 2002): 97-122.

Bernstein, David. "The Davis-Bacon Act: Let 's Bring Jim Crow to an End," Washington, DC: Cato Institute Briefing Paper No. 17, January 18, 1993.

Bilginsoy, Cihan. "The Hazards of Training: Attrition and Retention in Construction Industry Apprentice- ship Programs." Industrial and Labor Relations Review 57 (October 2003): 54--67.

Business Roundtable (The). Training Problems in Open Shop Construction. Washington, DC: A Construc- tion Industry Cost Effectiveness Project Report, Report D-4, September 1982.

Mangum, Garth L. and John Walsh. Union Resilience in Troubled Times: The Story of the Operating Engi- neers, AFL-CIO, 196~1993. Armonk, NY: M.E. Sharpe, 1994.

Marshall, F. Ray. The Negro and Organized Labor. New York: John Wiley, 1965.

and Vernon M. Briggs, Jr. "Negro Participation in Apprenticeship Programs." Journal of Human Resources 2 (Winter 1967): 51-69.

Mills, Daniel Quinn. Industrial Relations and Manpower in Construction. Cambridge, MA: MIT Press, 1972.

Northrup, Herbert R. and Howard G. Foster. Open Shop Construction. Philadelphia, PA: Industrial Research Unit, Wharton School, University of Pennsylvania, 1975.

Penn, Roger. "Social Exclusion and Modern Apprenticeships: A Comparison of Britain and the USA." Jour- nal of Vocational Education and Training 50 ( 1988): 259-76.

Polachek, Solomon William. "Occupational Self-selection: A Human Capital Approach to Sex Differences in Occupational Structure." Review of Economics and Statistics 63 (February 1981): 60-69.