Right-to-work laws and free riding.

Davis, Joe C. ; Huston, John H.

I. INTRODUCTION

Section 14(b) of the Taft-Hartley Act permits state right-to-work (right-to-work) laws which prohibit compulsory payment of union dues. By allowing employees to enjoy the benefits of a union contract without the costs of union membership, right-to-work laws can reduce not only the demand for memberships but also the supply as unions lose incentives to organize and maintain bargaining units.

The significance of right-to-work laws as a deterrent to membership, however, is open to dispute. Olson |1971, 88~ reasons that right-to-work laws must be ineffective because the incentive to free ride is so great that large unions could not survive if workers were free not to join. Olson considers the continued existence of large unions in states with right-to-work laws as evidence that these laws are not being enforced. Farber's |1984~ finding that the supply of union jobs relative to demand is comparable in right-to-work and non-right-to-work states also suggests non-enforcement. Hirsh |1980~ finds that while right-to-work laws do not affect coverage, they do affect membership. He infers that free riders account for the slippage, that is, right-to-work laws do not deter unions from organizing firms but they do interfere with unions signing up individual members. Freeman and Medoff |1984, 243~, Bloom and Northrup |1981, 227~ and others point to membership figures indicating that right-to-work laws create serious free rider problems for unions. Some 15.5 percent of employees covered by collective bargaining contracts free ride in states with right-to-work laws, about 2.4 times the 6.4 percent who free ride in other states.(1)

The literature on union membership, however, makes it clear that such bivariate comparisons neglect other factors that could affect free riding. Jones' |1982~ research using the National Longitudinal Survey reveals that nonmembers covered by union contracts are more likely to be younger, to live in the South, and to work in white-collar jobs than members are. Moore, Dunlevy, and Newman |1986~ contend that the same attitudes that give rise to right-to-work laws also affect union membership. When these attitudes are properly controlled for, right-to-work laws have no significant effect on membership. Moore and Newman |1985~ provide a good review of the literature on union membership. Although these studies and others have examined the effects of right-to-work laws on union membership of all workers, the impact on only those workers covered by union contracts has not been as thoroughly explored.

Previous analysis of free riding has been extensive but largely theoretical. This study fills a gap in the literature by developing an empirical model of free riding. Having created a statistical model of free riding we can test the marginal impact of right-to-work laws. We find that, after taking into account other factors that affect the decision to free ride, right-to-work laws remain statistically significant, although of a smaller magnitude than bivariate analysis would suggest.

II. CAUSES OF FREE RIDING

The theoretical model used to explain free riding is abstracted from the standard model of the demand for and supply of union membership that Moore and Newman |1985~ describe in detail. One of the advantages of focusing on workers covered by union contracts is that we avoid the simultaneity problem inherent in modeling the supply as well as the demand for union services. We limit our analysis to the demand for membership because the union, as a rule, must supply membership to all covered workers.(2)

The key variables that influence the demand for membership by covered workers include price (dues and fees), the union wage premium, job security and pressures from peers and community. As in other studies of union membership, data limitations do not allow us to gauge the effect of dues and fees. Where membership is not compulsory, free riders do not face immediate loss of the union wage because by law the union must negotiate the same wages and benefits for nonmembers as for members. Of course, if sufficient numbers of employees free ride, the survival of the union as bargaining agent may be endangered, but the marginal impact of one additional free rider is likely to be minimal. Still, some employees may be less willing to free ride than others. For instance, older workers who fear losing the union's seniority-based wages and job protections may free ride less than younger employees with better job opportunities elsewhere. Thus by controlling for age and other characteristics which reflect the costs and benefits of membership to individuals, we can more accurately measure the extent to which right-to-work laws, by removing the constraints on demand for membership imposed by union shops, affect free riding.

The sample used in estimating our empirical model of free riding consists of 4,606 full-time employees covered by union contract from the pooled April, May, and June 1985 tapes of the Current Population Survey; 378 of the employees were not union members. We exclude the self-employed; supervisors; and government, airline, railroad, and agricultural employees who are not subject to the Taft-Hartley Act.(3) Table I presents descriptions of the variables. The dependent variable FREE RIDE takes a value of one if the worker covered by union contract is a free rider and zero if the worker is a union member.

The key explanatory variable is RTW, which indicates whether or not an TABULAR DATA OMITTED individual's state of residence has a right-to-work law. Several factors may weaken the hypothesized positive relationship between RTW and FREE RIDE. First, residence in a non-right-to-work state does not necessarily mean that a covered employee's union has been able to win a union shop in collective bargaining. In fact, only 77 percent of workers covered by major collective bargaining agreements in non-right-to-work states work in union shops, although some of the remaining 23 percent work in modified union shops that exempt from compulsory payment of dues those workers employed when the firm is organized.(4) Second, free riding is discouraged in right-to-work states. For example, peer pressure, as Moore and Newman |1985~ observe, has a strong negative effect, and unions may seek contract clauses requiring employers to recommend union membership and to allow the union to make presentations to new employees.(5) Third, as Haggard |1977, 153,57~ points out, unions bargain harder in right-to-work states for checkoff clauses that are not subject to right-to-work laws. Some 92 percent of workers covered by major contracts in right-to-work states are subject to dues checkoff, compared to 82 percent in other states.(6) Employee authorization of dues withholding deters membership attrition, because the authorization is often irrevocable for the term of the contract.

The impact of our control variables (race, gender, marital status, age, education, region, occupation, industry and firm size) on union membership in general (see Moore and Newman |1985~; Farber |1985~; Olson |1971~) help us form expectations about their effect on free riding. The worker characteristics found to be positively correlated with union membership should be negatively correlated with free riding. (Workers who perceive greater benefits to membership should be less likely to free ride.) We expect nonwhite, married, less-educated and older individuals to free ride less because the wages and job security provided by unions is of greater benefit to them. In contrast, these union advantages may be less important to women if on average they expect to spend fewer years in the labor market, and they may free ride more. Free riding should also be more likely among those employed in the service industries (where labor-intensive production may limit the ability of unions to win substantial wage premiums) and in white-collar occupations (where some employees may view membership as a barrier to promotion opportunities). Residents of the South have traditionally been more likely to decline membership than those in other regions. Following Olson, we expect that free riding will more likely in large firms where monitoring by fellow employees is more difficult. Firm size was not available for the individuals in our sample. Instead, we have included the average firm size (calculated from U.S. Bureau of the Census, |1987~) for the three-digit industry in which the individual is employed. People employed in industries with a high average firm size are more likely to be a member of a large company and, if Olson is correct, more likely to free ride.

The prevalence of free riding may also be a function of public opinion toward unions. In areas where unions are viewed favorably, the likelihood of an individual choosing to free ride is hypothesized to be lower. The variable COPE is used in other studies (e.g., Carroll, |1986~; Moore, Dunlevy and Newman, |1986~) to capture the general climate of public opinion. Its inclusion is important because the attitudes that lead to the passage of right-to-work laws may also be associated with increased free riding. Regressing free riding on RTW without some measure of public attitudes toward unions would bias the coefficient on RTW upwards. COPE, calculated from Pressman |1985~, shows the rating given to the federal legislative delegation of the individual's state in 1984. For example, a rating of .65 indicates that 65 percent of votes cast by the delegation endorse the AFL-CIO's legislative program. We predict an inverse relationship between COPE and FREE RIDE.

III. EMPIRICAL RESULTS

To avoid the biases inherent in the linear probability model, a probit procedure is used to determine the effect of each of the variables on the probability of free riding. Results of ordinary least squares and probit regressions are shown in Table II. The mean predicted probability of free riding is .0821, which compares favorably to the true proportion of free riders, .0820. Of the eleven independent variables examined, seven are statistically significant at the 10 percent level or better with the expected sign. (Six are significant at the 1 percent level.) Free riding is positively related to living in a right-to-work state, living in the South, being female, having attended college, and working in white-collar or service occupations; it is negatively related to age. The lack of significance of the COPE coefficient is troubling but, in retrospect, not surprising given the high correlation between COPE and RTW. (If RTW is omitted, COPE becomes highly significant.)(7)

Olson's hypothesis that free riding would be more likely in large firms cannot be verified by these results. The coefficient on FIRM SIZE was not significantly different from zero. Alternative specifications such as a dummy variable for large firms or a cross-product term between firm size and RTW also generated insignificant firm-size effects. This may, however, simply reflect the insufficient specificity of our data with regard to firm size. The hypothesis that nonwhites would find union membership more valuable and free ride less was also not confirmed. Nonwhites appear to free ride no more or less than whites.

The magnitude of the effect of right-to-work laws is distinct from its statistical significance. Using the above equation, we can forecast the effects of repealing right-to-work legislation by substituting 0 in place of 1 for the RTW variable for each observation in a right-to-work state. The estimated proportion of free riders in those states drops from .155 to .072. Note that the model predicts that these workers would still free ride somewhat more often than their counterparts in non-right-to-work states, where .064 of the covered workers free ride. The nationwide rate of free riding declines by only 1.6 percentage points, from .082 to .066.(8)

IV. CONCLUSION

Our goal in this research was to create an empirical model of free riding and to find direct evidence of the effects of Section 14(b) of the Taft-Hartley Act are open to dispute because they ignore other variables that may affect an individual's decision to free ride. When we control for these influences, the effect of right-to-work laws remains statistically significant but of a smaller magnitude than that found in previous studies. Although right-to-work laws remain an important factor explaining the difference in free riding between states with and without right-to-work laws, the overall impact is not remarkable since only 20 percent of workers reside in right-to-work states.

TABLE II

Estimates of Free Riding by Employees Covered by Union
Contracts

 PROBIT OLS
VARIABLE COEFFICIENT COEFFICIENT

RTW .493(**) .080(**)
 (.095) (.014)

COPE .122 .031
 (.222) (.030)

NONWHITE .077 .010
 (.075) (.011)

FEMALE .120(*) .019(*)
 (.068) (.010)

MARRIED -.082 -.010
 (.063) (.009)

AGE -.012(**) -.002(**)
 (.003) (.0003)

SOUTH .238(**) .040(**)
 (.081) (.012)

WHITE COLLAR .635(**) .114(**)
 (.071) (.011)

SERVICES .248(**) .039(**)
 (.066) (.010)

FIRM SIZE -.00005 -.000007
 (.0002) (.00002)

COLLEGE .221(**) .035(**)
 (.064) (.009)

INTERCEPT 1.535(**) .060(**)
 (.169) (.023)

N 4606 4606

Likelihood
Ratio Test 354(**) F-Test 39.1(**)

|R.sup.2~ -- 0.86

Notes: Standard errors in parenthesis.
* Significant at the 10 percent level
** Significant at the 1 percent level

Our results point to two lines of future inquiry. First, if right-to-work laws do promote free riding, how does one account for the limited impact on the supply of memberships as suggested by Farber and Hirsh? Second, to what extent does free riding curtail the ability of unions to win wage gains for covered workers?

JOE C. DAVIS and JOHN H. HUSTON Associates Professors, Trinity University. We wish to thank the editor and two anonymous referees for their valuable suggestions.

1. Calculated from a 1985 Current Population Survey sample of employees who are subject to the Taft-Hartley Act, Section 14(b). "Covered" employees include those represented in collective bargaining by traditional labor unions and employee associations.

2. Our concentration on demand is supported by the studies of Farber and Hirsh cited above which suggest that the supply of union jobs relative to demand is comparable in right-to-work and non-right-to-work states.

3. We are unable to identify two other groups of employees whose membership choice is not affected directly by Taft-Hartley: (1) employees in union-shop states who have not been employed long enough for mandatory membership and (2) employees of private firms on federal enclaves in right-to-work states, such as military bases and national parks. Had we been able to exclude such workers, the measured impact of right-to-work laws on free riding would be greater.

4. U.S. Bureau of Labor Statistics |1982, 84~.

5. U.S. Bureau of Labor Statistics |1982, 17~. The result can be de facto union shops, as predicted by Samuel Gompers |1921, 84~: "Even if all the courts in the country should decide that the union shop contract is illegal, an impossible situation, the union shop would not disappear. The only result would be that no such contract would be made; the condition would be enforced without written contract."

6. U.S. Bureau of Statistics |1982, 58~.

7. We were concerned that the lack of significance of COPE reflected its failure to capture states' attitudes toward unions. Because those attitudes are likely to affect both free riding and the passage of right-to-work laws, there is the potential for biasing the coefficient on RTW upwards. To examine this possibility, we performed a two-stage procedure in which RTW was modeled as a function of exogenous variables to remove the effects of local attitudes. The coefficient on RTW remained statistically significant at the 5 percent level but did decrease somewhat in magnitude. For a good discussion of the simultaneous nature of RTW laws and unionization, see Wessels |1981~.

8. Using the two-stage results, the decline in free riding is 1.0 percent.

REFERENCES

Bloom, Gordon F., and Herbert R. Northrup. Economics of Labor Relations, 9th ed. Homewood, Ill.: Irwin, 1981.

Carroll, Thomas M. "Do Right-to-Work Laws Matter? Reply." Southern Economic Journal, October 1986, 525-28.

Farber, Henry S. "Right-to-Work Laws and the Extent of Unionization." Journal of Labor Economics, July 1984, 319-52.

-----. "The Extent of Unionization in the United States." In Challenges Facing American Labor, edited by Thomas A. Kochan. Cambridge: MIT Press, 1985, 15-45.

Freeman, Richard B., and James L. Medoff. What Do Unions Do? New York: Basic Books, 1984.

Gompers, Samuel. "The Union Shop and the 'Open Shop'," in Selected Articles on the Closed Shop, edited by Lamar T. Berman. New York: H. T. Wilson, 1921.

Haggard, Thomas R. Compulsory Unionism, the NLRB and the Courts. Philadelphia: University of Pennsylvania, 1977.

Hirsch, Barry T. "The Determinants of Unionization: An Analysis of Interarea Differences." Industrial and Labor Relations Review, January 1980, 147-61.

Jones, Ethel B. "Union-Nonunion Differentials: Membership or Coverage?" Journal of Human Resources, Spring 1982, 276-85.

Moore, William J., and Robert J. Newman. "The Effects of Right-to-Work Laws: A Review of the Literature." Industrial and Labor Relations Review, July 1985, 571-85.

Moore, William J., James A. Dunlevy, and Robert J. Newman. "Do Right to Work Laws Matter? Comment." Southern Economic Journal, October 1986, 515-24.

Olson, Mancur. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge: Harvard University Press, 1971.

Pressman, Steven. "Interest Groups Give Members Their Grades." Congressional Quarterly, 20 April 1985, 741-49.

U.S. Bureau of the Census. County Business Patterns, 1985. Washington, D.C.: U.S. Government Printing Office, 1987.

U.S. Bureau of Labor Statistics. Major Collective Bargaining Agreements; Union Security and Dues Checkoff Provisions. Bulletin 1425-21. Washington, D.C.: Government Printing Office, 1982.

Wessels, Walter J. "Economic Effects of Right-To-Work Laws." Journal of Labor Research, Spring 1981, 55-75.