This preview shows page 1. Sign up to view the full content.
Unformatted text preview: 151 lectures Union Impacts on the Labor Market This week we look at the effects of unions on a number of labor market outcomes: wages, wage inequality, benefits, employment, productivity and profits. I will also touch on effects on worker turnover and savings rates. Recall that although union density has been falling, unions still represent a large number of workers and that they are diverse organizations—differing in the construction trades, in manufacturing, in services and in the public sector. I will be summarizing a large research literature that focuses on the average effect of unions—that is, on the impacts of an average union. But keep in mind that the effects may vary crucially with the type of union and with the quality of the employer-union relationship, which involves input and commitment from both sides. I will conclude by reviewing the voice model as summarized in E-S at the end of chapter 13. A. Wages 1. The theory of union wage effects: rent-sharing, superior worker, spillover and threat effects a. Rent-sharing model: Unions clearly try to raise wages of union members; in a rent-sharing model, their gains are at the expense of managerial or shareholder rents, not other workers. b. Superior workers: If union wages are higher, employers will try over time to hire better workers and such productive workers will flock to these good jobs. c. Spillover effect: lowers wages in the non-union sector by shifting labor to that sector. This effect is smaller insofar as there is “wait 1 unemployment”, which involves workers waiting for a job opening to return in the union sector rather than spilling into the nonunion sector. d. Threat effect: raises wages in the non-union sector. Employer wants to pre-empt the attraction of the union. Note the assumption that the employer has some wage-setting market power. e. Implication of the last two: any observed unadjusted union wage premium could overstate (if spillovers dominate) or understate (if threats dominate) actual positive effects on pay. 2. Findings for these models summarized --Rent-sharing variables (product market concentration ratio, profit rate, size and K/L ratio as well as unionization) are significant determinants of interindustry wage differentials. --Superior workers hard to measure: Union workers are somewhat better educated. More on this later. --Threats dominate spillovers when data compare cities of different densities: so nonunion workers’ pay is raised within a city—a local labor market that has more union members. This shows the positive effects of unions within a primary labor market. --Spillovers dominate threat effects when data compare industries of different densities. So there is more wage inequality within these industries. --Text says these findings are contradictory but I disagree. They can be reconciled. The industry-based finding reflects segmentation of industries and workers into primary and secondary labor markets. Many econometric studies of union wage effects hold city and industry constant, so they are looking at impacts beyond these. 3. Methodological problems: causality and data issues 2 a. Causality: need to control for other factors that also influence wages and that influence likelihood of unionism. --Usual individual demographic and locational characteristics: age, education, gender, race, ethnicity, region, industry, city. --Workplace characteristics: size and age of firm, industry, stability of employment (turnover rate), compensating wage differentials for e.g., injuries, firm’s market power, other workplace practices. Only some of these are observable in CPS datasets. b. Data issues: --Some of the above variables may be at least partly endogenous—both cause and effect of unionism. --In studies using cross-sectional data, the challenge is to find a truly exogenous (causal) instrument, will always have questions about omitted variables or about endogeneity. --With time-series data—that have wages before and after a union election and first contract—can still have some omitted variables problems: variables that both allowed union to win election and contract and that are correlated with higher wages. 4. Summary of econometric findings using above methods --union wage premium in the U.S. is between 10 to 20 percent, centered on 15 percent. --quantitative estimate is remarkably similar over different decades—1960s to 1990s—and over different methods and datasets. --Some variation: union premium was higher in the 1970s and lower in the 1990s, tends to be counter-cyclical and is much larger for construction workers than for the average union member. 3 --private union wage effect is larger than public union effect: 17 versus 12 percent. Note that some Federal public sector workers cannot bargain over wages. --union wage advantage is larger in U.S. than in many other countries. E.g., it is only about 6 percent in Germany. This reflects greater effects of union wage setting on nonunion wages in many countries, including Germany. B. Wage inequality --In all industries and in the U.S. economy as a whole, more unionism is associated with less wage inequality. Why? --Unions are associated with standardization of equal pay for equal work and for more rational wage hierarchies within a firm, so they tend to reduce inequality. --Industrial unions raise pay of the unskilled more than that of the skilled workers (and this dominates the effects of craft unions). --Many unionized workers are in the second and third quintiles of the wage distribution—in the middle or below-- so pay increases for union members reduce wage variance. --Unions are associated with reduced pay differences between black and white workers, as they fought racial differentials in the 1960s and 1970s especially. --Overall, a large (perhaps one-third) share of the increase in wage inequality in the U.S. in recent decades is associated with declining levels of unionism in the U.S. C. Benefits --Unionized workers’ benefits—especially health and pension benefits-exceed nonunion benefits by a much greater amount than the unionnonunion wage premium. Often by more than 60 percent. (Freeman and Medoff 1984; Ghilarducci and Reich 2001)
4 --This holds even holding worker age and education constant. --Unions have some positive effects on savings rates—by encouraging worker contributions to their pensions and by providing assistance for their members with navigating the bureaucratic mazes. --Greater benefit effects partly reflect interests of more active union members—who are older, partly reflects union impacts on reducing turnover, so workers have more interest in their benefits, and partly reflects preference for benefits over wage increases when benefits are not taxable. --Because of anti-discrimination rules in Erisa, many more nonunion white collar workers get more of these benefits when blue-collar workers are unionized than when they are not. D. Employment levels --E-S text suggests that a few studies show that unionism leads to slower employment growth. --This reading of the literature seems less clear to me, as there are very few studies. Unionized industries are in much more mature, even declining industries, so it is very difficult to get a good control group. One study of 42 metro areas (Montgomery JOLE 1989) found that a 10 percent increase in the union wage premium reduced employment by 0.06 percent, a very small effect. --In a rent-sharing model, wage increases do not necessarily produce employment declines. E. 1. Productivity E-S text cites two views --Negative managerial prerogatives view: unions interfere with managerial decisions. 5 --Positive worker voice view: unions and voice imply much lower level of quits, so training will be greater. --According to E-S, empirical results are conflicting and the quality of the relationship makes a difference. --Early econometric studies were based on specific industries: cement, school and highway construction, minimills in the steel industry. Most found worker voice effect dominated. 2. My reading is again different --The best studies (by Black and Lynch) have data that surveys from all the manufacturing industries and that combines information on the workers, the workplaces, capital-labor ratios, the quality of the union-management relationship and the use of innovative Japanese-style employee involvement techniques and union-management cooperation (in contrast to traditional Taylorism with just adversarial relations between unions and management). [See also Brown, Nakata, Reich and Ulman, Work and Pay in the U.S. and Japan. Oxford University Press, 1997] --Black and Lynch arrive at a four-way ranking of workplace productivity by: --Union vs. nonunion AND traditional vs. innovative high-performance workplace practices (HPW). --Productivity compared to a non-union and traditional workplace: U and HPW: Non-U and HPW: U and not HPW: +20 percent +11 percent -15 percent --Could say union effect is a wash if average the two union categories, but that would be misleading. With the additional dimension of HPW/not HPW we get results that are not conflicting and that make a lot of sense. F. Profits
6 Most studies suggest that profitability is lower in unionized environments, with or without increases in productivity. But this redistribution does not necessarily reduce economic efficiency. If higher union wages come at the expense of monopolistic rents, then there are no negative effects on efficiency. If higher wages come at the expense of competitive level profits, then efficiency may suffer as there is less investment and technological progress. If higher wages are associated with higher productivity because of positive voice effects, then there can be positive effects on efficiency even with lower profits. G. Conclusion: look again at the voice model 7 ...
View Full Document
This note was uploaded on 06/06/2010 for the course ECON 151 taught by Professor Staff during the Spring '08 term at University of California, Berkeley.
- Spring '08