Economists have long suggested that labor unions suffer a free rider problem. The argument is that, since union-set wages are available to all workers covered by unions irrespective of their union status, and union membership entails costs, workers will only join if they are coerced or are offered non-wage goods that they value above membership costs. Yet U.S. and British empirical research has found a substantial union membership wage premium among private-sector union-covered workers, implying that there is no free rider problem. The authors of this study hypothesize that these findings arise due to selectivity problems associated with identifying the union membership effect. Their analysis, which uses rich data from a new linked employer-employee survey for Britain and exploits the within-establishment variation in wages as a function of individual union membership status, demonstrates that the apparent wage premium for members is illusory. Hence, a potential free rider problem remains.