Labor Unions and Life Satisfaction: Evidence from New Data

We do not ignore the fact that recent dissenting opinions call into question the empirical usefulness of contemporary happiness/life-satisfaction research. Wilkinson’s (2007) thoughtful piece expounds the position that most happiness surveys do not in fact capture precisely what they intend to in respondents’ answers, and that better designed surveys will be necessary in the future to justify the often sweeping claims of happiness scholars. Nonetheless, for now the scholarly consensus is that the survey instruments hold up reasonably well provided one is careful not to attribute explanatory power to them beyond what they represent.

Some evidence in the literature suggests that union members are actually more dissatisfied than non-members, but also that they are far less likely to quit then non-members (Freeman and Medoff 1984). This puzzle was resolved by applying the “voice hypothesis,” in that unionization allows members to complain about their working conditions since they are in a position to ameliorate them through collective action. Workers thus sought to improve their working conditions rather then “exit” because they could, and presumably because they valued their jobs enough to try. There is also an endogeneity problem that Pfeffer and Davis-Blake (1990) successfully explain. Clark (1996, 202) nicely addresses the issue: “if unions address worker dissatisfaction, the more dissatisfied workers will be the most attracted to union membership,” so that union shops will emerge in those industries and under those employers, that create the most initial dissatisfaction. When controlling for this effect, Pfeffer and Davis-Blake demonstrate that “unionization has a significant positive effect on [job] satisfaction.” Similar evidence is provided by Bender and Sloane (1998).

When respondents who were former but not current union members are instead coded as non-members, the results are substantively identical to those we report below.

Data on social welfare expenditures and unemployment rate are from the OECD. Data on GDP are from Penn World Tables. Data on a country’s level of “individualism” are drawn from Triandis (1993).

Country-level measures of life satisfaction are computed by taking the mean response to the 1–10 scale life satisfaction item on the 2005 WVS.

These results are robust to the inclusion of additional variables that measure a country’s culture, political ideology, and general disposition toward labor unions. Specifically, we added country-level mean values for post-materialism (to control for the general culture of a country, y001), political ideology (to control for a progressive cultural outlook, v114), and general trust of labor unions (to control for political receptivity to the demands of organized labor, v135) from the 2005 wave of the WVS to the model specification for Table 1 and all subsequent individual-level models (Tables 2 and 3). In all models, the coefficient for union density and/or union membership remains statistically significant and its substantive impact is of a similar magnitude.