Leveraging the macro-level environment to balance work and life: an analysis of female entrepreneurs’ job satisfaction

The average variance inflation factor of Model 2 (excluding the squared age term) is 1.84, which is below the conservative threshold of 5, so there is no notable indication of multicollinearity.

Consistent with COR theory (Hobfoll 1989, 2001), our theoretical focus is on the negative linear relationship between work–life imbalance and job satisfaction, and then how this linear relationship might be mitigated by women-friendly macro-conditions. In a post hoc analysis, we also estimate a potential curvilinear direct relationship between work–life imbalance and job satisfaction, to test for shifting marginal effects. We find no such effects.

A follow-up analysis indicated that the difference in the mean values of job satisfaction at high versus low levels of emancipative values was not significant across the lower range of work–life imbalance values but became significant at high levels. The significant p values in Table 4 provide support for the hypothesized moderating effects, but these effects are weaker in the case of emancipative values.

To investigate the role of education specifically, in another post hoc analysis, we assessed the extent to which the interplay of female entrepreneurs’ work–life imbalance and the three macro-level factors in predicting job satisfaction depended on education levels. We thus added corresponding three-way terms (work–life imbalance × action resources × education, work–life imbalance × emancipative values × education, and work–life imbalance × civil entitlements × education) to Models 3–5, respectively. The findings indicated positive, significant three-way interactions for the first two macro-conditions. Highly educated female entrepreneurs thus appear better able to leverage women-friendly action resources and emancipative values to mitigate the hardships of work–life imbalances.

The detailed results are available on request.

In light of this caveat, we conducted two robustness tests to check for endogeneity. First, work–life imbalance and other individual-level variables might correlate with the random intercept, which represents the effects of omitted level 2 covariates. When we include the country-level means of all individual-level covariates, the estimated coefficients for the covariates, which vary at the individual level but are not susceptible to cluster-level confounding, are directly comparable with those reported in Table 4. Second, endogeneity could be caused by reverse causality, so we also ran the regressions in the opposite direction. Job satisfaction and work–life imbalance are still positively related, but we do not find any significant interaction effects. This outcome supports our basic premise that women-supportive environmental conditions help female entrepreneurs overcome the energy resource depletion that they experience in the presence of work–life imbalances (not that environmental conditions influence the impact of job satisfaction on work–life imbalance). The results of these two robustness checks are available upon request.