Women’s employment, income and divorce in West Germany: a causal approach

Journal for Labour Market Research

Table 3 Sample bounds for the income effect of divorce on daily gross income (regular employment) for the “always observed” under stochastic dominance and/or monotonicity

	P_1\|0	P_1\|1	Stochastic dominance		Monotonicity		Stochastic dominance and monotonicity
	P_1\|0	P_1\|1	Lower	Upper	Lower	Upper	Lower	Upper
Main sample
At day t₀	0.479	0.566	− 59.25	65.56	− 9.17	7.18	0.61	7.18
At day t₃₆₅	0.474	0.544	− 79.22	86.15	− 6.68	7.34	1.82	7.34
At day t₇₃₀	0.477	0.571	− 57.85	69.88	− 7.60	10.06	2.80	10.06
At day t₁₀₉₅	0.502	0.607	− 43.20	53.03	− 10.36	8.98	0.89	8.98
Group A—women not regularly employed in t₋₂₁₈₉ to t₋₇₃₀
At day t₀	0.134	0.268	− 110.45	38.70	− 18.41	7.47	− 5.29	7.47
At day t₃₆₅	0.158	0.331	− 129.05	37.16	− 18.51	7.29	− 5.77	7.29
At day t₇₃₀	0.179	0.415	− 96.87	37.78	− 17.66	13.88	− 1.87	13.88
At day t₁₀₉₅	0.220	0.472	− 131.59	48.40	− 16.70	15.02	− 1.12	15.02
Group B—women with strong labor market attachment (regular employment) in t₋₂₁₈₉ to t₋₇₃₀
At day t₀	0.847	0.857	− 3.89	11.40	1.82	4.74	3.55	4.74
At day t₃₆₅	0.827	0.813	− 4.04	15.00	5.19	7.54	5.94	7.54
At day t₇₃₀	0.794	0.777	− 3.15	20.77	7.69	10.60	8.72	10.60
At day t₁₀₉₅	0.795	0.768	− 5.85	19.76	5.37	10.36	7.34	10.36

Values are in € and P_1|0 and P_1|1 are the employment rates (regular employment) for the controls (column 1) and treated (column 2) with P_1|0 ≡ \( \frac{{\mathop \sum \nolimits_{i = 1}^{N} S_{i} *\left( {1 - T_{i} } \right)*w_{i} }}{{\mathop \sum \nolimits_{i = 1}^{N} \left( {1 - T_{i} } \right)*w_{i} }} \) and P_1|1 ≡ \( \frac{{\mathop \sum \nolimits_{i = 1}^{N} S_{i} *T_{i} }}{{\mathop \sum \nolimits_{i = 1}^{N} T_{i} }} \) with S_i\( \in \) (0,1) if non-employed or employed and T_i\( \in \) (0,1) if control or treated. T₀ is the day the divorce file was opened. Cumulative days of regular employment for group A in t₋₂₁₈₉ to t₋₇₃₀ equals zero and for group B is between 967 to 1460 days. Lower and upper bounds are derived with weights from the logit model and kernel matching. Note that differences in employment rates (between P_1|0 and P_1|1) are different to Table 2 (column 4) because in Table 2 I estimated the ATT-DiD. The difference here (between P_1|0 and P_1|1) represents the ATT