Table 3.

Propensity score nearest neighbour matching

Outcome	Hours of caring per week (more than 15 h)
	Number of observations treated (controls)	ATT (Std. Erra)
Employed	596 (692)	− 0.111 (0.029)***
Unemployed	596 (692)	0.007 (0.014)

Average treatment effects on the treated (ATT). The numbers of treated and controls refer to actual nearest neighbour matches. Treatment (T) is carer hours > 15 h p/w (1) > 15 h p/w (0). Outcome (Y) is employed (1, 0), unemployed (1, 0). Independent variables are gender, age, education, marital status, region (the variables ‘postsec’ and ‘southwest’ are dropped to satisfy the balancing property)

*** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level

^aBootstrapped standard errors