25
Table 5: The effect (hazard ratio) of being offered early retirement on the conditional
probability of dying, based on a discrete-time Cox regression model
Censoring at age 71
Censoring at age 66
(1)
(2)
(3)
(4)
(5)
(6)
Hazard ratio
0.7275*
0.7242*
0.7424†
0.5899* 0.5926* 0.6109*
(0.1107) (0.1102) (0.1148) (0.1215) (0.1222) (0.1283)
County
No
Yes
Yes
No
Yes
Yes
Control variables
No
No
Yes
No
No
Yes
Note. Estimation is performed with maximum likelihood. Standard errors are adjusted for 19, 906 clusters
(individuals): † p<.1; * p<.05; ** p<.01. Control variables are income, income interacted with dummy for cohort
1938-1939, and education. All models include a military dummy and dummy for cohort 1938-1939. The number of
observations is 283,208 and 232,087 for the estimates that are censored at age 71 and age 66 respectively.
6.4
Heterogeneous treatment effect
The question remains if we can identify which factors in the offer of early retirement
that protects from health problems. One might, for example, expect health problems
related to a hazardous work environment to be reduced, especially since the effect is
potentially larger at younger ages, i.e., at ages before the normal retirement age.
Another, complementary, hypothesis is that early retirement lowers the risk of stress-
related heart problems among recipients of the offer (e.g., Steptoe & Kivimäki, 2013).
On the other hand, there is no obvious link between working and cancer. There can be
an increased risk for cancer in occupations with specific risk exposures, but then the
neoplasm (tumor) should not directly be related to early retirement.
