Social Insurance and Truncated Benefits: Measuring the Impacts of Workers' Compensation

Table 4

Regression estimates by occupation for the comprehensive benefit index model. Model: OLS with state-specific comprehensive annualized benefit index⁺.

Estimated impact of $1 change in present value of expected benefits
Occupation	Coefficient	t-statistic	𝑁	𝑅^2
Cranemen	−$1.00	(−0.44)	878	0.56
Deliverymen	$3.15	(1.34)	1406	0.16
Machinists	$1.75^∗∗	(1.99)	5570	0.38
Meat cutters	$0.20	(0.19)	4618	0.35
Millwrights	−$2.66	(−0.76)	407	0.63
Painters	$1.02	(0.98)	3991	0.19
Rollers	−$1.66	(−0.36)	361	0.58
Roofers	$5.13^∗∗	(2.22)	1089	0.20
Sawyers	$4.34^∗∗	(1.97)	878	0.32
Welders	$2.07	(1.62)	1940	0.26

⁺This model uses the real annual wage as the dependent variable. Each coefficient comes from a separate regression that pools all years for each occupation to form a pseudopanel. The model is estimated in levels (i.e., the dependent variable and the benefit regressor are both in real US dollars), so coefficients can be interpreted as the change in real annual wages due to a one dollar increase in the real net present value of annual expected benefit. Each regression includes control variables for education, experience, race, and demographics as well as year dummy variables and state-fixed effects. (t-statistics are in parentheses.)