15
and uncertainty) to employers of using the courts and describe how 25-40 percent of any
compensation awarded employees might be swallowed up in legal fees. Accordingly,
they identify a broad consensus favoring reform and describe the outcome as a win-win
situation - for all but trial lawyers. Disputation was over the details: state-run versus
private insurance and the level of indemnity benefits. In the former area, organized labor
lined up against the insurance companies (employers offered mixed support for a state-
run insurance fund); in the latter organized labor and large employers were in obvious
contention (but see immediately below). The outcomes were determined by variations in
the political strength of these groups (see also Fishback and Kantor, 1996b).
Additional insight into employer support for workers’ compensation (and the speed with
which the laws were enacted across most states) can be gleaned from an analysis of
wages. Fishback and Kantor (1994) contend that although expected injury compensation
rose considerably with the passage of the state laws - both as a result of more individuals
receiving compensation and compensation levels that were considerably higher than
under negligence liability - much or all of the employers’ costs were shifted back onto
employees. The authors construct three panels for relatively dangerous industries -
coalmining (1911-22), lumber (1910-13, 1915, 1921, and 1923), and union contracts in
the building trades (1907-13) - in each case regressing hourly earnings in a state
regressed on an index of expected injury benefits (computed both prior to and after the
introduction of workers’ compensation) and a fairly wide set of controls (product prices
or demand index, output per man, unionization, strike activity, and occupation dummies,
etc). For coal mining the authors’ fixed effect estimates suggest that workers not only
paid for the sharply-stepped increase in their expected benefits but may also have fully
paid the employers’ costs of purchasing insurance to provide those benefits. In the lumber
industry, there appears to have been a full wage offset. But in unionized business
construction (and indeed unionized coalmining) the coefficient estimate for the expected
benefits variable was statistically insignificant, indicating an absence of any downward
adjustment. These results were broadly robust to specification and to sample (restricting
the sample to states and years when workers’ compensation was in effect.). Fishback and
Kantor (1994, p. 19) conclude: “The presence of wage offsets for nonunion workers also