finds that THS employment is positively associated with the implied contract exception
but not to the other exceptions in his favored fixed effects specification that also contains
a set of state-specific time trends. This result, which is consistent with his priors,6 is
robust to additional controls such as labor force demographics and the percentage of the
state workforce that is unionized. (Interestingly, the latter coefficient estimate is negative
and highly significant, indicating that temporary employment grew less rapidly in states
where unions declined less - given the decline in union density of over one-third over the
sample period - which is of course consistent with union opposition to THS
employment.) The bottom line is the finding that the implied contract exception
contributed about 500,000 additional jobs (or some 20 percent) to the growth of THS
employment. Independently, slower rates of union decline added to this total.
In Krueger (1991) the exceptions to at-will are taken to be exogenous. In a subsequent
study of the state employment effects of these legal incursions, 1980-87, Dertouzos and
Karoly (1992, 1993) argue that the probability of having one of the wrongful dismissal
doctrines is strongly related with a number of state characteristics. Their instruments are
whether a state had a right to work law (see subsection (iii) below), whether it had a
Republican governor, the level and change in union density, the change in
unemployment, the percentage of neighboring states recognizing a similar exception, the
percentage change in lawyers per capita, and year dummies. They find that right-to-work
states and those with a Republican governor (indicative of a conservative attitude toward
labor) are less likely to have either a tort-based or contract-based exception, while the
converse is true for the degree of unionization variable. There is also some evidence of
spillover: the higher the fraction of neighboring states that have recognized the respective
doctrine, the more likely is the state to have the doctrine. The effect of the other
instruments is either mixed or statistically insignificant.
Dertouzos and Karoly model the determinants of the exceptions to at-will in an attempt to
provide unbiased estimates of their employment effects, since they argue there is
simultaneous determination of the employment and the legal environment. Having
instrumented the doctrine/remedy, the authors use the predicted values in place of the