One of the major themes of the tort anti-reform movement has been the assertion insurance premiums have not fallen after tort reform. Ty and I are doing some research on this (which we will post after it is finished). However, one of the things I was reading was the paper put out by the Center for Justice & Democracy entitled Premium Deceits. In it, they count the number of major tort reforms in a state and then they look at whether the change in premiums over the period 1985-1998 is related to the number of major reforms. They find that price changes over the period are not related to the number of reforms. The report then makes the claim, that tort reform is deceptive and that it will not have an effect on lowering premiums.
I thought I'd talk about two things I noticed in the report. While the report was carefully done in terms of enumerating the states' reforms, it doesn't necessarily make sense just to add the reforms up into an index of tort reform. For example, one state may have a collateral source rule change and a damage cap enactment. Another state may have a damage cap enactment and a change in joint and several liability . ... both would count as a '2', but they may or may not have similar impacts. So, I thought this is a problem and is generally treated in academic papers by using a dummy variable for each significant tort reform.
So what if I disaggregated the Center's tort reform index and made a dummy variable for those states with two tort reforms and those with three tort reforms? To see the effect of this minor change in defining the index, I estimated two regressions. (If your eyes are glazing over just skip down to the § symbol below.)
I estimated a simple regression (something the Center did not do in the report) of the form: percentage change in products liability premiums = a + b*Number of Major Tort Reforms Enacted.
I obtained the following results:
*** significant at the 0.001 level.
This is essentially the conclusion that the authors made. Tort reform was not statistically related to reductions in the change in prices over the period.
However, if I were to break out the number of reforms into two dummy variables (TR2 =1 if state has enacted 2 tort reforms, and 0 otherwise, TR3 =1 if the state has enacted 3 tort reforms and 0 otherwise) I get a slightly different result:
***significant at the 0.001 level and ** significant at the 0.05 level.
§This result suggests the tort reform does have an effect on lowering prices, but the number of reforms that accomplished this goal is 2. This result refutes the Center's conclusion (at least for products liability) as the coefficient on TR2 is significant at the 0.05 level. In simple terms, two tort reforms were associated with a 26 percent reduction in the percentage change in product liability premiums over the period in question.
This is still unsatisfactory, though, as one can not really add the state's tort reforms up to get a simple additive index of tort reform. I don't even believe that two reforms make a difference but three do not. So as all good academics are wont to claim, more study is needed.
FYI: The data used are all available in the appendix of the Center for Justice and Democracy's paper which I linked to above.