Accounting for censoring in duration data: An application to estimating the effect of legal reforms on the duration of medical malpractice disputes

Using a sample of medical malpractice insurance claims closed between 1 October 1985 and 1 October 1989 in the USA, we estimate the impact of legal reforms on the longevity of disputes, via a competing risks model that accounts for length-biased sampling and a finite sampling horizon. We find that only the 'English rule'-a rule which requires the loser at trial to pay all legal expenses-shortens the duration of disputes. Our results for this law also show that failure to correct for length-biased sampling can incorrectly imply that the English rule lengthens the time needed for settlement and litigation. Our estimates also suggest that tort reforms that place additional procedural hurdles in the plaintiff s' paths tend to lengthen the time to disposition. Here, correction for a finite sampling horizon substantially changes the inferences with regard to the eff ect of this reform on duration.