Schap, Guest and Kraynak (2013) examines time series properties of medical net discount rates (MNDRs) based on three different, short-term Treasury securities. The article finds attributes of stationarity in each of the series examined and notes greater support for total offset of interest rates and medical cost growth rates than does previously published research.
The present study makes four adjustments to Schap, Guest and Kraynak (2013). First, the data set is updated by two years and four months. Second, the t-test applied in the original study and also used in earlier studies of total offset fails to account for autocorrelated error terms, so a substitute test is applied to both the original data set and the newly extended data set, with results reported. Third, Phillips-Perron testing is conducted in addition to Augmented Dickey-Fuller and Kwiatkowski-Phillips-Schmidt-Shin testing of stationarity of the various series analyzed, with results reported. Finally, Zivot-Andrews testing is applied diagnostically, to identify stationary sub-series of MNDRs, similar to an approach used successfully to identify stationary sub-series of wage net discount rates (Schap, Baumann and Guest, 2014).
The data analysis suggests two options available to practicing forensic economists who endorse the use of Treasury securities of short duration in formulating MNDRs. One option is to use total offset (i.e., a zero MNDR), based on modestly favorable empirical support in the period 1981:01 to 2014:10. The alternative option would be to rely on a body of evidence amassed over a shorter time frame, from 2001:01 to 2014:10, but which has remarkably strong stationarity properties and yields MNDRs between minus 1.5 and minus 2.0% (i.e., an implicit forecast of medical cost growth exceeding the discount factor).