Problems in the Use of Historical Data in Estimating Economic Loss in Wrongful Death and Injury Cases

Abstract

In his Comment, William Harris suggests an alternative to first differencing as a means to achieve stationarity when dealing with nominal time series on interest rates and earnings.' Specifically, he suggests that time series expressed in real terms may be stationary and eliminate the problems associated with using nonstationary series. The use of reliable data expressed in real terms also would eliminate much of the controversy surrounding the proper treatment of expected future inflation in accidental death and injury awards. Unfortunately, the issue is not as simple as Harris makes it appear. There are a number of problems with his analysis. First, his conclusion that nominal data can be used to calculate the present value of future earnings requires a stronger condition that stationarity in the real data. Second, the problems involved in estimating real data were not addressed. Third, his interpretation of his empirical results is not accurate. At the outset we should be clear about what constitutes stationarity in time series data. Throughout his Comment, Harris misinterprets stationarity as implying no trend or that a series is . . relatively stable . . . (constant?). However, that is too restrictive. A series is stationary if the stochastic process that generates the series is stable or invariant with respect to time. The values assumed by the generated variable can vary with respect to time; a nonstationary series may or may not exhibit a trend. For example, the ordinary random walk is a trendless nonstationary process. Moreover, a trendless stationary series need not be a constant (with only random error as in a white noise process), but may exhibit a cyclical structure.