Testing exponentiality versus a trend change in mean residual life (Q1819516)

Let F be a distribution function of a nonnegative random variable X and define the mean residual life function m by \(m(t)=E[X-t|\) \(X>t]\) for all t such that \(\bar F(t)>0\). The distribution F is said to be increasing then decreasing mean residual life (IDMRL) if there exists a \(\tau\geq 0\) such that m is nondecreasing on \([0,\tau]\) and is nonincreasing on \([\tau,\infty)\). The authors develop tests for testing the null hypothesis that F is exponential, versus the alternative hypothesis that F is IDMRL but not exponential. One test procedure is appropriate when \(\tau\) is known, the other test procedure is appropriate when \(F(\tau)\) is assumed known. The asymptotic normality of both test statistics is established. Two examples which illustrate the tests are given.