Estimation of the Jump-Point in a Hazard Function

We consider a piecewise constant hazard function with exactly one jump point, say τ. It uniquely determines an Exponential distribution whose density features a discontinuity of the first kind at the change point τ. Assuming that τ is the unknown parameter of interest, the maximum likelihood estimator is shown to be strongly consistent for τ. Its computation is very simple, because it requires merely a finite number of comparisons. Some graphics and calculations illustrate our results.