A Nonparametric Investigation of Duration Dependence in the American Business Cycle

Does the termination probability of a business expansion or contraction increase with age? This question may be formally addressed by analyzing the nature of duration dependence in aggregate economic activity. Our null hypothesis is that there is no duration dependence, which we test via intentionally nonparametric procedures. We also argue that common notion of business cycle periodicity can be usefully interpreted in terms of whole-cycle duration dependence. We find some evidence for duration dependence in whole cycles and in prewar expansions, but little evidence elsewhere.