Abstract: Forecasting mortality is of importance in managing longevity risks for insurance companies and pension funds. Some widely employed models are the so-called Lee-Carter model and its extensions. First we show that the proposed two-step inference procedure in Lee and Carter (1992) can not detect the true dynamics of the mortality index except when the index follows from a unit root AR(1) process. Second we propose a new method to test whether the index does follow from a unit root AR(1) model and then apply the new test to some mortality data to show that a blind application of an existing R package leads to different conclusions.
Testing for predictability of asset returns has been a long history in econometrics. Recently, based on a simple predictive regression, Kostakis, Magdalinos and Stamatogiannis (2015) proposed a Wald test derived from the IVX methodology for stock return predictability and Demetrescu (2014) showed that the local power of the standard IVX-based test could be improved in some cases when a lagged predicted variable is added to the predictive regression on purpose. Therefore an interesting question is whether a lagged predicted variable does appear in the model. Here we propose unified tests for testing the existence of a lagged predicted variable in a predictive regression and the predictability regardless of whether the predicting variable is stationary or nearly integrated or unit root. We further apply the proposed tests to some real data sets in finance.