Two sources of heterogeneity are often overlooked. On the one hand, time-varying hazard contributions of explanatory variables cannot be captured in the widely used Cox proportional hazard model. To this end, a dynamic survival model is investigated within a Bayesian framework. The specification allows parameters to evolve over time, thus accounting for time-varying effects gradually. On the other hand, unobserved heterogeneity across groups is often ignored, leading to invalid estimators. Accounting for such effects is made feasible for even large numbers of groups through a shared factor model, which picks up unexplained covariance in the error term. Building on a Markov Chain Monte Carlo scheme based on data augmentation allows the usage of shrinkage priors to avoid overfitting in such a highly parameterized model. In particular, the shrinkage priors are implemented to automatically detect which parameters should be included in the model and which should be allowed to vary over time. Finally, an R package that makes the routine easily available is introduced.
24 Juni 2021 → 26 Juni 2021
4th International Conference on Econometrics and Statistics
Österreichische Systematik der Wissenschaftszweige (ÖFOS)