Inference for multiple temporal transitions and higher-order temporal dependence

Develop estimation and model-selection procedures for the multivariate general nesting spatio-temporal regression framework that fully exploit observations at multiple time points and accommodate higher-order autoregressive dependence.

Background

The paper formulates the multivariate general nesting spatio-temporal (MGNST) model as a single-transition conditional model based on the density of the response at time t conditional on the response and covariates at time t−1. The authors note that observations at multiple time points would yield a product conditional likelihood and that higher-order AR(s) dependence could be incorporated by augmenting the design matrix with additional lagged responses.

The unresolved methodological issue is to develop estimation and model-selection procedures that fully use these extensions rather than treating them only as straightforward specification modifications. This is relevant for applications with longer temporal panels and dynamics extending beyond first-order temporal dependence.

References

Developing estimation and model-selection procedures that fully exploit these extensions remains an important topic for future research.

— Multivariate Spatio-Temporal Regression with Penalized Model Selection and an Empirical Application  (2608.19664 - Nishii et al., 20 Aug 2026) in Section 6, Conclusion and Discussion

Neither expression includes the additional uncertainty from selecting $(m,\gamma)$. Hence, the expansion above does not establish the distribution conditional on pAIC or pBIC selection; bias correction and selection-adjusted inference for the MGNST model remain topics for future research.

— Multivariate Spatio-Temporal Regression with Penalized Model Selection and an Empirical Application  (2608.19664 - Nishii et al., 20 Aug 2026) in Appendix C, Section 'Penalized likelihood and asymptotic approximation'