A constrained robust Markov regime-switching model for long-term risk evaluation

Shanshan Qin, Beibei Guo, Yuehua Wu, Hong Xie and Jingjing Dong

Journal of Applied Statistics, 2026, vol. 53, issue 4, 574-589

Abstract: Markov Regime-Switching (MRS) models are widely used for modeling equity return time series. Yet standard MRS models may inadequately capture the mean reversion behavior of long-term equity returns and exhibit unstable parameter estimation due to their reliance on normality assumptions within each regime. These limitations in model adequacy can compromise the accuracy of risk exposure measurements for invested assets. To address these issues, we propose a constrained robust MRS (CRMRS) model, which integrates an order restriction and sparse constraints on regime means and transition probabilities to better capture mean reversion while employing a general ρ-based least favorable distribution to improve distributional flexibility across regimes. We assess the method's performance through finite-sample simulations under various scenarios in the presence or absence of atypical values. Furthermore, we empirically validate the improvements in model adequacy and risk exposure measurement using monthly returns from the S&P/TSX Composite Index, the benchmark for Canadian equity performance, where S&P and TSX stand for Standard & Poor's and the Toronto Stock Exchange, respectively. Our findings demonstrate that the proposed CRMRS-Huber produces stable parameter estimates and superior approximations of higher-order moments, such as skewness and kurtosis, and provides balanced intermediate risk evaluation across all cases.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
http://hdl.handle.net/10.1080/02664763.2025.2525880 (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:taf:japsta:v:53:y:2026:i:4:p:574-589

Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/CJAS20

DOI: 10.1080/02664763.2025.2525880

Access Statistics for this article

Journal of Applied Statistics is currently edited by Robert Aykroyd

More articles in Journal of Applied Statistics from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().