Abstract
We propose a sequential monitoring scheme to detect changes in dynamic semiparametric risk models that capture Value–at–Risk (VaR) and Expected Shortfall (ES) jointly. The monitoring scheme is based on a gradient–based detector and a boundary function, and a change is detected when the detector crosses the boundary function. We derive the asymptotic limit of the stopping time of detection under the null hypothesis of no change. Monte Carlo simulations show that the proposed test has reasonable size control under the null hypothesis and high power under alternative hypotheses of various change point scenarios in finite samples. Empirical applications based on the S&P 500 index and the GBP/EUR exchange rate illustrate that our proposed test is able to detect change points in real–time.
| Original language | English |
|---|---|
| Pages (from-to) | 482–496 |
| Number of pages | 15 |
| Journal | Journal of Business and Economic Statistics |
| Volume | 44 |
| Issue number | 2 |
| Early online date | 11 Nov 2025 |
| DOIs | |
| Publication status | E-pub ahead of print - 11 Nov 2025 |
Data Availability Statement
The authors confirm that the data supporting the findings of this study are available within its supplementary materials.Keywords
- Change point analysis
- Risk measures
- Sequential monitoring
ASJC Scopus subject areas
- Statistics and Probability
- Social Sciences (miscellaneous)
- Economics and Econometrics
- Statistics, Probability and Uncertainty
Fingerprint
Dive into the research topics of 'Sequential Monitoring for Changes in Dynamic Semiparametric Risk Models'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS