Mean-preserving variance response in nonlinear Hawkes processes
We study the variance response of nonlinear Hawkes processes under interventions that preserve the stationary event rate exactly. A response functional separates the counting field, which probes fluctuations, from the causal input, which changes feedback. At a Poisson reference, the derivative of the long-run count-variance rate is the integrated expected effect of inserting a past event into the feedback history. We establish this identity for bounded finite-memory feedback, derive its consequences for general bounded Hawkes links, and evaluate the saturating case explicitly. Excitatory feedback produces a positive variance response, with saturation controlling its magnitude. The corresponding moderate-deviation rate describes the change in count fluctuations, while conditional event-time Fisher information measures the distinguishability of the compensated intervention. These results connect mean-preserving variance response to temporal dependence through the response functional.
Authors
- Daniel E. Clark
Institutions
- University of Southampton (GB)
Publication Details
- Journal
- Statistics & Probability Letters
- Published
- 2026-11-18
- DOI
- https://doi.org/10.1016/j.spl.2026.111020
- Primary Topic
- Point processes and geometric inequalities
- Type
- article
- Field-Weighted Citation Impact
- 0.00