{"id":"arxiv/2610.25237","url":"https://arxiv.org/abs/2610.25237","tags":["math.ST","econ.EM"],"score":0,"title":"Conventional Inference under Nonstationary Heavy Tailed Fractional Processes & Arrays","domain":"math.ST","source":"arxiv","authors":["Ioannis Kasparis"],"reviews":0,"subject":"math.ST","summary":"Existing regression inference methods for nonstationary heavy tailed data are largely limited to unit roots tests. Functional central limit theorems (FCLTs), which are commonly used for the asymptotic analysis of nonstationary models, are inapplicable to nonstationary data characterized by a memory parameter $d<1$ and a stability parameter $α<2$. Furthermore, it is well known that inference under nonstationarity is often non-conventional, and this issue is exacerbated in the presence of heavy-tailed data. The interaction of long memory and heavy tails significantly enlarges the nonstationary…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-18","reactions":{}}