{"id":"arxiv/2609.39355","url":"https://arxiv.org/abs/2609.39355","tags":["math.ST"],"score":0,"title":"Invariant Measures as Estimators: Second-Order Stochastic Expansions and Bias Reduction","domain":"math.ST","source":"arxiv","authors":["Shota Yano"],"reviews":0,"subject":"math.ST","summary":"We develop a general second-order asymptotic framework for estimators constructed from invariant probability measures of data-dependent Markov processes. The framework yields a universal bias correction for smooth functionals, computed from the same invariant measure used to construct the estimator, without the resampling and repeated estimation involved in bias-correction methods such as the bootstrap and jackknife. The statistical input is represented by a random one-form, allowing differentials of log-likelihoods and general random criteria, as well as estimating functions, to be treated…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-30","reactions":{}}