Sieve-based confidence intervals and bands for L\'{e}vy densities

TL;DR

This paper develops sieve-based methods to construct confidence intervals and bands for Lévy densities from discrete data, providing theoretical guarantees and convergence rates for the estimators.

Contribution

It introduces central limit theorems for sieve estimators of Lévy densities, enabling the construction of confidence intervals and bands with near-optimal convergence rates.

Findings

Pointwise estimators converge close to minimax rates.

Uniform confidence bands achieve near-logarithmic and polynomial rates.

Results depend on smoothness, sampling frequency, and sieve dimension.

Abstract

The estimation of the L\'{e}vy density, the infinite-dimensional parameter controlling the jump dynamics of a L\'{e}vy process, is considered here under a discrete-sampling scheme. In this setting, the jumps are latent variables, the statistical properties of which can be assessed when the frequency and time horizon of observations increase to infinity at suitable rates. Nonparametric estimators for the L\'{e}vy density based on Grenander's method of sieves was proposed in Figueroa-L\'{o}pez [IMS Lecture Notes 57 (2009) 117--146]. In this paper, central limit theorems for these sieve estimators, both pointwise and uniform on an interval away from the origin, are obtained, leading to pointwise confidence intervals and bands for the L\'{e}vy density. In the pointwise case, our estimators converge to the L\'{e}vy density at a rate that is arbitrarily close to the rate of the minimax risk of estimation on smooth L\'{e}vy densities. In the case of uniform bands and discrete regular sampling, our results are consistent with the case of density estimation, achieving a rate of order arbitrarily close to $\log^{-1/2}(n)\cdot n^{-1/3}$, where $n$ is the number of observations. The convergence rates are valid, provided that $s$ is smooth enough and that the time horizon $T_n$ and the dimension of the sieve are appropriately chosen in terms of $n$.