Lévy matters IV. Estimation for discretely observed Lévy processes (Q476619)

This special volume deals with statistical inference problems for Lévy processes when what is available are discrete data. The main assumption is that the Lévy process \(X_t\), \(t\geq 0\), being a continuous time process, is observed at discrete times so the observations are the increments \(X_{i\Delta_n}- X_{(i-1)\Delta_n}\), which are independent and identically distributed random variables. Some, or all, of the characteristics \((b,\sigma^2,F)\) (drift, diffusion and Lévy measure) of the process \(X\) are unknown and the problem is to use the available discrete data to `identify' or estimate `well' what is unknown. There are parametric and/or non-parametric problems and they are all far from being trivial. While there is a long list of papers published over the last few decades, there are some new developments. The goal of editors and authors is exactly to reflect different aspects of the recent progress in this area. The volume contains three chapters. The first chapter, ``Estimation and calibration of Lévy models via Fourier methods'', is written by \textit{D. Belomestny} and \textit{M. Reiß}. It covers the low frequency situation. The idea of using the empirical characteristic functions is efficient. It leads to finding `optimal' estimators for the unknown characteristics including the rate of convergence. The second chapter, ``Adaptive estimation for Lévy processes'', authored by \textit{F. Comte} and \textit{V. Genon-Catalot}, deals with non-parametric inference problems mainly in the case of high-frequency observations. The adaptive estimators constructed here obey a higher rate of convergence than the estimators in the previous chapter. Each approach, however, has its own merits. In the third chapter, ``Parametric estimation of Lévy processes'', prepared by \textit{M. Masuda}, all three characteristics \((b,\sigma^2,F)\) depend on a parameter \(\theta\), one-dimensional or multidimensional. This author uses the maximum likelihood method and the local asymptotic normality property to derive estimators with various rates of convergence in the high-frequency case. It is also shown that the method of moments leads to rate-efficient estimators when using the median of an appropriate function of the observed increments. Useful concrete examples are provided. As a whole, this is a remarkable collection presenting different but important aspects in statistical inference problems for Lévy processes based on discrete observations. The authors of the three chapters have made essential contributions in this area. All three chapters are carefully written and referenced. PhD students and professionals in stochastic modelling will benefit a lot from this volume.