Department of Economics and Business Economics

A local Gaussian bootstrap method for realized volatility and realized beta

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

This article introduces a local Gaussian bootstrap method useful for the estimation of the asymptotic distribution of high-frequency data-based statistics such as functions of realized multivariate volatility measures as well as their asymptotic variances. The new approach consists of dividing the original data into nonoverlapping blocks of M consecutive returns sampled at frequency h (where h -1 denotes the sample size) and then generating the bootstrap observations at each frequency within a block by drawing them randomly from a mean zero Gaussian distribution with a variance given by the realized variance computed over the corresponding block. Our main contributions are as follows. First, we show that the local Gaussian bootstrap is first-order consistent when used to estimate the distributions of realized volatility and realized betas under assumptions on the log-price process which follows a continuous Brownian semimartingale process. Second, we show that the local Gaussian bootstrap matches accurately the first four cumulants of realized volatility up to o(h), implying that this method provides third-order refinements. This is in contrast with the wild bootstrap of Goncalves and Meddahi (2009, Econometrica 77(1), 283-306), which is only second-order correct. Third, we show that the local Gaussian bootstrap is able to provide second-order refinements for the realized beta, which is also an improvement of the existing bootstrap results in Dovonon, Goncalves, and Meddahi (2013, Journal of Econometrics 172, 49-65) (where the pairs bootstrap was shown not to be second-order correct under general stochastic volatility). In addition, we highlight the connection between the local Gaussian bootstrap and the local Gaussianity approximation of continuous semimartingales established by Mykland and Zhang (2009, Econometrica 77, 1403-1455) and show the suitability of this bootstrap method to deal with the new class of estimators introduced in that article. Lastly, we provide Monte Carlo simulations and use empirical data to compare the finite sample accuracy of our new bootstrap confidence intervals for integrated volatility with the existing results.

Original languageEnglish
JournalEconometric Theory
Volume35
Issue2
Pages (from-to)360-416
Number of pages57
ISSN0266-4666
DOIs
Publication statusPublished - 2019

    Research areas

  • COVARIANCE, ECONOMETRIC-ANALYSIS, FUNCTIONALS, INFERENCE, MICROSTRUCTURE NOISE, MODELS, REGRESSION, SEMIMARTINGALES, STOCHASTIC VOLATILITY, WILD BOOTSTRAP

See relations at Aarhus University Citationformats

ID: 142215565