Arbitrary Spearman’s Rank Correlations in Maximum Entropy Bootstrap and Improved Monte Carlo Simulations

Abstract

The R package for maximum entropy bootstrap (meboot) is widely used for numerous applications involving statistical inference for time series data without having to do differencing or de-trending. We report some simulations confirming its effectiveness. It has been used for simulating time series, especially the function "flexMeboot," in the package, which uses non-overlapping blocks while avoiding allowing for random trend reversals. This paper describes a newer extension called "mebootSpear," which permits the user to specify arbitrary Spearman's rank correlation coefficient rho between original data xt and resampled series. We use simulations to study the properties of the new option and extensions to the traditional Monte Carlo simulations.

Keywords: Dependent Data, Statistical Inference, Unit Root, Monte Carlo Simulations, Value at Risk

JEL Classification: C15, C32, C58, C63