The Performance of Double Bootstrap Method for Large Sampling Sequence

Abstract

Studies on the iteration procedure in double bootstrap method have given a great impact on confidence interval performance. However, the procedure was claimed to be complicated and demand intensive computer processor. Considering this problem, an alternative procedure was proposed in this research. Despite of using small sampling sequence, this research was aimed to increase the accuracy estimation using a second replication number which resulted in a large sampling sequence of double bootstrap. In this paper, the alternative double bootstrap method was hybrid onto an example model and its performance was based on Studentised interval. The performance was examined in simulation study and real sample data of sukuk Ijarah. The result showed that hybrid double bootstrap model gave more accurate estimation in terms of its shorter length when dealing with various parameter values and has shown to improve the single bootstrap estimation.