SAMPLE SIZE SIMULATION FOR UNIT ROOT, STRUCTURAL BREAK AND REGIME SHIFTS

Authors

  • V.A. Micheal Department of Statistics, Federal Polytechnic Bida, Niger State,
  • M.O. Adenomon Department of Statistics & NSUK-LISA Stat Lab, Nasarawa State University, Keffi (NSUK), Nasarawa State,
  • N.O. Nweze Department of Statistics, Nasarawa State University, Keffi (NSUK), Nasarawa State,

Abstract

Unit root test is an important means to determine the integration order of a variable which has involved different methods of testing for stationarity. Simulation method is adopted in this study to verify whether unit root, structural breaks and regime shifts exist in the sample considered. For sample sizes of 20 and 50 as small, 100 and 250 as medium, and 2500 and 5000 as large, the enhanced Dickey-Fuller test and Zivot-Andrews test were used. The experiment was conducted 5000 times for each sample size, and the results demonstrated that there is presence of unit root at level for all sample sizes taken into consideration, but they were integrated of order 1. This implies that they are stationary at first difference. The results also showed that there are structural breaks at various levels depending on sample size, but it was noted that the breaks remained stable regardless of size when the sample size was large. The MSVAR results demonstrated that regime 1 is more resilient than regime 2, and that regime 1 is projected to last longer than regime 2. As a result, we draw the conclusion that simulation can be utilized to verify a real-world situation.

Downloads

Published

2023-07-04

Issue

Vol. 18 No. 2 (2023)

Section

ARTICLES