Title: An approach to arrive at stationarity in time series data

Authors: M.I. Nafeesathul Basariya; Punniyamoorthy Murugesan

Addresses: National Institute of Technology, Tiruchirappalli 620 015, Tamil Nadu, India ' National Institute of Technology, Tiruchirappalli 620 015, Tamil Nadu, India

Abstract: This paper presents a framework for converting non-stationary data into stationary data in a systematic way. Initially, the data is fitted with a unit root model to check the presence of unit root. The model which confirms the absence of the unit root is considered for stationarity check. The error term of the model that has passed the unit root test is checked for randomness and homoscedasticity. The data is considered to be stationary if it satisfies the conditions like the absence of unit root, presence of error randomness and homoscedasticity of error term. The non-stationary data has to be differenced and is checked for absence of unit root, presence of error randomness and homoscedasticity of error term. This process is continued to ensure the stationarity. This framework has been elucidated in this paper for macroeconomic variables namely consumption, gross domestic product and consumer price index.

Keywords: Dickey-Fuller; Durbin-Watson; Lagrangian multiplier test; augmented Dickey-Fuller; stationarity; randomness of error term; homoscedasticity of error term.

DOI: 10.1504/IJAMS.2022.10050102

International Journal of Applied Management Science, 2022 Vol.14 No.3, pp.221 - 245

Received: 24 Sep 2019
Accepted: 06 Dec 2019

Published online: 31 Aug 2022 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article