Open AccessForecasting GDP Movements in Nepal Using Autoregressive Integrated Moving Average (ARIMA) Modelling Process
Author(s) -
Surya Bahadur Rana
Publication year - 2019
Publication title -
journal of business and social science research/journal of business and social sciences research
Language(s) - English
Resource type - Journals
eISSN - 2631-2433
pISSN - 2542-2812
DOI - 10.3126/jbssr.v4i2.29480
Subject(s) - autoregressive integrated moving average , box–jenkins , univariate , moving average , econometrics , time series , autoregressive model , statistics , sample (material) , series (stratigraphy) , mathematics , multivariate statistics , paleontology , chemistry , chromatography , biology
This study attempts to test the ARIMA model and forecast annual time series of GDP in Nepal from mid-July, 1960 to mid-July, 2018. The annual time series on GDP used in this study consists of total 59 observations. Out of them, three years’ data from mid-July 2016 to mid-July 2018 have been used for in-sample forecasting and evaluation. The study uses univariate Box-Jenkins ARIMA modelling process to identify the best fitted model that describes the sample data set. The study examines a number of ARIMA family models and recommends ARIMA (0,1,2) as the most appropriate model that best describes the annual GDP series of the sampled period. The ARIMA (0, 1, 2) model incorporates zero lag order for autoregression, integrated with 2 lag order for moving average model using first difference operator. The ARIMA model forecasts documented in this study are not significantly different from actual because the actual annual GDP series observed in forecast period fall within 95 per cent confidence interval of estimates. Hence, ARIMA (0,1,2) model can best capture the GDP movement in Nepal for the sample period.