Forecasting Malaysia’s unemployment rate: a comparative study of ARIMA and interrupted time series with ARIMA models

Malaysia, like other countries, was experiencing the effect of the outbreak of the COVID-19 pandemic on the economic sector, which led to a hike in unemployment. High unemployment could interfere with the economic progress, which may impact Malaysia’s long-term performance. Hence, to maintain economic stability in the long run, the issue of unemployment should be addressed strategically. In a previous study, the ARIMA and ARFIMA models were used to model and forecast the unemployment rate in Malaysia. Since then, the outbreak of the COVID-19 pandemic which saw a sudden spike resulting in an abrupt change of trend in the unemployment rate may cause the forecast ability of existing models to be not as good. Thus, this study aimed to model Malaysia’s unemployment rates using ARIMA and ARIMA interrupted time series models besides comparing the forecast performance between these two models. The monthly data obtained from of the Department of Statistics Malaysia consisted of 176 observations of Malaysia’s unemployment rate from January 2010 to August 2024. The ARIMA (0,1,1) was found to be the best ARIMA model in its class for forecasting Malaysia’s unemployment rate since it has the lowest BIC and AICc values. Further, the ARIMA model also gave a better forecast performance compared to the ARIMA with interrupted time series model based on the error measures of RMSE, MAE and MAPE values.