Backward Stepwise Modelling for Unemployment Rate in Malaysia

Although the post-COVID-19 unemployment rate is expected to decline further, in many economies, unemployment remains one of the most challenging issues to overcome. In addition, the unemployment rate is not always constant as many economic determinants are accountable for its variability. Thus, this study examines 15 economic determinants from the literature which are manufacturing production index (MPI), total salaries and wages, public consumption, total deposits, total loans, exports of goods and services, imports of goods and services, total employment, gross domestic product (GDP), consumer price index, producer price index (PPI), USD/MYR exchange rate (EXRT), population, overnight policy rate, and foreign direct investment. All data were collected from economic reports compiled by the Economic Planning Unit, Prime Minister's Department of Malaysia for the longest common period of 21 years beginning in 2001. A backward stepwise regression is implemented to determine a well-specified model to represent unemployment rates in Malaysia. In addition, this study focuses on a general-to-specific modelling procedure and its diagnostic tests using R programming. Multicollinearity existed, and the VIF table was used to examine the modelling transitions. The empirical evidence indicates that MPI, public consumption, total loan, GDP, PPI, EXRT, and total employment account for 94.3% of the variability in unemployment rates. We suggest that the use of machinery and technology in the manufacturing sector may contribute to the positive effects of MPI and PPI on unemployment. Findings indicate that higher growth in GDP, public consumption, total loans, and total employment would lead to a lower unemployment rate, unlike EXRT. This study reveals that GDP growth is the most significant economic determinant of the unemployment rate.