Multilevel Panel Data Modelling for Unemployment Rate in Indonesian

Abstract

Indonesia as a developing country is often faced with the problem of unemployment. Low unemployment rates can reflect good economic growth and reflect the well-being of the population, thus each country will attemp to lower the unemployment rate, including Indonesia. The unemployment rate of provinces in Indonesia from 2005 to 2015 observed every 6 months can be viewed as panel data. This study aims to model the trend of unemployment rate in Indonesia and external factors that influence it through multilevel modeling approach, where unemployment rate value at any point of time (level 1) is nested within the province (level 2). Selected models are models with random intercept and random time slopes which indicate there are varieties of unemployment rate between provincies that occurs at initial values and the rate of change over time. The average decline is about 0.05% every 6 months. Provinces with a high initial unemployment rate value were more likely to decline the unemployment rate compared to provinces with low initial unemployment rate values. After being controlled by the effect of time, the fixed effect that significantly affects unemployment rate is investment and population density. In addition, unemployment rate in August tended to be higher than unemployment rate in February.