Forecasting Indonesia's Unemployment Rate with Macroeconomic & Big Data: A MIDAS Approach

Abstrak

Research Originality: The current model is unable to forecast the unemployment rate utilizing varying periods of predictor variables. Furthermore, the use of official statistics and big data in previous studies to forecast Indonesia's unemployment rate has been limited. Research Objectives: This study forecasts Indonesia's biannual unemployment rate (UR) by utilizing monthly Google Trends Index (GTI) data, quarterly Gross Domestic Product (GDP) data, and monthly inflation data. Research Methods: The unrestricted mixed data sampling (U-MIDAS) model is applied to forecast Indonesia's UR using data from the second semester of 2006 to the first semester of 2024. Empirical Results: This study finds that the best model for predicting UR is one that utilizes a combination of big data and official statistics. Using 34 GTI keywords relevant to job seekers' cultural and behavioral patterns in Indonesia, Indonesia's UR in February 2024 was 4.7%. Implications:  This study demonstrates that employing GTI and macroeconomic variables for forecasting unemployment enhances predictive accuracy compared to utilizing either variable independently.  JEL Classification: C55, E24, J64