A1213
Title: Forecasting Philippine economic growth using mixed frequency data: MIDAS versus MF-DLFM
Authors: Rutcher Lacaza - University of the Philippines (Philippines) [presenting]
Abstract: Assessing the economic impact of COVID-19 in most developing countries like the Philippines has been hampered by the delayed publication of official statistics, such as GDP. Traditional forecasting models for economic growth rely on aggregating economic or financial indicators observed at higher frequencies than quarterly GDP growth, which can lead to a loss of useful forward-looking information and less accurate forecasts. To address this issue, mixed-frequency models have been developed, such as the Mixed Data Sampling (MIDAS) regression technique and the Mixed-frequency Dynamic Latent Factor Model (MF-DLFM), to incorporate high-frequency data into prediction models. The aim is to compare the performance of MIDAS and MF-DLFM in forecasting quarterly GDP in the Philippines using monthly and weekly data from 2000 to 2023. The results indicate that both models outperform traditional models that use only quarterly data. However, the MF-DLFM provides slightly more accurate forecasts than the MIDAS model. The findings demonstrate the usefulness of mixed frequency models in providing timely and accurate information to policymakers, enabling informed decisions, especially during the COVID-19 pandemic.
