Estimation and Forecasting of Industrial Production Index

Abstract

It is essential that policymakers consider cyclical changes in output. Monthly industrial production is one of the most important and commonly used macroeconomic indicators for this purpose. However, monthly estimates of industrial production are not available for Pakistan. Instead, policymakers rely on a large-scale manufacturing (LSM) index that accounts for only 10 percent of GDP. Another limitation of this index is that it accounts primarily for private sector industry, leaving out the direct public sector presence in industrial production. Economic policymakers rely heavily on the LSM index to gauge economic activity in Pakistan. In this study, we compute a new industrial production index (IPI) that extends to the whole industrial sector in Pakistan, incorporating additional information that the LSM index misses. Post-estimation, we build seven econometric models reflecting conditions in the real, financial, and external sectors to estimate year-on-year changes in the new IPI. Our results show that the root mean square error of the ARDL model reflecting financial conditions is lowest of the models tested, which included AR, VAR, and BVAR, across all horizons.