Description

This paper studies the informational content of synthetic indicators of economic activity for projecting the monthly index of economic activity (Imacec) for Chile. Comparing them to the Urrutia-Sánchez (2008) model, which employs energy production, and calendar and seasonal components to forecast the Imacec, models employing synthetic leading indicators and financial conditions indicators are competitive in terms of the MSFE. Moreover, we show that combinations of different forecasting strategies with small bias present improvements in terms of the MSFE with respect to individual models. We also show that, due to their inertial behavior, projections with synthetic indicators display errors that last for many periods after an exogenous event (such as an earthquake). Specifications with variables that quickly adjust to economic activity, such as energy consumption, do not have this problem.