14

We develop monthly indicators for tracking short-run trends in real GDP growth in 32 advanced and emerging-market economies. To do so, we use a two-stage procedure inspired by Giannone et alii (2008) that combines the estimation of dynamic factor models on and the application of the Kalman filter to heterogeneous data in terms of frequency and publishing delays. We test the historical performance of our indicators and find that they do a good job at describing the business cycle. In a recursive out-of-sample forecasting exercise, we find that the indicators generally produce good real GDP growth forecasts relative to a range of time series models.