Forecasting economic activity with mixed frequency BVARs |
| |
Institution: | 1. Federal Reserve Board (retired), 100 Old Stage Road, Woolwich, Maine 04579, United States;2. Reserve Bank of Australia - Research Department, 65 Martin Place, GPO Box 3947, Sydney, New South Wales 2000, Australia;1. Bank of England, Threadneedle Street, London EC2R 8AH, UK;2. Federal Reserve Bank of New York, 33 Liberty Street, NY NY 10045, USA;3. London Business School, Sussex Pl, London NW1 4SA, UK;1. Norges Bank, Norway;2. European University Institute, Italy;3. Bocconi University, Italy;4. CEPR, United Kingdom |
| |
Abstract: | Mixed frequency Bayesian vector autoregressions (MF-BVARs) allow forecasters to incorporate large numbers of time series that are observed at different intervals into forecasts of economic activity. This paper benchmarks the performances of MF-BVARs for forecasting U.S. real gross domestic product growth against surveys of professional forecasters and documents the influences of certain specification choices. We find that a medium–large MF-BVAR provides an attractive alternative to surveys at the medium-term forecast horizons that are of interest to central bankers and private sector analysts. Furthermore, we demonstrate that certain specification choices influence its performance strongly, such as model size, prior selection mechanisms, and modeling in levels versus growth rates. |
| |
Keywords: | Mixed frequency Bayesian VAR Real-time data Nowcasting Forecasting Economic activity |
本文献已被 ScienceDirect 等数据库收录! |
|