A1359
Title: Real-time nowcasting growth at risk
Authors: Manuel Schick - Heidelberg University (Germany) [presenting]
Christian Conrad - Heidelberg University (Germany)
Abstract: Macroeconomic and financial indicators are exploited to nowcast Growth at Risk (GaR) in a Mixed-Data Sampling (MIDAS) framework. Indicators are used at high-frequency with exact timing, i.e., as they become available to forecasters in real-time according to exact publication dates. A small number of factors for both mean and conditional variance are incorporated in the MIDAS model, resulting in a parsimoniously parameterized prediction model that provides daily updates of GaR of the current quarter. In particular, the conditional variance features a long-term volatility component of the S\&P 500 index that is extracted by means of the MF2-GARCH model. Therefore, financial conditions are monitored on a daily basis which provides a timely signal for downside risk. The results show that the new model provides good out-of-sample GaR nowcasts.
