Real-Time Forecasting Revisited: Letting the Data Decide

Real-time GDP forecasting, also often known as “nowcasting,” produces estimates for current-quarter real GDP growth, typically based on a centered value from a set of estimates from incoming indicators. These real-time measures are usually intended to be data-based and to not be based on forecaster judgment or add factors. Even so, estimation methodologies in this research area—and prior versions of the system we use—typically have been constrained by using various “fixed” relationships, such as a fixed historical sample horizon and fixed empirical specifications for the indicator variables. This paper describes the methodology, estimation, and software code for a more flexible real-time GDP system that allows the data to decide the best real-time GDP forecast for varying sample horizons and varying specifications for each indicator variable through time. Our system uses data on key indicators as they become available (accounting for the “jagged-edge” nature of the data in the current quarter) to generate an estimate of current-quarter real GDP growth, with weights for combining the indicator-specific estimates as determined by the strength of the indicators’ historical relationships to GDP growth. The improved system searches across a variety of specifications and across sample horizons to choose the best specification as determined by a minimum Schwarz criterion test while also searching for the best sample horizon for minimizing the mean absolute error for a recent prediction period. We illustrate the operation of the system for real-time estimates of real GDP growth over a specific quarter, and examine the properties of the estimates and the implications for predictions. We also discuss potential additional applications and demonstrate a specific application for real-time predictions of the monthly change in payroll jobs.