The 24th Federal Forecasters Conference

Forecasters often need to predict the present—or “nowcast’’—to help inform policy decisions. The onset of the pandemic highlighted the importance of nowcasting the economy and the pandemic itself. Conditions altered rapidly and policymakers responded to those changes with unprecedented fiscal and monetary stimulus and with public policies such as lockdowns and mask mandates. To aid in nowcasting, forecasters turn to machine-learning techniques and high-frequency data, which are often drawn from alternative data sources such as social media, rather than from standard sample surveys and administrative data. Nowcasting poses difficult challenges, particularly because data about the present may be unavailable, incomplete, or inaccurately measured. While high-frequency data can be a great source of timely and detailed information, it can come with its own dynamics, noise, and structural breaks. Those features can arise as a byproduct of individual decisions (as with Twitter) or the customer composition for private businesses (as with credit card transactions). The conference will consider nowcasting and its roles in decision-making. How have recently developed tools and data sources contributed to nowcasting, and how do nowcasts serve policymakers?