Economics Seminar - The Macroeconomics of Firm Forecasts

We document a systematic increase in the forecasting ability of US firms overtime and show that this increase is accounted for largely by the change in the firm-size distribution. We develop a macroeconomic framework of firm information production that is consistent with this evidence. We show that firms' size-dependent incentives to use `data-driven decision-making' can rationalize the size-forecasting-ability relationship documented in the data. Consistent with the data, our framework implies that firms that use data more intensely allocate inputs more efficiently, adopt more efficient technologies, are more profitable and grow faster and larger. Our framework further suggests that data-driven decision making has important macroeconomic consequences: in a calibration exercise, we find that total factor productivity (household welfare) in the US would have been up to 13 percent (29 percent) lower in 2022 absent the increase in the forecastiing ability of US firms over the past two decades.