Working Paper: NBER ID: w19567
Authors: Steven L. Scott; Hal R. Varian
Abstract: We consider the problem of short-term time series forecasting (nowcasting) when there are more possible predictors than observations. Our approach combines three Bayesian techniques: Kalman filtering, spike-and-slab regression, and model averaging. We illustrate this approach using search engine query data as predictors for consumer sentiment and gun sales.
Keywords: Bayesian; Nowcasting; Time Series; Variable Selection
JEL Codes: C11; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Kalman filtering, spike-and-slab regression, and model averaging (C32) | forecasting accuracy of economic indicators (F37) |
| Google Trends data (Y10) | relevant predictors for consumer sentiment (D12) |
| financial planning queries (D14) | consumer sentiment (D12) |
| investing queries (G11) | consumer sentiment (D12) |
| BSTS model (C51) | forecasting accuracy compared to AR(1) models (C53) |
| gun stores (L81) | mean absolute error in gun sales forecasting (C29) |