Working Paper: NBER ID: w29849
Authors: William N. Goetzmann; Akiko Watanabe; Masahiro Watanabe
Abstract: This paper tests the retrieved context model of Wachter and Kahana (2019) using a long-term panel of economic forecasts by participants in the Livingston Survey. Events in historical time contribute additional explanatory power to a relative time series model. Historical precedents for current macroeconomic conditions appear to be more relevant for extreme quantile forecasts. The results are consistent with the use of the retrieved context mechanism for formulating expectations about asset prices. They also suggest that historical events, not just lagged variables in relative time, matter in economic forecasting.
Keywords: No keywords provided
JEL Codes: G02; G17
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Historical macroeconomic events (N14) | Memory retrieval processes (D87) |
| Memory retrieval processes (D87) | Current economic forecasts (F37) |
| Historical macroeconomic events (N14) | Current economic forecasts (F37) |
| Historical macroeconomic events (N14) | Extreme quantile forecasts (C53) |
| Extreme quantile forecasts (C53) | Current economic forecasts (F37) |
| Historical macroeconomic events (N14) | Tail forecasts sensitivity (C53) |
| Tail forecasts sensitivity (C53) | Current economic forecasts (F37) |