Working Paper: CEPR ID: DP14469
Authors: Massimiliano Marcellino; George Kapetanios; Yiannis Dendramis
Abstract: In the aftermath of the recent financial crisis there has been considerable focuson methods for predicting macroeconomic variables when their behavior is subject toabrupt changes, associated for example with crisis periods. In this paper we proposesimilarity based approaches as a way to handle parameter instability, and apply themto macroeconomic forecasting. The rationale is that clusters of past data that matchthe current economic conditions can be more informative for forecasting than the entirepast behavior of the variable of interest. We apply our methods to predict bothsimulated data in a set of Monte Carlo experiments, and a broad set of key US macroeconomicindicators. The forecast evaluation exercises indicate that similarity-basedapproaches perform well, in general, in comparison with other common time-varyingforecasting methods, and particularly well during crisis episodes.
Keywords: macroeconomic forecasting; forecast comparison; empirical similarity; parameter time variation; kernel estimation
JEL Codes: No JEL codes provided
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| trigger variable (z_t) (C22) | estimation of current model parameters (β_n) (C51) |
| estimation of current model parameters (β_n) (C51) | forecasting accuracy (C53) |
| similarity-based forecasting methods (C53) | forecasting outcomes (C53) |
| similarity-based forecasting methods (C53) | forecasting performance during crisis episodes (F37) |
| kernel weighting and local averaging (C45) | adaptability to changing economic conditions (O00) |