Working Paper: CEPR ID: DP12716
Authors: Juan J. Dolado; Liang Chen; Jesus Gonzalo
Abstract: Quantile factor models (QFM) represent a new class of factor models for high-dimensional panel data. Unlike approximate factor models (AFM), which only extract mean factors, QFM also allow unobserved factors to shift other relevant parts of the distributions of observables. We propose a quantile regression approach, labeled Quantile Factor Analysis (QFA), to consistently estimate all the quantile-dependent factors and loadings. Their asymptotic distributions are established using a kernel-smoothed version of the QFA estimators. Two consistent model selection criteria, based on information criteria and rank minimization, are developed to determine the number of factors at each quantile. QFA estimation remains valid even when the idiosyncratic errors exhibit heavy-tailed distributions. An empirical application illustrates the usefulness of QFA by highlighting the role of extra factors in the forecasts of US GDP growth and inflation rates using a large set of predictors.
Keywords: factor models; quantile regression; incidental parameters
JEL Codes: C31; C33; C38
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Quantile Factor Models (QFM) (C22) | Extraction of quantile-dependent factors (C32) |
| Extraction of quantile-dependent factors (C32) | Shift of other distributional characteristics of observables (D39) |
| Quantile Factor Analysis (QFA) (C38) | Consistent estimators for quantile-dependent factors and loadings (C51) |
| Quantile Factor Analysis (QFA) (C38) | Estimate factors that capture extreme values and volatility (C51) |
| Quantile Factor Analysis (QFA) (C38) | Enhance forecasts of U.S. GDP growth and inflation rates (E37) |
| Quantile Factor Analysis (QFA) (C38) | Improve predictive power of models (C52) |
| Quantile Factor Analysis (QFA) (C38) | Inherit robustness properties from quantile regression (C32) |