Forecasting Financial and Macroeconomic Variables Using Data Reduction Methods: New Empirical Evidence

Abstract

In this paper, we empirically assess the predictive accuracy of a large group of models that are speci…ed using principle components and other shrinkage techniques, including Bayesian model averaging and various bagging, boosting, least angle regression and related methods. Our results suggest that model averaging does not dominate other well designed prediction model speci…cation methods, and that using "hybrid"combination factor/shrinkage methods often yields superior predictions. More speci…cally, when using recursive estimation windows, which dominate other "windowing" approaches, "hybrid" models are mean square forecast error "best" around 1/3 of the time, when used to predict 11 key macroeconomic indicators at various forecast horizons. Baseline linear (factor) models also "win" around 1/3 of the time, as do model averaging methods. Interestingly, these broad …ndings change noticably when considering di¤erent sub-samples. For example, when used to predict only recessionary periods, "hybrid" models "win" in 7 of 11 cases, when condensing …ndings across all "windowing" approaches, estimation methods, and models, while model averaging does not "win" in a single case. However, in expansions, and during the 1990s, model averaging wins almost 1/2 of the time. Overall, combination factor/shrinkage methods "win" approximately 1/2 of the time in 4 of 6 di¤erent sample periods. Ancillary …ndings based on our forecasting experiments underscore the advantages of using recursive estimation strategies, and provide new evidence of the usefulness of yield and yield-spread variables in nonlinear prediction model speci…cation.

The Role of Indicator Selection in Nowcasting Euro Area GDP in Pseudo Real Time

“…In the literature, alternative techniques and methods have been introduced to perform this information reduction: targeting indicators with thresholding rules (Bai and Ng, 2008, Schumacher, 2010, Bulligan et al, 2012, and Kim and Swanson, 2013, estimating weighted principal components and preselecting indicators with rules that eliminate irrelevant information Ng, 2006, andCaggiano et al, 2011), estimating factors under a sparse prior (e.g. Kaufmann and Schumacher, 2013), and selecting one "representative" indicator of each category in which the large panel can be classified (Alvarez et al, 2012).…”

“…Kim and Swanson, 2013). In fact, higher-quantiles PFM (using fewer indicators) reduce the degree of freedom for the researcher and the unavoidable arbitrariness of subjective choices.…”

Testing for structural stability of factor augmented forecasting models

“…Castle, Clements and Hendry (2012) use Autometrics to select the most accurate factor (augmented) models. The use of recursive and rolling techniques for both factor estimation and factor augmented forecasting model estimation is analyzed in a series of prediction experiments by Kim and Swanson (2012).…”

Assessing the Macroeconomic Forecasting Performance of Boosting: Evidence for the United States, the Euro Area and Germany

“…However, only recently has the method found its way into the macroeconometric literature. Apart from several financial applications (Audrino and Barone-Adesi, 2005;Gavrishchaka, 2006;Audrino and Trojani, 2007;Andrada-Félix and Fernández-Rodríguez, 2008), there are only few macroeconometric studies on the forecasting per-formance of boosting (Bai and Ng, 2009;Shafik and Tutz, 2009;Buchen and Wohlrabe, 2011;Robinzonov, Tutz, and Hothorn, 2012;Kim and Swanson, 2014). Results with respect to the predictive accuracy of boosting are promising.…”

“…2 They are used, for instance, by Bai andNg (2009), Shafik andTutz (2009), and Kim and Swanson (2014).…”