Predicting GDP growth across different Quantiles Open Access

Abstract

We study the relationship between real GDP growth and economic and financial conditions across different quantiles. To overcome the invalidity of quantile regression models rising from persistent regressors, we apply IVXQR method that handles size distortion while preserving discriminatory powers. Increasingly severe financial conditions are associated with an increase in conditional volatility, indicating that financial conditions are informational for vulnerability predictions. The GDP growth forecast for a tight economy becomes more conservative after addressing the invalidity for QR methods. To compare out of sample forecast performance across models, we define final prediction error (FPE) using the quantile loss comparison. Generalized Random Forest has the best performance among different methods, implying that splitting the data into splitting and estimation sets would help GDP growth forecast.

Table of Contents

1 Introduction 1

2 Data 3

3 Method 4

3.1 QuantileRegression.................................................................. 4

3.2 Methods robust to the existence of persistent regressor .............. 5

3.3 TreeBasedMethods ................................................................... 7

3.3.1 QuantileRegressionForest.................................................... 7

3.3.2 GeneralizedQuantileRandomForest ...................................... 8

4 Results 9

4.1 UnivariateAnalysis.................................................................... 9

4.2 BivariateAnalysis...................................................................... 11

5 Out-of-Sample Prediction 19

5.1 Visualizations........................................................................... 22

6 Data Simulation 27

7 Conclusion 29

References 29

Appendix 31 

About this Honors Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Emory College
Degree	B.S.
Submission	Honors Thesis
Language	English
Research Field	Economics, General Statistics Applied Mathematics
Keyword	Tree based methods Quantile Regression IVXQR VAR
Committee Chair / Thesis Advisor	Rho, Seunghwa, Emory University
Committee Members	Ettinger, Bree, Emory University Xiong, Ruoxuan, Emory University

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Predicting GDP growth across different Quantiles ()	2022-05-06 21:51:07 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions