Mind the gap: forecasting euro-area output gaps with machine learning

ABSTRACT

In this paper, we use the Eurozone yield curve in an effort to forecast the deviations of the euro-area output (IPI) from its long-run trend. We use various short- and long-term interest rates spanning the period from 2004:9 to 2020:6 in monthly frequency. The interest rates are fed to three machine learning methodologies: Decision Trees, Random Forests, and Support Vector Machines (SVM). These Machine Learning methodologies are then compared to an Elastic-Net Logistic Regression (Logit) model from the area of Econometrics. According to the results, the optimal SVM model coupled with the RBF kernel outperforms the competition reaching an in-sample accuracy of 85.29% and an out-of-sample accuracy of 94.74%.

JEL CLASSIFICATION:

• C53
• E43
• E47

KEYWORDS:

• Output gaps
• interest rates
• euro area
• machine learning
• forecasting

Acknowledgments

«This research is co-financed by Greece and the European Union (European Social Fund – ESF) through the Operational Programme “Human Resources Development, Education and Lifelong Learning” in the context of the project “Strengthening Human Resources Research Potential via Doctorate Research” (MIS-5000432), implemented by the State Scholarships Foundation (ΙΚΥ)”

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Optimal in the sense of the model generalization to unknown data

² Our implementation of SVM models is based on LIBSVM, Chang and Lin (2011)

³ From Bootstrap AGGregatING.

⁴ The measure we use is: $\frac{TruePositive+TrueNegative}{Positive+Negative}$, since the data are balanced between the two classes (52.6%-47.4%).

⁵ To address concerns of possible break-points after the 2008 crisis, we used a Chi-squared test of the hypothesis that the model’s accuracy is different in the period 2004:9 to 2007:12 with respect to 2008:1 to 2020:6. The test has a p-value of 0.0646 thus the forecasting performance does not change over time.

Additional information

Funding

This work was supported by the State Scholarships Foundation [MIS-5000432].