Research Article

When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets

Volume: 1 Number: 1 August 11, 2026
EN TR

When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets

Abstract

This paper asks not whether tree-based machine-learning models beat a random walk in forecasting energy inflation, but under which conditions they fail to. We use monthly energy consumer price index data for Türkiye, South Africa, Mexico and Poland for 2002–2022, keeping the information set deliberately small and univariate. One-step-ahead forecasts from a random walk, an AR(1), two linear models (ordinary least squares and ridge) and two tree-based models (gradient boosting and random forest) are compared using the Diebold-Mariano and conditional Giacomini-White tests. On average no estimated model beats the random walk, and in some countries the tree-based models are significantly worse. More important, performance depends on the inflation regime: tree-based models lose accuracy against the random walk when inflation is high. The pattern runs in the same direction across countries and estimation windows, though its pooled significance is sensitive to how cross-country dependence is treated. The evidence fits a structural explanation: tree-based methods cannot predict outside the range of their training data, so their flexibility becomes a disadvantage once energy inflation rises far above past values. During inflation crises in emerging markets, simple benchmarks remain hard to beat, and models should be evaluated conditionally rather than on average.

Keywords

References

  1. Alquist, R., Kilian, L., & Vigfusson, R. J. (2013). Forecasting the price of oil. In G. Elliott & A. Timmermann (Eds.), Handbook of economic forecasting (Vol. 2A, pp. 427–507). Elsevier.
  2. Atkeson, A., & Ohanian, L. E. (2001). Are Phillips curves useful for forecasting inflation? Federal Reserve Bank of Minneapolis Quarterly Review, 25(1), 2–11.
  3. Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32.
  4. Clark, T. E., & West, K. D. (2007). Approximately normal tests for equal predictive accuracy in nested models. Journal of Econometrics, 138(1), 291–311.
  5. Clements, M. P., & Hendry, D. F. (2006). Forecasting with breaks. In G. Elliott, C. W. J. Granger, & A. Timmermann (Eds.), Handbook of economic forecasting (Vol. 1, pp. 605–657). Elsevier.
  6. Diebold, F. X. (2015). Comparing predictive accuracy, twenty years later: A personal perspective on the use and abuse of Diebold–Mariano tests. Journal of Business & Economic Statistics, 33(1), 1–9.
  7. Diebold, F. X., & Mariano, R. S. (1995). Comparing predictive accuracy. Journal of Business & Economic Statistics, 13(3), 253–263.
  8. Faust, J., & Wright, J. H. (2013). Forecasting inflation. In G. Elliott & A. Timmermann (Eds.), Handbook of economic forecasting (Vol. 2A, pp. 2–56). Elsevier.

Details

Primary Language

English

Subjects

Machine Learning Algorithms, Applied Macroeconometrics, Inflation

Journal Section

Research Article

Publication Date

August 11, 2026

Submission Date

June 18, 2026

Acceptance Date

August 4, 2026

Published in Issue

Year 2026 Volume: 1 Number: 1

APA
İldeş, E. C. (2026). When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets. Economic & Financial Analysis Letters, 1(1), 27-41. https://izlik.org/JA66FL92MK
AMA
1.İldeş EC. When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets. EFL. 2026;1(1):27-41. https://izlik.org/JA66FL92MK
Chicago
İldeş, Erol Can. 2026. “When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets”. Economic & Financial Analysis Letters 1 (1): 27-41. https://izlik.org/JA66FL92MK.
EndNote
İldeş EC (August 1, 2026) When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets. Economic & Financial Analysis Letters 1 1 27–41.
IEEE
[1]E. C. İldeş, “When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets”, EFL, vol. 1, no. 1, pp. 27–41, Aug. 2026, [Online]. Available: https://izlik.org/JA66FL92MK
ISNAD
İldeş, Erol Can. “When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets”. Economic & Financial Analysis Letters 1/1 (August 1, 2026): 27-41. https://izlik.org/JA66FL92MK.
JAMA
1.İldeş EC. When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets. EFL. 2026;1:27–41.
MLA
İldeş, Erol Can. “When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets”. Economic & Financial Analysis Letters, vol. 1, no. 1, Aug. 2026, pp. 27-41, https://izlik.org/JA66FL92MK.
Vancouver
1.Erol Can İldeş. When Do Tree-Based Models Fail? Evidence from Energy Inflation Forecasting in Emerging Markets. EFL [Internet]. 2026 Aug. 1;1(1):27-41. Available from: https://izlik.org/JA66FL92MK

Economic & Financial Analysis Letters (EFL) | e-ISSN:xxxx Published by the Faculty of Political Sciences, Ankara Yıldırım Beyazıt University. Contact: efljournal@aybu.edu.tr The views expressed in published work are those of the authors and do not represent the official position of the journal. Work published in this journal is licensed under CC BY-NC 4.0.