Inflation Forecasting - Are ML Models Superior? Evidence from India

• 발행기관: 서울대학교 경제연구소
• 저자명: Sudipta Dutta
• 간행물 정보: 『Seoul Journal of Economics』Volume 37 No.4, 243~272쪽, 전체 30쪽
• 주제분류: 경제경영 > 경제학
• 파일형태: PDF
• 발행일자: 2024.11.30

국문 초록

Disruptions in the channels of production, distribution and sale of agricultural and industrial products driven by the pandemic outbreak affected the linkages of inflation across the world as well as the accuracy of traditional inflation forecasting models. The validity of linear econometric models, which assume a linear and static linakge between the variable of interest and its regressors, have long been a subject of scrutiny. As a result, alternative models, especially, machine learning (ML) based predictive models have emerged in an attempt to more accurately capture the evolving dynamics of inflation. ML models have the capability to capture non-linear connections between inflation and its determinants. The study compares the forecasting performance of various ML models with popular econometric models for both the period prior to the pandemic as well as the period post the pandemic. The findings substantiate the superiority of ML models over linear econometric models in terms of improved predictive performance when forecasting inflation in India over various horizons.

목차

I. Introduction

II. The Indian Context

III. Literature Review

IV. Research Methodology

V. Empirical Analysis

VI. Conclusion

References

키워드

Inflation Econometric Machine learning Neural network Long-short term memory