이자율 기간구조가 주가지수에 미치는 영향 분석: ARDL과 기계학습 모형들 간의 분석 효율성 비교

This study aims to analyze the impact of the term structure of interest rates on the stock market with the daily data from 2013 to 2022. The efficiency criteria based on the ARDL regression that captures the financial econometric characteristics is compared with the learning outcomes of major machine learning models. The relationship was estimated using the ARDL model, and the resulting RMSE and MAE were used as comparative statistics with the learning outcomes of machine learning models. The learning outcomes of the decision tree-based XGBoost model and the SVR model were superior to those of the NN-based LSTM model, but the analysis efficiency was lower than that of the ARDL regression estimates. This result suggests that when there are limitations in data size or when the specific econometric relationships between variables are assumed, machine learning analysis is not necessarily superior to traditional financial econometric analysis. However, it cannot be ruled out that the efficiency of analysis may change when the analysis is conducted based on the relevant big data.