Семинар «Институциональные проблемы российской экономики»: доклад Ш. Сикхвал

26 мая состоялся очередной семинар возобновленной серии ИАПР «Институциональные проблемы российской экономики».

Новый стажер-исследователь МЦИИР Швета Сикхвал представила доклад "Can machine learning models provide a better forecast of money demand? Evidence from the Indian economy".

Аннотация. Money demand has been a subject of empirical research due to its instability in the money demand function. The study attempts to find a better forecast model by exploring machine learning models. An accurate prediction of money demand can serve the goal of inflation targeting of the central bank of India. Specifically, the research focuses on the random forest regression, gradient boosting, and Xtreme-gradient boost method for the prediction of money demand for the Indian economy for a monthly dataset from 1997 to 2021. The results indicate the machine learning models perform fairly well for predicting both narrow money demand and broad money demand. The results from the ML models are then compared with the traditional econometric approach, namely ARDL. The forecasted values from ARDL deviate significantly from the actual values, proving that ML models could provide a better forecast of money demand.