Intelligent Method of Predicting the Discount Rate Trend

Анотація

The predictability of National Bank of Ukraine (NBU) monetary policy is an important point to gain credibility as an inflation targeter. Basing on the theoretical and empirical literature the direct tool-expert expectations about up-coming interest rates decision - was chosen as approach on quantitative assessment of NBU's predictability. It is found that generally economic agents do not make a systematic mistake in interest rates expectations. But, predictability is an imperfect that is normal for emerging market countries. Some asymmetries in predictability across interest rates cycle is found. However, economic uncertainty is taken in similar way between experts and policy-makers meaning that transparency policy is generally effective. An intelligent method for predicting the discount rate trend has been developed, based on machine learning methods: ARIMA, LSTM and Prophet. Conducted RMSE Errors, trend: bond rate - <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{ARIMA} = 8.15,\ \text{LSTM} =6.39$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{Prophet} =4.15$</tex> ; expert assessment of the bond rate - <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{ARIMA} =8.18,\ \text{LSTM} =7.01$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{Prophet} = 5.99$</tex> .

Класифікація

Ідентифікатори

Схожі роботи