Forecasting Russia's Key Macroeconomic Indicators with the VAR-LASSO Model

Fokin, Nikita; Polbin, Andrey

doi:10.31477/rjmf.201902.67

Cited by 5 publications

(3 citation statements)

References 21 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This was especially evident in the forecasting of key macro indicators for up to two years. Also, [Fokin, Polbin, 2019] estimated VAR-Lasso models for making scenario forecasts of the Russian GDP and its components to 2019-2024. The researchers have chosen as a basis for comparison a couple of models, such as ARIMA, VAR, and BVAR by Pestova and Mamonov, as well as the official predictions of the Ministry of Economic Development and IMF.…”

Section: Literature Review Of the Use Of Bayesian Models In Macroecon...mentioning

confidence: 99%

Comparing Forecasting Accuracy between BVAR and VAR Models for the Russian Economy

Rudakouski

2023

HSE Economic Journal

View full text Add to dashboard Cite

This paper investigates variations in the accuracy of forecasting key macroeconomic indicators through the comparison of Frequentist and Bayesian vector autoregression (VAR) models. The primary aim of the study is to identify the most effective prior type in minimizing forecast errors for the key macroeconomic indicators in the context of the Russian economy. A significant aspect of this research involves elucidating the theoretical foundations of Bayesian methods and delineating the roles of different priors in the prediction of macroeconomic indicators. A pivotal consideration in the application of the Bayesian approach is the diversity of priors, such as Jeffreys and Minnesota, which may overlook economic considerations like inflation targeting and neutral money. Conversely, certain priors, such as steady-state or independent normal-Wishart priors, are grounded in economic policy. The study delves into the nuanced interplay between these priors and their implications for forecasting accuracy. The empirical findings reveal that all Bayesian VARs exhibit superior forecasting accuracy compared to the classical VARs. Furthermore, expanding the model's scope from a limited number of variables (4) to a more comprehensive set (18) enhances forecast precision, as evidenced by the escalating log-predictive scores, Model Confidence Sets, and The Diebold-Mariano test. Simultaneously, the BVAR with the steady-state prior has demonstrated the lowest forecast error over a two-year period, but the prediction with the Minnesota prior looks relatively stable in all horizons.

show abstract

Section: Literature Review Of the Use Of Bayesian Models In Macroecon...mentioning

confidence: 99%

Comparing Forecasting Accuracy between BVAR and VAR Models for the Russian Economy

Rudakouski

2023

HSE Economic Journal

View full text Add to dashboard Cite

show abstract

“…The quality metric used in the research is RMSFE (Root Mean Squared Forecast Error): (15) This metric can often be seen when solving regression tasks (it is used, for instance, in Fokin and Polbin, 2019;Baybuza, 2018). All calculations 4 were performed in R using the following packages:…”

Section: Generating Forecastsmentioning

confidence: 99%

“…A paper by Fokin and Polbin (2019) is dedicated to the alignment of vector autoregression and a LASSO-regularisation (VAR-LASSO) model to forecast the major indicators of the Russian economy: GDP, consumption, and fixed capital investment based on exogenous oil price shock. Based on the testing results, the model offered by the authors demonstrates good predictive power as compared to an ordinary VAR model and to the forecasts of the MED.…”

Section: Introductionmentioning

confidence: 99%

Use of Machine Learning Methods to Forecast Investment in Russia

Gareev¹

2020

RJMF

View full text Add to dashboard Cite

This work forecasts the growth rate of quarterly gross fixed capital formation in Russia using machine learning methods (regularisation methods, ensemble methods) over a horizon of up to eight quarters. The methods tested show higher quality in terms of RMSFE than that of simple alternative models (autoregressive model, random walk model), with ensemble methods (boosting and random forest) leading in quality. The last statement is consistent with the results of other research on the application of big data in macroeconomics. It was found that removing observations from the sample which relate to the time before the 1998 crisis and that are atypical for the subsequent period of time does not worsen the short-term forecasts of machine learning methods. Estimates of the coefficients of generally accepted key investment factors obtained using regularisation methods are, on the whole, consistent with economic theory. The forecasts of the author's models are superior in quality to the annual forecasts of growth rates of gross fixed capital formation published by the Ministry of Economic Development.

show abstract