The proposed article is devoted to forecasting the exchange rate using time series analysis methods. This study was conducted in 2 stages. At the first stage, the theoretical basis for forecasting the exchange rate based on the views of scientists of the post-Keynesian economic school was considered. Representatives of this school developed the Quarterly Projection Model, which is used for small open economies with a floating exchange rate and inflation targeting. In the Quarterly Projection Model, the factors affecting the formation of the exchange rate are considered. Such factors are GDP growth, the state of the balance of payments, NBU FX interventions, and the inflation rate. We separately investigated the impact of these factors on the exchange rate of the USD/UAH currency pair. For example, changes in GDP can directly affect the relative value of the hryvnia. The Key Policy Rate is a key instrument of the NBU's monetary policy. An increase in the Key Policy Rate can strengthen the hryvnia, while its decrease can lead to its weakening. Another factor affecting the exchange rate is the current account of the balance of payments. A small imbalance of this indicator (2–5% of GDP) is considered a standard phenomenon for many countries. An excessive current account deficit increases the risk of devaluation of the national currency. High inflation forces hryvnia holders to abandon it and switch to other liquid means of preserving real wealth, such as foreign currency. Therefore, a rather strong, rapid and constant increase in prices becomes the reason for a significant increase in the exchange rate. It is worth noting that the influence of these indicators occurs with a significant time lag. We used the Granger causality test to empirically analyze the causal relationships between the exchange rate and influencing factors. Empirical analysis has shown that the volume of NBU interventions, the balance of payments and the Key Policy Rate are significant and influential factors in the formation of the exchange rate in Ukraine. At the second stage, exchange rate forecasting was carried out using the vector autoregression (VAR) model. The application of the VAR model made it possible to investigate the complex interrelationships between economic indicators. Also, this model made it possible to take into account time lags and exogenous variables. The obtained results indicate the high adequacy of the vector autoregression (VAR) model in forecasting the exchange rate in the short- and medium-term planning periods.
Read full abstract