International linkages and external shocks: A Global VAR perspective for Belarus. Evidence from different model specifications
Belarus as a small, open economy is closely integrated into the global economy via complex commodity, trade and financial channels. The empirical investigation of the impact of different external shocks, e.g. changes in the global price of crude oil, or a recession in Belarus’ main trading partners on its main macroeconomic variables like GDP, inflation and the exchange rate is of concern to policymakers and researchers alike.
In this policy study, we use a very modern and sophisticated econometric approach to investigate these issues further. The Global Vector Autoregression (GVAR) as a multi-country model has been specifically developed to study such global macroeconomic issues, and has to our knowledge not previously been applied to the case of Belarus.
In our empirical GVAR work, we use two distinct modelling approaches. In the first step, we analyse the propagation of shocks in a small model of the Eurasian Economic Union (EAEU), which comprises the five member countries, including Belarus, as well as some selected external variables. The second step involves a much larger data-set of 35 countries, which allows a much deeper analysis of economic cross-border links. What are our main results, and what policy conclusions can be drawn? We use generalized impulse response functions (GIRF) to analyse the transmission of external as well as domestic shocks on Belarus. In general, empirical reactions follow theoretical considerations. In both versions of the model, the strong effect of the exchange rate on prices (“pass-through”) is clearly observed, as is the pattern of inflation following a shock to broad money aggregates. We also find only weak evidence of the interest channel of monetary transmission. External shocks like a drop in the price of oil, or a fall in Russian or global GDP have also the negative impact on Belarusian output that one would expect. What is perhaps surprising is that the link of Belarus’ GDP appears to be stronger with oil rather than with Russia’s GDP. For Russia, we can identify the typical reaction to a fall in the price of oil, which is a fall in its GDP.
How do the GVARs we estimated perform in comparison with a standard single country Structural Vector Autoregression (SVAR)? For this comparison, we estimated an SVAR model of the Belarusian economy and investigated similar external and domestic shocks. In general, the results obtained from all three different models (large-scale GVAR, small-scale GVAR, and SVAR) point in the same direction, which supports the robustness of the results. What specific model should be applied depends largely on the concrete context of the question.