Spillover Effects in the Customs Union of Russia, Kazakhstan and Belarus

A Common Economic Area (CEA) formed by Russia, Kazakhstan and Belarus since January 1st 2012, following creation of the Customs Union between these countries in 2007 (and in operation since mid-2010), raises a number of topical questions on whether it can be sustainable, trade-stimulating, efficient in terms of long-run economic growth etc. However, an important part in the effects of inter-country influence within such a union is played by the degree the countries are connected through other than trade policies - monetary policy in general and exchange rate policy in particular. The nature of such inter-connectedness is influenced by the proneness of these countries to ‘resource curse' (or ‘Dutch disease' broadly understood as reallocation of production inputs in the economy due to its dependence on natural resource exports). Our objective is to check how monetary policies (including exchange rate policies) in these countries influence the others in the Customs Union.We use quarterly data for 1996-2010 for Russia, Belarus and Kazakhstan on the following economic indicators: real GDP, inflation, bilateral and effective exchange rates, interest rates. Additional data are collected on oil prices, oil-related real GDP in all three countries (estimated using the method suggested by Masaaki 2009), capital flows and a proxy for ‘world GDP’ in real terms. We build a small inter-country forward-looking simultaneous equations model based on the New Keynesian Phillips curve in which economies of Russia, Belarus and Kazakhstan are described using a number of equations. The model contains two layers of links between countries: explicit one through real effective exchange rates (that rely also on trade intensity between the countries) and implicit one using inter-country averages suggested by GVAR methodology (Chudik and Pesaran 2007). Unlike GVAR, our model uses a small number of countries only, and contains one dominating country (Russia). However, as Monte-Carlo experiments described in Charemza et al (2009) suggest, GVAR methodology can be successfully used in case of small number of countries with a dominating country. Our version of the inter-country model is an adaptation of the model built in Charemza et al (2009) with a number of changes suitable for a different set of countries. While Belarus can be considered small opened economy and microfoundations for the type of model used can be found in e.g. Gali and Monacelli (2005) and Benigno and Benigno (2006), for Russia microfoundations have to be different and are loosely derived following the argument from Sosunov and Zamulin (2007) and Charemza et al (2009). Kazakhstan might be regarded as a somewhat ‘middle’ case, since in terms of economy size it is closer to Belarus, but in terms of expected macroeconomic dependencies might be reasonably considered closer to Russia with a potential threat of the Dutch disease. The first equation for each country describes output gap depending on its own lagged values, real effective exchange rate (REER), world output gap, base interest rate and inter-country averages of output gaps (GVAR methodology). The second equation describes dynamics of non-systematic part of current inflation through lagged inflation, output gap (alternatively – oil-related real GDP), REER, expected deviation of future inflation from its target level (forward-looking equation) and inter-country averages of inflation. The REER is modeled as consisting of two parts – external (proxied by REER with USD and Euro) and internal (REER with the other two countries from the model) with trade shares used as weights (third equation of the model). Bilateral exchange rates are modeled as related to output gaps and REERs of corresponding countries (fourth equation of the model). The model is closed by requiring that bilateral exchange rates (after proper transformations) are inversely related to each other. The last equation of the model describes monetary policy rule for each country, allowing for different modifications depending on the previous research on the topic, Central Banks announcements etc. Bilateral exchange rates modeling allows also to reflect the degree of exchange rate control by the Central Banks of the relevant countries. The equations are estimated using GMM method. The model includes 18 equations in total, being quite parsimonious in terms of parameters – 77 in all equations. Being heavily inter-related through both exchange rates inter-influence and trade inter-influence, the model allows us to simulate spillover effects of various policy measures and pass-through effects of the ‘Dutch disease' between the CEA countries. Simulation experiments (modeling changes in exchange rate regimes, base interest rate (monetary policy changes) and external changes reflected in external part of REER) demonstrate that Belarus is most dependent on the other two counterparts of the union.