Applications of Cointegration Analysis in Economy of the Czech Republic

The purpose of this paper is to introduce and examine the possibilities of using cointegration analysis based on VAR models for modelling the Czech Republic economy.(1)) At first, the used methodology and its characteristics and advantages are described. Then, two different cases with conclusions for the economy of the Czech Republic are given. In the first example, long-run sustainability of the current account (and thus also sustainability of the foreign debt) is investigated. Cointegration relationship is tested by using a two-equation VAR(4) model, which reflects the relationships between variables which quantify export and import. According to the results of the tests and especially with respect to estimated values of the parameter we were interested in, we can submit that, in spite of the potential risk that the imports are growing faster than exports, the current account of the Czech Republic is sustainable in the long-run. The second example examines long-run tendencies of the monetary sector of the Czech Republic. At first, a two-dimensional cointegration relationship is stated (using a four-equation VAR model) among the variables which are usually used for describing the monetary sector. Thus, this relationship can be divided into two unique ones. The first of them is stationarity of real long-term interest rate ("Fischer effect") with the inflation parameter too large according to the expected value, It can be interpreted as higher sensitivity of interest rate to expected inflation rate. The second relationship can be interpreted as equality between money supply and demand (where opportunity costs can be expressed either by interest or inflation rate) with income elasticity positive but higher than one, interest rate elasticity negative and inflation rate elasticity also negative. Thus, cointegration analysis can be used for modelling of the economy of the Czech Republic to give valuable results. It is suitable especially for the cases where we don't suppose any a priori dependences among modelled variables and where we want to examine a relationship among them generally.