Tue dole of National Bank of Rwanda from 1995 to 2010

3.4. Data presentation, analysis and interpretation

The data obtained were analyzed using econometrics methodology. Money stock constitutes the dependent variable. The study employs an econometric technique of cointegration methodology.

To formulate a monetary reaction function for National Bank of Rwanda, the Taylor rule equation was adopted to the context of monetary policy in Rwanda.

The original Taylor rule can be expressed as following:

FFR=f (YG, IG) (1)

Where: FFR= the federal funds rate, YG= the output gap, IG= the inflation gap.

Because of data availability problems for the monetary base series of Rwanda, the monetary stock aggregate (Mt) will be used as the instrument policy. Indeed, the monetary stock aggregate (M) plays an important role in Rwanda monetary policy since the National Bank of Rwanda assumed its responsibility to regulate liquidity in the economy and the data of the monetary base are frequently referred to Mt.

In respect of goal variables, inflation and output will be used. The former variable has emerged from many economists as the real goal of monetary policy in order to maintain price stability and the latter is considered as a historically objective of monetary policy in various countries.

In the context of Rwanda, the strategy used by the Central Bank is to ensure that liquidity expansion is consistent with target inflation and GDP growth levels. Thus, the modified version of Taylor's rule to be estimated can be written as:

Mt= B₀+ B₁ (IG_t) + B₂ (YG_t) + _t (2)

However, recently, with number of empirical studies related to the Taylor rule, economists argue that the exchange rate would also be an essential state variable that has to be included in the model in the case of a small and open economy. On this basis, the equation (2) is extended as follow:

Mt= B₀+ B₁ (IG_t) + B₂(YG_t) + B₃DEX_t + _t (3)

EX_t = the change in exchange rate in terms of the Rwandan Francs per US Dollars,

B₀ is constant term and B_1, B_2, B₃ are coefficients respectively to be estimated empirically,

_t = the error term that presents all other variables which can have effects on Money stock in Rwanda.

The equation (3) can be seen as a function in which the money stock aggregate reacts to the Inflation gap, Output gap and the change in Exchange rate.

The version of the equation (3) to be empirically estimated can take a dynamic form since there is the lag response of Monetary Authority. On this basis, the equation (3) is expressed as follows:

LnMt= B₀+ lnM_t-1 + lnIG_t + lnIG_t-1 +lnYG_t + lnYG_t-1 +lnDEX_t +lnDEX_t-1 + _t (4) and finally the equation (4) become:

M_t= B₀+ B₁M_t-1 + B₂IG_t + B₃IG_t-1 +B ₄YG_t + B₅YG_t-1 +B ₆DEX_t +B₇DEX_t-1 + _t (5)

(Mukunzi, 2004: 34)

The Taylor Rule has been considered as a representation of Central Bank behavior in various countries. It provides information about the responsiveness of the monetary policy instrument to the monetary variables. Therefore, estimating the policy behavior of the Rwandan Central Bank and determining the target the Central Bank followed, is essential to the different policy implications, especially to the implementation of an accurate and successful monetary policy.

According to Mishkin (1997) six basic goals are continually mentioned by Central Banks when they discuss the objective of monetary policy:

- High employment level, - Economic growth, - Price stability, - Interest- rate stability, - Stability of financial markets, - Stability in foreign exchange rate markets.

The primary objective is to establish a model that explains well the relationship between the money stock and some macroeconomic variables in Rwanda. In addition, the qualitative and quantitative impacts of each of these variables on money stock are determined. A lot of other information is also obtained. For instance, it will be possible to know what the different partial elasticity that pertained by each variable is.

The augmented Duckey-Fuller Unit Root Test will be used for the purpose of data analysis throughout the research. According to Gujarati (1999: 455-467), it is this test which detects the stationary of a variable. Many other tests will also be conducted.

According to Gujarati (1999: 377-398), the Durbin-Watson test, the Runs test or the examination of the residuals are techniques that will be used in relation the problem of serial correlation.