Granger causality within monetarist transmission in the Czech republic.
Granger methodology (Granger and Engle 1987) is applied for monthly data that covers the period from January 2001 until September 2011. This period was chosen due to data availability. Relevant measures and objectives of the monetary policy are expressed via monthly monetary base, monetary aggregate (M2), consumer price index, and index of industrial production. Index of industrial production substitutes GDP indicator because GDP data is only available quarterly.
The relevant time series are seasonally adjusted, using the Census One method. This method is used for determination of multiplicative seasonal coefficients, which allow for an eliminating seasonal component of the time series. Because the regression approach can be used only for stationary time series, stationarity of the time series is tested (Hill et al. 2011). If the time series do not fulfill a weak stationarity condition, the stationary time series will be defined due to first differences of the relevant variables. Finally, the stationary series, selected based on theoretical assumptions of the monetary policy transmissions, are tested by using Granger causality tests. In this case, the delay is specified on the basis of the Akaike information criterion. Maximum delay is limited by an 18-month period.
According to Revenda et al. (2011), the monetarists define the beginning of the monetary transmission general form as a change in the monetary base, which is determined by the implementation of monetary policy instruments, especially the open market operations. These operations are realized from the central bank initiative. Changes in the monetary base will affect the money supply, which can stimulate real macroeconomic variables in the short run, like output and employment. In the long run, the money supply changes only affect the change in price level.
Granger causality testing can be used as an effective tool to identify causal relationships between variables which represent different instruments and objectives of the monetarist transmission. According to the monetarist concept, there are relationships tested between monetary base (MB) and monetary aggregate (M2), monetary aggregate and industrial production index (IP_CR), and monetary aggregate and consumer price index (CPI). The test results are presented in Table 1.
Table 1 Granger causality testing results Observations F-statistics p-value MS does not Granger Cause MB 125 2.18650 0.0933 MB docs not Granger Cause M2 125 2.68666 0.0497 IP_ CR does not Granger Cause M2 126 4.46532 0.0135 MS does not Granger Cause IP_CR 126 4.75212 0.0103 CPI does not Granger Cause MS 127 3.18623 0.0767 MS dues not Granger Cause CPI 127 8.48658 0.0042
The causal relationships that are bolded indicate the rejected null hypothesis, which occurs when we refuse negative claims about the causal relationships within the meaning of the Granger causality concept. That means that if p-values in the table are lower than 0.05, then (at the appropriate level of significance we reject the null hypothesis) and we accept an alternative hypothesis. The null hypothesis is formulated as follows: the certain variable does not affect the other variable in the causal sense, according to Granger. Therefore, if the null hypothesis is rejected, i.e., if the p-value is low (the appropriate level of significance), the first-mentioned variable affects another variable in the sense of Granger causality.
In terms of the test results, causality of the relationship between a monetary base and money aggregate in the Granger concept has been proven, but a causal link between a money aggregate and real (industrial) production was mutual. This leads to the conclusion that the money supply cannot be considered for the exogenous variables in macroeconomic models.
On the other hand, due to the complexity of the problem, the simplicity of the Granger causality test methodology does not allow for formulating strong conclusions. Interpretation of results is also limited by the fact that the time series cover only relatively short periods. Test results, therefore, can only be considered indications, which signal discrepancy problems between the trivial economic theories assumptions and economic reality. Granger causality analysis results will be used for creating the more sophisticated econometric models (vector autoregressive models, respectively vector error correction models) which can be useful for advanced research of the monetary policy's consequences.
S. Burian (*) * J. Brcak
Faculty of Economics and Management, Department of Economic Theories, Czech University of Life Sciences, Prague, Czech Republic
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||RESEARCH NOTE|
|Author:||Burian, Stanislav; Brcak, Josef|
|Publication:||International Advances in Economic Research|
|Date:||Feb 1, 2013|
|Previous Article:||Income, inequality, and social welfare: a study of the U.S. states.|
|Next Article:||Migration and welfare.|