Testing the joint hypothesis of rationality and neutrality under seasonal cointegration: the case of Korea

Abstract

The monetary policy of Korea of the last two decades was apparently conducted in a systematic and predictable manner, and believed to be effective. However, according to the monetary policy ineffectiveness proposition, predictable monetary policy has no real effect on output. The purpose of this research is to investigate whether the monetary policy of Korea has been effective or not. The investigation is carried out by testing the so-called joint hypothesis of rationality and neutrality of monetary policy. The test is performed with both seasonally adjusted and seasonally unadjusted data, using a general error-correction multivariate system of money, interest rate, price and output. Upon establishing that these four variables are integrated of order one, cointegrating relations were investigated with seasonally adjusted data, and one cointegrating relation was found. similarly, with seasonally unadjusted data, one cointegrating relation at the zero frequency (long-run component) and one at the biannual frequency were found. with seasonally adjusted data, the joint hypothesis cannot be rejected at the 5 percent confidence level, but the test result is ambiguous at the 10 percent level. with seasonally unadjusted data, the hypothesis cannot be rejected at levels much higher than 10 percent level. The difference between the two test results, although somewhat marginal, is the consequence of exploiting the additional information contained in seasonally unadjusted data. The result of this research is relevant for both macroeconomic policy and econometric practice. From the macroeconomic viewpoint, it provides evidence of money neutrality and rationality in Korean data. Its policy implication is that a fixed money supply rule such as Friedman's k percent rule is preferable to more complex countercyclical money supply rules aimed at fine-tuning the economy. From the econometric viewpoint, it offers evidence that seasonal adjustment procedures, such as the commonly used X-11 procedure, may eliminate useful information contained in the original data. It follows that using seasonally unadjusted data rather than adjusted data for empirical work may be more appropriate.