A Study on the Estimation of Interest Rate Prediction Model Using VECM
This paper estimates an interest rate prediction model using monthly data on the New Wage Base (NWB) and Balance Standard (BS) of mortgage rate, and KORIBOR provided by the Bank of Korea's Economic Statistics System and the Korea Federation of Banks. The vector error correction (VECM) model was used as a research model, and the model was set up and estimated by performing the Granger Causality Test and Cointegration Test. The Multivariate Portmanteau Test was performed to test the goodness of fit of the model. The causal test results show that there is a bidirectional linear dependency in which each time series variable is affected by the past values of itself and two other time series variables. The model was set up as a vector autoregressive (VAR) model using the Schwarz Bayesian Criterion (SBC) statistic based on the Minimum Information Criterion. Then, as a result of cointegration test using trace statistics, a model with a constant intercept in the error correction term was selected. The prediction model was estimated using the selected model, and as a result of testing the goodness of fit of the prediction model with the multivariate Portmanteau test, the cross-correlation no longer existed because the P-values of the chi-square statistics were all greater than the significance level of 0.05 until the maximum delay of 3 to 12 lags. These results imply that the interest rate prediction model presented in this paper is suitable and, therefore, the predictive value of interest rate could be presented using the model. The results of this paper can be very important for providing useful information for forecasting future real economy and analyzing the effectiveness of monetary policy of government and financialinstitutions.