Fitting and forecasting of nonlinear Taylor rule in China

Abstract

Purpose– In this paper, six forms of non-linear Taylor rule have been applied to compare the fitting and prediction of response function of monetary policy of China, in an attempt to figure out a form of non-linear Taylor rule that accords with Chinese practices. The paper aims to discuss this issue.Design/methodology/approach– In this paper, the authors will conduct in-sample fitting and out-of-sample prediction on the response function of monetary policy of China by introducing the factor of exchange rate and by applying forward-looking, backward-looking and within-quarters non-linear Taylor rule with data from the first quarter of 1994 to the second quarter of 2011, with a view to provide reference for formulation and implementation of monetary policies of China.Findings– By analyzing the experimental data, the authors find that first, after introducing the factor of exchange rate, both the implementation effect and prediction ability of the monetary policies improve. Exchange rate has a relatively greater influence on the effect of the monetary policies during low inflation period. Introduction of exchange rate can improve the prediction accuracy of our monetary policies significantly. Second, as the implementation effect of monetary policy under different macro-background varies greatly, the situation should be correctly appraised when formulating and implementing monetary policies. According to the empirical results, the monetary policies have obvious non-linear characteristics, and transit smoothly with the change of inflation rate. On the two sides of inflation rate of 2.174 percent, there is an asymmetry response.Research limitations/implications– Surely, the conclusions are reached on the basis of quarterly data and one-step prediction method. It is no doubt that use of frequency mixing data including quarterly and monthly data will provide more sample information for studying relevant issues. And the use of multiple-step prediction method may cause a dynamic change of prediction indicators of models, which will help choose more appropriate prediction models. That is what the authors will study next.Originality/value– First, by introducing exchange rate, this paper will extend non-linear Taylor rules and test its applicability and fitting effect in China. Second, figure out a non-linear Taylor rule that conforms to Chinese practices with data. In this paper, multiple forms of non-linear Taylor rules and actual macro date will be adopted for fitting and finding out a non-linear Taylor rule that fits Chinese practices. Third, empirical basis will be provided for further perfecting monetary policies prediction models. As there are few studies in connection with the prediction accuracy of non-linear Taylor rules so far, this paper will compare and study the prediction accuracy of non-linear Taylor rules by utilizing multiple advanced prediction techniques, so as to offer a beneficial thinking for predicting and formulating monetary policies by the central bank.