Inflow and outflow forecasting of currency using multi-input transfer function

Abstract

Time series is a series of observations taken sequentially by the same time interval, can be daily, weekly, monthly, yearly or others. A time series can be affected by other variables (input series). Preparation of a model with one or more input series and a series output can be done with the transfer function. Bank of Indonesia through Open Market Committee has monthly agenda to make projections net flow currency in circulation to control liquidity. The purpose of this study was to examine the effectiveness of multi input transfer function in improving the accuracy of the realization of the provision of the inflow and outflow of currency in Indonesia. This study uses the inflow and outflow data of currency by Bank of Indonesia, CPI, the tourists visit data, and the middle rate of the US Dollar. The result are multi-input transfer function model for the inflow series have accuracy values (RMSE) of 319.16, while the final model for the series outflow is 456.78. Based on the criteria of out-sample, the best forecasting model is by using a multi-input transfer function. The addition of input variables can improve the accuracy of the model by 38% for inflow and outflow, compared to naïve models