Comportamiento de la UVR en el largo plazo

Abstract

In this paper, we implemented different techniques to predict the Colombian real value unit (UVR), based solely on the history of its behavior. The UVR represents the purchasing power based on the variation of the consumer price index (CPI) for the calendar month immediately before the calculated one,, in addition, many housing credits are indexed to UVR in Colombia, allowing financial institutions to preserve the purchasing power of money borrowed. For the study, forecasting techniques are considered using SARIMA models, linear regression, additive and multiplicative Holt-Winters smoothing, and artificial neural networks. According to the analysis, it is obtained that the best forecasting technique to make predictions at the end of the seasonal cycle was the multiplicative Holt-Winters smoothing and to predict in the middle of the seasonal cycle was the SARIMA (1,1,1) (0, 1,1).