Regressão dinâmica aplicada na previsão de demanda de PVC no mercado brasileiro

Abstract

This work addresses the estimation of polyvinyl chloride (PVC) consumption in Brazil. PVC is one of the most produced polymers in the world, with a wide range of applications. The civil construction area is the main sector demanding the product for the production of final items such as pipes, fittings, wires, cables, and windows. The goal of the study is to develop a predictive model of demand increase for the civil construction segment according to the effects of the main economic and industrial indices for the manufacture of plastic pipes and fittings, based on a dataset from a Brazilian petrochemical company, between the first quarter of 2014 and the second quarter of 2023. The dynamic linear regression model obtained included the civil construction industrial production index of the current and previous quarter, the exchange rate of the previous quarter and the variable that represents the quarters of the year. The model managed to explain 59% of the sales volume variability according to the adjusted R², with a residual analysis showing that this was an adequate fitting. Controlling for other factors, it is estimated that the first and fourth quarters present similar average PVC sales volumes. The highest rate of variation in the previous quarter presents characteristics of current demand and the level of industrial production in civil construction in the quarter tends to increase the average sales volume. The study highlights the important role of statistical analysis and its potential for companies in the sector to better understand the production process and its effects on product demand, allowing the generation of useful projections for its evolution.