Análise regional do volume de vendas no varejo brasileiro via regressão linear

Abstract

The influence of macroeconomic indicators on retail sector performance refers to their capacity to explain variations in sales and commercial activities over time, which is essential for defining business strategies and formulating regional public policies. Using monthly data collected between 2012 and 2022, this study developed a predictive linear regression model linking economic variables to the Retail Sales Volume Index (IVV) across the five regions of Brazil. The Regional Economic Activity Index (IBCR), unemployment rate, interest rate, seasonal indicators, extraordinary effects from the COVID-19 pandemic, and IVV lags were included in the model as covariates. The final model demonstrated high predictive power, with an adjusted R² of approximately 0.79, and showed practical consistency in the signs of estimated coefficients. The results indicated that, controlling for other variables, retail performance varies among regions and tends to improve with increases in IBCR and reductions in unemployment and interest rates, while seasonal and atypical events, significantly impact performance. Residual analysis confirmed the model’s adequacy, revealing near-normal distribution, homoscedasticity, and absence of autocorrelation. Predictive evaluation demonstrated greater model efficiency in the Northeast, Southeast, and Central-West regions, while the North and South regions presented greater predictive challenges, suggesting the need for future methodological refinements. Overall, the analyses provide critical insights for optimizing retail sector strategies and developing public policies aimed at promoting regional.