Modeling and Forecasting a Firm's Financial Statements with a VAR-VECM Model

Abstract

This paper details efforts at developing and estimating a Vector Autoregressive (VAR) econometric model representative of the financial statements of a firm. Although the model can be generalized to represent the financial statements of any firm, this work was carried out as a case study, where the chosen company is the largest firm in Brazil: Petrobras S/A. The methodology utilized makes use of correlation analysis, unit root tests, cointegration analysis, VAR modeling, Granger causality tests, in addition to impulse response and variance decomposition methods. Besides the endogenous financial statement variables, an exogenous variable vector was utilized, namely, Brazilian GDP, domestic and foreign interest rates, the international oil price, the exchange rate, and the country risk. The final version of the model is a Vector Error Correction Model (VECM), which takes into account the cointegrating relationships among the endogenous variables. After estimation and validation, the model is used to produce forecasts of the financial statements of the firm under study. Estimates for the exogenous variables and dividend forecasts were also used to estimate the firm's market value. The results are apparently robust and the authors expect they might contribute to the field of financial planning and forecasting.