Risk management and value creation: new evidence for Brazilian non-financial companies

ABSTRACT

The practice of financial risk management with derivatives has received attention both from the academia and the market. In Brazil, there is a growing use of these instruments by companies, in line with the growth of such market in the global economy. This article aims to investigate the relationship between the financial risk management and the value creation to the shareholder for non-financial Brazilian companies. The sample was made up of 1794 firm-year observations from 2006 to 2014. Results obtained via panel data, including the GMM, point out that companies which used derivatives did not add value during the period analysed. An explanation for this result is that most of the companies use derivatives in order to manage the cash flow and not to add value.

KEYWORDS:

• Risk management
• derivatives
• Hedge Accounting
• GMM

JEL CLASSIFICATION:

• G19
• G31
• M41
• C23

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ CVM has powers to discipline, standardize and oversee the performance of the various market players.

² The Economatica System is used by thousands of analysts following Latin America’s stock markets, government bonds, the fund industry and various indicators.

³ Wald test returned a Prob>chic2 = 0.000, rejecting, thus, the null hypothesis of homoscedasticity. Wooldridge test presented Prob>F = 0.000, which makes the null hypothesis of autocorrelation absence to be rejected. Between the two approaches of panel data, robust Hausman test (p-value = 0.000), to 5% significance level, indicated that the most appropriate approach is the one of Random Effects. The R² to be analysed is the between.

⁴ Wald test returned a Prob>chic2 = 0.000, rejecting, thus, the null hypothesis of homoscedasticity. Wooldridge test presented Prob>F = 0.000, which makes the null hypothesis of autocorrelation absence to be rejected. Between both approaches of panel data, the robust Hausman test (p-value = 0.000), at 5% significance level, indicated that the most appropriate approach is the one of Fixed Effects. The R² to be analysed is the within.

⁵ Wald test returned a Prob>chic2 = 0.000, rejecting, thus, the null hypothesis of homoscedasticity. Wooldridge test presented Prob>F = 0.000, which makes the null hypothesis of autocorrelation absence to be rejected. Between both approaches of panel data, the robust Hausman test (p-value = 0.000), at 5% significance level, indicated that the most appropriate approach is the one of Fixed Effects. The R² to be analysed is the within.

⁶ Wald test returned a Prob>chic2 = 0.000, rejecting, thus, the null hypothesis of homoscedasticity. Wooldridge test presented Prob>F = 0.4365, which makes the null hypothesis of autocorrelation absence to be rejected. That is, there was no autocorrelation in this model. Chow returned with Prob>F = 0.000, indicating, therefore, that the panel data method is preferable to the OLS. Between the two panel data approaches, the robust Hausman test (p-value = 0.0074), at 5% significance level, indicated that the most appropriate approach is the one of Fixed Effects. The R² to be analysed is the within.

⁷ Wald test returned a Prob>chic2 = 0.000, rejecting, thus, the null hypothesis of homoscedasticity. Wooldridge test presented Prob>F = 0.5224, which makes the null hypothesis of autocorrelation absence not to be rejected. That is, there was no autocorrelation in this model. Chow returned with Prob>F = 0.000, indicating, thus, that the panel data model is preferable to the OLS. Between the two panel data approaches, the robust Hausman test (p-value = 0.000), at 5% significance level, indicated that the most appropriate approach is the one of Fixed Effects. The R² to be analysed is the within.