How are industry concentration and risk factors related? Evidence from Brazilian stock markets

Abstract

We isolate the two factors pointed out by previous literature as potential causes for the documented relation between returns and industry concentration: innovation risk (Creative Destruction Hypothesis, Dominant Replacement Effect Hypothesis) and distress risk (Market Power Hypothesis). Brazilian law allows us to isolate the two effects. Using data from Brazilian firms listed in BOVESPA, we estimate a regression model that uses State-Controlled Enterprises as a control group to separate the concentration/returns trade-off in two risk components. We find little evidence of the existence of a relation between industry concentration and stock returns. However, we do find evidence of the existence of innovation and market power effects. Oligopolistic firms are responsible for most of the investment ininnovation. Our findings do not support the Creative Destruction Hypothesis. They support the Dominant Replacement Effect Hypothesis.

KEYWORDS:

• Concentration
• competition
• premium
• oligopoly
• product markets

JEL classification::

• G32
• L11.

Acknowledgements

I would like to thank Sheri Tice, Robert Hansen, Paul Spindt, Jaideep Shenoy, David Lesmond, José Luís Rossi Jr., Antonio Mazali, the editors Gail Pacheco and Steffen Lippert, three anonymous referees, and participants at the 2011 Midwest Finance Association Meetings, 2012 World Finance Conference, and 2012 Financial Management Association Meeting, and seminars at the Catholic University of Brasilia and INSPER for helpful comments. I would like to thank Cynthia Fransen, Will Dunn, and Tom Connor for the English editing. All remaining errors are my own responsibility.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. For instance, check Section III of Harris and Raviv (1991) for the classic literature on the matter. For more recent developments on the field, check Istaitieh and Rodríguez-Fernández (2006).

2. See, for instance, von Stackelberg (1934), Spence (1977), Dixit (1979), Dixit (1980).

3. For example, firms could use debt to improve their positions in bargains with suppliers or labor unions (Bronars & Deere, 1991; Kale & Shahrur, 2007; Subramaniam, 1996; Titman, 1984). Several studies documented similar interactions between product and financial markets in a variety of different contexts. For a broad review of the literature on product market/financial market interactions, see Istaitieh and Rodríguez-Fernández (2006).

4. The legal term that characterize firms in which both the State and private citizens hold part of the firms’ equities is Sociedades de Economia Mista (Mixed Economy Societies). These are the entities we characterize as SCEs in our study.

5. Numbers of 27 May 2012.

6. See DiPietro (2007), p. 427.

7. See DiPietro (2007), p. 427.

8. The only problem period is between 2001 and 2005, the period in between the approval of laws No. 10,303 and No. 11,101. During this period, the bankruptcy of SCEs is not mentioned in any of the written legal codes that regulated corporations. However, many law specialists still think that, even in this period, SCEs could not go bankrupt. Ulhôa (2003), who was the official Ministry of Finance Chief of Legal Affairs during that period, discussed this issue and came to the conclusion that in 2003, SCEs could not go bankrupt.

9. See Giambiagi and de Além (2009), Chapter 3.

10. SELIC rates are daily averages of all government financing operations that involved bonds issued by the Brazilian Treasury Department and the Brazilan Central Bank.

11. See, for instance, Fama and French (1996) and Ferguson and Shockley (2003).

12. See Ho, Imai, King, and Stuart (2007).

13. Using patent application data, Rocha and Ferreira (2001) show that there was a substantial gap between R&D investments from SCEs and non-SCEs. Firms that underwent privatization in the 1990s have considerable reduced the R&D investment gap. These were firms that were protected from distress risk, but after privatization they were no longer protected. If this is the case, then firms in more competitive industries that were privatized should be the ones investing the most in R&D. We could be attributing this growth in R&D investment to market power when in fact it would be due to innovation risk. This issue should also be addressed by our Matched Sampling algorithm. If we selected our Propensity Score variables correctly, each firm in the treatment group population would be matched with a firm in the control group population that behaves the same way as she does, and this endogeneity issue should not be of concern.

14. See Rosenstein-Rodan (1943) and Murphy, Shleifer, and Vishny (1989) for a more detailed discussion on “Big Push” policies.

15. See Abreu (1990).

16. In unreported results, we used the (Hausman, 1978) test in order to test for the existence of correlation between our covariates and the firm-effect term for the whole sample. This test is often used to select between Fixed and Random effects model. We were not able to reject the null hypothesis that there is no correlation between our covariates and the firm-effect term, which indicated that Random Effects yields the more efficient estimates of Equation (1(1) ). However, we would like to allow for correlation between the unobserved heterogeneity term and our covariates, which would control for endogeneity in unobservable heterogeneity, giving us unbiased estimators in this case. Therefore, all estimations were conducted using the Fixed Effects specification. These results are available upon request.

17. See Di Pietro (2007), p. 427.