Family Control, Socioemotional Wealth, and Governance Environment: The Case of Bribes

For the relevant research, see Anderson et al. (2003), Anderson and Reeb (2003), Burkart et al. (2003), Fernandez and Nieto (2006), Chrisman et al. (2004), Carney (2005), Le Breton-Miller and Miller (2006), Villalonga and Amit (2006), Wu et al. (2007), and Chua et al. (2011), among others.

For instance, Ali et al. (2007) analyze the typical agency problems faced by family-controlled firms in the U.S. and find that these firms report higher quality earnings; in another accounting study, Chen et al. (2010) find that family-controlled firms are less tax aggressive. Anderson and Reeb (2003), Villalonga and Amit (2006), and Zhang and Zhang (2004) examine the financial performance and firm value of family-controlled firms versus non-family-controlled firms. Recent studies in international business show that family ownership affects a firm’s strategy to operate globally (e.g., Fernandez and Nieto 2005, 2006; Gómez-Mejía et al. 2010).

One referee pointed out that corruptions, unethical behaviors, bribery, and frauds are overlapping concepts, but not exactly the same. We acknowledge this point. Since prior studies used all these terms in their studies, we used them as well in our study, but our focal point is the extent of bribery.

Evidence on social influence is presented by prior studies. For example, self-reported unethical behavior among marketing managers and ad agency account executives was found to be associated with perceptions of their peer’s behavior (Zey-Ferrell and Ferrell 1982; Zey-Ferrell et al. 1979).

We thank one anonymous referee for raising this important point.

Standardized survey instruments and uniform sampling methodology were used to ensure sampling efficiency.

Mature companies have different ethical problems from entrepreneurial start-ups. We focus on the young firms only to prevent firm maturity from confounding our tests; as discussed later, we also controlled for firm age in our tests.

The distribution of firms that are covered by WBES is uneven across countries. For some small countries, very few firms were surveyed and therefore restricting our sample to young firms that reported their control status results in a reduction of the number of countries in our data set.

There may exist a downward bias if firms tend to conceal their bribing behavior. The 2000 World Business Environment Survey (WBES) utilized a fairly uniform survey methodology in 80 countries and one territory to generate indicators that allow comparisons across countries and over time. The data collection process was solely administered and carried out by the World Bank staff and the identity of the firms surveyed and their responses to the questionnaire are not released to any government of the countries covered by the survey or other related parties. This ought to significantly reduce the tendency for the surveyed firms to underreport their bribing behavior. Furthermore, there is no evidence that such underreporting (if exists) differs systematically between family- and non-family-controlled firms given that there is no special economic or other gains for both family-controlled firms and non-family-controlled firms to underreport bribing information in the survey process. Therefore, we believe that the downward bias may not be a big concern and it should not affect our estimation results systematically in this paper. We thank one anonymous referee for raising this point.

We eliminate observations with a value of 7 for this variable because it means that the respondents do not know the percentage of contract value paid as bribes.

Due to data limitation, we are not able to distinguish successful from unsuccessful firms, or to take intention for transgenerational succession into consideration.

The mean size of family firm and non-family firm is 6.378 and 6.629, respectively, and they are not statistically different. Also, the average age of family firm and non-family firm is 4.515 and 4.417, respectively, and they are not statistically different, either.

We removed those observations with answering 7 which indicates that the respondent did not know the answer.

According to information reported in Table 3, there are missing values in most of the variables included in the analysis. For example, only 1,813 observations have values of the variable OWNERSHIP. For another example, only 1,915 observations have values of the variable AWARE. Since various variables included in the analysis have missing values in different observations, the number of observations presented in Table 4 is 1,209 only.

We also used variance analysis to replace regressions for the robustness tests. No qualitative change was found, either. Since variance analysis is not statistically different from regressions (Maijs 2010), we do not tabulate their results in order to save space.

The results are available from the authors upon request.

We thank an anonymous referee for raising this point.

We thank an anonymous referee for raising this point.

We thank an anonymous referee for raising this point.