To better understand the nature of family-owned firms, one must distinguish management from control rights (Villalonga and Amit, 2006). Also not all controlling sh-areholders in family firms serve as CEOs. When professional managers serve as CEOs of family firms, the controlling shareholders often assume the role of monitoring the hired managers. Therefore, these firms end up bearing some agency costs that incur between the owner and professional managers. While the owner-manager conflict can be mitigated in firms in which family members serve as CEOs (henceforth, ‘family CEO'), the agency problem between controlling shareholders and minority shareholders can be also intensified (Shleifer and Vishny, 1997). Taken all factors into consideration, family CEOs can affect firm performance in both positive and negative ways. By definition, controlling shareholders of any business groups face the dilemma whether to get involved in the firm management. And yet, not much research has been conducted on this issue as to address the question which is better for the firm. This must have been so perhaps because of data limitations. In fact, Lin and Hu (2007) is the only study that directly addresses the selection issue. However, their sample is limited to listed Taiwanese firms, not sufficient to account for general characteristics of business groups including unlisted firms. To the best of my knowledge, this paper is the first study of its kind to analyze the determinants of family CEO using information of all affiliatesboth listed and unlisted- belonging to the same business groups. When a family keeps control of a firm, its controlling shareholders can get involved in the management as CEOs or the like to maximize the sum of the value of the retained block and the private benefits obtained (Burkart, Panunzi, and Shleifer, 2003). If maximizing the former is intended, family CEO can positively affect firm performance as long as he/she has good management skills. But if maximizing the latter is intended, family CEO can be negatively related to firm performance. Stein (1989) and James (1999) develop a model illustrating the positive effect of long-term investors on firms' investment efficiency especially when they are the members of the family that owns the firm. The study explains that such long-term investors can minimize myopic investment decisions by managers. Anderson and Reeb (2003) empirically find that firms with family CEOs outperform those with professional CEOs for this reason. On the other hand, controlling shareholders of family firms don't tend to take enough advantage of competitive managerial labor market since they are likely to limit management control to their own family circles. Such practice of ‘nepotism' can harm firm performance (Perez-Gonzalez, 2006). As such, existing theoretical and empirical studies about the effect of family CEO on firm performance are contradictory. To draw its own conclusion, this paper first investigates the precise determinants of family CEO before probing the question of family CEO's effect on firm performance. Finally to reflect the endogenous nature inherent in family CEOs, the self-selection or reverse causality issue is addressed through modeling family CEO as an endogenous choice. Using the Korean business group data from 2001 through 2011, the following results are found. First, firms with higher cash flow rights of controlling shareholders and larger equity portion in affiliates are more likely to have family CEOs. This means that controlling shareholders tend to serve as CEOs to maximize their control rights over the business group as well as the value of their stake, supporting a proposition of Burkart et al. (2003). Second, family CEO negatively affects firm profitability measured by EBITDA. This implies that when controlling shareholders serve as CEOs, agency costs between themselves and minority shareholders increase. Also, only for firms with non-family CEOs, their investments have a significantly positive impact on EBITDA, indicating that professional CEOs commit more to efficient investments than family CEOs do. Third, for firms with family CEOs, higher cash flow rights and larger equity portion in affiliates are associated with lower firm values measured by Tobin's q. This implies that the concentration of ownership and management by controlling shareholders induces agency problems due to the entrenchment effect, making them commit to maximizing control rights over the business group although the value of their shares is damaged. On the other side, for non-family CEO, larger equity portion in affiliates is associated with higher firm value, implying that professional CEOs invest more efficiently in affiliates' equity than family CEOs do. But the effect of cash flow rights on firm values is insignificant for these firms. Just like the results on EBITDA, investments have a significantly positive impact on firm values only for the sample with non-family CEO. Fourth, controlling the endogeneity of family CEO through Heckman (1979)'s model, family CEO still negatively affects value as well as profitability of firm. Fifth, through a robustness check we found that firm profitability worsens only when family CEO runs the firm; if professional managers are CEOs and family members are not involved in the management position, those firms show higher value. This finding suggests that securing an independent professional management system without family members can be the most effective way to maximize corporate value in the business group. Overall, the findings indicate that when controlling shareholders with potentially powerful influence over the business group directly manages the firm, they are highly like to be driven by the self-serving incentive to maximize their controlling rights over the business group, rather than to commit themselves to increasing firm value through efficient investments. By drawing this conclusion, this study sheds more light on this unexplored yet salient question of which organizational structure, with or without family CEOs, is more efficient.