Founding Family Firms and Bank Loan Contracts

Notably, although the characteristics of the Taiwanese market are conducive to the study of founding family firms, they also limit the generalizability and robustness of any resulting findings. For example, our findings may not be able to generalize to countries with better creditor protection, such as the U.S.

These perspectives include the borrower reputation (Diamond 1991; Sufi 2007), borrower risk (Strahan 1999), information rents (Mattes et al. 2013), political connections (Chen et al. 2014), differences in legal or institution (Qian and Strahan 2007; Bae and Goyal 2009; Haselmann et al. 2010), and role of borrower accounting quality (Bharath et al. 2008).

Anderson et al. (2003) suggest that a dummy variable approach could be more robust because of the potential concern that some founding families are able to utilize control with minimal fractional ownership. Hence, our main analyses are based on this approach, which does not require a minimum threshold for family ownership or control above those imposed by reporting requirements.

Jensen (1993) also suggests that small boards are more effective because of the higher coordination costs and free rider problems associated with large boards.

Recent work by Boone et al. (2007), Coles et al. (2008), and Linck et al. (2008) argue that the optimal board size should depend on firm-specific characteristics, such as firm size, age, and complexity of business.

Taiwanese law requires directors of listed firms to disclose their stocks collateralized in banks.

Similarly, Jensen (1993) suggests that the CEO can control information available to other board members in the duality structure. Shivdasani and Yermack (1999) find that duality makes managers more entrenched and more likely to manipulate a firm’s earnings.

Bhojraj and Sengupta (2003) and Anderson et al. (2004) also find that board independence is negatively related to the cost of public debt.

We adopt TEJ rating score because it provides the most complete rating data for all listed firms during our sample periods. We also try to use rating data from international rating agencies, such as Standard and Poor’s (S&P). However, the samples are considerably reduced.

To lessen the effect of these outliers on our results, the loan spreads as well as all governance and accounting variables are winsorized at the 1th percentile on both sides.

Following the studies of Larcker et al. (2007) and Fields et al. (2010), we use principal component analysis (PCA) to develop a board quality index.

We confine our sample to those issuers who are also listed in the Taiwan Stock Exchange Corporation, and we exclude financial institutions.

We obtain a founding family’s cash-flow rights and voting rights from TEJ Corporate Governance Database, if the founding family is also the firm’s controlling shareholder. TEJ calculates the cash-flow rights by the summing the products of the ownership stakes along control chains as in La Porta et al. (1999). For example, the Lin family in 2010 owns 32.76 % of the shares in Chunghwa Picture Tubes, LTD and 16.7 % of the shares in Tatung, which in turn owns 34.1 % of the shares in Chunghwa Picture Tubes. It follows the Lin family owns 32.76 % (direct ownership) + 16.7 %*34.1 % (indirect ownership through Tatung Co.) = 38.5 % of the cash-flow rights in the Chunghwa Picture Tubes. TEJ calculates the voting rights by summing up the ownership stakes in the last control link over each control chain (La Porta et al. 1999). Accordingly, the Lin family owns 32.8 % (the direct control) + 34.1 % (the indirect control through Tatung Co.) = 66.9 % of the voting rights in the Chunghwa Picture Tubes. Because TEJ only provides the cash-flow rights for controlling shareholders, we hand-collect necessary information to calculate the cash-flow rights from the company’s proxy statement for founding families who are not the controlling shareholders.

We obtain similar results when adding D_Foundingx Deviation and D_Foundingx D_CEO to the model at the same time.

The credit risk index refers to a score of 1 to 10. Higher scores indicate greater default risk.

We obtain similar results when including five macroeconomic variables (GDP, GDP growth rate, budget surplus, inflation rate, and exchange rate) in the regression model as in Dinç (2005).

Although we control for some firm characteristics, these variables are clearly insufficient in ruling out the endogenous problem stemming from the omitted variable problem.