Bank Loan Supply in the Financial Crisis: Evidence from the Role of Political Connection

Abstract

We investigate the changes in bank loan supply during the 2007–2008 financial crisis, with particular focus on the influence of political connections. We demonstrate that although political connections can help firms obtain lower loan rates during the precrisis period, such benefits disappear in the postcrisis period. Moreover, the loan acceptance ratio for politically connected firms is enhanced in the postcrisis period, especially for the politically connected firms with high risks. Evidence reveals that the focus of the benefits for politically connected firms is more likely to shift from the loan rate to the loan acceptance ratio during the postcrisis period.

KEY WORDS:

• bank loan supply
• loan acceptance ratio
• loan rate
• political connections

Notes

^1. See, for example, Cooper, Gulen, and Ovtchinnikov (2010), Ferguson and Voth (2008); and Goldman, Rocholl, and So (2009).

^2. In the bank loan market, we find the total loan sizes (in million New Taiwan [NT] dollars) in the market are 4,040, 5,521, 4,215, 4,427, and 4,836 in 2006, 2007, 2008, 2009, and 2010, respectively. Hence, the bank loan market significantly decreased after the 2007–2008 financial crisis occurred. In addition, we observe a significantly different trend of non–politically and politically connected firms in the bank loan market. We find that the loan growth ratios of non–politically connected firms are 26.79 percent, −3.60 percent, 2.20 percent, 19.51 percent compared with their loan sizes in 2006. The loan growth ratios of politically connected firms are 72.46 percent, 22.81 percent, 36.40 percent, 20.38 percent compared with their loan sizes in 2006. The evidence suggests that the politically connected firms can continue to easily obtain bank financing, but non–politically connected firms cannot, supporting our hypothesis that the success rate of bank loans (loan acceptance ratio) for politically connected firms is enhanced in the postcrisis period.

^3. Following the suggestion of Ivashina and Scharfstein (2010), we select the year 2007 as the beginning of the financial crisis; the precrisis and postcrisis periods are defined as 2004–06 and 2007–10, respectively. Shen, Chu, and Wang (2012) use Taiwanese bank-level loan data to study the changes during the recessions in the granting of loans to enterprises by banks; Abor and Biekpe (2007) and Talavera, Xiong, and Xiong (2012) investigate the issue of bank financing in Ghana’s enterprises and Chinese enterprises, respectively.

^4. See Chen, Shen, and Lin (2014), Claessens, Feijen, and Laeven (2008), and Khwaja and Mian (2005).

^5. Examples include Acharya and Naqvi (2012), Ivashina and Scharfstein (2010), and Puri, Rocholl, and Steffen (2011).

^6. In this article, we consider a firm to be politically connected only when the incumbent top managers are politically connected.

^7. Charumilind, Kali, and Wiwattanakantang (2006), Faccio (2006), Faccio and Parsley (2009), Fisman (2001), and Johnson and Mitton (2003) also use this definition to measure political connection.

^8. We exclude two firms that have been connected to both KMT and DPP parties in the same period. We also drop three firms switching their political connections from one party to another within our sample periods. However, our results remain the same when our sample includes these firms.

^9. For brevity, the detailed information of this loan transaction data can be found in Chen, Shen, and Lin (2014), Shen and Wang (2005), Yen et al. (2014), and Yen et al. (2015).

^10. All firm characteristic variables are estimated one year prior to the initiation of a loan contract. We also use White’s (1980) heteroskedasticity consistent standard errors and Petersen’s (2009) approach to correct for clustering at the firm level.

^11. The recent financial crisis in Taiwan maybe started in 2008, so we also use the end of 2007 to classify pre- and postcrisis periods for the robustness check. The new results are similar and support our hypothesis.

^12. We also employ Heckman’s two-stage regression to control the endogeneity question of political connections. The newly estimated results still significantly support our hypotheses.

^13. We also use Heckman’s two-stage regression to control the endogeneity question of political connections. The newly estimated results also significantly support our hypotheses.