Political Uncertainty and Bank Loan Contracts: Does Government Quality Matter?

Transcription

1 Political Uncertainty and Bank Loan Contracts: Does Government Quality Matter? Iftekhar Hasan Fordham University, Bank of Finland and University of Sydney 45 Columbus Avenue, 5th Floor New York, NY Ying-Chen Huang Department of Finance National United University Maoli, Taiwan Yin-Siang Huang College of Management Yuan Ze University Taoyuan, Taiwan Chih-Yung Lin * College of Management Yuan Ze University Taoyuan, Taiwan d @ntu.edu.tw Abstract This research investigates whether political uncertainty influences bank lending decisions, using bank loan contracts from 47 countries between 1982 and We find that, in this period, banks charged firms higher loan spreads to compensate for their domestic political risks during election years. This political effect is stronger when borrowers are located in countries with poor government quality. We also observe that banks ask higher spreads on international loan contracts to compensate for political risks, especially with a high differential in government quality between lender and borrower. Better government quality can thus reduce the impact of political risk on the cost of bank loans. JEL: G32, G33, G34, L14 Key words: Political uncertainty, bank loan contracts, government quality, country governance, financing cost. i

2 Political Uncertainty and Bank Loan Contracts: Does Government Quality Matter? Abstract This research investigates whether political uncertainty influences bank lending decisions, using bank loan contracts from 47 countries between 1982 and We find that, in this period, banks charged firms higher loan spreads to compensate for their domestic political risks during election years. This political effect is stronger when borrowers are located in countries with poor government quality. We also observe that banks ask higher spreads on international loan contracts to compensate for political risks, especially with a high differential in government quality between lender and borrower. Better government quality can thus reduce the impact of political risk on the cost of bank loans. JEL: G32, G33, G34, L14 Key words: Political uncertainty, bank loan contracts, government quality, country governance, financing cost. ii

3 1. Introduction A burgeoning literature examines the real effect of political uncertainty on a variety of corporate decisions (Khwaja and Mian, 2005, Julio and Yook, 2012, Pastor and Veronesi, 2012, and Baker, Bloom and Davis, 2016). Despite abundant evidence of the real effect of political uncertainty on firm operations and performance, there has been much less discussion of the role of political uncertainty in bank lending decisions. This paper extends the boundaries of the literature to encompass bank lending decisions under different levels of government quality and political uncertainty. 1 In recent years, the determinants of bank loan contracts has been a popular topic for research (Qian and Strahan, 2007; Bae and Goyal, 2009; Graham, Li, and Qiu, 2008; Bharath, Sunder, and Sunder, 2008; Haselmann, Pistor, and Vig, 2010; Lin, Ma, Malatesta and Xuan, 2011; Hasan, Hoi, Wu and Zhang, 2014). Specifically, using U.S. experience, Francis, Hasan, and Zhu (2014) suggest that political uncertainty influences the cost of borrowing. 2 For domestic bank loan contracts, firms have faced different degrees of political uncertainty depending on their nation s election results. If an election results in a change of ruling party, economic policies, investor expectations, or market behaviors might also change (Francis, Hasan, and Zhu, 2014), 1 In this paper, we focus on the bank loan market for two reasons. First, private debt financing has become a prevailing source of external funding around the world (Chava, Livdan, and Purnanandam, 2009; Graham, Li, and Qiu, 2008). For example, in 2005, while the amount of the newly issued corporate bonds was about 700 billion dollars, US firms raised new external funding of 1,500 billion dollars in total through the syndicated loan market (Bharath, Sunder, and Sunder, 2008). Second, compared to public bondholders, banks have greater capabilities and incentives to collect information and monitor borrowers (e.g., Fama, 1985; Roberts and Sufi, 2009; Roberts, 2015). 2 Julio and Yook (2012) define the period of political uncertainty surrounding presidential elections as 60 days prior to and 274 days after the election date. They find that corporate investment is reduced during the time of national elections, especially when national leadership changes. Thus, we follow their study and employ the same definition for political uncertainty. In this paper, we mainly focus on the financing costs of borrowers, which are measured by loan spreads. Therefore, we regard political uncertainty as a period of high political risk for the country of the borrower. 1

4 and firms will face higher political uncertainty. Thus, we propose that banks would charge additional loan spreads during election years to compensate for the domestic political risks they face. When international businesses and global enterprises emerge in large numbers, the demand for transnational loan contracts subsequently increases (Chen, Huang, Lobo, and Wang 2016; Chui, Kwok, and Zhou 2016). Giannetti and Yafeh (2012) emphasize that differences between transnational cultures should be considered in borrower-lender relationships. 3 They suggest that cultural differences lead to more burdensome negotiations, increase contracting costs, and make information-gathering less efficient. Recently, La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999) constructed a government quality index for each country, which could be a key factor affecting political risk. To date, however, no studies have discussed the effect of government quality on bank loan spreads under conditions of political uncertainty. In this paper, we investigate the role of differences in government quality on bank loan spreads when the borrower s country is undergoing a period of political uncertainty. We propose that the greater the difference in government quality between the countries of the lending bank and the borrowing firm, the more harmful it is to the borrower. When the borrower functions under political uncertainty and the government quality differential is large, the firm must pay higher bank loan spreads. In this study, we focus on the effects of the interaction between political uncertainty and government quality on bank loan contracts and seek to answer the following research questions: 1) Do banks charge firms higher bank loan spreads under political uncertainty?; 2) Does government quality influence the impact of political uncertainty on domestic loan contracts?; and 3) Does the interactive effect of 3 Greater cultural differences may lead more costs and higher risk (Hofstede, 1980). 2

5 political uncertainty and government quality extend to international bank loans? We inquire as to whether political uncertainty across countries is perceived by lenders in international markets as being similar, and examine whether political uncertainty in the U.S. is similar to that in an emerging market, which is relatively new to democratic elections. To measure political uncertainty, we follow Julio and Yook (2012) and hand-collect bank loan contract data from 47 countries in major election years between 1982 and We assemble the data sample from the DealScan database, which comprises 73,920 bank loan contracts. 4 Accounting variables are obtained from Compustat and the Compustat Global databases. Macroeconomic variables are gathered from the World Bank. Our empirical results indicate that, on average, firms charge higher loan spreads during election years relative to non-election years (the political-uncertainty effect). Our empirical results reveal that banks ask an increase of in bank loan spreads during election years. In addition, we find that the political-uncertainty effect is mainly driven by domestic bank loans, confirming that domestic political risk affects domestic, but not necessarily international, loans. How do we measure government quality? From the literature, we could use government quality and country governance indices to proxy government quality. A government quality index, proposed by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999), can be used to measure the level of government quality for each country. We expect that if a borrower s country has very poor property rights, business freedom and tax compliance track records, firms should have to undertake 4 In our sample, loan characteristics include Spread, Maturity, Loan size, Collateral, and GenCov and FinCov (numbers of general/financial covenants). 3

6 higher bank loan costs during election years, suggesting that firms with better government quality would obtain lower financing cost during such years. Regarding international loan contracts, banks strengthen their lending standards after considering differences in government quality. Our empirical results show that firms are more likely to be charged a higher cost for loans during election years when a high differential in government quality exists between borrower and lender. These results imply that banks evaluate current risk through government quality factors while formulating their transnational lending decisions. That is, banks factor into their loan contracts the possibility of losses resulting from different levels of government quality. Finally, we perform various additional experiments and robustness tests. We first use a country governance index as an alternative measure of government quality. 5 We find similar results that banks view country governance as a critical consideration when designing loan contracts. Second, we also control for international cultural differences in our analysis, ensuring that our results do not derive from their effect. Third, we conduct a regression analysis that substitutes total cost of borrowing for our spread, which has been proposed by Berg, Saunders and Steffen (2016). Fourth, to deal with the endogeneity problem of reverse causality and omitted variables in our political uncertainty measurement, we adopt a fixed-effect methodology of executive legislation, election type, and election timing to re-examine our main results (Julio and Yook, 2012). For all tests, the main results remain qualitatively unchanged. These tests show that our results are robust to all the above-discussed concerns. 5 Kaufman, Kraay, and Mastruzzi (2009) employ six dimensions of governance: Voice and Accountability (VA), Political Stability and Absence of Violence (PV), Government Effectiveness (GE), Regulatory Quality (RQ), Rule of Law (RL) and Control of Corruption (CC), and aggregate these six dimensions into an index of Country Governance (CG). In this paper, we employ GE, CC and CG to capture country governance. 4

7 Our work contributes to the literature in two ways. First, our research complements several recent studies on the determinants of bank loan contracting. These determinants, such as borrower risk (Strahan, 1999), borrower reputation (Diamond, 1991; Sufi, 2007), legal system or institution (Qian and Strahan, 2007; Bae and Goyal, 2009), misreporting announcements (Graham, Li, and Qiu, 2008), accounting quality (Bharath, Sunder, and Sunder, 2008), creditor rights (Liberti and Mian, 2010), financial law improvement (Haselmann, Pistor, and Vig, 2010), ownership structure (Lin, Ma, Malatesta and Xuan, 2011), and tax avoidance and political connection (Hasan, Hoi, Wu and Zhang, 2014) have been investigated by in-depth analysis. We find that banks view political uncertainty as an extra risk-taking cost and thus charge higher interest rates to firms located in countries with poor government quality. Second, our work relates political uncertainty to lending decision-making. Previous studies have typically focused on the borrower firm s condition and loan characteristics. Few studies have focused on the influence of the interaction between political uncertainty and government quality in bank loan contracts. Our paper complements these studies by exploring how political uncertainty influences firms loan financing costs in countries with different levels of government quality. Our study provides a new perspective and finds that banks consider political uncertainty and government quality as an extra cost. This finding also provides a good reason for countries governing bodies to improve their levels of government quality. Therefore, our results provide a reference for scholars, policymakers, and market investors to assess the role and effect of political uncertainty vis-a-vis government quality and, at the same time, supplies a reference for managers to understand the conditions underlying their financing costs. 5

8 The remainder of the paper is organized as follows: We develop our main hypotheses in Section 2. Section 3 describes our data and presents descriptive statistics. Empirical results and additional supporting evidence are presented in Sections 4 and 5. Section 6 concludes the paper. 2. Hypothesis development 2.1. Political uncertainty and bank lending decisions A growing number of research studies shed light on the effect of political uncertainty. Baker, Bloom and Davis (2016) use newspaper coverage frequency and construct an index of economic policy uncertainty (EPU), which proxies for movement in policy uncertainty. Several studies (Dinc, 2005; Brown and Dinc, 2005; Khwaja and Mian, 2005; Leuz and Oberholzer Gee, 2006; Micco, Panizza, and Yanez, 2007; Julio and Yook, 2012) use national election data to specify political uncertainty and find that political uncertainty indeed has significant influence during election years. 6 According to Julio and Yook (2012), the uncertainties associated with possible changes in national leadership or government policy have implications for firms. They find that firms reduce investment expenditures during election years, since political uncertainty can affect firm s future cash flows or discount rates. Nini, Smith, and Sufi (2012) find that firms default risks are affected by their future cash flow change. Qi, Roth and Wald (2010) use corporate bonds issued by firms incorporated in 39 countries and state that the impact of country-level political rights on bond 6 Firms that are exposed to higher political uncertainty would be charged a higher cost of debt, because of a higher default rate and loan size (Khwaja and Mian, 2005). Politicians may change the uncertainty policy and cause anxiety to investors during election years. At the micro-economic level, political uncertainty can affect a firm s financing cost, lending amount, and bank loan spread (Francis, Hasan, and Zhu, 2014). Political uncertainty also impacts macroeconomics, firms and lending behaviors in the banking system, and can lead to market volatility increase (Baker, Bloom and Davis, 2016). 6

9 yields are negatively significant. Gao and Qj (2013) also emphasize that yields of municipal bonds increase before U.S. gubernatorial elections and are subsequently reversed. While the above-cited papers all mention the impact of political uncertainty on bond yields, we are concerned with the influence of political uncertainty on the bank loan market. Francis, Hasan and Zhu (2014) use the EPU index and find that firms face a higher cost of debt under conditions of political uncertainty. In this paper, we investigate the causal relationship between political uncertainty and bank loan contracts. Firms have faced different degrees of political uncertainty depending on their nation s election results. If election results change the ruling party, economic policies, investor expectations, and/or market expectations may also change, leading firms to face higher levels of political uncertainty. Thus, we propose that banks would charge higher loan spreads during election years to compensate for their increased risks. Moreover, the banks should only account for the domestic political risk of borrowers in their lending decisions, because only that will increase their domestic borrowers default risks. Thus, we propose our first hypothesis: H1: Banks charge firms higher loan spread for domestic loans during election years (the political-uncertainty effect) Government quality Banks charge higher spreads when borrowers and lenders face different cultural values under individual country system (Hilary and Hui, 2009; Giannetti and Yafeh, 2012). Giannetti and Yafeh (2012) adopt World Values Survey (WVS) data to measure cultural values, which includes corruption, political interest and participation, and political culture and regime. In addition, La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999) construct government quality indexes for each country, which could be 7

10 a vital factor influencing political risk. In this paper, we further examine whether government quality impacts bank lending decisions. We use the following indexes of government quality from La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999): The index of property rights protection (PR), of business freedom (BF), and of tax compliance (TC). 7 The index of property rights protection (PR) measures of the degree to which a country s laws protect private property rights and the extent to which those laws are respected. The index of business freedom (BF) measures the extent to which the regulatory and infrastructure environments constrain the efficient operation of businesses; the score is a number from 0 to 100 for each country, with 100 indicating the most free business environment. The index of tax compliance (TC) indicates whether a taxpayer complies with the tax rules of his/her country. For domestic loans, lender banks and borrower firms are located in the same country. Better government quality might alleviate the political uncertainty faced by banks. When the government quality of the county is stronger, the relationship between political uncertainty and bank loan spreads is weaker. Thus, we propose our second hypothesis: H2: The political-uncertainty effect in domestic loans is reduced in countries with higher government quality Differences in transnational country-specific factors Difference between countries cultures also affects bank lending decisions. Cultural distance between countries can capture trust between nations, and easily measure the future cooperation probability and communication process (Hofstede, 7 The infrastructure index is also a sign of a well-functioning government proposed by LLSV, and indicates efficiency of communication between headquarters and other operations within a country. Government performance of a given country should be assessed in part by evaluating the quality of public goods provisions. 8

11 2001; Schwartz, 2006; Guiso, Sapienza, and Zingales, 2009; Giannetti and Yafeh, 2012; Chui, Kwok, and Zhou 2016). 8 Giannetti and Yafeh (2012) find that a one-standard-deviation increase in cultural distance, approximately the distance between Canada (Japan) and the United States (South Korea) is associated with a 6.5-basis-point higher loan spread. Similar to their findings, we expect that greater government quality differences will also lead to higher loan costs for borrowers. 9 Therefore, for international bank loans, we focus on differences in government quality between lenders and borrowers under conditions of political uncertainty. Banks facing a greater differential of government quality between their and their borrowers countries will undertake higher political risk, and will consequently charge higher loan spreads. We thus examine the impact of government quality differences on international bank loan contracts and propose our third hypothesis: H3: The political-uncertainty effect in international bank loans will be greater when the difference in government quality between lender and borrower is greater. 3. Data and descriptive statistics 3.1. Data In this paper, we follow Julio and Yook (2012) and use an election dummy to measure political uncertainty. The election dummy variable equals one if a presidential election is held within 60 days prior to and 274 days after a scheduled election date, and zero otherwise. We collect transaction details of bank loan contracts 8 Adair, Brett and Okumura (2011) proposed that people facilitate communication and the exploration of alternatives when they share similar norms and codes, e.g., a borrower may need a long period of maturity, but may be willing to concede on the loan amount. 9 In this paper, we consider government quality as one type of national culture. 9

12 from the DealScan database, which includes the transaction records of loan amount, loan spread, loan maturity, and collateral. Accounting variables come from Compustat and the Compustat Global database. 10 Appendix A presents definitions and sources of all variables. The final sample encompasses 73,920 bank loan contracts from 47 different countries between 1982 and Following the study of La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999), we employ the property rights index (PR), business freedom index (BF), and tax compliance indicators (TC) to measure the government quality. We also use country governance measures (Kaufman, Kraay, and Mastruzzi, 2009) as an alternative government-quality measure, including control of corruption (CC), government effectiveness (GE), and a country governance index (CG) Summary statistics Table 1 presents summary statistics of all variables. First, the mean of Spread is , or basis points (bps) with a standard deviation of The average of Election is frequencies with a standard deviation of The summary statistics for other control variables are also reported in Table [Insert Table 1 here] Table 2 lists the political factors of countries, which represent the basis of the borrowers of our sample. For government quality, the averages of PR, BF, and TC are , , and 3.181, respectively. For the country governance index, the averages of CC, GE, and CG are 0.863, 0.972, and 0.780, respectively. 10 We exclude observations with missing information on accounting data or loan spread. 11 In Appendix B, we present the Pearson correlation coefficient matrix of the variables. The correlation coefficients between Election and the other variables are less than 0.029, indicating that multicollinearity is not a concern. 10

13 [Insert Table 2 here] 3.3. Differences in election and non-election year groups Next, we divided our sample into Election and Non-election subsamples. Table 3 compares the mean differences of all variables between the two groups, in election and non-election years. The average Spread in the election year group is 4.956, and that in the non-election year group is The differences between them are all significantly positive, which means that banks charge higher spreads under conditions of higher political uncertainty. The average Collateral is in the election year group, in the non-election year group, and the differences between them are all significantly positive. Similarly, Maturity and Loan size are reduced in the election year group. All the results are consistent with our expectation that banks charge firms higher loan costs during election years. [Insert Table 3 here] 4. Empirical results 4.1. Bank loan spread and political uncertainty We use linear regressions to investigate the effect of political uncertainty on bank loan spreads. The analysis focuses on the coefficients of the political uncertainty dummy variables. A positive coefficient supports our hypothesis that banks charge firms higher loan spreads within president election campaign timeframes to account for greater political uncertainty. Following Graham, Li, and Qiu (2008), Hasan, Hoi, Wu and Zhang (2014) and Francis, Hasan, and Zhu (2014), we use an ordinary least squares (OLS) regression to investigate the effect of political uncertainty on loan spreads: 11

14 Spread i,t = α 1 + α 2 Election k,t + β Firm i,t 1 + θ Z i,t + ν j + ω k + μ t + ε i,t (1) where Spread i,t is the natural logarithm of bank loan spread for firm i in year t. The main explanatory variable of Election k,t equals one if a presidential election is held within 60 days prior to and 274 days after the scheduled election date, and zero otherwise. Firm i,t 1 is the vector of the control variables for borrower firm i in year t 1, Z i,t is the vector of the control variables for loan and macroeconomic factor i in year t, ν j, ω k and μ t capture the industry, country and year fixed effects, respectively, and ε i,t is the random error. Firm i,t 1 represents firm characteristics including Asset, Market-to-book ratio, Tangibility, Profitability, Leverage, Z-score and Cash flow volatility (CFV). Z-score includes six loan characteristics: Maturity, Loan size, Collateral, Performance, GenCov and FinCov, three macroeconomic factors: GDP and GDP_G, and Inflation, and the purpose and type of loan. In all the equations, we report t-values based on standard errors adjusted for heteroskedasticity and sample clustering at the country level (White, 1980, and Petersen, 2009). To save space, we do not report the coefficients of loan purpose, loan type, and industry, country and year dummies. Table 4 presents the regression results for the effect of political uncertainty on loan spreads. There are four specifications. The first controls for the loan purpose, loan type, industry, country and year fixed effects. The second adds controls for firm characteristics. The third adds controls for loan characteristics, and the fourth adds controls for macroeconomic factors. Across all specifications, the proxy for political uncertainty, Election, is significantly positively related to loan spread, after controlling for borrower characteristics, loan conditions, loan types, loan purposes, year, country, and industry 12

15 fixed effects. For example, in Model (1), the coefficient of Election is , and is statistically significant at the 1-percent level, which shows that, on average, firms facing political uncertainty are charged higher loan spreads than firms that do not face political uncertainty. The empirical results are consistent with those of Francis, Hasan, and Zhu (2014) that political uncertainty increase firms bank loan costs. [Insert Table 4 here] Regarding firm characteristics, loan spread is significantly negatively associated with Assets, Market-to-book ratio, Tangibility, Profitability and Z-score. The results are consistent with the notion that firms reduce their financing costs when they are large and exhibit higher profitability (Graham, Li, and Qiu, 2008), as well as higher market-to-book ratio, tangibility and Z-score. Higher loan spreads are charged to firms with higher leverage, consistent with the notion that firms facing higher risks are charged higher loan spreads (Chava and Roberts, 2008; Hasan, Hoi, Wu, and Zhang, 2014). The results also show that loans with longer maturities, larger amounts, and performance pricing have lower loan spreads. GenCov and Collateral are positively associated with loan spreads, consistent with the notion that firms with higher risk will be asked to guarantee more loan covenants or collaterals (Chava and Roberts, 2008; Graham, Li, and Qiu, 2008). In terms of macroeconomic factors, the results show that firms pay higher financing costs when GDP growth is lower and Inflation is higher. In summary, after controlling for other potential factors that may affect loan spreads, a presidential election is generally associated with a higher cost of bank loan financing Political uncertainty effect: Domestic and international bank loans 13

16 The question remains, however, as to whether banks treat domesic and international loans differently during an election year. Because political uncertainty could bring domesic borrowers greater political risk, we expect that its effect on loan spreads could have different influences on domestic and international loan contracts. Empirically, we split our sample into two parts: domestic and international loan contracts. Our expectation is that political uncertainty would have a stronger effect on domestic than on international loans, while neglecting transnational specific variables (see Table 5). The result in Model (1) of Table 5 indicates that political uncertainty tends to result in higher bank loan spreads for domestic bank loan contracts. We find that the coefficient of Election in domestic loan loans is positive and significant at the 1-percent level. Borrowers are charged average additional spreads of during election year compared to a non-election year, even when we control for all other potential affecting factors. However, the evidence concerning international loans shows that no political-uncertainty effect exists in Model (2) of Table 5. These results are consisting with our Hypothesis 1 that banks charge firms for higher loan spreads on domestic loans during election years (the political-uncertainty effect). [Insert Table 5 here] 4.3. Political uncertainty effect and government quality: Domestic bank loans For domestic bank loan contracts, lenders charge firms a higher cost for loans during periods of political uncertainty. In this section, we further examine whether government quality influences domestic bank lending costs. Following Graham, Li, and Qiu (2008), La Porta, Lopez-de-Silanes, Shleifer, and Vishny, LLSV (1999), we use OLS regression and consider government quality factors: Spread i,t = α 1 + α 2 Election k,t + α 3 GQ k,t + α 4 Election k,t GQ k,t 14

17 +β Firm i,t 1 + θ Z i,t + ν j + ω k + μ t + ε i,t (2) where GQ k,t is government quality of country k at time t, including the property rights index (PR), business freedom index (BF), and tax compliance (TC). Table 6 presents the regression results of political uncertainty on bank loan spreads for firms when various levels of government quality are considered. In Models (1), (2) and (3) of Table 6, we use PR, BF, and TC to proxy governance quality, respectively. In all models, the coefficients of Election GQ are negative and statistically significant, suggesting that banks charge lower spreads for firms located in countries with better government quality under conditions of political uncertainty. Thus, greater property rights, broader business freedom, and higher tax compliance can reduce the impacts of political uncertainty in bank loan contracts. These results are consistent with our Hypothesis 2 that the political-uncertainty effect in domestic loans will be reduced in countries with better government quality. [Insert Table 6 here] 4.4. Political uncertainty and government quality: International bank loans Our results so far examine the effect of political uncertainty and government quality on domestic bank loans. In this section, we focus on international bank loans and we use the differences in governance quality between lender and borrower to test this issue. Following Graham, Li, and Qiu (2008) and Giannetti and Yafeh (2012), we use OLS regression and consider the impact of government quality factors on international bank loans: Spread i,t = α 1 + α 2 Election k,t + α 3 GQ k,t + α 4 Election k,t GQ k,t 15

18 +β Firm i,t 1 + θ Z i,t + ν j + ω k + μ t + ε i,t (3) where GQ k,t denotes an absolute value of the difference of government quality between borrower and lender, including the property rights index (PR), business freedom index (BF), and tax compliance (TC). Table 7 presents regression results for the interaction impact of political uncertainty and differences in government quality on loan spreads. In all models, the coefficients of Election PR, Election BFand Election TC are positive and statistically significant, confirming our Hypothesis 3 that the political-uncertainty effect in international bank loans will be stronger when the difference in government quality between lender and borrower is greater. For example, in Model (1), the coefficient of the interaction term is , which indicates that firms suffer the higher loan spreads during election years when a greater difference in property rights exists between lender and borrower. Thus, banks facing a greater disparity in government quality with their borrowers usually undertake higher political risk and, thus, will charge higher loan spreads. [Insert Table 7 here] In sum, the results presented in Tables 6 and 7 indicate that poor government quality increases the strength of the political-uncertainty effect. Conversely, better government quality can reduce the influence of political risk in firms financing costs. 5. Additional supporting evidence 5.1. Country governance We use country governance indexes measured by Kaufman, Kraay, and Mastruzzi (2009) as government quality to test this issue. In the literature, poor 16

19 country governance negatively affects bank performance (Boehmer, Nash, and Netter, 2005; Shen and Lin, 2012). Therefore, we further examine whether the level of country governance influences the political-uncertainty effect on bank loan contracts. 12 We use government effectiveness (GE), control of corruption (CC) and the country governance index (CG) in our regression. We then use OLS regression: Spread i,t = α 1 + α 2 Election k,t + α 3 CG k,t + α 4 Election k,t CG k,t +β Firm i,t 1 + θ Z i,t + ν j + ω k + μ t + ε i,t (4) where CG k,t represents the country governance factor of country k at time t, including GE, CC and CG. All the other variables are defined as before. Table 8 presents the additional influences of control of corruption, government effectiveness, and the country governance index on the political-uncertainty effect. In Models (1), (2), and (3) of Table 8, we find that the coefficients of Election CC, Election GE, and Election CG are all significantly negative. Thus, the evidence shows that additional spreads would be eliminated if the borrower s country possesses features such as uncorrupted elites, good governance effectiveness, and a better country governance index. [Insert Table 8 here] contracts. Furthermore, we verify the country-governance effect on international loan 12 Spread i,t = α 1 + α 2 Election k,t + α 3 CG k,t + α 4 Election k,t CG k,t The effect of poor control of corruption (CC) with political uncertainty leads to bank underperformance that is easily realized through the public power for private gain. Government effectiveness (GE) also affects the lender-borrower relationship in terms of bargaining power. GE emphasizes the credibility of the government s commitment, measures the quality of civil service and public services, and evaluates the quality of policy formulation and implementation. Country governance (CG) reflects the aggregate governance index for six broad dimensions in Kaufman, Kraay, and Mastruzzi (2009). 17

20 +β Firm i,t 1 + θ Z i,t + ν j + ω k + μ t + ε i,t (5) where CG k,t denotes an absolute value of the difference of country governance between borrower and lender, including GE, CC and CG. Models (1), (2) and (3) of Table 9 report the estimated results of the CC, GE, and CG interaction with Election, respectively. The coefficients of interaction in all the models are significantly positive at the 1-percent level. Overall, these results also confirm that the political-uncertainty effect on international bank loans will strengthen when the difference in level of country governance between lender and borrower is greater. [Insert Table 9 here] 5.2. International cultural differences Due to the global tendency toward multinational enterprise development, international business firms in host countries have greater need for external financing. Giannetti and Yafeh (2012) propose that greater disparity in national culture between borrower and lender will lead to higher loan spreads. In this section, we additionally control for international cultural differences in our analysis, making sure our results do not derive from the effect of such differences. Table 10 present the regression results after controlling for international cultural differences. In the Models (1), (2), and (3), we additionally control for culture distance, which is calculated by Inglehart and Welzel (1981) and includes two dimensions. The first is traditional versus secular/rational values, which reflects the contrast between societies in which traditional values are most important in people s lives and those in which rational values predominate. The second dimension is the 18

21 central component of survival versus self-expression, which shows the polarization between materialist and postmaterialist values. Postmaterialists possess a trustful and tolerant attitude toward life, and materialists do not. In Models (4), (5) and (6), we control for New culture distance, which is calculated by Welzel (2013) and includes the overall secular and emancipative value dimensions. In the all models, the coefficients of Election PR, Election BF and Election TC are still significantly and positively related to bank loan spread, even when we take into consideration cultural-distance effects, thereby supporting our Hypothesis 3 that the political-uncertainty effect in international bank loans will strengthen when the difference in government quality between lender and borrower is greater. In Models (4), (5) and (6), we control for the dimensions of new cultural distance from Welzel (2013). After considering the Euclidean distance between the cultures of borrowers and lead banks countries, our results still support Hypothesis 3. [Insert Table 10 here] 5.3. Robustness check Total cost of borrowing The total cost of corporate borrowing has been considered an important measure for loan costs. Berg, Saunders and Steffen (2016) propose that more than 80 percent of U.S. syndicated loans contain at least one fee type, a menu of spread that contracts specify. Thus, following their study, we employ total cost of borrowing (TCB) to replace bank loan spread in Equation (1). Model (1) in Table 11 presents the regression results on the effect of political 19

22 uncertainty on total cost of borrowing (TCB). The results show that the impact of political uncertainty on the total cost of borrowing is significant and positive, which is consistent with the results in Table 4. Thus, borrower firms under conditions of political uncertainty experience high TCB. These results support our Hypothesis 1. [Insert Table 11 here] Executive legislation, election type, and election timing LLSV (1998) suggests that legal origins generate a striking influence on economic freedom and exert a highly statistically significant effect. Glaeser and Shleifer (2002) also explain that common law influences economic freedom through judicial independence. Specifically, for firms located in countries with strong laws and reliable legal enforcement, lenders create more concentrated syndicated bank loans and low-cost contracting. Julio and Yook (2012) state that the type and timing of elections are not decided by any individual firm, but are fixed in time by constitutional rules. Following their study, we deal with the endogeneity problem of reverse causality and omitted variables in political uncertainty measurement by adopting a fixed-effect methodology of executive legislation, election type, and election timing in our regressions, based on Equation (1). Model (2) in Table 11 presents the regression results on the effect of political uncertainty on bank loan spread by considering executive legislation, election type, and election timing. The coefficients of Election are significantly positive. The results once again support our Hypothesis 1. [Insert Table 11 here] 6. Conclusion 20

23 This study investigates the effect of political uncertainty on bank loan spread using a sample of loan contracts from 47 countries around the world. Prior studies have shown that banks require a higher cost of debt under conditions of political uncertainty in the U.S. (Francis, Hasan, and Zhu, 2014). Our paper extends their study to include global data and further investigates whether government quality plays an important role in bank lending decisions. In addition, we also investigate whether banks charge higher spreads to compensate for political uncertainty risks when the difference in government quality between lender and borrower is higher. First, our results show that banks charge firms higher interest rates in periods of political uncertainty, after controlling for firm characteristics, loan conditions, and macroeconomic factors. These results are consistent with the extensive literature showing that banks consider political uncertainty as one type of political risk. Second, we find that, while political uncertainty increases loan spreads for domestic bank loans, it does not necessarily do so for international bank loans. Third, our results show that the political-uncertainty effect on domestic loans will be greater in countries with lower or weaker government quality. Fourth, we find that lender banks ask higher international loan spreads under conditions of political uncertainty when differences in government quality between borrower and lender countries are larger. Therefore, better government quality can reduce the influence of political risk on firms financing costs. Specifically, government monitoring and efficiency in terms of property rights, business freedom, and tax compliance are instrumental for economic growth. This empirical research supports the idea, stressed by economists, that strong economic institutions in the public sector are necessary for lower lending costs. Finally, this study performs four additional experiments and robustness tests by applying country governance to replace government quality, and by taking into 21

24 consideration international cultural differences, total cost of borrowing, and other election-related variables, respectively (Berg, Saunders and Steffen, 2016; La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1998). We confirm that the effect of political uncertainty on loan spreads still holds and our main results are not driven by other possible factors. In sum, our study provides evidence that banks consider political uncertainty and poor government quality as an extra cost. These findings not only have significance for policymakers, but also provide a solid reason for governmental institutions to improve their governance quality. Therefore, our results provide a reference for scholars, policymakers, and market investors to assess the impacts of the interaction between political uncertainty and government quality. The paper also supplies a reference for managers firm financing policies. 22

25 Appendix A The Definition for Variables Variable Definition Source of data Dependent variables Spread Natural logarithm of loan spread. Loan spread is measured as the all-in spread drawn in the DealScan database. All-in spread drawn is the amount the borrower pays in basis points over LIBOR or DealScan LIBOR equivalent for each dollar drawn down. TCB Fees play the role of price option, drawdown option, and cancellation option in bank loans, and are used to screen borrowers based on the likelihood of exercising these options. The measure of this total cost of borrowing variable includes various Berg, Saunders and Steffen (2016) fees charged by lenders. Political uncertainty variable Election Government quality variables Property rights index (PR) Business Freedom index (BF) Tax Compliance (TC) A dummy variable that equals one if presidential elections are held within 60 days prior to 274 days after the election date, and zero otherwise (Julio and Yook (2012)). The property rights component assesses the extent to which a country s legal framework allows individuals to freely accumulate private property, secured by clear laws that are enforced effectively by the government. Relying on a mix of survey data and independent assessments, it provides a quantifiable measure of the degree to which a country s laws protect private property rights and the extent to those laws are respected. It also assesses the likelihood that private property will be expropriated by the state. The business freedom component measures the extent to which the regulatory and infrastructure environments constrain the efficient operation of businesses. The quantitative score is derived from an array of factors that affect the ease of starting, operating, and closing a business. Assessment of the level of tax compliance. Scale from 0 to 6, where higher scores indicate higher compliance. 23 Hand collect Economic Freedom Economic Freedom La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999) Firm characteristics Leverage Long-term debt plus debt in current liabilities divided by total assets. Compustat Asset The natural logarithm of total assets of the firm. Compustat Tangibility Net property, plant and equipment/total assets. Profitability The Earnings before interest, taxes, depreciation and amortization (EBITDA) divided by the total assets. Compustat Market-to-book Total Assets Book value of Equity + Price Common Shares ratio (MB ratio) Outstanding) / Total Assets. Compustat Z-score Modified Altman s (1968) Z-score = (1.2 working capital+1.4 retained earnings+3.3 EBIT+0.999sales)/total assets. We use a modified Z-score, which does not include the ratio of market Compustat value of equity to book value of total debt, because a similar term, market-to-book, enters the regressions as a separate variable. Cash flow Volatility (CFV) Standard deviation of quarterly cash flows from operations over the four fiscal years prior to the loan initiation year scaled by the total debt. Compustat Loan characteristics Maturity Natural logarithm of loan maturity in months. DealScan Loan size Natural logarithm of amount of loan in US$ million. DealScan Collateral A dummy variable that takes a value of 1 if a loan is secured, and DealScan

26 0 otherwise. Performance A dummy variable that equals 1 if the loan facility uses performance pricing, and 0 otherwise. DealScan Loan type Dummy variables for loan types, including term loan, revolving loans longer than 1 year, revolving loans shorter than 1 year, and DealScan 364-day facility. These IDs start from 1 to 62. Loan purpose Dummy variables for loan purposes, including corporate purposes, debt repayment, working capital, takeover, etc. IDs start DealScan from 1 to 41. GenCov Number of general covenants. DealScan FinCov Number of financial covenants. DealScan Macroeconomic variables GDP Log of GDP per capita is gross domestic product divided by midyear population. Data are in current U.S. dollars. World Bank GDP_G Annual percentage growth rate of GDP per capita based on constant local currency. World Bank Inflation Country s inflation rate. Inflation as measured by the consumer price index reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services that World Bank may be fixed or changed at specified intervals, such as yearly. The Laspeyres formula is generally used. Country governance variables Government Effectiveness (GE) Control of Corruption (CC) Country Governance index (CG) Measures the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government s commitment to such policies. The score is between -2.5(weak) to 2.5(strong) Measures the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as capture of the state by elites and private interests. Aggregate Governance Index aggregates the above six broad dimensions which included the voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law and control of corruption using an unobserved components model. Cultural difference variables Cultural distance Euclidean distance between the cultures of the borrower s and the lead bank s countries/wvs (Inglehart Welzel 1981~2015) New cultural distance Law and election related variables Election type Executive Legislation Elections Timing Considering the measurement equivalence of statistically mathematical problem and calculating Euclidean distance in their overall secular values (y-axis) and emancipative values (x-axis) between the cultures of the borrower s and the lead bank s countries Identifies the types of elections in each country. There are three obviously different types: (1) Presidential, (2) Legislative and (3) Prime Ministerial. Identifies the legal origin of the Company Law or Commercial Code of each country. There are five possible origins: (1) English Common Law; (2) French Commercial Code; (3) German Commercial Code; (4) Scandinavian Commercial Code; and (5) Socialist/Communist laws. Identifies the types of election timing of each country. There are two obviously different types: (1) Fixed and (2) Flexible. An election is classified as regular if it is held within 6 months before or after the election date, which is calculated by adding the nominal term of the chief executive to the previous election date. Otherwise, an election is classified as irregular. 24 Kaufman,Kraay, and Mastruzzi Kaufman,Kraay, and Mastruzzi Kaufman,Kraay, and Mastruzzi Giannetti and Yafeh (2012) Welzel (2013) Hand-collect La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998) Julio and Yook (2012)

27 Appendix B The Correlation Matrix This table shows that correlation coefficients for the variables use in the regressions. The sample is from 1982 to (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (1) Spread (2) Election 0.009* (3) Asset ** ** (4) MB ratio 0.148** ** (5) Tangibility ** * 0.030** 0.222** (6) Profitability ** ** 0.072** 0.122** (7) Leverage 0.201** ** 0.326** 0.152** ** (8) Z-score ** ** ** ** 0.562** ** (9) CFV 0.016** ** ** ** ** ** 0.023** (10) Maturity 0.074** ** 0.040** 0.094** 0.070** 0.097** 0.131** * ** (11) Loan size ** ** 0.577** ** 0.075** 0.146** 0.071** 0.026** ** 0.155** (12) Collateral 0.522** 0.012** ** 0.102** ** ** 0.153** ** 0.021** 0.129** ** (13) Performance ** * ** 0.050** ** ** 0.083** ** 0.123** 0.184** 0.089** (14) GenCov 0.241** ** ** 0.066** ** 0.048** 0.114** ** ** 0.151** 0.082** 0.389** 0.554** (15) FinCov 0.212** ** 0.125** ** 0.051** 0.056** ** ** 0.337** 0.507** 0.730** (16) GDP 0.104** 0.015** ** 0.015** ** 0.035** ** ** 0.036** 0.168** 0.098** 0.215** 0.299** 0.224** (17) GDP_G ** ** 0.081** 0.014** ** ** ** ** ** ** ** (18) Inflation 0.028** 0.029** ** 0.013** 0.040** ** 0.023** ** ** * ** ** ** ** 0.132**