View
4
Download
0
Category
Preview:
Citation preview
Does Social Trust affect International Contracting? Evidence from Foreign Bond Covenants*
Paul Brockmana , Sadok El Ghoulb, Omrane Guedhamic, Ying Zhengd
aLehigh University, Bethlehem, PA 18015 USA pab309@lehigh.edu
bUniversity of Alberta, Edmonton, AB T6C 4G9, Canada elghoul@ualberta.ca
cUniversity of South Carolina, Columbia, SC 29208 USA omrane.guedhami@moore.sc.edu
dBryant University, Smithfield, RI 02917 USA czheng@bryant.edu
* The authors thank Narjess Boubakri, Chuck Kwok, Ruiyuan Chen, Kevin J. Maloney, Sattar Mansi, Peter J. Nigro, John Wald, He Wang, Xiaolan Zheng for their helpful comments and suggestions. Sadok El Ghoul acknowledges financial support from Canada’s Social Sciences and Humanities Research Council [grant number 435-2014-1904].
Contact author: Paul Brockman, Perella Chair Professor of Finance, College of Business and Economics, 407A Rauch Business Center, 621 Taylor Street, Lehigh University, Bethlehem, PA 18015, Phone: +1.610.758.2914, Email: pab309@lehigh.edu.
Does Social Trust affect International Contracting? Evidence from Foreign Bond Covenants
Abstract
This paper investigates the influence of social trust on security-level contract design. Using a
sample of foreign (non-U.S.) firms issuing bonds in U.S. debt markets (i.e., Yankee bonds), we
find that U.S.-based creditors impose fewer covenants on bond issuers domiciled in countries
with a high degree of social trust. We further show that the inverse relation between social trust
and debt covenants is more pronounced for firms from countries with weak institutions, for firms
with poor corporate governance, and for more opaque firms. These findings are robust to
endogeneity tests, within-country analysis, alternative measures of trust, alternative explanations,
different samples, and alternative empirical models. In an additional analysis, we find that a
higher level of social trust is associated with lower borrowing costs, but this relationship
weakens when covenants are included in the debt contract. Overall, our findings provide new
insight into the significant role that social trust plays in security-level investor protection.
Key Words: Trust; Contracts; Covenants; Institutions; Corporate Governance
JEL Classification: F30; G30; M41
1
1. Introduction
Motivated by Arrow’s (1972, p. 357) insight that “virtually every commercial transaction
has within itself an element of trust,” recent empirical research examines the role of social trust
in economic transactions and outcomes. The empirical evidence to date finds significant links
between social trust and merger and acquisition (M&A) activities (Ahern, Daminelli, and
Fracassi, 2015), stock market participation and ownership structure (Guiso, Sapienza, and
Zingales, 2008), bilateral trade in goods, financial assets, and foreign direct investment (Guiso,
Sapienza, and Zingales, 2009), and venture capital investment decisions (Bottazzi, Da Rin, and
Hellmann, 2016). In this study we extend this line of research by examining the link between
social trust and debt contracting. Since creditor-debtor contracts are inherently incomplete, few
economic transactions are likely to be more sensitive to the level of trust between parties than
debt contracts. We expect a higher level of social trust to affect debt contracting by reducing
creditor demand for restrictive covenants; that is, provisions that restrict borrowers from taking
actions that may be harmful to creditors.
To test this prediction, we first build on Williamson’s (2000) four-level social analysis.
Level 1 comprises a country’s informal customs, traditions, and norms – social patterns that are
embedded in the social fabric, are highly persistent over time, and can be viewed as inherent to
the country’s underlying culture. These social patterns are effectively exogenous and not subject
to social engineering. Level 2 comprises a country’s political, judicial, and economic institutions
that determine the formal rules of the game. Level 3 comprises a country’s governance
structures, which influence individuals’ incentives in playing the game. Level 4 comprises
economic transactions that together determine prices, quantities, and the allocation of resources.1
Importantly, under this framework, higher-level social structures have a direct impact on lower-
level structures (i.e., Level 1 has a direct effect on Levels 2, 3, and 4; Level 2 on Levels 3 and 4;
and Level 3 on Level 4), but lower levels do not influence higher levels except through weak
feedback loops.
1 Williamson’s (2000) four levels of social analysis can also be thought of in terms of the economic research that takes place at each level. As rough approximations, research on economic history tends to focus on Level 1, property rights and legal institutions on Level 2, agency theory and transaction cost economics on Level 3, and pricing theory on Level 4.
2
Using this framework, we examine the mechanisms through which a fundamental Level 1
characteristic (social trust) affects Level 4 transaction-level quantities (bond covenants) and
prices (borrowing costs). We first test for a direct causal relation from social trust to debt
contracting, as Level 1 trust imposes constraints on Level 4 transactions. We then test for
indirect (interaction) effects of social trust on debt contracting through Level 2 institutions
(investor protection, political institutions) and Level 3 governance structures (firm-level
governance and information environment). Just as Williamson’s (2000) framework suggests that
a country’s cultural traditions (Level 1) influence the development of its financial institutions
(Level 2) to a greater degree than its financial institutions influence the development of its
cultural traditions, we expect Level 1 social trust to have a direct influence on Level 4 creditor-
debtor transactions – with little if any effect running in the opposite direction.
We also build on Miller and Reisel (2012), who use the setting of Yankee bonds (bonds
of non-U.S. companies issued in U.S. debt markets) to investigate the impact of country-level
creditor rights on issue-level debt covenants. Miller and Reisel (2012) find that firms from
countries with weak creditor protection rights use more restrictive bond covenants when they
issue Yankee bonds, and that restrictive bond covenants are associated with lower bond spreads
in weak creditor rights environments. In terms of Williamson’s framework, these results are
consistent with creditor rights, which are determined at Level 2, having a direct effect on
international debt contracts, which are negotiated at Level 4. We follow Miller and Reisel (2012)
in employing a sample of Yankee bonds. Our study differs, however, in that we move the level
of analysis up to a Level 1 characteristic embedded in a country’s social fabric – social trust.
More specifically, we investigate the interaction effects between social trust and country-level
institutional environment as well as firm-level governance and information environment.
Our main empirical analysis proceeds as follows. To capture the level of social trust in
the country of origin, we employ a widely used measure from the World Values Survey (WVS).
Specifically, we use the country-level average percentage of affirmative responses to the
question “Generally speaking, would you say that most people can be trusted or that you need to
be very careful in dealing with people?” Next, to capture the use of restrictive covenants, we
collect information on Yankee bond issues from the Fixed Investment Securities Database
(FISD). The data in the FISD include the number of covenants associated with each Yankee
3
bond issue as well as a breakdown of the number of covenants into the number of financing,
investing, and payout covenants.2 We construct Number of Covenants as the sum of restrictive
covenants (all, financing, investing, or payout covenants) and Intensity of Covenants as the sum
of three indicator variables for the presence of covenants in the three respective categories.
Finally, using a sample of 1,033 Yankee bond issues from 34 countries over the period 1989 to
2014, we conduct Poisson regressions in which we examine the effect of the WVS-based social
trust measure on the number or intensity of restrictive covenants written into Yankee bond
contracts, while controlling for issue-, firm-, and country-level variables, including creditor
rights.
The results show that social trust plays a significant role in debt contracting. In particular,
using both the total number and intensity of covenants as the dependent variable, we find that the
coefficient on social trust is negative and statistically significant. The coefficient on social trust
is also economically significant. In our baseline specification, for example, a one-standard-
deviation increase in social trust around the mean leads to a 0.29 decrease in the total number of
covenants written into international bond issues. These results suggest that when a Yankee bond
originates in a country with a high (low) degree of social trust, U.S. creditors demand relatively
fewer (more) restrictive covenants. When we separately examine the number of covenants used
in the financing, investing, and payout categories, we find that social trust helps alleviate creditor
concerns about firms’ investing and payout behavior, but is less effective in mitigating creditor
concerns about financing behavior.
We next examine interaction effects between social trust and the country-level
institutional environment as captured by the strength of investor protection (e.g., debt recovery
efficiency, rule of law), the political environment (e.g., freedom of the press, political stability),
and the presence of interlocking business groups. The results show that social trust plays a more
important role when a country’s institutional environment is weak. Specifically, we find positive
and significant interaction effects between social trust and investor protections, positive and
significant interaction effects between social trust and political institutions, and negative and
significant interaction effects between social trust and interlocking business groups. Since social
2 There are seven possible financing covenants (e.g., restrictions on senior debt issuance), four possible investing covenants (e.g., M&A restrictions), and two possible payout covenants (e.g., restrictions on dividend payouts), for a total of 13 possible bond covenants.
4
trust operates at a more fundamental level of society than formal institutions (i.e., Level 1 versus
Level 2), these results suggest that strong institutions can act as a substitute for social trust.
In addition to country-level interaction effects, we examine interaction effects between
social trust and the firm-level governance and information environment. We expect that at the
firm level, stronger governance and a more transparent information environment can act as a
substitute for social trust. To test this conjecture, we use the corporate governance index,
earnings forecast errors, and Amihud’s (2002) illiquidity (or price impact) measure. The results
reveal a positive (negative) and significant effect of the interaction between social trust and the
corporate governance index (forecast errors, the illiquidity measure). These findings suggest that
managers are able to mitigate the negative effects of low levels of social trust by improving firm
governance and reducing information asymmetry.
One potential concern with cross-country analyses is endogeneity. Endogeneity may arise
due to the non-random sample selection, whereby firms are not randomly assigned into countries
with different levels of trust. In a cross-country study, the non-random sample selection problem
is further complicated because it opens up possibilities for country-level variations to confound
the effects of trust. For example, country-specific historical forces (such as institutional
structures) that produce specific trust environments can also directly impact firms’ contract
designs as well. An endogeneity bias may arise from reverse causality. That said, reverse
causality is unlikely to drive our main findings because, as Williamson (2000) argues, Level 4
outcomes (i.e., the use of debt covenants) are unlikely to influence a Level 1 characteristic (i.e.,
social trust). Nevertheless to ensure the reliability of our inferences we mitigate potential
endogeneity concerns in two ways. First, we use instrumental variables analysis. To capture the
exogenous variations in social trust, we follow prior research (e.g., Tabellini, 2008; Guiso,
Sapienza, and Zingales, 2009; Bjørnskov and Méon, 2013) and employ the following three
instrumental proxies: climate, genetic dissimilarity of populations, and language (or linguistic
rules).3 We empirically validate that these instruments are relevant and exogenous to social trust.
The instrumental variables estimation results continue to show a negative and significant relation
3 We provide a more detailed discussion of these instruments in Section 4.4.
5
between social trust and the use of debt covenants, providing additional assurance that our
findings reveal a causal effect from trust to debt covenants.
Second, in addition to controlling for various country-specific factors shown to matter for
debt contracting, we further mitigate concerns about omitted country-specific traits by providing
out-of-sample evidence based on a single country. By doing this, we have a clean setting that
excludes all country-level variations that might contaminate the effect of trust. Using covenants
attached to public corporate bonds issued by U.S. firms and WVS-based social trust across U.S.
regions, we find that more restrictive covenants are placed on bonds issued by firms located in
U.S. geographic regions with lower social trust. These single-country results are less susceptible
to concerns about omitted country-level variables since all of the sample firms share the same
political, legal, social, and economic environment.
In addition to these endogeneity analyses, we further validate our empirical findings by
conducting a series of robustness tests using alternative proxies for trust, alternative
explanations, alternative samples, and alternative empirical models. In all such tests, we continue
to find a negative and significant relation between social trust and the use of debt covenants.
Finally, we build on our establishment of a causal link from social trust to debt
contracting by examining the pricing effects of social trust. We expect higher levels of social
trust to lead to lower borrowing costs, all else equal. To test this prediction, we estimate the
effects of social trust and our set of control variables on Yankee bond interest rates over
maturity-matched Treasury rates (i.e., Treasury spreads) using two specifications, one without
and one with interaction effects between social trust and the use of debt covenants. The results
without interaction effects show that higher levels of social trust lead to significantly lower
borrowing costs. The results with interaction effects show that higher levels of social trust have a
lower marginal effect on borrowing costs when covenants are included in the debt contract.
Thus, issue-level debt covenants substitute for social trust when the level of social trust is low.
In summary, our empirical findings consistently document an inverse relation between
the level of social trust and the use of debt covenants. This result is consistent with the view that
when the level of trust is high, incomplete contracts pose less of a concern to contracting parties,
whereas when the level of trust is low, incomplete contracts pose major stumbling blocks that
result in higher transaction costs. Our results also document significant interaction effects
6
between social trust and both the country-level institutional environment and the firm-level
governance and information environment. Finally, our results show that social trust has both a
direct effect on borrowing costs and an indirect effect through its influence on the number (and
intensity) of debt covenants.
Our study contributes to the literature on culture and finance. One important question that
researchers in this literature often face is how to distinguish between the direct and indirect
effects of culture (Aggarwal et al., 2016). Under Williamson’s (2000) framework, for example,
Level 1 characteristics can operate on Level 4 characteristics either directly or indirectly through
the middle levels, which raises the question of how to disentangle these various effects. Our
study shows how a fundamental social characteristic such as trust (Level 1) interacts with the
country-level institutional environment (Level 2) and firm-level governance structure (Level 3)
to influence transaction-level prices and quantities (Level 4). As such, our study provides a
comprehensive picture of the multi-level links connecting social trust and bond covenants.
Our study also contributes to the literature by identifying the role that social trust plays in
debt contracting. Extant literature on debt contracting considers the role of the institutional
environment (Level 2) but largely overlooks the influence of more fundamental social factors
(Level 1) such as trust. For example, previous studies show that weak creditor rights increase the
use of bond covenants (Miller and Reisel, 2012) and that investor protection influences the terms
of bank loans (Qian and Strahan, 2007; Bae and Goyal, 2009) and project financing (Esty and
Megginson, 2003). Our results show that the effect of social trust on debt contracting is
comparable to that of institutions: in our baseline model, a one-standard-deviation increase in
creditor rights (trust) around the mean is associated with a 0.34 (0.29) decrease in the total
number of covenants.
Our study provides new insights on the use of contract adaptations to increase investor
protections. Miller and Reisel (2012) note that while it is difficult to improve inefficient legal
rules in weak investor protection countries, firms and investors can substitute for legal protection
of investors by strengthening investor protection at the security level. Similarly, our results
suggest that while a shift in the social landscape of a country would be especially difficult to
bring about (Williamson, 2000), firms and investors can use security-level investor protections to
circumvent inefficiencies arising from a low level of social trust.
7
The remainder of the paper is organized as follows. In Section 2, we provide a brief
summary of related literature and develop our primary hypotheses. In Section 3, we describe our
sample, variables, and empirical design and we present descriptive statistics. In Section 4, we
present our empirical results. Section 5 concludes.
2. Background and Hypotheses
2.1. Background
The idea that social trust plays a role in determining economic outcomes has a long and
venerable history.4 As quoted in Zak and Knack (2001, p. 295), in the 1651 edition of The
Leviathan, Thomas Hobbes observes that, “If a covenant be made, wherein neither of the parties
perform presently, ... he that performeth first, has no assurance the other will perform after;
because the bonds of words are too weak to bridle men’s ambition, avarice, anger, and other
Passions, without the fear of some coercive Power ....” This passage succinctly captures the
modern economic notion of moral hazard, and identifies the underlying agency problem as a lack
of social trust. Adam Smith and John Stuart Mill more fully develop the connection between
trust and economic outcomes by observing cross-country variation in social trust. Smith (1763, p.
538), for example, states that “Of all the nations of Europe, the Dutch, the most commercial, are
the most faithful to their word,” while Mill (1848, p. 132) argues that “[t]here are countries in
Europe … where the most serious impediment to conducting business concerns on a large scale,
is the rarity of persons who are supposed fit to be trusted with the receipt and expenditure of
large sums of money.”
Although scholars have long observed a connection between social trust and economic
outcomes, formal theoretical analysis and empirical verification began in the 1980s and 1990s.
Several studies in this period examine the effects of social capital on the real economy, including
the development of human capital (Coleman, 1988), differences in economic performance
between Northern and Southern Italy (Putnam, Leonardi, and Nanetti, 1993), and general
4 Recent survey articles on the impact of culture (including social capital and social trust) on economics and finance include Aggarwal et al. (2016), Karolyi (2016), and Bjørnskov (2018).
8
economic prosperity (Fukuyama, 1995). Much of this research can be summarized by Knack and
Keefer’s (1997, p. 1251) finding that “trust and civic norms are stronger in nations with higher
and more equal incomes, with institutions that restrain predatory actions of chief executives, and
with better-educated and ethnically homogeneous populations.” In the context of Williamson’s
(2000) framework, these studies include connections across all four levels of social analysis.
More recent research has refined the theoretical and empirical relations between trust and
specific economic outcomes. Zak and Knack (2001) develop a theoretical model predicting that
societies with a higher level of trust have higher long-run economic growth because of reduced
risk and a higher investment rate. Stulz and Williamson (2003) show that a country’s main
religion can explain cross-country variation in creditor rights better than legal origin, openness to
trade, and GDP per capita. Similar studies document that social capital/trust plays a significant
role in explaining cross-country differences in financial development (Guiso, Sapienza, and
Zingales, 2003), national savings, entrepreneurship, and income redistribution (Guiso, Sapienza,
and Zingales, 2006), stock market participation (Guiso, Sapienza, and Zingales, 2008), venture
capital development (Bottazzi, Da Rin, and Hellman, 2016), stock market reactions to earnings
announcements (Pevzner, Xie, and Xin, 2015), and the availability of trade credit during
financial crises (Levine, Lin, and Xie, 2017).
Using variations at the US county level, researchers have documented that social trust
plays a significant role in the cost of bank loans and public bonds (Hasan et al., 2017), as well as
agency costs, profitability, and firm valuations (Hilary and Huang, 2016). And at the firm level, a
related line of research examines the effect of cross-sectional variation in corporate social
responsibility (CSR) – an element in Williamson’s (2000) Level 3 – on stock returns (Lins,
Servaes, and Tamayo, 2017) and the pricing of public debt (Amiraslani et al., 2017) during the
2008–2009 financial crisis. Two general themes that run through all of these empirical studies
are that i) both informal institutions (Level 1) and formal institutions (Levels 2 and 3) have a
significant effect on economic outcomes (Level 4), and ii) the connection between institutions
and economic outcomes can be studied formally using economic theory and empirical research
methods, consistent with the two main propositions of new institutional economics.
9
2.2. Hypotheses
We develop four main hypotheses based on the literature reviewed above. The first three
are motivated by the argument that Level 1 informal institutions significantly affect Level 4
economic outcomes through both a direct effect and indirect effects that operate via interactions
with Level 2 formal institutions and Level 3 governance mechanisms. Our fourth hypothesis
relates the direct and indirect effects of social trust on borrowing costs.
Our first hypothesis holds that Level 1 social trust has a significant direct effect on Level
4 debt contracting. In particular,
H1: A higher level of social trust reduces the use of restrictive covenants in debt contracts.
The rationale is that a higher level of social trust in the firm’s country of origin (i.e., our WVS-
based measure of trust) leads US bond investors to demand fewer and less intense safeguards
against contract breaches. There are at least two channels through which higher levels of trust
within the country of origin can increase US-based Yankee bondholders’ trust in the firm. First,
higher levels of trust within a particular country reduce agency problems5 in the country, which
reduces expected (ex ante) costs of financial distress for US-based Yankee bondholders. Second,
US-based creditors will infer higher levels of trustworthiness for firms from countries in which
creditors and debtors display a higher level of trust among themselves. We test H1 by regressing
the number (and intensity) of Yankee bond debt covenants on our measure of social trust and a
set of issue-, firm-, and country-level control variables.
Our second hypothesis holds that Level 1 social trust also affects Level 4 debt contracting
through its interaction with Level 2 formal institutions (i.e., legal system, political system,
prevalence of family/interlocking firms). Formally,
H2: There is a significant substitution effect between social trust and country-level formal
institutions in determining the use of restrictive covenants in debt contracts.
5 There are two types of agency costs in firms with risky debt outstanding. First, the underinvestment problem arises when shareholders have incentives to forgo profitable investment opportunities if creditors can capture a disproportionate share of firm value resulting from such investments (Myers, 1977). Second, the asset substitution problem arises when firms that issue debt on the premise of taking low-risk projects pursue high-risk projects instead (Jensen and Meckling, 1976).
10
Here we posit that social trust and formal institutions function as substitutes in terms of their
impact on debt covenants; specifically, when trust is high (low), the marginal impact of formal
institutions on covenants is low (high). We test H2 by regressing the number (and intensity) of
Yankee bond debt covenants on our measure of social trust, various measures of the country-
level institutional environment, the interaction between social trust and these measures of the
country-level institutional environment, and our set of control variables.
Our third hypothesis holds that Level 1 social trust affects Level 4 debt contracting
through its interaction with Level 3 governance mechanisms (i.e., firm-level governance and
asymmetric information). Formally,
H3: There is a significant substitution effect between social trust and the firm-level governance
and information environment in determining the use of restrictive covenants in debt contracts.
Following the same logic as H2, we posit that social trust and governance structures operate as
substitutes in terms of their impact on debt covenants; specifically, when trust is high (low), the
marginal impact of governance structures on covenants is low (high). We test H3 by regressing
the number (and intensity) of Yankee bond debt covenants on our measure of social trust, various
measures of firm-level governance and information environment, the interaction between social
trust and these measures of firm-level governance and information environment, and our set of
control variables.
While firm value is independent of the choice of firm financing instruments in perfect
capital markets, in practice frictions such as agency costs and asymmetric information problems
between shareholders and creditors constrain firms’ access to external capital. Previous research
(Myers, 1977; Smith and Warner, 1979; Barclay and Smith, 1995) shows that agency and
asymmetric information problems can be alleviated by an appropriate choice of debt maturity,
covenants, and seniority (i.e., debt that is repaid before subordinated debt if the borrower enters
bankruptcy). Focusing on the role of covenants, the rationale is that in the presence of agency
and asymmetric information problems, creditors require a larger risk premium on covenant-free
debt to compensate for the risk that the borrower’s credit quality will decline. Thus, debt without
(with) covenants is overpriced (underpriced). Because low-quality borrowers prefer the
overpriced debt while high-quality borrowers prefer the underpriced debt, creditors interpret the
choice of debt covenants as a signal of firm quality (Gârleanu and Zwiebel, 2009). We extend
11
these arguments by positing that trust contributes to systematic differences in debt covenants by
shaping creditors’ perception of agency and asymmetric information problems. In particular, we
predict that higher social trust reduces borrowing costs by decreasing agency and asymmetric
information problems (a direct effect). In addition, higher social trust can reduce borrowing costs
through its impact on the use of covenants (as examined in H1; an indirect effect). Our fourth
hypothesis is thus stated as follows:
H4: A higher level of social trust lowers borrowing costs by mitigating agency and information
problems (direct effect) and by reducing the use of restrictive covenants in debt contracts
(indirect effect).
We test H4 using two specifications. We first regress Yankee bond interest rates over maturity-
matched Treasury rates (i.e., Treasury spreads) on social trust and our set of control variables. In
the second regression, we re-run this model after adding an interaction term between social trust
and a debt covenants indicator. H4 predicts that higher levels of social trust will have less of an
impact on borrowing costs when covenants are added to the debt contract (i.e., debt covenants
substitute for social trust in countries where social trust is low).
3. Sample, Variable Construction, Empirical Design, and Descriptive Statistics
In this section we describe our sample construction, the main variables used in our
analysis, and our empirical design. We then report descriptive statistics.
3.1. Sample
We begin the sample selection process with the sample of foreign bonds issued in the US
(Yankee bonds) contained in the Fixed Investment Securities Database (FISD), which is widely
used in finance studies on bond covenants (e.g., Reisel, 2014; Chava, Livdan, and Purnanandam,
2009; Qi, Roth, and Wald, 2011; Miller and Reisel, 2012). Our initial sample comprises 3,064
issues by borrowers from 52 countries between 1989 and 2014. We exclude bonds issued before
1989 because of the poor quality of covenant data before 1989 (Miller and Reisel, 2012). To
ensure the validity of covenant information, we also exclude from this sample medium-term
notes, private placements, bonds with special characteristics such as convertible, pass-through, or
12
payment-in-kind features, as well as issues with an offering amount that exceeds $50 billion or
an offering yield that is greater than 50 basis points. Our initial sample represents an update to
Miller and Reisel’s (2012) sample, which ends in 2007. Next, we merge the FISD data with
Compustat (North America and Global) by manually matching company names. This yields a
sample of 1,650 bonds across 38 countries. We then merge with the WVS to obtain data on trust.
We exclude observations with missing values for variables used in our main regression and
observations with a missing Campbell (1996) industry classification. This procedure yields a
final sample of 1,033 bond issues from 34 countries.
3.2. Variable Construction
3.2.1. Covenants
Our main dependent variable is the total number of covenants written into bonds issued
by non-US firms. Following Miller and Reisel (2012), we count the number of restrictive
covenants in each outstanding debt issue. However, while the total number of covenants gives
one measure of covenant use, it obscures the fact that some types of firm activities face more
potential restrictions than others. To address this concern, we use as an alternative dependent
variable the intensity of covenants, which is equal to the sum of three indicators equal to 1 if the
debt contract contains a covenant in the given category. In effect, this alternative measure assigns
equal weight to three categories of restrictions on firm activities, namely, restrictions on a firm’s
financing, investment, and payout activities.
Covenants on financing activities are designed to protect bondholders from claim dilution
by limiting the firm’s ability to enter into future obligations that could lead to bankruptcy. This
category contains seven covenant terms: (1) the issuer is not allowed to issue secured debt unless
it secures the current issue on a pari passu basis; (2) the issuer is not allowed to issue debt with
initial maturity of one year or longer; (3) the issuer is not allowed to incur additional debt with
limits on the absolute dollar amount of debt outstanding or the percentage of total capital; (4) the
issuer is restricted on the amount of senior debt; (5) the issuer is not allowed to issue
subordinated debt; (6) the issuer is not allowed to issue additional debt unless the issuer achieves
or maintains a certain profitability level; and (7) the issuer is restricted on sale-leaseback
transactions, which involve raising capital by selling an asset and simultaneously leasing the
asset back for a fixed term at an agreed-upon rate.
13
Covenants on investment activities are designed to protect bondholders from asset
substitution problems by limiting the firm’s investments in excessively risky projects. This
category contains four covenant terms: (1) the issuer is restricted on investment policy to reduce
risky investments; (2) the issuer is restricted on mergers or consolidations; (3) the issuer is
restricted on selling its assets; and (4) the issuer is required to use proceeds from the sale of
assets to redeem bonds at par or at a premium.
The last category of covenants relates to corporate payouts. These restrictions are
designed to protect bondholders from wealth transfers by limiting the firm’s ability to transfer
assets (through cash dividends) to shareholders. This category contains two covenant terms: (1)
the issuer is not allowed to make dividend-related payments to shareholders or other entities
above a certain percentage of net income or some other accounting ratio; and (2) the issuer is
restricted on non-dividend-related payments to shareholders or other entities.
3.2.2. Trust
Following prior literature, we construct our trust variable based on the WVS question
“Generally speaking, would you say that most people can be trusted or that you need to be very
careful in dealing with people?” Specifically, we calculate the level of trust in a country as the
mean percentage response in the country, where a response is recorded as 1 if the participant
reports that most people can be trusted and 0 otherwise. This question is documented in all six
waves of the survey: 1981, 1990, 1995, 1999, 2005, and 2010. We match the trust variable to
firm-year observations in commensurate periods. For country-years with missing trust values, we
fill in the missing values based on the previous wave. In robustness tests, we adopt alternative
measures of trust, including 100% + (the mean percentage response to “most people can be
trusted”) – (the mean percentage response to “can’t be too careful”) as well as the mean response
to the survey question asking the extent to which the participant trusts their neighborhood.
Detailed variable definitions are reported in the Appendix.
3.3. Empirical Design
We estimate the following model (subscripts omitted) using a Poisson regression:6
6 In a sensitivity check, we also estimate a negative binomial regression. The results are qualitatively unchanged.
14
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝛼𝛼0 + 𝛼𝛼1𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶 + 𝛼𝛼2𝐼𝐼𝐼𝐼𝐼𝐼 + 𝛼𝛼3𝐹𝐹𝐼𝐼𝐼𝐼 + 𝛼𝛼4𝐶𝐶𝐼𝐼𝐼𝐼 + 𝐹𝐹𝐹𝐹𝐹𝐹𝐶𝐶𝐹𝐹 𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐸𝐸𝐶𝐶𝐶𝐶 + 𝜀𝜀. (1)
The dependent variable, Covenants, is either Number of Covenants or Intensity of Covenants.
The treatment variable of interest, Trust, is our WVS-based measure of trust. ILV is a vector of
issue-level control variables: Amount (natural logarithm of the bond issue’s offering amount) and
Maturity (the maturity of the bond issue in days). These variables come from FISD. FLV is a
vector of firm-level control variables: Size (natural logarithm of total assets), Leverage (total debt
over total assets), Interest Coverage (earnings before interest and taxes over interest expenses),
Profitability (earnings before interest and taxes over total assets), Capital Expenditure (capital
expenditure over total assets), and Tangibility [following Berger, Ofek, and Swary (1996),
(0.715×receivables + 0.547×inventories + cash + 0.535×net property, plant, and equipment) /
total assets]. These variables are collected from Compustat Global. CLV is a vector of country-
level control variables: GDP (logarithm of GDP per capita), GDP Growth (growth of GDP per
capita), Financial Development (private credit to GDP), Inflation, Creditor Rights, Rule of Law,
and Control of Corruption. These variables are obtained from the World Development Indicators
(WDI) database, Djankov, McLiesh, and Shleifer (2007), and the World Governance Indicators
(WGI) database. We follow extant literature in choosing the controls to include in ILV, FLV, and
CLV (e.g., Barclay and Smith, 1995; Chava, Kumar, and Warga, 2010; Qi, Roth, and Wald,
2011; Miller and Reisel, 2012). Finally, we control for year and industry effects in all
regressions, and we cluster standard errors at the country level.
3.4. Descriptive Statistics
Table 1 summarizes our sample of bond issues in the initial sample and the Compustat
subsample. Our initial sample comprises international public debt offerings from 1989 to 2014
and represents an update to Miller and Reisel’s (2012) sample. In general, our initial sample
displays the same characteristics as in Miller and Reisel (2012). Panel A, which reports bond
characteristics and frequency by period, shows that the average offering amount is $573.93
million, consistent with the notion that the Yankee bond market is one of the few markets in
which a foreign entity can issue large amounts of long-term fixed-rate debt (Karolyi and
Johnston, 1998). Moreover, the use of covenants appears to be common, with a median of three
covenants per bond. Most of the bonds are senior, callable, have a maturity between 5 and 15
15
years and are from the later part of the sample (1998–2006 or 2007–2014). Similar patterns
obtain in the Compustat subsample.
Panel B of Table 1 shows the frequency by country for the initial sample, which covers
52 countries. Consistent with prior studies (e.g., Miller and Puthenpurackal, 2002; Miller and
Reisel, 2012), this sample is dominated by Canada (29.54%) followed by the UK (16.81%). At
the other end of the spectrum, countries such as Costa Rica and Denmark have relatively few
Yankee bond issues, with 0.03% of observations each.7
Panel C of Table 1 summarizes firm and country characteristics for the Compustat
subsample. Our main variable of interest, Trust, ranges from a low of 0.06 for the Philippines to
a high of 0.69 for Norway, with an average of 0.35 and a standard deviation of 0.13. Importantly
for our purposes, these figures suggest that there is a large degree of variation in the level of
within-country trust across the globe.
------------------------------------------
Insert Table 1 about here
-------------------------------------------
Table 2 presents Pearson correlation coefficients for our key regression variables. We
find that Trust is significantly negatively related to our proxies for restrictive covenants,
providing preliminary evidence that firms from countries with high levels of trust tend to issue
fewer covenants. The correlation coefficients among the key explanatory variables are low,
suggesting that multicollinearity is unlikely to affect our results.
------------------------------------------
Insert Table 2 about here
-------------------------------------------
7 In robustness tests, we show that our main findings hold after excluding countries with an extremely large or small number of observations.
16
4. Empirical Results
4.1. Main Evidence
Table 3 presents our main evidence on the relation between trust and debt covenants. In
Model 1 we regress the total number of covenants across all categories on Trust, that is, we
regress Equation (1) without controls and fixed effects. We find that the coefficient on Trust is
negative and significant at the 1% level, consistent with H1, which holds that covenants are more
prevalent in bonds issued by firms from countries with a low level of trust. In Model 2, our
baseline specification, we add to Model 1 controls for issue-, firm-, and country-level factors that
may influence debt covenants as well as industry and year fixed effects, that is, we estimate
Equation (1). The coefficient on Trust continues to be negative and statistically significant. This
effect is also economically significant: a one-standard-deviation increase in Trust around its
mean leads to an average decrease in the number of covenants of 0.29.8 By comparison, we find
that a one-standard-deviation increase in Creditor Rights around its mean results in an average
decrease in the number of covenants of 0.34 (consistent with the finding in prior literature that
country-level legal protection of creditors can substitute for contract-level covenants; see, e.g.,
Qi, Roth, and Wald, 2011; Miller and Reisel, 2012). These results support the view that social
trust is an important determinant of the use of debt covenants.
In Models 3, 4, and 5 of Table 3, we investigate the effect of trust on the number of
covenants used in the financing, investment, and payout categories, respectively. We find that
trust significantly reduces the use of restrictive covenants on investment and payout activities,
which suggests that trust mitigates opportunistic behavior such as risk-shifting and wealth
transfers. In Model 6, we use our alternative measure of debt covenants, Intensity of Covenants,
which assigns equal weights to the three categories of restrictive covenants. We again document
a negative and significant coefficient on Trust.
Overall, the results in Table 3 support H1, thereby confirming that social trust
significantly affects debt contracting by reducing the number (and intensity) of debt covenants.
These results also suggest that since a high level of trust alleviates the need for restrictive bond
8 When we address endogeneity below, the economic impact of Trust becomes stronger than Creditor Rights.
17
covenants, bond covenants can act as a substitute mechanism for trust in firms operating in low
trust environments.9
------------------------------------------
Insert Table 3 about here
-------------------------------------------
4.2. Cross-country Variation in the Effects of Trust on Debt Covenants
In this section, we examine cross-country variation in the relationship between social
trust and debt covenants. Specifically, we test whether our main finding (i.e., a higher level of
social trust reduces the use of debt covenants) is more pronounced in countries with high agency
costs (e.g., La Porta et al., 1997, 1998; Johnson et al., 2000; Bae, Kang, and Kim, 2002; Dyck,
Volchkova, and Zingales, 2008; Shleifer and Vishny, 1997), where we capture agency costs
using the strength of investor protection, the presence of interlocking business groups, and the
political environment. H2 predicts significant substitution effects between social trust and these
country-level institutions in determining the number (and intensity) of debt covenants.
4.2.1. Investor Protection
The seminal work of La Porta et al. (1997, 1998) documents substantial cross-country
variation in the legal protection of investors. Johnson et al. (2000) and Dyck and Zingales (2004)
further show that private benefits of control are greater in countries with weaker legal protection
of investors. We capture the strength of investor protection in a country using five measures.
Debt Recovery (Djankov et al., 2008b) is the recovery rate of secured creditors; higher recovery
rates indicate better creditor protection. Efficiency of Debt Enforcement (Djankov et al., 2008b)
is the present value of the terminal value of the firm upon bankruptcy; higher values suggest
greater bankruptcy procedure efficiency. Efficient Judiciary (Djankov et al., 2008a) is the
logarithm of the number of days needed for the judiciary to collect on a bounced check. Rule of
9 According to Mansi, Qi, and Wald (2013), the use of debt covenants may reflect a herding behavior, whereby the use by many firms of a particular type or group of covenants in bond issues increases the probability of these covenants being used again. Accordingly, we follow Mansi et al. (2013) and capture the herding behavior in the use of bond covenants by controlling for the average number of covenants included in bonds issued in each country in the previous year. In untabulated results, we continue to find a negative and statistically significant coefficient at the 1% level on Trust, while the measure of the herding is statistically insignificant for our sample.
18
Law is an index that captures perceptions of the degree to which people have confidence in and
abide by the rules of society, particularly the quality of contract enforcement, property rights, the
police, and the courts, as well as the likelihood of crime and violence. Finally, Control of
Corruption is an index that captures perceptions of the degree to which public power is exercised
to benefit private interests, including petty and grand forms of corruption, as well as “capture” of
the state by elites. The Appendix provides further details on these variables.
In Panel A of Table 4, we examine the interaction between these measures of investor
protections and Trust. In Models 1 to 5, we regress Number of Covenants on Trust, the investor
protection variables, the interactions between the investor protection variables and Trust, and the
control variables. The coefficients on the interaction terms are positive and significant, which
suggests that, in line with H2, investors rely more on social trust in countries where investor
protection is weak. In Models 6 to 10, we repeat this analysis after replacing the dependent
variable with Intensity of Covenants. The results are qualitatively unchanged, which provides
additional evidence in support of the effect of social trust being more pronounced in countries
with a weak institutional environment.
4.2.2. Political Institutions
A country’s political environment can also influence firms’ agency costs. For instance, if
the media can be easily influenced by political pressure, expropriation activities may be more
easily concealed from the public, whereas independent media can increase the costs of diverting
firm resources (Dyck, Volchkova, and Zingales, 2008) and discourage corporate fraud (Miller,
2006; Dyck, Morse, and Zingales, 2010). In addition, a large literature shows that political
interference in corporate decision-making, including state-owned enterprises and politically
connected firms, can lead to expropriation at the cost of minority shareholders’ interests (Shleifer
and Vishny, 1997; Megginson and Netter, 2001; Faccio, Masulis, and McConnell, 2006;
Boubakri, Cosset, and Saffar, 2008, 2013; Chen et al., 2017).
To examine the influence of political institutions, we first use Press Freedom, an index
that assesses the degree of print, broadcast, and internet media independence; we adjust this
index so that a higher score represents a higher level of press freedom. Next, we use Kaufmann,
Kraay, and Mastruzzi’s (2009) Voice and Accountability, which measures the perceived degree
19
to which a country’s citizens can participate in selecting their government as well as their
freedom of expression and association, and Political Stability, which measures the perceived
likelihood that the government will be destabilized or overthrown by unconstitutional or violent
means (e.g., politically motivated violence and terrorism). Finally, we employ Government
Effectiveness, which measures the perceived degree of government effectiveness, including the
quality of public services, quality of the civil service, degree of independence from political
pressure, and quality of policy design and implementation.
In Panel B of Table 4, we run an analysis similar to that in Panel A but we use the
political institution measures above rather than the investor protection measures. Consistent with
the results in Panel A, we find that the interaction terms are positive and statistically significant,
which further supports H2 by showing that the effect of social trust on the use of debt covenants
is significantly influenced by the country-level political environment.
4.2.3. Interlocking Business Groups
Prior research suggests that business groups, and particularly family groups, are
associated with private benefits of control due to the divergence of cash flow and control rights
(La Porta et al., 1999; Johnson et al., 2000; Bae, Kang, and Kim, 2002; Bertrand, Mehta, and
Mullainathan, 2002; Baek, Kang, and Lee, 2006; Anderson et al., 2011). Bae, Kang, and Kim
(2002), for instance, argue that the wedge between cash flow and control rights can lead to
intragroup transactions within a business group that result in wealth transfers from a firm for the
benefit of its controlling shareholders.10 Such transactions are referred to as tunneling. Consistent
with creditors’ higher expropriation risk under a business group structure, Boubakri and Ghouma
(2010) and Lin et al. (2011) find that the agency costs embedded in business groups are
associated with higher costs of debt financing.
To capture the degree of intragroup tunneling among business groups, we employ four
measures that come from Masulis, Pham, and Zein (2011). Intragroup Tax captures the extent to
which a country’s tax law regulates intragroup transactions. Consolidation is a dummy variable
equal to 1 if a parent firm can consolidate its subsidiary with an ownership stake of less than
10 For example, under the chaebol “LG Group,” profitable member firm LG Securities acquired unprofitable member firm LG Merchant Bank because of the transaction’s positive effect for controlling shareholders. As expected, this acquisition led to substantial losses for the non-controlling investors of LG Securities.
20
90% and 0 otherwise. Prevalence of Family Group is the percentage of total market
capitalization held by firms that belong to a family group. Prevalence of Business Group is the
percentage of total market capitalization held by firms that belong to a business group, including
family and non-family business groups. The Appendix provides further details on these
variables.
In Panel C of Table 4, we re-run the analysis in Panels A and B using the four business
group measures rather than the investor protection and political institutions variables. The
interaction terms have positive loadings on Intragroup Tax and negative loadings on
Consolidation, Prevalence of Family Group, and Prevalence of Business Group. Thus, consistent
with H2, the results show that the effect of social trust on debt contracting is more pronounced in
the presence of interlocking business groups. Our business group results in Panel C represent an
intermittent or transition stage between country-level (Level 2) formal institutions and firm-level
(Level 3) governance structures. While business group membership is a firm-level measure, the
degree to which business groups dominate the national ownership structure is a country-level
institutional characteristic.
------------------------------------------
Insert Table 4 about here
-------------------------------------------
4.3. Cross-firm Variation in the Effect of Trust on Debt Covenants
In this section we shift focus from Level 2 country-level factors (i.e., H2) to Level 3
firm-level factors (i.e., H3) that influence the relationship between trust and the use of restrictive
covenants. To do so, we use three measures to capture the quality of firm-level governance and
information environment: Corporate Governance, a composite index created by ASSET4 that
assesses a company's ability to create incentives as well as checks and balances to enhance long-
term shareholder value; Error, the percentage forecast error, which is equal to the percentage
difference between actual and forecasted earnings per share; and Illiquidity, the price response to
one dollar of trading volume (Amihud, 2002), which is equal to the average of (|r|/Volume),
where r is the daily stock return and Volume is daily trading volume. H3 predicts significant
21
substitution effects between social trust and firm-level corporate governance and information
environment in determining the number (and intensity) of covenants.
Table 5 presents regression results on the influence of firm-level corporate governance
and information asymmetry on the relationship between social trust and the use of debt
covenants. We use Number of Covenants as the dependent variable in Models 2 and 3, and
Intensity of Covenants in Models 5 and 6. In line with H3, the coefficients on the interaction
terms are consistently statistically significant at the 1% level with the expected signs, which
shows that stronger corporate governance and lower information asymmetry can compensate for
low levels of social trust.
------------------------------------------
Insert Table 5 about here
-------------------------------------------
Taken together, the results in Tables 4 and 5 show that country-level institutions and
firm-level governance and transparency can substitute for social trust. These results suggest that
while it may be difficult if not impossible to reduce the use of restrictive covenants by improving
Level 1 social trust within a country, policy makers and firm managers can achieve this aim
through improvements to Level 2 institutions and Level 3 governance structures and information
environment, respectively.
4.4. Endogeneity
Reverse causality is unlikely to explain our results because, as Williamson (2000) argues,
Level 4 outcomes (i.e., the use of debt covenants) are unlikely to influence a Level 1
characteristic (i.e., social trust). It is possible that an omitted variable can simultaneously
influence both social trust and the use of debt covenants, leading to a spurious relation between
the two. However, an omitted variable is unlikely to account for our results since any such
variable would have to explain not only the relationship between social trust and debt covenants,
but also cross-country and cross-firm interaction effects (i.e., substitution mechanisms) on this
relationship, as documented in Sections 4.2 and 4.3. Nevertheless, in this section we conduct a
two-stage least squares (2SLS) analysis to further alleviate concerns about endogeneity that may
undermine the inference of a causal effect of social trust on the use of debt covenants.
22
To capture the exogenous variations in social trust, we use three predetermined
instrumental variables motivated by prior research. The first is Tropical, a dummy variable equal
to 1 if the country is in a tropical climate zone, and 0 otherwise (Beck, Demirgüç-Kunt, and
Levine, 2003). The motivation for this variable is that a country’s climate can affect its level of
social trust. As Bjørnskov and Méon (2013) point out, survival through winters in cold climates
was historically dependent on help from strangers. Therefore, the extension of trust towards
unfamiliar people was an evolutionary strategy in cold countries, such as those in Scandinavia.
The second instrument Genetic Distance measures the genetic distance between the population
of one country and that of the US (where Yankee bonds are issued). This instrument, which is
obtained from Spolaore and Wacziarg (2009), measures the probability that random selection by
two alleles at a given locus from the population of one country and the population of the U.S.
will be different. Genetic distance between two populations is a reflection of their common
linguistic and cultural roots. Therefore, genetically similar countries should exhibit higher levels
of trust (Guiso, Sapienza, and Zingales, 2009). The third instrument is the 2nd Person
Differentiation, a dummy variable equal to 1 if the number of second person pronouns used in
spoken language differs by the type of person with whom one interacts, and 0 otherwise.
Tabellini (2008) argues that the structure of language, in particular its specific grammatical rules,
can exert an independent causal effect on concept formation and cultural traits, including trust
and respect. Specifically, some languages use multiple personal pronouns according to the type
of interpersonal relationship between speakers, whereas others do not differentiate. The
distinction among personal pronouns is associated with a hierarchy of power and social distance,
and therefore reflects less respect for the individual and greater individual mistrust.11
Accordingly, a country exhibiting a greater variation in the use of second person pronouns is
associated with a lower level of trust. In general, we expect trust to be negatively associated with
Tropical, Genetic Distance, and 2nd Person Differentiation.
Table 6 reports the results of 2SLS regressions using these three instruments. In Model 1,
we report results from the first-stage regression in which the dependent variable is Trust. Overall,
we find that our instruments are significantly correlated with Trust, as expected. More
11 In unreported results, we also use another grammar rule suggested by Tabellini (2008) (i.e., the license to drop pronouns) and find that our results are qualitatively similar. In addition, using proxies for religious affiliation as instruments for trust yields similar findings.
23
specifically, Tropical, Genetic Distance, and 2nd Person Differentiation are associated with
lower levels of trust. These results suggest that our instruments satisfy the relevance requirement.
The relevance of our instruments finds further support from the first-stage F-statistic of 59.15,
which is well above the threshold value of 10.
In Models 2 and 3 of Table 6, we report results from second-stage regressions of debt
covenant use on social trust. We find significantly negative coefficients of -2.249 (number of
covenants, Model 2) and -1.728 (intensity of covenants, Model 3) on Trust, thus confirming that
creditors demand fewer restrictive covenants from firms domiciled in countries with a high level
of trust (after controlling for endogeneity). The bottom of Table 6 reports the results of Hansen’s
(1982) over-identification test for the null hypothesis that our instrumental variables are
exogenous. We find that the p-values exceed 10% in both Model 2 and Model 3, which suggests
that the null hypothesis cannot be rejected. In general, our main findings continue to hold after
addressing endogeneity concerns.
------------------------------------------
Insert Table 6 about here
-------------------------------------------
4.5. Out-of-sample Evidence
Although our cross-country setting allows us to examine the effects of social trust on debt
contracting across a wide variety of societies, it is also susceptible to concerns about potential
omitted country-specific traits. We further mitigate this concern by performing an out-of-sample
analysis using covenants in public corporate bonds issued within a single-country setting (i.e., by
U.S. companies operating in the United States). In addition to providing out-of-sample evidence,
the use of U.S. bond covenants helps to alleviate any concerns about omitted country-level
variables since these U.S. firms share a similar institutional (e.g., political, legal, social, and
economic) environment.
Following our main analysis, we use WVS to construct Trust US at the U.S. region level.
Consistent with our baseline model, we control for the same set of issue- and firm-level
variables. We replace the country-level control variables with state-level variables, namely, State
24
GDP (logarithm of the state-level GDP per capita) and State GDP Growth (growth of the state-
level GDP per capita). The state-level GDP information is collected from the U.S. Bureau of
Economic Analysis. Standard errors are clustered at the firm level.
Table 7 reports the results. Reinforcing our main evidence, we find that Trust US loads
negatively and significantly on both the Number of Covenants and the Intensity of Covenants,
thereby alleviating any concerns that omitted country-level characteristics are driving the
documented relation between trust and the use of debt covenants.
------------------------------------------
Insert Table 7 about here
-------------------------------------------
4.6. Robustness Tests
We conduct a series of tests to check the robustness of our main findings. Specifically,
we examine whether our main findings continue to hold after considering alternative proxies for
trust, alternative explanations, alternative samples, and alternative empirical methods.
We begin by examining whether our results are sensitive to using two alternative
measures of Trust. The first measure is based on the formula 100 + (% of WVS participants who
respond “most people can be trusted”) – (% of WVS participants who respond “can’t be too
careful”) (Pevzner, Xie, and Xin, 2015). The second measure is the country-year average of a
rescaled response to the WVS question “Do you trust your neighborhood completely, somewhat,
not very much, or not at all?” The results are reported in Panel A of Table 8. We find that social
trust continues to have a negative and significant effect on both Number of Covenants and
Intensity of Covenants, which demonstrates that our main results are robust to using alternative
proxies for Trust.
Next, we examine whether alternative stories can explain the relation between social trust
and the use of debt covenants. One could argue that trust is a manifestation of cultural values and
that these (non-trust) cultural values are the real factors that drive the relation we document. We
investigate this possibility in Panel B of Table 8, where we add various proxies for (non-trust)
25
cultural values to our baseline regression using Number of Covenants. Our first set of tests
employs Hofstede’s (2001) cultural dimensions (individualism, uncertainty avoidance, power
distance, and masculinity), which are widely used in studies of cross-cultural values (e.g.,
Karolyi, 2016). Models 1 to 4 consider these four dimensions separately, while Model 5 runs a
horserace on all four of Hofstede’s dimensions. We find that the loadings on Hofstede’s (2001)
cultural dimensions are insignificant, suggesting that (non-trust) cultural values as captured by
Hofstede (2001) cannot explain the effect of social trust on the use of debt covenants that we
document. In contrast, Trust continues to load significantly negatively in all specifications, in
line with our earlier results. In Models 6 to 8 we employ the embeddedness and mastery
measures of Schwartz (1994). Model 9 considers the power distance, institutional collectivism,
uncertainty avoidance, and assertiveness measures of House et al. (2004), which are based on the
GLOBE database. In Model 10 we use the tightness measure of Gelfand et al. (2011). Results of
these additional tests are qualitatively similar to our main findings. When we repeat these tests
after replacing Number of Covenants with Intensity of Covenants (Models 11 to 20), we again
find that the results are qualitatively similar to our main findings. We therefore find strong
evidence that (non-trust) cultural values cannot explain the relation between social trust and the
use of debt covenants that we document.12
We also consider alternative samples. Focusing on our initial sample of Yankee bonds,
we note that similar to other studies of Yankee bonds (e.g., Miller and Puthenpurackal, 2002;
Miller and Reisel, 2012), our sample is unbalanced. Some countries account for a large
percentage of our sample, for instance, Canada (29.54%) and the UK (16.81%), while Costa Rica
(0.03%) and Denmark (0.03%) account for a small percentage. To examine whether our main
findings could be driven by sample composition bias, we re-estimate our baseline regressions
after excluding Canadian firms (Models 1 and 5), British firms (Models 2 and 6), firms from
both countries (Models 3 and 7), and firms from countries with no more than five observations
(Models 4 and 8). The results, reported in Panel C of Table 8, continue to show significant
negative loadings on Trust, mitigating concerns about sample composition bias.
12 As additional robustness tests, we control for shareholder protection using the revised anti-director rights index from Djankov et al. (2008b). The unreported results show that the effect of social trust on debt covenants persists when we control for shareholder protection, which is statistically insignificant.
26
Finally, we employ alternative empirical methods. The use of country fixed effects can
mitigate the influence of time-invariant country-level unobserved variables. Our regressions
above do not include country fixed effects, however, because Trust varies little over time. As
Wooldridge (2002, p. 286) points out, fixed effects can generate imprecise estimates for
explanatory variables that are relatively stable over time. Nevertheless, in Models 1 and 3 of
Panel D we re-run our baseline model after including country fixed effects. We find that Trust
loads with a negative and significant coefficient in both models, consistent with our main
findings. We also re-run our baseline specification using a negative binomial regression – an
alternative approach to estimating count variables – rather than a Poisson regression. As can be
seen in Models 2 and 4 of Panel D, the coefficients on Trust are negative and significant at the
1% level, which shows that our main results do not depend on the choice of estimation method.
------------------------------------------
Insert Table 8 about here
-------------------------------------------
4.7. Additional Analysis: The Pricing Effect of Covenants
In this section we test our final hypothesis, H4, which holds that social trust reduces
borrowing costs by reducing agency and information problems (direct effect) and by reducing
the use of debt covenants (indirect effect).
To test H4, we employ the treatment-effects model of Greene (2000) to examine the
pricing effects of Yankee bond covenants and social trust (particularly for low trust countries) on
the issuing firm’s cost of debt. Following Miller and Reisel (2012), our treatment-effects model
comprises a system of two equations: a covenant selection equation (Equation (2)) and a bond
pricing equation (Equation (3)). Note that both covenant selection and bond pricing can be
endogenously determined. The treatment-effects model mitigates this concern by simultaneously
estimating both equations as follows:
27
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝛿𝛿0 + 𝛿𝛿1𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶 + 𝛿𝛿2𝑍𝑍 + 𝐹𝐹𝐹𝐹𝐹𝐹𝐶𝐶𝐹𝐹 𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐸𝐸𝐶𝐶𝐶𝐶 + 𝜑𝜑 (2)
𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝑇𝑇 𝑆𝑆𝑆𝑆𝑇𝑇𝐶𝐶𝐶𝐶𝐹𝐹 = 𝛽𝛽0 + 𝛽𝛽1𝐷𝐷(𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶) + 𝛽𝛽2𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶 +
𝛽𝛽3𝐼𝐼𝐶𝐶𝐹𝐹𝐹𝐹𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝑇𝑇 𝐶𝐶𝐸𝐸 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 × 𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶 + 𝛽𝛽4𝑋𝑋 + 𝐹𝐹𝐹𝐹𝐹𝐹𝐶𝐶𝐹𝐹 𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐸𝐸𝐶𝐶𝐶𝐶 + 𝜀𝜀, (3)
where D(Covenants) is a dummy variable equal to 1 if the debt contract contains a covenant of
any type, and 0 otherwise. In the covenant selection equation (Equation (2)), Z is a vector that
includes all of the control variables (i.e., ILV, FLV, and CLV) in our main model (Equation (1))
as well as year and industry indicators. In the bond pricing equation (Equation (3)), Treasury
Spread, or the cost of debt, is the difference between the bond yield and the benchmark Treasury,
and X is a vector that contains bond, firm, and macroeconomic factors that determine the
Treasury Spread: the previously defined Amount, Maturity, Size, Leverage, Interest Coverage,
Profitability, and Capital Expenditure, as well as Term Spread (the 10-year Treasury rates minus
the 2-year Treasury rate), Credit Spread (the yield spread between Aaa- and Bbb-rated corporate
bonds), Log Exchange Rate Volatility (the natural logarithm of exchange rate volatility), ADR
(an indicator of whether a firm has previously issued an ADR), Secured (an indicator of whether
the security is a secured issue), Subordinated (an indicator of whether the security is a
subordinated issue), Junk (an indicator of whether the security is a junk bond), Sinking Fund (an
indicator of whether the issue has a sinking fund), Callable (an indicator of whether the security
issue is callable), and Gross Spread (the difference between the price received by the underwriter
and the price that investors pay, which captures bond liquidity). Equations (2) and (3) also
control for year and industry effects.
Table 9 presents results of the pricing equation. In Model 1, we first examine the
independent pricing effects of D(Covenants) and Trust. We find that the presence of covenants is
associated with a 55.63 basis point reduction in Treasury Spread. If we increase Trust by one
standard deviation, Treasury Spread decreases by 13.46 basis points. These results confirm our
prediction that restrictive covenants and social trust are priced by investors. When we compare
our results to those of Miller and Reisel (2012),13 who also show that Creditor Rights is priced
by investors, we find that the effect of Creditor Rights is significantly lower when Trust is
included in the pricing effect model. This suggests that the pricing effect of Creditor Rights is
13 Miller and Reisel’s (2012) pricing model (which does not consider Trust) shows that a one-standard-deviation increase in Creditor Rights leads to a 4.41 basis point reduction in Treasury Spread.
28
subsumed by Trust, that is, investors may weigh the level of social trust more heavily than the
extent of legal protections when pricing debt. Thus, consistent with Williamson’s (2002)
hierarchical model, we find that Level 1 social trust has greater influence on Level 4 transaction
prices than does Level 2 creditor rights.
Next, we investigate the interaction effects of D(Covenants) and Trust on the cost of debt.
If both debt covenants and social trust can reduce managerial opportunism, the protection
provided by covenants should be more valuable when the level of social trust is low. Consistent
with this conjecture, Model 2 shows that the coefficient on the interaction between D(Covenants)
and Trust is positive and significant, with an impact level of 161.69. Economically speaking, by
moving from a country with a trust value half a standard deviation above the mean to a country
with a trust value half a standard deviation below the mean, the presence of covenants can reduce
the cost of debt by 19.31 basis points. In extreme cases such as Norway (the country with highest
level of trust in our sample), the presence of covenants gives investors no incentive to reduce the
cost of debt, while in the Philippines (the country with lowest level of trust in our sample), the
presence of covenants can reduce the cost of debt by 101.06 basis points.
Overall, the results in this section improve our understanding of the pricing of debt
contracts. High levels of social trust at the country level, as well as the use of debt covenants at
the firm level, can reduce a firm’s cost of debt. More importantly, we show that the role of debt
covenants in reducing the cost of debt is more important when the issuing firm is located in a
country with a low level of social trust, that is, we show that social trust and debt covenants act
as substitutes in assuring the firm’s creditors.
------------------------------------------
Insert Table 9 about here
-------------------------------------------
5. Conclusion
While previous studies on debt contracting highlight the role of formal institutions (e.g.,
creditor rights), the role of informal institutions has received little attention. Using Williamson’s
29
(2000) four levels of social analysis, we examine whether informal institutions such as social
trust (a Level 1 deep-rooted cultural characteristic) can directly influence the issuing firm’s use
of bond covenants and borrowing costs (transaction-level quantities and prices at Level 4). We
also examine whether social trust can indirectly influence the use of covenants and borrowing
costs through its impact on formal institutions (Level 2) and governance structures (Level 3).
Building on extant research suggesting that social trust is a fundamental determinant of
economic outcomes (Arrow, 1972; Ahern, Daminelli, and Fracassi, 2015; Guiso, Sapienza, and
Zingales, 2008, 2009), we posit that a high level of social trust reduces borrower opportunism
and in turn creditor demand for restrictive covenants. Using a sample of 1,033 Yankee bond
issues from 34 countries from 1989 to 2014, we find that firms from high trust countries issue
bonds with fewer bond covenants. The effect of social trust on the use of debt covenants is
moderated, however, by the presence of strong institutions (Level 2) and strong firm-level
governance mechanisms and information transparency (Level 3), suggesting that these factors
can substitute for social trust in influencing debt contracting. These results are robust to a series
of robustness tests to alleviate concerns about potential endogeneity, omitted variable bias, and
the use of an unbalanced sample. The results are also robust to using alternative proxies for
social trust and alternative model specifications. Finally, we examine the direct and indirect
pricing effects of social trust and find that high social trust is related to significantly lower
borrowing costs, but this effect is reduced in the presence of covenants. We conclude that issue-
level debt covenants act as a substitute for social trust in countries where the level of trust is low.
Taken together, our findings lend strong support to the argument that informal
institutions can have significant effects on economic outcomes. These effects can be both direct
(i.e., from Level 1 to Level 4) and indirect (i.e., from Level 1 to Level 4 through Levels 2 and 3).
In particular, our results show that a deep-rooted cultural characteristic – social trust – plays a
significant role in debt contracting over and above the role played by formal institutions. Our
study thus contributes to the contracting literature by identifying social trust as an important
determinant of bond covenant use and bond contract pricing. In addition, our study addresses a
common challenge faced by the culture and finance literature, namely, the difficulty of
disentangling the direct and indirect effects of a country’s cultural characteristics (Aggarwal et
al., 2016) on economic outcomes. By using the setting of the Yankee bond market, we are able to
30
identify both direct and indirect effects of social trust on debt contracting. Overall, we provide a
comprehensive picture of the multilayered connections running from social trust to bond
covenants, incorporating mechanisms from all four levels of Williamson’s (2000) social analysis.
31
Appendix. Variable Description
Variable Definition Source: Authors’ calculations based on
Panel A. Trust variables Trust mean percentage of response to the WVS question: Generally speaking, would you say that most people
can be trusted or that you need to be very careful in dealing with people? WVS
Trust (Alternative) two alternative trust variables are adopted. The first alternative trust index is calculated as 100% + (% most people can be trusted) – (% can’t be too careful). The second alternative trust index is the country-year average of rescaled response to the WVS question: Do you trust your neighborhood completely, somewhat, not very much, or not at all?
As above
Trust US mean U.S. region-level percentage of response to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?
As above
Panel B. Covenant variables Number of Covenants the total number of restrictive covenants. The covenants are grouped into three major categories, namely
restrictions on a firm’s financing, investment, and payout activities. FISD
Restrictions on Financing the total number of restrictive covenants on financing activities. This category contains seven covenant terms: (1) the issuer is not allowed to issue secured debt unless it secures the current issue on a pari passu basis; (2) the issuer is not allowed to issue debt with initial maturity of one year or longer; (3) the issuer is not allowed to incur additional debt with limits on the absolute dollar amount of debt outstanding or the percentage of total capital; (4) the issuer is restricted on the amount of senior debt; (5) the issuer is not allowed to issue subordinated debt; (6) the issuer is not allowed to issue additional debt unless the issuer achieves or maintains a certain profitability level; and (7) the issuer is restricted on sale-leaseback transactions, which involve raising capital by selling an asset and simultaneously leasing the asset back for a fixed term at an agreed-upon rate.
As above
Restrictions on Investment the total number of restrictive covenants on investment activities. This category contains four covenant terms: (1) the issuer is restricted on investment policy to reduce risky investments; (2) the issuer is restricted on mergers or consolidations; (3) the issuer is restricted on selling its assets; and (4) the issuer is required to use proceeds from the sale of assets to redeem bonds at par or a premium.
As above
Restrictions on Payouts the total number of restrictive covenants on payout activities. This category contains two covenant terms: (1) the issuer is not allowed to make dividend-related payments to shareholders or other entities above a certain percentage of net income or some other accounting ratio; and (2) the issuer is restricted on non-dividend-related payments to shareholders or other entities.
As above
32
Intensity of Covenants sum of the following three covenant indicators. Indicator of financing covenants is a dummy equal to 1 if the debt contract contains a covenant of financing activities, and 0 otherwise. Indicator of investment covenants is a dummy equal to 1 if the debt contract contains a covenant of investment activities, and 0 otherwise. Indicator of payout covenants is a dummy equal to 1 if the debt contract contains a covenant of payout activities, and 0 otherwise.
As above
D(Covenants) a dummy variable equal to 1 if the debt contract contains a covenant of any type, 0 otherwise As above Panel C. Other variables
Amount natural logarithm of the bond issue’s offering amount, expressed in US dollars FISD Maturity the maturity of the bond issue in days, namely number of days between the maturity date and the offering
date As above
Size natural logarithm of total assets Compustat Leverage total debt over total assets As above Interest Coverage earnings before interests and taxes over interest expenses As above Profitability earnings before interests and taxes over total assets As above Capital Expenditure capital expenditures over total assets As above Tangibility equals (0.715*receivables+0.547* inventories+cash+0.535*net property, plant, and equipment) / total
assets Berger, Ofek, and Swary (1996)
GDP logarithm of GDP per capita WDI GDP Growth growth of GDP per capita (%) As above State GDP logarithm of the U.S. state-level GDP per capita the U.S. Bureau of
Economic Analysis State GDP Growth growth of the U.S. state-level GDP per capita (%) As above Financial Development private credit to GDP As above Inflation annual inflation rate (%) As above Creditor Rights creditor rights index, the sum of four binary variables. A score of 1 is assigned if an automatic stay on the
assets of the firm is not imposed upon filing the reorganization petition (No Automatic Stay), if secured creditors are ranked first in the distribution of the proceeds payment as opposed to government or workers (Secured Creditor Paid First), if the reorganization procedure imposes restrictions, such as creditors' consent or minimum dividends for a debtor to be able to file for reorganization (Restrictions on Reorganization), and if management does not retain administration of its property pending the resolution of the reorganization process (No Management Stay). This index ranges from 0 to 4.
Djankov et al. (2007)
Rule of Law an index capturing perceptions of the degree to which people have confidence in and abide by the rules of society, particularly the quality of contract enforcement, property rights, the police and the courts, as well as the likelihood of crime and violence.
WGI; Kaufmann et al. (2009)
Control of Corruption an index capturing perceptions of the degree to which public power is exercised for private interests, including petty and grand forms of corruption, as well as “capture” of the state by elites.
As above
33
Debt Recovery the recovery rate for the secured creditor, or (100*GC+70*(1-GC)-12*(P-1)-100*c) / (1+r)^t. GC is equal to 1 if Mirage continues as a going concern and 0 otherwise. P is the order of priority in which claims are paid. P=1 if the secured creditor is paid first out of the proceeds from the insolvency proceeding. P=2 if the secured creditor is ranked second after another claimant group. P=3 if it is ranked third. P=4 if it is ranked fourth. c denotes COST, represented by the estimated cost of the insolvency proceeding for Mirage, reported as a percentage of the value of the insolvency estate. t is TIME, represented by the estimated duration from the moment of Mirage's default to the point at which the secured creditor receives payment. r is the nominal lending rate.
Djankov et al. (2008b)
Efficiency of Debt Enforcement the present value of the terminal value of the firm after bankruptcy costs, or (100*GC+70*(1-GC)-100*c) / (1+r)^t. GC is equal to 1 if Mirage continues is a going concern and 0 otherwise, c denotes COST discussed above, t is TIME discussed above, and r is the nominal lending rate.
As above
Efficient Judiciary logarithm of the number of days needed for the judiciary to collect on a bounced check Djankov et al. (2008a)
Intragroup Tax a country’s tax law regulating intra-group transactions. The measure is a sum of four indicators: 1) whether the law regulates transfer pricing, 2) whether the law regulates the use of thin-capitalization companies, 3) whether the law regulates registration of holding companies in tax havens, 4) whether there are explicit reporting requirements.
Deloitte International Taxation Guide
Consolidation a dummy variable equal to 1 if a parent firm can consolidate its subsidiary with an ownership stake of less than 90% and 0 otherwise
PwC Corporate Taxes: A Worldwide Summary
Prevalence of Family Group the percentage of total market capitalization held by firms that belong to a family group Masulis et al. (2011)
Prevalence of Business Group the percentage of total market capitalization held by firms that belong to a business group, including family and non-family business groups
As above
Press Freedom an index that assesses freedom of the press. The lower value of the initial raw score represents a higher level of press freedom. We rescale it using 100 subtracting the raw score so that a higher score represents a higher level of press freedom.
Freedom House
Voice and Accountability measuring the degree to which a country’s citizens perceive themselves being able to participate in selecting their government, as well as freedom of expression, freedom of association, and free media.
WGI; Kaufmann et al. (2009)
Political Stability measuring the likelihood that the government is perceived to be destabilized or overthrown by unconstitutional or violent means, e.g., politically motivated violence and terrorism.
As above
Government Effectiveness measuring the degree to which government effectiveness is perceived, including 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.
As above
Corporate Governance measuring a company's systems and processes that ensure that its board members and executives act in the best interests of its long-term shareholders. It represents a company's ability, through its use of best
ASSET4
34
management practices, to direct and control its rights and responsibilities by creating incentives, as well as checks and balances in order to generate long-term shareholder value.
Error percentage forecast error, measured as the percentage difference between actual and forecasted earnings per share
I/B/E/S
Illiquidity a measure of illiquidity, which is equal to average (|r|/Volume). The measure is calculated over all positive-volume days, since the ratio is undefined for zero-volume days.
Amihud (2002)
Tropical a dummy variable equal to 1 if the country is in a tropical climate zone, and 0 otherwise. Beck, Demirgüç-Kunt, and Levine (2003)
Genetic Distance the genetic distance between the population of one country and that of the U.S. The measure gauges the probability that random selection by two alleles at a given locus from the population of one country and the population of the U.S. will be different
Spolaore and Wacziarg (2009)
2nd Person Differentiation a dummy variable equal to 1 if the number of second person pronouns used in spoken language differs by the type of person with whom one interacts, and 0 otherwise.
Kashima and Kashima (1998)
Individualism cultural index on individualism scaled by 100 Hofstede (2001)
Uncertainty Avoidance cultural index on uncertainty avoidance scaled by 100 As above
Power Distance cultural index on power distance scaled by 100 As above
Masculinity cultural index on masculinity scaled by 100 As above
Embeddedness cultural index on embeddedness Schwartz (1994)
Mastery cultural index on mastery As above
Collectivism_Globe cultural index on institutional collectivism from the GLOBE study House et al. (2004)
Uncertainty Avoidance_Globe cultural index on uncertainty avoidance from the GLOBE study As above
Power Distance_Globe cultural index on power distance from the GLOBE study As above
Assertiveness_Globe cultural index on assertiveness from the GLOBE study As above
Tightness cultural index on tightness Gelfand et al. (2011)
Treasury Spread the difference between the yield of the bond and benchmark Treasury FISD Term Spread the 10-year Treasury rates minus the 2-year Treasury rates Federal Reserve
database Credit Spread (AAA-BAA) the yield spread between Aaa- and Bbb-rated bonds As above Log Exchange Rate Volatility natural logarithm of exchange rate volatility Pacific Exchange
Rate Service
35
ADR a dummy variable equal to 1 if a firm has previously issued an ADR, and 0 otherwise. Compustat
Secured a dummy variable equal to 1 if the security is a secured issue of the issuer, 0 otherwise FISD Subordinated a dummy variable equal to 1 if the security is a subordinated issue of the issuer, 0 otherwise As above Junk a dummy variable equal to 1 if the security is a junk bond, 0 otherwise As above Sinking Fund a dummy variable equal to 1 if the security issue has a sinking fund, 0 otherwise As above Callable a dummy variable equal to 1 if the security issue is callable, 0 otherwise As above Gross Spread the difference between the price received by the underwriter and the price that investors pay, which
captures bond liquidity As above
36
References
Aggarwal, R., Faccio, M., Guedhami, O., Kwok, C., 2016. Culture and finance: An introduction. Journal of Corporate Finance 41, 466-474. Ahern, K., Daminelli, D., Fracassi, C., 2015. Lost in translation? The effect of cultural values on mergers around the world. Journal of Financial Economics 117(1), 165-189. Amihud, Y., 2002. Illiquidity and stock returns: Cross-section and time-series effects. Journal of Financial Markets 5(1), 31-56. Amiraslani, H., Lins, K. V., Servaes, H., Tamayo, A., 2017. A matter of trust? The bond market benefits of corporate social capital during the financial crisis. Available at SSRN: https://ssrn.com/abstract=2978794. Anderson, C. W., Fedenia, M., Hirschey, M., Skiba, H., 2011. Cultural influences on home bias and international diversification by institutional investors. Journal of Banking & Finance 35, 916–934. Arrow, K. J., 1972. Gifts and exchanges. Philosophy & Public Affairs 1(4), 343-362. Bae, K. H., Goyal, V. K. 2009. Creditor rights, enforcement, and bank loans. Journal of Finance 64(2), 823-860. Bae, K. H., Kang, J. K., Kim, J., 2002. Tunneling or value added? Evidence from mergers by Korean business groups. Journal of Finance 57(6), 2695-2740. Baek, J. S., Kang, J. K., Lee, I., 2006. Business groups and tunneling: Evidence from private securities offerings by Korean chaebols. Journal of Finance 61(5), 2415-2449. Barclay, M. J., Smith, C. W. Jr., 1995. The priority structure of corporate liabilities. Journal of Finance 50(3), 899-916. Beck, T., Demirgüç-Kunt, A., Levine, R., 2003. Law and finance: why does legal origin matter? Journal of Comparative Economics 31(4), 653-675. Berger, P. G., Ofek, E., Swary, I., 1996. Investor valuation of the abandonment option. Journal of Financial Economics 42(2), 257-287. Bertrand, M., Mehta, P., Mullainathan, S., 2002. Ferreting out tunneling: An application to Indian business groups. Quarterly Journal of Economics 117(1), 121-148. Bjørnskov, C., 2018. Social trust and economic growth. In: Uslaner, E. M. (Ed.), Oxford Handbook of Social and Political Trust. Available at SSRN: https://ssrn.com/abstract=2906280. Bjørnskov, C., Méon, P. G., 2013. Is trust the missing root of institutions, education, and development? Public Choice 157 (3-4), 641-669. Bottazzi, L., Da Rin, M., Hellmann, T., 2016. Editor's choice: The importance of trust for investment: Evidence from venture capital. Review of Financial Studies 29(9), 2283-2318. Boubakri, N., Cosset, J. C., Saffar, W., 2008. Political connections of newly privatized firms. Journal of Corporate Finance 14(5), 654-673. Boubakri, N., Cosset, J. C., Saffar, W., 2013. The role of state and foreign owners in corporate risk-taking: Evidence from privatization. Journal of Financial Economics 108(3), 641-658.
37
Boubakri, N., Ghouma, H., 2010. Control/ownership structure, creditor rights protection, and the cost of debt financing: International evidence. Journal of Banking and Finance 34(10), 2481-2499. Campbell, J. Y., 1996. Understanding risk and return. Journal of Political Economy 104(2), 298-345. Chava, S., Kumar, P., Warga, A., 2010. Managerial agency and bond covenants. Review of Financial Studies 23(3), 1120-1148. Chava, S., Livdan, D., Purnanandam, A., 2009. Do shareholder rights affect the cost of bank loans? Review of Financial Studies 22(8), 2973-3004. Chen, R., El Ghoul, S., Guedhami, O., Nash, R., 2017. State ownership and corporate cash holdings. Journal of Financial and Quantitative Analysis, forthcoming. Coleman, J., 1988. Social capital in the creation of human capital. American Journal of Sociology 94, 95-120. Djankov, S., Hart, O., McLiesh, C., Shleifer, A., 2008a. Debt enforcement around the world. Journal of Political Economy 116(6), 1105-1149. Djankov, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2008b. The law and economics of self-dealing. Journal of Financial Economics 88(3), 430-465. Djankov, S., McLiesh, C., Shleifer, A., 2007. Private credit in 129 countries. Journal of Financial Economics 84(2), 299-329. Dyck, A., Morse, A., Zingales, L., 2010. Who blows the whistle on corporate fraud? Journal of Finance 65(6), 2213-2253. Dyck, A., Volchkova, N., Zingales, L., 2008. The corporate governance role of the media: Evidence from Russia. Journal of Finance 63(3), 1093-1135. Dyck, A., Zingales, L., 2004. Private benefits of control: An international comparison. Journal of Finance 59(2), 537-600. Esty, B. C., Megginson, W. L., 2003. Creditor rights, enforcement, and debt ownership structure: Evidence from the global syndicated loan market. Journal of Financial and Quantitative Analysis 38(1), 37-59. Faccio, M., Masulis, R. W., McConnell, J., 2006. Political connections and corporate bailouts. Journal of Finance 61(6), 2597-2635. Fukuyama, F., 1995. Trust: The Social Virtues and the Creation of Prosperity. Free Press, New York. Gârleanu, N., Zwiebel, J., 2009. Design and renegotiation of debt covenants. Review of Financial Studies 22(2), 749-781. Gelfand, M. J., Raver, J. L., Nishii, L., Leslie, L. M., Lun, J., Lim, B. C., ... Aycan, Z., 2011. Differences between tight and loose cultures: A 33-nation study. Science 332(6033), 1100-1104. Greene, W., 2000. Econometric Analysis. Prentice Hall, Upper Saddle River, NJ. Guiso, L., Sapienza, P., Zingales, L., 2003. People's opium? Religion and economic attitudes. Journal of Monetary Economics 50(1), 225-282.
38
Guiso, L., Sapienza, P., Zingales, L., 2006. Does culture affect economic outcomes? Journal of Economic Perspectives 20(2), 23-48. Guiso, L., Sapienza, P., Zingales, L., 2008. Trusting the stock market. Journal of Finance 63(6), 2557-2600. Guiso, L., Sapienza, P., Zingales, L., 2009. Cultural biases in economic exchange? Quarterly Journal of Economics 124(3), 1095-1131. Hansen, L. P., 1982. Large sample properties of generalized method of moments estimators. Econometrica 50, 1029-1054. Hasan, I., Hoi, C. K., Wu, Q., Zhang, H., 2017. Social capital and debt contracting: Evidence from bank loans and public bonds. Journal of Financial and Quantitative Analysis 52(3), 1-31. Hilary, G., Huang, S., 2016. Trust and contracting. INSEAD Working Paper No. 2015/42/ACC. Available at SSRN: https://ssrn.com/abstract=2604974. Hofstede, G. H. (Ed.), 2001. Culture's Consequences: Comparing Values, Behaviors, Institutions, and Organizations Across Nations. Sage, Thousand Oaks, CA. House, R. J., Hanges, P. J., Javidan, M., Dorfman, P. W., Gupta, V. (Eds.), 2004. Culture, Leadership, and Organizations: The GLOBE study of 62 societies. Sage, Thousand Oaks, CA. Jensen, M. C., Meckling, W. H., 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3(4), 305-360. Johnson, S., Boone, P., Breach, A., Friedman, E., 2000. Corporate governance in the Asian financial crisis. Journal of Financial Economics 58(1), 141-186. Karolyi, G. A., 2016. The gravity of culture for finance. Journal of Corporate Finance 41 610-625. Karolyi, G. A., Johnston, R., 1998. Hutchison Whampoa’s Yankee Bond Issue. Ivey Publishing, Case No. 9A98N023. Kashima, E. S., Kashima, Y., 1998. Culture and language: The case of cultural dimensions and personal pronoun use. Journal of Cross-Cultural Psychology 29(3), 461-486. Kaufmann, D., Kraay, A., Mastruzzi, M., 2009. What do the worldwide governance indicators measure? Viewed online: http://info.worldbank.org/governance/wgi/#home. Knack, S., Keefer, P., 1997. Does social capital have an economic pay-off? Quarterly Journal of Economics 112(4), 1251-1288. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1997. Legal determinants of external finance. Journal of Finance 52(3), 1131-1150. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. Journal of Political Economy 106(6), 1113-1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 1999. The quality of government. Journal of Law, Economics, and Organization 15(1), 222–279. Levine, R., Lin, C., Xie, W., 2017. Corporate resilience to banking crises: The roles of trust and trade credit. Journal of Financial and Quantitative Analysis, forthcoming.
39
Lin, C., Ma, Y., Malatesta, P., Xuan, Y., 2011. Ownership structure and the cost of corporate borrowing. Journal of Financial Economics 100(1), 1-23. Lins, K. V., Servaes, H., Tamayo, A., 2017. Social capital, trust, and firm performance: The value of corporate social responsibility during the financial crisis. Journal of Finance 72(4), 1785-1824. Mansi, S., Qi, Y., Wald, J. K., 2013. Debt covenants, bankruptcy risk, and issuance costs. Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1786709. Masulis, R. W., Pham, P. K., Zein, J., 2011. Family business groups around the world: Financing advantages, control motivations, and organizational choices. Review of Financial Studies 24(11), 3556-3600. Megginson, W. L., Netter, J. M., 2001. From state to market: A survey of empirical studies on privatization. Journal of Economic Literature 39(2), 321-389. Mill, J. S., 1848. Principles of Political Economy with some of their Applications to Social Philosophy. John W. Parker, London. Miller, G., 2006. The press as a watchdog for accounting fraud. Journal of Accounting Research 44(5), 1001-1033. Miller, D. P., Puthenpurackal, J. J., 2002. The costs, wealth effects, and determinants of international capital raising: Evidence from public Yankee bonds. Journal of Financial Intermediation 11(4), 455-485. Miller, D., Reisel, N., 2012. Do country-level investor protections impact security-level contract design? Evidence from foreign bond covenants. Review of Financial Studies 25(2), 408-438. Myers, S. C., 1977. Determinants of corporate borrowing. Journal of Financial Economics 5(2), 147-175. Pevzner, M., Xie, F., Xin, X., 2015. When firms talk, do investors listen? The role of trust in stock market reactions to corporate earnings announcements. Journal of Financial Economics 117(1), 190-223. Putnam, R. D., Leonardi, R., Nanetti, R. Y., 1993. Making Democracy Work: Civic Institutions in Modern Italy. Princeton University Press, Princeton, NJ. Qi, Y., Roth, L., Wald, J. K., 2011. How legal environments affect the use of bond covenants. Journal of International Business Studies 42(2), 235-262. Qian, J., Strahan, P. E., 2007. How laws and institutions shape financial contracts: The case of bank loans. Journal of Finance 62(6), 2803-2834. Reisel, N., 2014. On the value of restrictive covenants: Empirical investigation of public bond issues. Journal of Corporate Finance 27, 251-268. Schwartz, S.H., 1994. Beyond individualism/collectivism: New cultural dimensions of values. In: Kim, U., Triandis, H. C., Kagitcibasi, C., Chol, S.-C., Yoon, G. (Eds.), Individualism and Collectivism: Theory, Method, and Applications. Sage, Thousand Oaks, CA. Shleifer, A., Vishny, R. 1997. A survey of corporate governance. Journal of Finance 52(2), 737-783.
40
Smith, A., 1763. Lectures on Jurisprudence. Oxford University Press. Smith, C. W. Jr., Warner, J. B., 1979. On financial contracting: An analysis of bond covenants. Journal of Financial Economics 7(2), 117-161. Spolaore, E., Wacziarg, R., 2009. The diffusion of development. Quarterly Journal of Economics 124(2), 469-529. Stulz, R., Williamson, R., 2003. Culture, openness, and finance. Journal of Financial Economics 70(3), 313-349. Tabellini, G., 2008. Institutions and culture. Journal of the European Economic Association Papers and Proceedings 6 (2-3), 255-294
Williamson, O. E., 2000. The new institutional economics: Taking stock, looking ahead. Journal of Economic Literature 38(3), 595-613.
Wooldridge, J. M., 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press, Cambridge, MA.
Zak, P., Knack, S., 2001. Trust and growth. Economic Journal 111(470), 295-321.
41
Table 1. Descriptive statistics for bond issues
Panel A. Bond characteristics and frequency by time period Initial sample (N=3,064) Compustat subsample (N=1,650)
Mean Median Std. dev. Mean Median Std. dev. Offering amount (million) 573.93 350.00 692.70 628.61 450.00 736.47 Offering yield (%) 6.50 6.46 2.88 6.07 6.08 2.73 Number of covenants per bond 3.14 3.00 2.29 2.96 3.00 2.10 Percentage of transactions: Senior secured 5.61
3.21
Senior 84.11
85.76 Subordinated 8.06
8.67
Callable 59.79
62.79 Low maturity (< 5 years) 11.95
12.79
High maturity (> 15 years) 14.62
17.15 1989–1997 22.29
18.55
1998–2006 41.55
38.79 2007–2014 36.19
42.67
Panel B. Frequency by country, initial sample
Number of Observations Number of Covenants Intensity of
Covenants Trust
No. Percentage Total Finance Invest Payout Total Argentina 39 1.27 5.00 2.10 2.31 0.59 2.33 0.17 Australia 129 4.21 2.29 0.83 1.39 0.08 1.26 0.48 Austria 15 0.49 1.13 0.47 0.53 0.13 0.47 . Belgium 9 0.29 4.22 1.33 2.22 0.67 2.22 . Brazil 99 3.23 2.96 1.25 1.51 0.20 1.68 0.07 Canada 905 29.54 3.51 1.38 1.82 0.32 1.82 0.39 Chile 35 1.14 3.71 1.69 1.97 0.06 1.86 0.20 China 43 1.40 1.49 0.47 0.95 0.07 0.93 0.56 Colombia 18 0.59 2.78 0.89 1.72 0.17 1.61 0.08
42
Costa Rica 1 0.03 0.00 0.00 0.00 0.00 0.00 . Croatia 3 0.10 0.67 0.67 0.00 0.00 0.67 0.23 Denmark 1 0.03 1.00 1.00 0.00 0.00 1.00 . Ecuador 1 0.03 7.00 3.00 3.00 1.00 3.00 0.07 Finland 5 0.16 3.20 1.60 1.60 0.00 1.60 0.52 France 117 3.82 2.83 1.14 1.48 0.21 1.57 0.19 Germany 102 3.33 1.65 0.66 0.76 0.23 0.81 0.33 Greece 20 0.65 3.20 1.20 1.50 0.50 1.50 . Guatemala 1 0.03 0.00 0.00 0.00 0.00 0.00 0.16 Hong Kong 25 0.82 3.60 1.60 1.72 0.28 1.72 0.40 Hungary 1 0.03 6.00 2.00 3.00 1.00 3.00 0.22 India 16 0.52 1.75 0.75 1.00 0.00 1.25 0.30 Indonesia 23 0.75 4.70 1.83 2.13 0.74 2.22 0.44 Ireland 47 1.53 3.49 1.26 1.79 0.45 1.77 . Israel 13 0.42 2.15 0.46 1.54 0.15 1.00 0.23 Italy 61 1.99 1.72 0.89 0.75 0.08 1.23 0.27 Japan 41 1.34 0.88 0.34 0.51 0.02 0.63 0.37 Kazakhstan 2 0.07 1.00 0.50 0.00 0.50 1.00 0.39 Korea, Rep. 62 2.02 2.42 1.26 1.05 0.11 1.40 0.30 Kuwait 2 0.07 0.00 0.00 0.00 0.00 0.00 0.29 Malaysia 14 0.46 2.79 1.50 1.29 0.00 1.64 0.09 Mexico 146 4.77 4.30 1.83 1.96 0.51 2.10 0.21 Netherlands 260 8.49 3.23 1.21 1.75 0.27 1.75 0.47 New Zealand 7 0.23 3.00 1.29 1.71 0.00 1.57 0.47 Norway 46 1.50 3.72 1.83 1.78 0.11 1.93 0.69 Panama 27 0.88 2.04 0.33 1.70 0.00 1.07 . Peru 8 0.26 3.25 1.75 1.50 0.00 1.50 0.08 Philippines 24 0.78 3.25 1.29 1.42 0.54 2.04 0.06 Poland 11 0.36 5.73 2.09 3.00 0.64 2.55 0.17 Portugal 1 0.03 1.00 1.00 0.00 0.00 1.00 . Romania 2 0.07 0.50 0.50 0.00 0.00 0.50 0.18 Russian Federation 5 0.16 2.00 1.20 0.80 0.00 1.20 0.24 Singapore 30 0.98 3.90 1.27 2.07 0.57 1.90 0.19
43
South Africa 2 0.07 2.00 0.00 2.00 0.00 1.00 0.14 Spain 37 1.21 2.51 0.97 1.46 0.08 1.65 0.25 Sweden 23 0.75 4.35 1.70 2.22 0.43 2.00 0.61 Switzerland 20 0.65 3.45 1.40 1.90 0.15 1.70 0.45 Taiwan, China 14 0.46 2.14 0.86 1.29 0.00 1.50 0.24 Thailand 11 0.36 1.64 0.91 0.73 0.00 1.18 0.41 Turkey 2 0.07 2.50 1.50 1.00 0.00 1.50 0.12 United Kingdom 515 16.81 3.21 1.12 1.79 0.29 1.64 0.29 Uruguay 2 0.07 1.00 1.00 0.00 0.00 1.00 0.21 Venezuela, RB 21 0.69 1.00 1.00 0.00 0.00 1.00 0.15
Panel C. Firm and country characteristics for the Compustat subsample Mean Median Std. dev. N Size 9.63 9.68 2.19 1650 Leverage 0.35 0.30 0.37 1649 Interest Coverage 4.20 2.55 69.29 1626 Profitability 0.10 0.10 0.14 1618 Capital Expenditure 0.09 0.08 0.07 1412 Tangibility 0.45 0.47 0.14 1423 Trust 0.35 0.37 0.13 1624 GDP 31,593.67 28,620.41 16,525.39 1611 GDP Growth (%) 1.57 1.55 2.40 1611 Financial Development 132.81 128.72 54.90 1331 Inflation 4.08 2.17 49.91 1611 Creditor Rights 1.93 1.00 1.36 1650 Rule of Law 1.36 1.68 0.68 1650 Control of Corruption 1.77 2.23 0.85 1650
The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD) between 1989 and 2014. The Compustat subsample includes bonds from issuing companies with non-missing total assets in Compustat. Offering amount and Size are denominated in U.S. dollars. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm’s financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Detailed variable definitions are provided in the Appendix.
44
Table 2. Correlations
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 1. 1.00 2. 0.88 1.00 3. 0.86 0.57 1.00 4. 0.74 0.58 0.51 1.00 5. 0.95 0.82 0.84 0.68 1.00 6. -0.07 -0.07 -0.07 -0.03 -0.07 1.00 7. -0.30 -0.28 -0.20 -0.29 -0.26 -0.03 1.00 8. -0.03 -0.04 0.04 -0.15 -0.02 0.03 -0.06 1.00 9. -0.47 -0.45 -0.32 -0.47 -0.42 -0.03 0.64 -0.01 1.00
10. 0.39 0.34 0.28 0.42 0.39 -0.01 -0.24 -0.01 -0.48 1.00 11. -0.19 -0.18 -0.13 -0.18 -0.18 0.05 0.25 -0.03 0.31 -0.34 1.00 12. -0.20 -0.11 -0.17 -0.32 -0.18 -0.07 0.24 0.07 0.30 -0.24 0.53 1.00 13. 0.13 0.08 0.14 0.10 0.16 0.04 -0.09 0.08 -0.16 0.13 -0.05 -0.05 1.00 14. 0.21 0.24 0.15 0.09 0.23 0.03 -0.24 0.08 -0.36 0.10 0.06 0.04 0.33 1.00 15. -0.31 -0.33 -0.25 -0.15 -0.31 0.52 0.27 -0.09 0.36 -0.25 0.21 0.07 -0.09 -0.28 1.00 16. 0.29 0.25 0.25 0.19 0.28 -0.12 -0.17 0.07 -0.26 0.17 -0.06 -0.05 0.12 0.13 -0.35 1.00 17. -0.27 -0.32 -0.18 -0.12 -0.27 0.41 0.20 -0.07 0.27 -0.17 0.11 -0.06 -0.11 -0.25 0.71 -0.37 1.00 18. 0.08 0.12 0.02 0.07 0.06 -0.16 -0.07 -0.06 -0.08 0.03 0.01 0.09 -0.00 0.05 -0.35 0.18 -0.43 1.00 19. -0.19 -0.22 -0.12 -0.11 -0.19 0.20 0.18 -0.10 0.27 -0.04 0.07 -0.07 -0.02 -0.19 0.32 -0.05 0.40 -0.26 1.00 20. -0.05 -0.10 0.01 -0.01 -0.03 0.57 0.03 0.05 0.06 -0.01 -0.06 -0.21 -0.03 -0.14 0.65 -0.25 0.65 -0.60 0.38 1.00 21. -0.03 -0.10 0.03 0.00 -0.02 0.59 0.01 0.05 0.03 -0.00 -0.08 -0.22 -0.04 -0.13 0.63 -0.24 0.65 -0.57 0.40 0.96 1.00
This table reports pairwise correlations for the key variables in the main regression. Correlation coefficients depicted in bold are significant at the 10% level or better. 1=Number of Covenants, 2=Restrictions on Financing, 3=Restrictions on Investments, 4=Restrictions on Payouts, 5=Intensity of Covenants, 6=Trust, 7=Amount, 8=Maturity, 9=Size, 10=Leverage, 11=Interest Coverage, 12=Profitability, 13=Capital Expenditure, 14=Tangibility, 15=GDP, 16=GDP Growth, 17=Financial Development, 18=Inflation, 19=Creditor Rights, 20=Rule of Law, 21=Control of Corruption. Detailed variable definitions are provided in the Appendix.
45
Table 3. Main evidence: effect of trust on the use of debt covenants
s
Number of Covenants Intensity of
Covenants
All All Finance Invest Payout All
(1) (2) (3) (4) (5) (6)
Trust -0.396** -0.741*** -0.502 -0.761*** -2.59** -0.583***
(-2.73) (-3.33) (-1.61) (-3.68) (-2.09) (-3.11)
Issue-level controls
Amount 0.003 0.009 0.007 -0.027 0.008 (0.27) (0.69) (0.39) (-1.40) (0.55)
Maturity -0.000* -0.000 -0.000 -0.000*** -0.000* (-1.73) (-1.01) (-1.38) (-7.56) (-1.71)
Firm-level controls
Size -0.076*** -0.084** -0.030 -0.356*** -0.031 (-2.95) (-1.99) (-1.58) (-3.35) (-1.17)
Leverage 0.618*** 0.575*** 0.405*** 1.906*** 0.662*** (4.43) (5.28) (2.73) (3.38) (5.65)
Interest Coverage 0.003 -0.000 0.004* 0.010** 0.003 (1.06) (-0.12) (1.86) (1.96) (1.17)
Profitability -0.216 0.580 -0.363 -2.241*** -0.207 (-0.71) (1.63) (-1.12) (-3.02) (-0.83)
Capital Expenditure -0.279 -0.472 0.007 -0.698 0.050 (-1.03) (-0.94) (0.05) (-0.93) (0.22)
Tangibility 0.334** 0.606** 0.193 -0.174 0.407*** (2.01) (2.15) (1.40) (-0.60) (2.60)
Country-level controls GDP 0.000 0.000* 0.000 0.000 0.000**
(1.41) (1.67) (0.52) (0.28) (1.97)
GDP Growth 0.036** 0.034* 0.034** 0.060* 0.026* (2.27) (1.67) (2.28) (1.85) (1.86)
Financial Development -0.001* -0.002** -0.001 0.000 -0.001* (-1.94) (-2.35) (-1.58) (0.18) (-1.82)
Inflation 0.002 0.000 0.002 0.010 -0.001 (0.45) (0.01) (0.55) (0.40) (-0.13)
Creditor Rights -0.082*** -0.093*** -0.065*** -0.204*** -0.074*** (-3.98) (-3.30) (-3.96) (-3.95) (-4.45)
Rule of Law -0.085 0.045 -0.160 -0.338 -0.139 (-0.53) (0.24) (-1.15) (-0.56) (-1.26)
Control of Corruption 0.165 -0.010 0.261*** 0.634 0.170** (1.41) (-0.07) (2.85) (1.56) (2.24)
Industry and year dummies No Yes Yes Yes Yes
Yes
Constant 1.349*** 1.544*** -0.223 0.683*** -12.609*** 0.316 (26.50) (4.44) (-0.44) (2.87) (-11.59) (1.27)
Observations 1,033 1,033 1,033 1,033 1,033 1,033
46
This table presents the main evidence on the relation between trust and debt covenants using Poisson regressions. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. Models 1 to 5 present our main evidence on the relation between Trust and Number of Covenants. Model 6 reports the relation between Trust and Intensity of Covenants. The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
47
Table 4. Cross-country variation in the effect of trust on the use of debt covenants
Panel A. Interaction between trust and investor protection Number of Covenants s Intensity of Covenants s
Deb
t Rec
over
y
Effic
ienc
y of
Deb
t Enf
orce
men
t
Effic
ient
Jud
icia
ry
Rule
of L
aw
Con
trol
of C
orru
ptio
n
Deb
t Rec
over
y
Effic
ienc
y of
Deb
t Enf
orce
men
t
Effic
ient
Jud
icia
ry
Rule
of L
aw
Con
trol
of C
orru
ptio
n
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Trust × Interaction Variable 0.022*** 0.030*** 0.550** 0.631** 0.453** 0.014*** 0.020*** 0.717*** 0.486** 0.214
(3.31) (4.10) (2.33) (2.25) (2.18) (3.02) (3.86) (3.20) (2.14) (0.96)
Interaction Variable -0.000 -0.001 -0.265** -0.154 -0.002 -0.000 -0.001 -0.319*** 0.082 0.108 (-0.16) (-0.96) (-2.36) (-1.02) (-0.02) (-0.03) (-0.61) (-3.00) (0.82) (1.13)
Trust -2.718*** -3.424*** -3.426*** -1.590*** -1.525*** -1.892*** -2.413*** -4.056*** -1.235*** -0.892* (-5.08) (-5.68) (-3.02) (-3.79) (-3.59) (-4.52) (-5.29) (-3.77) (-3.48) (-1.75)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes
Observations 1,032 1,032 1,033 1,033 1,033 1,032 1,032 1,033 1,033 1,033
48
Panel B. Interaction between trust and political institutions
Number of Covenantss Intensity of Covenantss
Pr
ess F
reed
om (s
core
)
Voic
e an
d Ac
coun
tabi
lity
(WG
I)
Polit
ical
Sta
bilit
y (W
GI)
Gov
ernm
ent E
ffect
iven
ess (
WG
I)
Pres
s Fre
edom
(sco
re)
Voic
e an
d Ac
coun
tabi
lity
(WG
I)
Polit
ical
Sta
bilit
y (W
GI)
Gov
ernm
ent E
ffect
iven
ess (
WG
I)
(1) (2) (3) (4) (5) (6) (7) (8)
Trust × Interaction Variable 0.026** 0.768** 0.517*** 0.689*** 0.027*** 0.613** 0.369** 0.443**
(2.10) (2.38) (2.69) (3.19) (3.05) (2.17) (2.04) (2.36)
Interaction Variable -0.010** -0.117 -0.069 0.067 -0.013*** -0.117 -0.005 0.071 (-2.11) (-0.76) (-0.65) (0.35) (-3.88) (-0.78) (-0.05) (0.42)
Trust -2.783*** -1.635*** -1.364*** -1.756*** -2.639*** -1.286*** -1.077*** -1.228*** (-2.82) (-4.52) (-4.41) (-4.54) (-3.75) (-3.85) (-4.07) (-3.66)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes Observations 987 1,033 1,033 1,033 987 1,033 1,033 1,033
49
Panel C. Interaction between trust and interlocking business groups
Number of Covenants s Intensity of Covenants s
Intr
agro
up T
ax
Con
solid
atio
n
Prev
alen
ce o
f Fam
ily G
roup
Prev
alen
ce o
f Bus
ines
s Gro
up
Intr
agro
up T
ax
Con
solid
atio
n
Prev
alen
ce o
f Fam
ily G
roup
Prev
alen
ce o
f Bus
ines
s Gro
up
(1) (2) (3) (4) (5) (6) (7) (8)
Trust × Interaction Variable 0.345** -1.612* -0.044** -0.040* 0.332 -1.293* -0.045** -0.031
(2.06) (-1.96) (-2.00) (-1.95) (1.60) (-1.93) (-2.33) (-1.48)
Interaction Variable -0.070** 0.290 0.007 0.007 -0.083* 0.285* 0.008* 0.007 (-2.02) (1.53) (1.51) (1.60) (-1.94) (1.74) (1.88) (1.52)
Trust -1.654** -0.474** 0.037 0.187 -1.484** -0.389** 0.144 0.118 (-2.49) (-2.20) (0.10) (0.40) (-2.02) (-2.12) (0.45) (0.27)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,025 1,025 1,025 1,025 1,025 1,025 1,025 1,025
This table presents the cross-country variation in the effect of trust on debt covenants using Poisson regressions. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. Panel A shows the interaction between trust and investor protection. Panel B shows the interaction between trust and political institutions. Panel C shows the interaction between trust and interlocking business groups. The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
50
Table 5. Cross-firm variation in the effect of trust on the use of debt covenants (number and intensity)
Number of Covenants s Intensity of Covenants s
Corporate
Governance Error Illiquidity Corporate
Governance Error Illiquidity (1) (2) (3) (4) (5) (6)
Trust × Interaction Variable 0.052*** -26.547*** -17.195*** 0.052*** -20.474*** -11.986*** (4.37) (-3.48) (-3.62) (3.93) (-2.91) (-2.81)
Interaction Variable -0.020*** 9.555*** 5.659*** -0.022*** 7.073*** 3.885*** (-4.97) (3.44) (3.57) (-5.04) (2.76) (2.72)
Trust -5.740*** 4.217*** 0.381 -5.955*** 4.553*** 0.163 (-5.76) (3.91) (0.58) (-5.62) (4.21) (0.26)
Issue-level controls Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Observations 334 125 233 334 125 233
This table presents the cross-firm variation in the effect of trust on debt covenants using Poisson regressions. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. Models 1 to 3 examine the cross-firm variation in the effect of trust on Number of Covenants. Model 1 shows the interaction between trust and the quality of firm-level governance. Models 2 and 3 show the interaction between trust and firm-level information environment. Models 4 to 6 repeat the results of Models 1 to 3 by replacing the dependent variable with Intensity of Covenants. The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
51
Table 6. Endogeneity
First Stage Second Stage Number of Covenants Intensity of Covenants (1) (2) (3)
Trust -2.249*** -1.728***
(-2.97) (-2.80) Instruments
Tropical -0.074** (-2.56)
Genetic Distance -0.001*** (-8.14)
2nd Person Differentiation -0.067*** (-9.51)
Issue-level controls Yes Yes Yes Firm-level controls Yes Yes Yes Country-level controls Yes Yes Yes Industry and year dummies Yes Yes Yes Observations 972 972 972 First Stage F-statistic 59.15 overid Hansen’s J-Statistic - 1.17 2.47 overid p-value - 0.56 0.29
This table presents the 2SLS regressions examining the effect of trust on debt covenants. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Tropical is a dummy variable equal to 1 if the country is in a tropical climate zone, and 0 otherwise. Genetic Distance is the genetic distance between the population of one country and that of the U.S. 2nd Person Differentiation is a dummy variable equal to 1 if the number of second person pronouns used in spoken language differs by the type of person with whom one interacts, and 0 otherwise. Additional variable definitions are provided in the Appendix. The first-stage regression is reported in Model 1. Models 2 and 3 report the second-stage regressions using the fitted values of Trust. The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
52
Table 7. Out-of-sample evidence
Number of Covenants Intensity of Covenants (1) (2)
Trust US -0.431*** -0.431*** (-2.81) (-3.03)
Issue-level controls
Amount 0.062*** 0.037*** (4.46) (3.33)
Maturity -0.000*** -0.000*** (-6.28) (-5.73)
Firm-level controls
Size -0.042*** -0.011 (-4.62) (-1.34)
Leverage 0.554*** 0.517*** (11.51) (11.79)
Interest Coverage 0.000 0.000 (0.49) (0.07)
Profitability 0.538*** 0.543*** (5.48) (6.16)
Capital Expenditure 0.397*** 0.310*** (3.20) (2.99)
Tangibility -0.626*** -0.537*** (-9.25) (-8.74)
State-level controls State GDP -0.000 0.000
(-0.53) (0.16) State GDP Growth -0.005 -0.050
(-0.02) (-0.18) Industry and year dummies Yes Yes Constant 0.971*** 0.278**
(5.82) (2.08) Observations 10,392 10,392 This table presents a within-country analysis on the relation between trust and debt covenants using Poisson regressions. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust US is the U.S. region-level mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. Model 1 reports the relation between Trust and Number of Covenants. Model 2 reports the relation between Trust and Intensity of Covenants. The sample consists of bonds issued by U.S. companies and issued in the U.S. market contained in the Fixed Investment Securities Database (FISD). The merged data cover 10,392 bond issues over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the firm level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
53
Table 8. Robustness checks
Panel A. Alternative measures of trust Number of Covenants Intensity of Covenants
100 + (% of participants who respond “most people
can be trusted”) – (% of participants who respond
“can't be too careful”)
How much do you trust your
neighborhood?
100 + (% of participants who respond “most people
can be trusted”) – (% of participants who respond
“can't be too careful”)
How much do you trust your
neighborhood? (1) (2) (3) (4)
Trust (Alternative) -0.370*** -0.315* -0.287*** -0.287**
(-3.51) (-1.88) (-3.21) (-2.31) Issue-level controls Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Observations 1,033 1,021 1,033 1,021
Panel B. Additional controls for culture
Number of Covenants (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Trust -0.735*** -0.741*** -0.704*** -0.845*** -0.793** -0.781*** -0.786*** -0.800*** -0.898** -0.740** (-3.23) (-3.48) (-2.73) (-3.16) (-2.36) (-3.41) (-3.34) (-3.35) (-2.17) (-2.41)
Individualism 0.014 0.117 (0.07) (0.46)
Uncertainty Avoidance -0.005 -0.054 (-0.03) (-0.29)
Power Distance 0.123 0.128 (0.37) (0.32)
Masculinity -0.118 -0.146 (-1.04) (-1.28)
54
Embeddedness 0.131 0.116 (0.66) (0.56)
Mastery 0.117 0.059 (0.53) (0.26)
Power Distance_Globe -0.010 (-0.06)
Collectivism_Globe 0.255* (1.92)
Uncertainty Avoidance_Globe -0.085 (-0.84)
Assertiveness_Globe 0.107 (0.64)
Tightness -0.009 (-0.46)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,033 1,033 1,033 1,033 1,033 1,033 1,033 1,033 983 605 Intensity of Covenants (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Trust -0.601*** -0.580*** -0.509** -0.612** -0.537* -0.640*** -0.586*** -0.609*** -0.895** -0.604** (-3.06) (-3.21) (-2.19) (-2.51) (-1.67) (-3.74) (-3.01) (-3.44) (-2.08) (-2.39)
Individualism -0.048 -0.019 (-0.27) (-0.09)
Uncertainty Avoidance 0.021 0.058 (0.12) (0.31)
Power Distance 0.251 0.285 (0.78) (0.74)
Masculinity -0.034 -0.018 (-0.32) (-0.16)
55
Embeddedness 0.191 0.216 (1.20) (1.31)
Mastery 0.008 -0.098 (0.05) (-0.59)
Power Distance_Globe -0.086 (-0.54)
Collectivism_Globe 0.110 (0.92)
Uncertainty Avoidance_Globe -0.039 (-0.52)
Assertiveness_Globe -0.021 (-0.12)
Tightness 0.003 (0.15)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,033 1,033 1,033 1,033 1,033 1,033 1,033 1,033 983 605
Panel C. Alternative samples
Number of Covenantss Intensity of Covenantss
No Canada No UK No Canada & UK >5 obs No Canada No UK No Canada & UK >5 obs (1) (2) (3) (4) (5) (6) (7) (8)
Trust -0.773*** -0.705*** -0.721** -0.740*** -0.611*** -0.628*** -0.603** -0.615*** (-3.15) (-2.81) (-2.44) (-3.12) (-3.39) (-2.94) (-2.51) (-3.02)
Issue-level controls Yes Yes Yes Yes Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Yes Yes Yes Yes Observations 657 822 446 1,013 657 822 446 1,013
56
Panel D. Alternative models
Number of Covenants Intensity of Covenants Country fixed effect Negative binomial Country fixed effect Negative binomial (1) (2) (3) (4) Trust -1.762** -0.741*** -2.054** -0.583*** (-2.44) (-3.33) (-2.30) (-3.11) Issue-level controls Yes Yes Yes Yes Firm-level controls Yes Yes Yes Yes Country-level controls Yes Yes Yes Yes Industry and year dummies Yes Yes Yes Yes Country dummies Yes No Yes No Observations 1,033 1,033 1,033 1,033
This table presents the robustness checks of the effect of trust on debt covenants. Number of Covenants refers to the total number of restrictive covenants based on three categories, namely, restrictions on a firm's financing, investment, and payout activities. Intensity of Covenants refers to the sum of three covenant indicators, with each of the indicators equal to 1 if at least a restrictive covenant exists in one of the three categories. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. We respectively estimate whether our main findings continue to hold after considering alternative proxies for trust (Panel A), alternative explanations (Panel B), alternative samples (Panel C), and alternative empirical methods (Panel D). The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
57
Table 9. Pricing effect of trust and covenants
Treasury Spread (1) (2)
D(Covenants) -55.629* -110.758***
(-1.90) (-2.77) Trust -103.505** -215.587***
(-2.51) (-3.13) D(Covenants) × Trust 161.688**
(2.03) Creditor Rights -4.158 -3.615
(-1.09) (-0.95) Amount 0.852 1.885
(0.10) (0.23) Maturity 0.002 0.002
(1.56) (1.53) Size -20.094*** -20.581***
(-3.95) (-4.06) Leverage 16.846 13.306
(0.44) (0.35) Interest Coverage 0.226 0.220
(1.20) (1.17) Profitability -311.591*** -310.384***
(-4.21) (-4.21) Capital Expenditure 507.785*** 517.626***
(5.48) (5.60) Term Spread -24.146 -15.501
(-0.39) (-0.25) Credit Spread (AAA-BAA) -151.702 -136.071
(-0.61) (-0.55) Log Exchange Rate Volatility -1.351 -1.145
(-0.54) (-0.46) ADR 3.028 0.244
(0.17) (0.01) Secured 0.537 1.459
(0.01) (0.02) Subordinated 21.005 22.846
(0.27) (0.29) Junk 94.603*** 92.402***
(5.88) (5.75) Sinking Fund -84.259* -83.884*
(-1.66) (-1.66) Callable 13.835 13.400
(1.18) (1.15)
58
Gross Spread 0.211 0.256 (0.19) (0.24)
Industry and year dummies Yes Yes Constant 528.414 525.892
(1.25) (1.25) Observations 500 500
This table presents the pricing effect of trust and covenants using the treatment-effects model of Greene (2000). Treasury Spread measures the difference between the yield of the bond and benchmark Treasury. D(Covenants) is a dummy variable equal to 1 if the debt contract contains a covenant of any type, 0 otherwise. Trust is the mean percentage of affirmative responses to the WVS question: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Additional variable definitions are provided in the Appendix. We respectively estimate whether our main findings continue to hold after considering alternative proxies for trust (Panel A), alternative explanations (Panel B), alternative samples (Panel C), and alternative empirical methods (Panel D). The sample consists of foreign bonds issued in the U.S. (Yankee bonds) contained in the Fixed Investment Securities Database (FISD). The merged data cover 1,033 bond issues from 34 countries over the 1989–2014 period. t-statistics, reported in parentheses, are based on standard errors that are heteroskedasticity-consistent and allow for clustering at the country level. ***, **, and * denote statistical significance at the 1, 5, and 10% levels, respectively.
Recommended