St. Petersburg State University
Graduate School of Management
Master in Corporate Finance
COVENANTS OF A CREDIT AGREEMENT AS A
REAL OPTION
Master’s thesis by the 2nd year student
Valeriy Kurepanov
Research advisor:
Vitaly Okulov, Associate professor
St. Petersburg
2016
ЗАЯВЛЕНИЕ О САМОСТОЯТЕЛЬНОМ ХАРАКТЕРЕ ВЫПОЛНЕНИЯ
ВЫПУСКНОЙ КВАЛИФИКАЦИОННОЙ РАБОТЫ
Я , Курепанов Валерий Дмитриевич, с т уд е н т второго курса магистратуры
направления «Менеджмент», заявляю, что в моей магистерской диссертации на тему
«Ковенанты кредитного договора как реальный опцион», представленной в службу
обеспечения программ магистратуры для последующей передачи в государственную
аттестационную комиссию для публичной защиты, не содержится элементов плагиата.
Все прямые заимствования из печатных и электронных источников, а также из
защищенных ранее выпускных квалификационных работ, кандидатских и докторских
диссертаций имеют соответствующие ссылки.
Мне известно содержание п. 9.7.1 Правил обучения по основным образовательным
программам высшего и среднего профессионального образования в СПбГУ о том, что
«ВКР выполняется индивидуально каждым студентом под руководством назначенного ему
научного руководителя», и п. 51 Устава федерального государственного бюджетного
образовательного учреждения высшего образования «Санкт-Петербургский
государственный университет» о том, что «студент подлежит отчислению из СанктПетербургского университета за представление курсовой или выпускной
квалификационной работы, выполненной другим лицом (лицами)».
_______________________________________________ (Подпись студента)
____________25 мая 2016_________________________ (Дата)
STATEMENT ABOUT THE INDEPENDENT CHARACTER OF
THE MASTER THESIS
I, Valeriy Dmitrievich Kurepanov, second year master student, program «Management»,
state that my master thesis on the topic “Covenants of a Credit Agreement as a Real Option”,
which is presented to the Master Office to be submitted to the Official Defense Committee for
the public defense, does not contain any elements of plagiarism.
All direct borrowings from printed and electronic sources, as well as from master theses,
PhD and doctorate theses which were defended earlier, have appropriate references.
I am aware that according to paragraph 9.7.1. of Guidelines for instruction in major
curriculum programs of higher and secondary professional education at St.Petersburg University
«A master thesis must be completed by each of the degree candidates individually under the
supervision of his or her advisor», and according to paragraph 51 of Charter of the Federal State
Institution of Higher Education Saint-Petersburg State University «a student can be expelled
from St.Petersburg University for submitting of the course or graduation qualification work
developed by other person (persons)».
________________________________________________(Student’s signature)
___________25th May 2016_________________________ (Date)
2
АННОТАЦИЯ
Автор
Название магистерской
диссертации
Факультет
Направление подготовки
Год
Научный руководитель
Описание цели, задач и
основных результатов
Ключевые слова
Курепанов Валерий Дмитриевич
Ковенанты кредитного договора как реальный опцион
Высшая школа менеджмента
Менеджмент
2016
Окулов Виталий Леонидович
В работе применяется метод реальных опционов для анализа
кредитных договоров с ковенантами с целью определить
потенциал снижения процентной ставки для подобного
долга. В работе были рассмотрены основные теории
использования ковенант в контрактах, была построена
теоретическая модель заемщика с использованием
инструментария имитационного моделирования и
количественно оценены стоимости реальных опционов и
потенциал снижения процентной ставки. Результаты
исследования показывают, что для стандартного заемщика
процентная ставка по кредиту может быть снижена
приблизительно на 0,2% при включении ковенанты
коэффициента обслуживания долга в условия договора.
Кредитный договор, ковенанты, реальные опционы,
стоимость займа, коэффициент обслуживания долга
ABSTRACT
Master Student's Name
Master Thesis Title
Faculty
Main field of study
Year
Academic Advisor's Name
Description of the goal,
tasks and main results
Keywords
Valeriy D. Kurepanov
Covenants of a Credit Agreement as a Real Option
Graduate School of Management
Management
2016
Vitaly L. Okulov
We apply methods of real option analysis to the loan contract
with covenants in order to determine the interest rate reduction
potential that borrower can refer to in order to negotiate lower
cost of debt. In the Thesis, we discuss different lines of research
on debt covenants, build a theoretical model of a borrower, and
apply simulation procedure in order to obtain future states of the
firm and calculate the value of real options. The findings of the
work indicate that for a normally operating borrower the interest
rate of a loan with DSCR covenant can be reduced by
approximately 0,2%
Loan contracts, covenants, real options, costs of debt, debt
service coverage ratio
3
Table of Contents
Introduction..................................................................................................................................... 5
1. Background and Literature Review............................................................................................. 8
1.1 Debt covenants: definition, classification, and participation in bank monitoring process...8
1.2 Theoretical concepts of covenants: Agency Theory of Covenants..................................... 13
1.3 Theoretical concepts of covenants: Accounting perspective..............................................18
1.4 Theoretical concepts of covenants: Incomplete Contracts approach.................................. 21
1.5 The effect of covenant loan structure on borrowing terms................................................. 25
2. Methodology and Data.............................................................................................................. 28
2.1 Propositions.........................................................................................................................28
2.2 Model description............................................................................................................... 31
2.3. Covenant violations........................................................................................................... 37
2.4. Real options analysis..........................................................................................................39
3. Results....................................................................................................................................... 43
3.1 Industry............................................................................................................................... 43
3.2 Covenant-free debt..............................................................................................................46
3.3 Loan with DSCR covenant................................................................................................. 54
3.4 Abnormal leverage and volatility cases.............................................................................. 61
3.5 Managerial implications......................................................................................................64
4. Conclusion and Areas for Future Research............................................................................... 66
5. References................................................................................................................................. 69
Appendix 1. Russian oil industry players...................................................................................... 75
Appendix 2. Simulation Assumptions........................................................................................... 76
Appendix 3. Default probabilities of the firm (i=9%, various principal)...................................... 78
Appendix 4. Distribution of covenant violations over time (various principal)............................79
Appendix 5. Default probabilities for loan with and without covenant........................................80
Appendix 6. Expected values and volatilities for covenant and covenant free loan.....................81
4
Introduction
When the credit agreement between lender and borrower is signed, it is common to find a
set of covenants included in the terms. We define loan covenants as specific clauses written in
the loan contract that impose certain actions on the borrower or restrict them. The diversity of
these conditions is limited only to the creativity of the contract parties – from specifying
maximum amount of dividends to obliging the borrower to keep a particular management team.
When the borrower violates a particular covenant, the lender has a right to impose certain
sanctions.
In recent years, the topic of covenants has gained importance in the financial literature as
academics recognize the impact of these provisions on firm value, capital structure, managerial
behavior, etc. as well as their prominent role in addressing specific problems like contract
effectiveness or agency conflicts. Covenant theories began to emerge in the late 1970-s as a part
of the agency approach to business. Later covenants were introduced to a large body of financial
literature, from various accounting theories to studies of contracts. Numerous research on the
subject explored the ways covenants may provide certain benefits or help tackle particular
problems. The scientific importance of the subject is undoubted; yet, academics are far from
having a full picture. While empirical research finds positive relations between the presence of
covenants and other factors, like costs of debt or firm value, it is yet impossible to find a
common theory to explain all these relations.
Covenants are frequently mentioned in the business news reports. When borrower
violates loan contract clauses, a clear negative signal is sent to the markets. The consequences
may be bad, from possible credit rating revision to share price drop. When Standard & Poor’s
agency downgraded Russian credit rating to non-investment levels, it created a risk of early
repayment of corporate debt, as there were some covenants tied to the credit rating. (Metelitsa,
Sharoyan, and Nemtsova 2015). Early repayment of outstanding debt may be difficult for some
borrowers, or even impossible, as necessary liquidity levels were not maintained. In May 2015
the AvtoVAZ company violated covenants on loans totaling 36,6 bln. rubles, so one of the
creditors, Sberbank, demanded to have a meeting with the management after the accounting
statements are issued (Togliatti news May 22, 2016). In the same month, another borrower,
KOKS Group, violated covenants on Gazprombank loan totaling 4,5 bln rubles due to
revaluation of USD debt (Petlevoy 2016). Analysts wonder whether the firm would be able to
withstand other covenants or the default is inevitable. While the economic situation is Russia is
becoming less attractive, such news reports are becoming more common.
5
The diversity of covenant studies in literature and frequency with which covenants are
mentioned in news reports indicate the relevance of the topic and its high practical importance
for modern business. Unfortunately, the amounts of studies of covenants in Russian are far from
being substantial. While Russian borrowers are suffering from high costs of debt and are facing
possible covenant violations, the academic research with positive managerial implications
becomes extremely relevant.
In this Thesis, we decided to focus on the connection between the presence of covenants
in a loan contract and potential costs of debt. Among multiple theories on the subject, the
majority is quite clear that both lender and borrower enjoy specific benefits that covenants
provide and these benefits come at a price for both parties. The most obvious price the bank can
pay is decreased interest on debt, and we find support for this claim in the academic literature
(Bradley and Roberts 2004; Bazzana and Broccardo 2009; etc.). The empirical research on the
subject also confirms that (e.g. Deng et al. 2015). Given the fact that almost all commercial loans
contain covenants in some form – Roberts and Sufi estimate this share being 96% (Roberts and
Sufi 2009) – calculating the potential discount on interest rate when certain clauses are included
in the contract should be the top priority for the borrowers. If a company can actually negotiate
lower interest, rather than simply accept the covenants that bank offers, then research on the
subject is quite important not only for the management, but for financial theory overall.
In the Thesis, we used an unusual approach to estimate whether the preceding statement
holds for Russian borrowers. Some of the researchers (e.g. Gorton and Kahn 1993) viewed
covenants as options for the lender. We decided to employ the real options approach in our
analysis of covenants to estimate whether they have a certain option value and whether
borrowers can refer to this value and negotiate lower interest.
The aim of this research is to calculate the potential of an interest rate decrease that is
available for loan with a specific covenant. We define “potential for an interest rate decrease” as
the ability of borrower to obtain lower interest rate on loan through negotiation by referring to
the extra option value that covenant provides the bank. Particular research objectives include:
Define the term “loan covenant”, classify existing covenants and determine their
roles in loan process and debt contracts;
Identify theoretical approaches to covenants and related borrower and lender
activity;
Evaluate the effects of covenants on both lender and borrower;
Create a model of a firm that has a loan with a covenant in its debt structure, and
estimate the value of real options this covenant provides;
Calculate how much the interest rate on loan with covenant can be reduced
compared to the covenant-free loan
6
To achieve the research goal we used modelling, simulation and real options analysis in
the Thesis. Our work is different from existing theoretical and empirical research on the subject.
On the one hand, while theoretical research explains existing phenomena, we employ the current
theory of covenants to justify our claims. On the other hand, while modern empirical studies
estimate significant relations between covenants and other factors with a sample of existing
loans, we create a model of a company using the real-life industry data, incorporate a loan with
covenants in it and simulate the firm’s future state. With this methodological framework, we
achieve greater variability and are able to answer the main research question from the viewpoint
of how it should be, not how borrowers generally do that.
The Thesis is divided into three parts. In the first part, we discuss the main existing
theories of covenants. The second part is dedicated to methodological framework of our
research. The third part contains results of the analysis and managerial implications of the main
findings.
7
1. Background and Literature Review
1.1 Debt covenants: definition, classification, and participation in bank monitoring process
The financial literature has recently experienced a boom of interest to debt covenants.
The issues of academic journals dated 2010-2015 are filled with research particularly dedicated
to covenants, even though theoretical fundamentals were created a while ago. The rising interest
to the topic is no surprise – for years, quantitative research on firm value and capital structure
ignored the specifics of debt contracts. At the same time, analysis of 3,603 private credit
agreements done by of Robert and Sufi (2009) indicates that almost 97% of loans contain at least
one financial covenant. Covenants can be categorized broadly by the accounting measures on
which they are based – debt to cash flow (58%), debt to other balance sheet items (29%),
coverage ratios (74%), net worth (45%), liquidity (15%), and cash flow (13%) (Robert and Sufi
2009, p. 1662-1663). Thus, a covenant should not be considered as an insignificant anomaly in
the contract, but as an important and valuable part of the financial theory.
The literature generally agrees upon the definition of covenants. For instance, one of the
most common: “loan covenants are specific clauses designed to protect the bank and prohibit the
borrower from taking actions that could adversely affect the likelihood of repayment”
(Greenbaum and Thakor 2007, p. 212). Quite similarly, Smith and Warner (1979) state that “A
bond covenant is a provision, such as a limitation on the payments of dividends, which restricts
the firm from engaging in specified actions after the bond are sold”. Another, slightly different
approach to covenants suggests they “specify minimum standards for a borrower’s future
conduct and performance” (Paglia and Mullineaux 2006). The violation of covenants is a default
event and it allows the debtholders to intervene through either forced bankruptcy, renegotiation
of debt terms or additional forced constraints on firm’s behavior. In this context, covenants serve
the role of controlling mechanisms for debtholders as they restrict the ability of shareholders to
engage in strategies that are harmful for the wealth of debtholders.
One can encounter many different types of covenants in lending agreements across the
globe. The main difficulty of gathering the data and exploring the relations between covenants
and other factors is the secrecy of the loan agreement terms. Banks usually do not disclose what
covenants they set for the borrower for the sake of privacy or competition. Analysts would need
to guess, what is going on between a company and a bank during the negotiation process.
However, specific critical values of some accounting-based covenants are considered “standard”
in the financial world. The research by Demerjian (2007) showed that in a sample of 16,364
loans 78% of the deals included at least one financial ratio covenant. This proves again that
covenants are important part of loan agreement terms, financial ratio covenants are the
8
foundation of covenant structure of the deal and they can be more or less standard. Author
explores the types of ratios used and determines five being most common – minimum coverage
ratio (earnings/periodic debt-related expense), maximum debt to cash flow (total debt/earnings),
minimum net worth (assets –liabilities equals to shareholders’ equity), maximum leverage (total
debt/total assets) and minimum current ratio (current assets/current liabilities). Author claims
that all these are linked to a credit risk of the borrower (Demerjian 2007). Achleitner, Bock, and
Tappeiner in their 2012 work “Financial Covenants and Their Restrictiveness in European LBOs
– An assessment in the Aftermath of the Financial Crisis” find other common covenants –
leverage, interest cover, cash flow cover, and capex on the basis of the survey conducted among
25 managers.
Credit risk is the probability that borrower will fail to make required principal and
interest payments over the time of a loan (Demerjian, 2007). The credit risk of the borrower is
evaluated at the inception of the loan. The higher it is, the stricter the loan terms will be.
Unexpected rise of the credit risk is toxic for the borrower-lender relations. Altman in 1968
article “Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy”
discussed the issues of company default and built the first multivariate scoring model (z-score) to
determine the default probability. Almost all modern default models are variations of this.
Covenants were also explored in the paper: Altman concluded that by setting up covenants,
creditors reduce the credit risk by demanding the borrower to stick to certain thresholds. (Altman
1968)
The main problem with credit risk is that it is unobservable: you cannot measure it
directly or create the “ultimate ledger”, so lenders need to use covenants as a proxy. Demerjian
(2007) claims that previously listed covenants are indeed good proxies for the credit risk.
Minimum coverage ratio, maximum debt to cash flow and minimum net worth use the operating
performance measure. Operating performance is a good indicator of the credit risk – debt
payments are made out of cash flows and earnings can be used as a stable predictor of these cash
flows (Dechow et al. 1998; Barth et al. 2001). All things equal, companies with good earnings
performance have lower credit risk and are less likely to default; coverage ratios, debt to cash
flow and net worth covenants capture this rule. Current ratio captures the short-term-liquidity.
Lack of short-term liquid assets may be the cause of technical default on payment. Leverage is
an indicator of credit risk as well – the higher the proportion of debt is, the higher are the
possible distress costs are.
Gorton and Winton (2003) stress that covenants facilitate the dynamic relations between
lender and borrower. These relations are based on a fact that future is uncertain and many factors
that may affect the position of the borrower positively or negatively are hard to predict. Both
9
parties may include the specific provisions in the contracts in order to compensate their inability
to know the future for sure. Violations of these provisions will be recorded in future on the basis
of the available information. In case of the violation, lender may act to lower the credit risk level.
In order to proceed with various theoretical and empirical studies of covenants it is
necessary to provide some degree of covenants classification. R. Belyaev from Graduate School
of Management in 2015 “Covenants as Risk Management tool in Credit Relation” was one of the
few to provide a comprehensive classification of debt covenants and I am going to cite results of
his research here. One way to classify is to divide them in four groups: affirmative covenants –
require certain actions to take, like sending the bank audited financial statements on quarterly
basis; restrictive clauses – restrict certain borrower’s actions, like limiting the share of dividends
in profits; negative covenants – forbid the borrower from taking certain actions before bank
approves them, like prohibition to engage in M&A activity until the lender approves the deal;
default provisions – when lender has a right to declare a technical default. (Greenbaum and
Thakor 2007) Another classification suggests that covenants can be divided in financial;
management, control and ownership; reporting and disclosure covenants. (Lancett 2014). H.
Christensen and V. Nikolaev in 2011 article “Capital Versus Performance Covenants in Debt
Contracts” in line with the Agency Theory of Covenants (will be discussed later) divide
covenants in capital-based and performance-based. They argue that capital-based covenants
control the agency problems by aligning debt-shareholder interests, while performance-based
“serve as trip wires that limit the agency problems via transfer of control to lenders.” Companies
trade-off these mechanisms. While capital covenants impose costly restrictions on capital
structure, performance covenants require the accounting information to be available. Capital
covenants are defined as covenants formulated in terms of information about sources and uses of
capital, that is, balance sheet information only. Performance-based covenants, on the other hand,
are formulated in terms of current-period performance or efficiency ratios. They include interest
coverage, fixed charge coverage, debt-to-earnings, and debt-to-cash flow ratios as well as
earnings (cash flow) itself. Authors also find evidence that use of performance-based covenants
is more likely for firms that experience financial constraints of the borrower, high credit risk
portrayed by accounting information, are likely to renegotiate the contract and have restrictions
on specific managerial actions. (Christensen and Nikolaev 2011) Other classifications of
covenants can be also found in literature (Achleitner, Bock, Tappeiner, 2012; Saavedra, 2014;
Karapetov, 2011; etc.)
The topic of covenant tightness should also be discussed separately. Terms like “covenant
stringency,” “covenant tightness” or “strictness” are used in a large body of scientific literature
and usually without any adequate explanation. As no widely appreciated research distinguishes
10
these terms, we are going to treat them as synonyms. Belyaev (2015) provides a clear
explanation of the strictness of covenants. Covenants impose restrictions on borrower’s actions
thus giving the borrower less freedom in decision-making. Restrictions can vary: covenants may
require disclosing more or less information, demanding more or fewer managerial actions; set
the threshold value of financial ratios close to current values or not, etc. The more the covenants
restrict the borrower and require specific actions, the more they are tight or strict. (Belyaev 2015)
Strictness can be also defined through covenant slack which is the difference between threshold
and initial value of covenant ratio. The bigger the slack value is, the less strict is the covenant.
For example, a covenant may require a borrower to maintain earnings/interest ratio at level of at
least three. If the actual ratio is 3.1, further decline may be dangerous for the borrower. If the
ratio is around six, then value of the ratio can decline by three before the default, if it is nine –
even more. That is more slack before default. (Demerjian and Owens 2014)
James and Demiroglu (2010) in “The Information Content of Bank Loan Covenants”
study the determinants of thresholds for financial covenants in bank loan agreements and
specifically focus on covenant tightness. They discover that firms with fewer investment
opportunities or that are generally riskier usually prefer tighter covenants. Moreover, selection of
tight covenants in the sample was associated with improvements in the covenant variables.
Overall, the findings suggest that “… selection of tight covenants conveys information
concerning future changes in covenant variables, investment and financial policies, and the
outcome of covenant violations.” (James and Demiroglu 2010)
Murfin (2012) also studied debt covenants through the focus of their strictness. He is not
giving the definition of strictness, but providing the empirical loan-specific measure that aims to
capture the ex ante probability of a forced renegotiation. As a result, stricter contracts making
trip wires more sensitive thereby providing the lender contingent control in more states of the
world. (Murfin 2012). Author finds that particular banks tend to write tighter covenants than
other lenders after payment default occurs in their loan portfolios, even if the defaulted entity is
industrially or geographically distant from current borrower. The findings may suggest that
recent defaults alter the bank’s perception of its screening expertise, thereby changing the
contracting behavior.
We discussed the nature of covenants, explored the existing approaches to their
classification, and highlighted the “tightness” as an important feature. As we mentioned, when
covenants are breached lender needs to engage and take action in order to lower the risk level.
However, this is not a single-time action – lender should constantly track the performance of the
client and covenant financial ratio in order to be not caught off guard when violation happens.
11
Rajan and Winton (1995) suggest that loan contracts should be structured in a way to provide
incentives for lender to monitor the borrower, thus linking covenants and bank monitoring.
A bank as an intermediary serves to channel funds from individual investors to firms with
productive investment opportunities (Hoshi et. al. 1990). Banks exist for certain reasons. Theory
suggests that banks exist, in part, because they can exploit technologies to monitor the borrowers
more effectively in comparison to direct investors (Diamond 1984). The purpose of this
monitoring activity is to guide borrowers to make more value-generating investment decisions,
as banks usually do not want the client to default, and protect the bank’s rights on cash flows if
the firm actually defaults.
D. Diamond (1991) in “Monitoring and Reputation: the Choice Between Bank Loans and
Directly Placed Debt” examines the need for bank’s ex post monitoring and finds out that its
usefulness depends on credit reputation. In fact, banks monitor middle-tier risk companies more
intensively than low-tier or high-tier. Borrowers with low risk possess long track record of nodefault operations, they have an established reputation and their future profits are too high for
them to engage in doubtful investment projects. Low-risk clients, on the other hand, do not have
sufficient positive reputation to loose and without it they take the chance that monitoring may
reveal their risky investments. In response, banks will simply charge higher interest rates,
undertake less monitoring activities, and limit their exposure to debt. Therefore, “…the clientele
of borrowers, who rely on monitored bank loans, are the middle-rated borrowers, whose rating is
too low for reputational effects to eliminate moral hazard but is high enough for monitoring to
eliminate moral hazard” (Diamond 1991, p. 716)
The simplest way for a bank to monitor firm activities is through financial statements.
Financial statements enable banks to learn about the borrower’s investment decisions and
performance over time (Jensen and Meckling 1976). By assessing the quality of financial
statements bank can evaluate the borrower’s financial health and its actual risk level. Minnis and
Sutherland (2016) study the theory of use of financial statements as monitoring device in debt
contracting. They find that for half of the loans, financial statements are requested and variation
is related to credit risk of the borrower, length of the lender-borrower relationship, collateral, and
the provision of business tax returns. (Minnis and Sutherland 2016) Collectively, results of the
research provide evidence that fundamental demand for information from banks exists and
financial statements are indeed crucial channel for this information.
However, if the company is not required to produce financial statements, banks find
different ways to monitor the loans. Sometimes, lenders may not need financial statements – they
rely on faith and collateral. Collateral serves two roles after the originating of the contract: it
provides incentives for borrower to stay away from risky activities (to keep the value of the
12
asset) and bring compensation for bank in case of default. Moreover, financial statements (if
shared frequently) may be particularly useful when collateral is present in the agreement.
Collateral provides banks incentives to monitor financial statements to ensure that asset is intact
and has not been transferred to related parties. (Minnis and Sutherland 2016)
Banks may also use other information to monitor their loans: firm’s tax returns,
information on repeated activities, by approving major deals, etc. Tax returns can be a good
alternative to financial statements, as firms are required to report sales, expenses and other
figures in them.
Finally, covenants are included in the monitoring process of the bank as both means to
establish this process and as instruments to control the financial position and risk level of the
borrower. For instance, some affirmative covenants may require sending the financial figures to
lenders every month, which is way to establish the monitoring. Some restrictive covenants that,
for instance, prevent the borrower from investing a sum more than specified in a single project is
a way to establish control over the operations of the company and compensate the need for
monitoring. Finally, a covenant that requires the borrower to maintain a specific accounting ratio
is a way to monitor the financial position of the company. In case of covenant violation, lender
can reassess the borrower, its risk level, and financial health.
1.2 Theoretical concepts of covenants: Agency Theory of Covenants.
The ability of covenants to resolve the agency conflict between shareholders and
debtholders is the main implication of, how some authors call it, Agency Theory of Covenants,
which is primary built on the works of Jensen and Meckling (1976), Myers (1977) and Smith and
Warner (1979).
In their 1976 article, “Agency costs and the theory of the firm,” Jensen and Meckling
develop an ownership structure theory of the firm. They start by summarizing the progress to
date on the theory of property rights, agency and finance. By combining the elements of these
areas they develop a theory of firm’s ownership structure, that help to answer the questions like:
“Why an entrepreneur in a firm which has a mixed financial structure would choose a set of
activities … such that total value of the firm is less than it would be if he was a sole owner?” or
“Why lenders often place restrictions on the activities of the firms to whom they lend?”, etc.
Authors also defined the concept of agency costs, investigated their nature that is generated by
the existence of debt and outside equity, and showed their relationship to “separation and control
issue” (Jensen and Meckling 1976). The main focus of the paper is the nature of agency costs,
the owner of the costs and the way they are created. The inception of covenants by debtholders is
13
justified with their ability to limit the managerial behavior that results in reduction of bonds’
value.
In his 1977 work, “Determinates of Corporate Borrowing,” Steward Myers stated that
future growth opportunities of the firm should be viewed as real options and thus capital
structure (or the amount of corporate borrowing) is affected by the market value of these growth
options. In his analysis author did not rely on perfect or incomplete financial markets. The main
contribution of the paper is to partial theory of corporate borrowing, with Myers proving that the
amount of debt issued by the firm should be set equal to that amount that maximizes the market
value of the firm. He also suggested that firms should try to match loan maturity and asset life as
an attempt to “…schedule debt repayments to correspond to the decline of future value of assets
currently in place.” (Myers 1977)
The main contribution to future covenant theories is the
finding that managers of the firms with real options can engage in suboptimal investment
strategies and costs of these strategies are borne by firm’s creditors. The main reason why
borrowers actually accept the loan terms with embedded monitoring and renegotiation is that the
costs are offset by the increase in firm value due to reduction of poor investment decisions and
other agency costs. Myers also claims that managers usually complain about the restrictive
covenants as if they were evil, but in fact, they are rational from the viewpoint of both creditors
and debtors. Managers and shareholders “…freely choose to accept constraints today which rule
out the behavior that seems rational tomorrow”. (Myers 1977)
Finally, Smith and Warner in their 1979 article “On Financial Contracting: An Analysis of
Bond Covenants” fully cover the practical application of covenants in credit agreements. On the
basis of previously mentioned works by Myers and Jensen & Meckling they recognize the
conflict of interest between debtholders and shareholders and test the “irrelevance hypothesis”
(that is: firm’s total value is not affected by the conflict) and “Costly Contracting hypothesis”
(being: control of the bondholder/shareholder conflict through financial contracts can increase
the firm’s value). (Smith and Warner 1979). They find evidence for the second hypothesis and
suggest that use of covenants is the necessary controlling mechanism, even though it involves
costs. Value of the firm can be reduced via either conflict of interest between shareholders and
debtholders or managerial behavior that aims to protect the shareholders’ interests. Thus,
restrictive covenants are necessary to facilitate the firm’s value-maximizing strategies.
Analysis of Smith and Warner also shed light on relative costs of restrictions, which can
be written into the debt contract. For instance, authors conclude that production/investment
policy is too difficult to monitor. On the contrary, dividend policy and financing policy have
much lower monitoring costs. Finally, authors argue that it is in the interest of shareholders
14
themselves to include debt covenants in the agreement: with the increasing value of debt, the
associated agency costs will be reduced.
M. Bradley and M. Roberts in their 2004 article “The Structure and Pricing of Corporate
Debt Covenants” are one of the first to provide the thorough description of Agency Theory of
Covenants (ATC). It is worth mentioning, that this theoretical construct is not common in
literature. However, we decided that it is worth using in the Thesis as it sufficiently incorporates
major points and fundamentals of agency approach to covenants, which is represented in a huge
portion of financial literature (Billett, King and Mauer 2007; Chava and Roberts 2008;
Achleitner, Bock, Tappeiner, 2012, etc.) Bradley and Roberts root the ATC in the works of Smith
and Warner (1979), Myers (1977), and Jensen & Meckling (1976). The foundation of the theory
is the conflict of interest between shareholders and debtholders. It was briefly mentioned before,
but now we are going to provide a more a detailed description.
The management of the company is hired, because stockholders believe that it can
increase the value of the firm with the help of the skills and experience. Thus, management
should protect the wealth of shareholders and act in favor of their interests – which may not
always be the case in reality, by the way (Besanko et al. 2000). Debtholders and shareholders are
the two parties that have different relations to the company and different returns. Their opinions
concerning dividend policies, investment, and financing are conflicting due to the nature of
different relationships. For instance, shareholders have the higher overall required rate of return
and they are positive about the company taking more risky projects, while debtholders may see it
as a threat to the stability of their interest payments. Thus, managers are likely to engage in
strategies that are opportunistic to debtholders (and as ATC suggests, rational creditors should
expect that). According to Bradley & Roberts (2004), detrimental actions include, but are not
limited to:
Unauthorized distributions. Managers can liquidate the firm’s assets and
distribute the proceeds as a dividend or repurchasing shares at a premium. Black
(1976) illustrates it as “there is no easier way for a company to escape the burden
of a debt than to pay out all of its assets in the form of a dividend, and leave the
creditors holding an empty shell”. Although it is an extreme case, it shows that all
growth in payments to shareholders that is not supported by the increase in
external financing is detrimental to debtholders’ wealth. Stockholders, however,
prefer to have part of the firm’s profit to be distributed as dividends, as while they
are invested in projects, resulting cash flows may only end up in hands of
creditors. (Black 1976)
15
Claim dilution. If there is a certain debt structure within the firm with senior debt
being prioritized higher than subordinated debt, current debtholders face the risk
of subsequent debt given the higher claim. This risk affects the price of corporate
bonds (Smith and Warner 1979). When managers are issuing debt of higher
priority than existing debt, value of existing claims diminishes. (Nash et. al.,
2003)
Asset substitution. Managers can accept higher-risk projects than had been
anticipated by bondholders when they purchased their bonds.
(Bradley and Roberts 2004) Debtholders do not gain anything from projects with
increased risk level, but bear the increased costs of potential default, while
shareholders’ risk is limited to equity and they can win excess returns.
Over-Investment. Cash flows can be retained to fund negative net present value
projects. The overinvestment is generally associated with lower firm value
(Berger and Ofek 1995) and more related to conflict between ownership and
control. For example, managers can prefer to invest in negative-NPV projects to
increase their own human capital and make the firm success associated with their
personal skills. (Shleifer & Vishny, 1989). Still, this strategy can be employed by
managers to decrease the value of debtholders’ wealth due to information
asymmetries, as part of asset substitution, etc.
Under-Investment. Underinvestment happens when projects with positive net
present value are foregone. There are negative effects for the firm’s value and for
the debtholders if management redistributes wealth to stockholders by rejecting to
invest in certain positive-NPV projects, as they only benefit the firm’s
bondholders. (Begley and Felham 1999). Underinvestment is more likely to occur
when firm is in financial distress as stockholders would like to maintain dividend
payments thus reducing the amount of assets ready for investment. (Malitz 1986)
Here is some example of the real-life detrimental action towards the debtholders from the
article “Bond Covenants and Creditor Protection” by W. Bratton, 2006:
“Assume that Firm A is worth 100 and has borrowed 50 from Lender 1 pursuant to an
unsecured loan. The loan was priced on the assumption of no further borrowing by Firm A but
contains no explicit restrictions. Firm A then borrows 35 from Lender 2 and invests the proceeds
in a project that turns out to be worthless. Firm A emerges 85 percent levered. The interest rate
on the second borrowing will reflect that possibility, compensating Lender 2, while the interest
rate on the first borrowing does not compensate Lender 1. Lender 1’s investment declines in
value, with the benefits of the decrease accruing to Firm A and Lender 2. Lender 1 is even worse
off if Lender 2 makes a loan secured by assets worth 35. So long as the obligation to Lender 2
16
remains outstanding, the encumbered assets will not be available to satisfy Lender 1.
Accordingly, Lender 1 must look for repayment to an asset base of 65 rather than the base of 100
envisaged at the time of the loan.” (Bratton 2006, p. 43)
The above example of claim dilution proves that such managerial strategies are not hard
to perform. There is also a body of research suggesting that there is a positive association
between shareholder power and some of the aforementioned actions. For example, Cremers and
Nair (2005) show that a portfolio of firms with long positions on companies that are mostly
owned by active shareholders (like public pension funds) and low takeover protection, and short
position on companies with similar ownership, but low takeover vulnerability, earns positive
abnormal returns. A portfolio buying firms with low ownership participation that is vulnerable to
takeovers and shorting companies with same ownership and great takeover protection earns no
abnormal returns. This evidence is supporting the idea that active shareholders can encourage the
activity on the takeover market (John and Kedia 2006). The debtholders act rationally and they
can predict detrimental actions and price the corporate debt accordingly. Thus, stockholders will
pay ex ante for the potential opportunistic behavior ex post. As stockholders bear the agency
costs of debt, they will try to minimize them by including debt covenants in the loan agreement.
Thus, covenants will control the management from taking actions that are detrimental to
debtholders. Debtholders will be willing to pay more for the debt contract with covenants.
(Bradley and Roberts 2004)
It is clear that actions listed above will have certain effects on overall firm’s behavior and
performance. Bradley and Roberts (2004) further test the implications of the ATC, which
include:
Potential benefit for shareholders from actions of management is greater when the
firm is in financial distress.
Since the agency costs of debt are inversely related to a firm's financial condition,
the poorer the firm's financial condition, the more likely is it that the firm would
include a covenant in its debt contracts.
Since it is virtually impossible to renegotiate covenants with public bondholders,
firms that include covenants in their debt contracts would issue primarily private
as opposed to public debt.
Firms with significant growth opportunities will include covenants in their
indenture agreements.
These implications were tested on 2002 sample of private corporate debt agreements
from Dealscan. The findings are generally in line with theory presented above and support the
findigs obtained by Malitz (1986), Begley (1994), Nash et. al. (2003), etc. Ileen Malitz in the
1986 article “On Financial Contracting: The Determinants of Bond Covenants” tested the sample
17
of 252 public debt issues and found out that presence of covenants is negatively related to the
size of the company and positively related to the leverage ratio. At the same time, Begley (1994)
studied the sample of 130 public issues and discovered that restrictive covenants are more likely
to be found in debt contracts of firms close to bankruptcy, with low asset value and low projected
cash flows. While Bradley and Roberts successfully tested the past evidence of the ATC, they
also discovered that the decision to include covenants and promised yield of the debt security is
determined simultaneously. Moreover, bonds with covenants would have a yield lower than debt
securities without any provisions. (Bradley & Roberts, 2004)
F. Bazzana and E. Broccardo in their 2009 “The Role of Covenants in Public and Private
Debt” employ the ATC to investigate the efficiency of covenants in bond contracts. At first, they
summarize the previous research to outline the costs and revenues of introducing covenants in
the contract to decrease agency conflict. The borrower would enjoy lower spread on debt
securities that have covenants compared to covenant-free instruments for an exchange of
constraints of company’s activities. The lender would offer less interest on debt for an exchange
of an option to interfere in company’s activities and to have a more effective monitoring system.
The authors also found out that while covenants can be effective in reducing agency costs, lack
of coordination between debtholders might reduce this efficiency due to high amount of
renegotiation costs followed by covenant violations. (Bazzana & Broccardo, 2009)
Therefore, judging by all of the aforementioned, there is a significant volume of scientific
literature that explains the fundamental reasons of presence of covenants and their impact on the
firm behavior and performance from the agency perspective. The inclusion of covenants is
justified by the reasons of mitigating shareholder/debtholder agency conflict and providing
associated benefits. Their possible impact on quantitative parameters like debt yield, capital
structure, leverage, etc. is not random. Covenants can reduce the agency costs between
borrowers and lenders, but each party has to pay the price for it.
1.3 Theoretical concepts of covenants: Accounting perspective
The theoretical research on the subject of covenants may deviate from the mainstream
agency approach and utilize a different view on the issue. For example, Dichev and Skinner
(2002) in their article “Large-Sample Evidence on the Debt Covenant Hypothesis” run largesample tests of so-called debt covenant hypothesis. They approach the issue of covenants from
an accounting perspective.
In the accounting research, “debt covenant hypothesis” stand for the idea that managers
tend to make accounting choices to reduce the likelihood of their firms will violate loan
covenants. Hypothesis was proposed by Watts and Zimmerman (1986) – they predicted that
18
firms which are close to thresholds (have low covenant slack) will make income-increasing
accounting choices. The strength of managers’ incentives to prevent covenant breaches from
happening depend on the costs of these breaches – these are costs of technical default (Smith and
Warner 1979; Holthausen and Leftwich 1983). Thus, accounting literature on covenants mainly
seeks ways to estimate these costs and find out, whether accounting choices of managers are
specified to avoid covenant violation.
Dichev and Skinner (2002) provided solid evidence that debt covenant hypothesis is true
and decisions of managers indeed depend on debt covenants. Their findings are consistent with
some prior research by Sweeney (1994), DeFond and Jiambalvo (1994), but contradict the results
of DeAngelo et. al. (1994) research. It is worth mentioning, that this work is superior to its
predecessors as it uses larger and more representative sample. Previous literature usually focused
on the accounting choices of firms, ending up in technical default, so their support for debt
covenant hypothesis was questionable: if the firms from the sample mainly defaulted, they can
represent only unsuccessful attempts of avoiding covenant violation. The article by Dichev and
Skinner (2002) does not have this drawback. Furthermore, the work provides evidence that in
private lending agreements covenants are set relatively tightly, covenant violations in the sample
are quite common (almost 30% of loans) and mostly these violations are not a result of financial
distress (Dichev and Skinner 2002). Authors suggest that private debt agreements lenders
frequently use covenants as a screening device – that complements the aforementioned
Christensen and Nikolaev (2011) research.
Several studies have examined the covenant breaches and associated costs of debt
covenant violation. For instance, Chen and Wei in 1993 article “Creditors’ Decisions to Waive
Violations of Accounting-Based Debt Covenants” explore the violation of covenants from
creditor point of view and explain the creditors’ decision-making process after covenants are
violated. They find out that after the covenant breach creditors can either waive the violation or
impose sanctions like early repayment, interest rate negotiation, etc. They also build the creditor
decision model either to waive or call the debt with option pricing techniques. Beneish and Press
(1993) in “Costs of Technical Violation of Accounting-Based Debt Covenants” investigate the
technical violations of accounting-based covenants for the sample of 91 firms. Their findings
suggest that in case of obtained waiver costs range from 1.2 to 2 percent of market value of
equity, depending on the assumptions made. However, if the creditor wants to impose the
additional restrictions after the covenant breach, then losses can fall between 4.4 to 7.3 percent.
These two studies have a drawback, that is recognized by the authors: the samples are likely to
include only “the worst” covenant violations as according to SEC Regulation S-X, Rule 4-08,
firms are not required to report the violations that have been cured by the report date.
19
Sweeney (1992) examines how managers respond to tightening constraints and explores
the accounting decisions of managers before and after they violate debt covenants. Generally, her
findings are in line with previously mentioned Dichev & Skinner (2002) work. The article
provides evidence that managers are more likely to make “income-increasing discretionally
accounting changes” when approaching such constraints, than managers from control firms.
(Sweeney 1992) Managers of default firms are inflating the accounting earnings figures even
more often and earlier adopt the income-increasing changes to accounting policies. Sweeney also
mentions that accounting changes helped to delay the technical default for 5 out of 22 default
firms. However, she states that in some cases, even substantial default costs will not motivate
managers to make accounting changes, for example, when degree of managerial flexibility is
low. Thus, the relevance of default costs for managers can be argued in some cases.
Further research on debt covenant violations was conducted with some implications to
other areas of management. Clifford W. Smith in 1993 “A Perspective on Accounting-Based
Debt Covenant Violations” summarized the theoretical and empirical findings on the topic.
Anand Jha in 2013 work “Earnings Management Around Debt-Covenant Violations” uses a large
sample of quarterly data to investigate how earnings are managed around the debt covenant
violations. This can be seen as a modern follow-up to Sweeney’s research, with an exception that
more narrow time interval is examined. The main finding on the paper is that “managers manage
earnings upward in the quarters preceding a debt-covenant violation, but downward in the
quarter a violation occurs. And they continue to manage earnings downward while the firm
remains in violation” (Jha 2013) Author argues that earnings management is also done to
improve the bargaining power of the management in the follow-up renegotiation. Furthermore,
Jha finds no evidence of excessive earnings management done by high-levered firms done in
order to prevent the covenant breach.
Fargher et. al. (2001) use the preceding research on firm’s behavior around the covenant
violation (Beneish and Press 1995; Wilkins 1997) to make the proposition that technical
violations of covenants involve significant breaches in financing agreements of firms; thus
“these breaches are likely to be associated with significant increases in violating firms’ risk”.
(Fargher et. al. 2001). Authors test this proposition by evaluating the changes in systematic and
unsystematic risks associated with initial technical violations. They conclude that there is solid
supporting evidence that both systematic and unsystematic risks increase close to covenant
breach. These findings are open to interpretations. The investors can be using the financial
disclosures on technical default for pricing and risk assessment. It may also be the case that risk
measures and the violation announcements both reflect the same underlying weakening in the
firm’s financial condition.
20
E. Borgonovo and S. Gatti of Bocconi University (Italy) extend the traditional risk
analysis of investment projects in their 2013 “Risk Analysis with Contractual Default. Does
Covenant Breach Matter?” by implementing the consequences of covenant breach in the Monte
Carlo simulation that is used to obtain the NPV distribution. Authors present the framework of
modelling material breaches (result in possible firm default) and technical breaches (accidental
violation of accounting variables) and apply their models to a real case study of project financing
of 64-million Euro biomass plant. Findings suggest that both technical and material breaches
matter, with their impact on NPV distribution being more relevant with rising leverage and costs
of debt. Although their approach does not involve accounting theory directly, results of the
research are extremely relevant when we discuss the covenant violations later in the
methodology session. (Borgonovo and Gatti 2013)
To sum up, various accounting theories treat covenants as important factors of accounting
decisions of the firm. This body of literature focuses on different situations that involve
covenants and that would likely influence the firm’s reporting – mostly, covenant violations. As
covenant violation is the main factor that affects the accounting both ex ante and ex post, no
surprise it is the main issue of the accounting research on covenants. Different works prove that
managers tend to make revenue-generating accounting choices in order to prevent covenant
violations and that credit risk indeed depends on breaches – it increases significantly right before
the violation. The accounting theory also explains tight thresholds. They are used by lenders to
monitor closely the activity of the borrower (as “trip wires”) and violations of tight thresholds
are often waived.
1.4 Theoretical concepts of covenants: Incomplete Contracts approach
A set of academic literature provides research on covenants via the so-called incomplete
contracts approach. The contract between the bank and the borrower can be complete and
incomplete. The economic theory suggests that writing complete contracts (contingent on all
future states of nature) is a certain way to improve overall efficiency of a transaction as it
facilitates the full risk sharing in these future states of nature. Still, complete contracts are not
available in practice. Renegotiations, that sometimes happen after the contract is signed, are even
more at odds with this theory. (Freixas and Rochet 2006) When the lending agreement is signed,
it is impossible to predict all the outcomes and events that may happen to the company. If the
firm goes bankrupt, lenders will have to initiate the bargaining process in order to save at least
some of the assets they have a claim on. That is clearly a negative outcome for the lenders, but
for them it was too difficult to predict the exact chain of events that triggered the bankruptcy.
21
Incomplete contracts theory considers this fact and allows one to work with information
asymmetries in contracts. Bolton & Scharfstein (1990) and Hart & Moore (1988) recognize the
situation when different states of nature are observable only by two parties to the contract, but
these states are not verifiable (outsiders cannot observe it). An incomplete contract will typically
require delegating the decision-making power (limited to a predetermined set of actions) to one
of the parties. This power should be dependent on the verifiable signal, thus, for instance, a
contract may specify that creditors take over the firm in the event of default. ( Freixas and Rochet
2006) Belyaev (2015) illustrates this concept with a following example:
“…Suppose that lending contract has a covenant that limits the borrowing by two times
of the amount of net capital of the firm. If the borrower tries to raise more money or his net
capital value depreciates, then the covenant violation will occur. Bank can demand early
repayment of the debt or change of loan conditions in order to lower the risk and increase
income… As we can observe from this situation, the lending agreement does not include all
possible scenarios that result in increased lender’s risk level – like borrower investing in more
risky project or unexpected losses of the company. Instead, bank and borrower fix the financial
ration in the loan agreement that serves as an indicator of borrower’s financial health and risk
level”
Nobel Prize winner of 2014 J. Tirole discusses information asymmetries, contracts and
covenants in his 2009 “Cognition and Incomplete Contracts.” He states that “incomplete contract
is a contract specifying the available design, which is renegotiated whenever this design turns out
not to be appropriate”. Tirole claims that analyzing the outcomes, choosing the full terms of the
agreement and assessing the results is costly for the bank. Parties to a contract consciously use
heuristics and leave it incomplete. Covenants protect the lender from unpredictable situations
that might happen (Tirole 2009)
Theory suggests that renegotiations between parties play a key role in incomplete
contracts. Two parties know that ex post the contract is not optimal. Therefore, “since they are
better off renegotiating the terms of the contract, they can’t credibly commit not to renegotiate.”
(Freixas & Rochet 2006, p. 114) Renegotiation is a way to improve efficiency of the contract.
Gorton and Kahn in their 1993 “The Design of Bank Loan Contracts, Collateral, and
Renegotiation” explore this idea in order to find the fundamental difference between public debt
(bonds) and private loans. One of the easily observable differences between debt types is the
presence of covenants: loans usually have much heavier covenant structure than public debt.
According to Gorton and Kahn, private loans have the embedded options to call the loan back.
This is a trigger for renegotiation and renegotiation will automatically increase the effectiveness
of the contract. Covenants are valuable as they allow efficient loan renegotiation. Authors also
22
claim that they have introduced new and only justification of bank existence via renegotiation –
while ex ante screening of borrowers and ex post monitoring of the borrowers is important, it
does not explain the seniority of bank claims, why banks are not equal to junior claimants or
even equity investors. (Gorton & Kahn, 1993)
The connection between covenants and debt renegotiation is a significant topic in
academic literature. Empirical studies show that it is actually a common practice to renegotiate
loans, for instance Zinbarg (1975) claim that loans in the portfolio of Prudential Insurance
Company of America are renegotiated at least once a year. Aghion and Bolton (1992) also utilize
the incomplete contracts approach to covenants. They view covenants as tools to allocate control
rights between managers and creditors optimally. Research shows that if the borrower faces the
liquidity constraints, then renegotiation will not help to achieve the first best optimum (Aghion
and Bolton 1992). That is contradicting the Kahn and Huberman (1989) findings: renegotiation
can achieve an efficient outcome, while no other simple contract can. Aghion and Bolton place
the trade-off in controlling agency problems versus giving the borrower leeway to act efficiently
as their main priority, which is similar to focus of Berlin and Mester in their 1992 work “Debt
Covenants and Renegotiation.” In the article, they explored the connections between stringency
of covenants’ restrictions and ease of renegotiation of debt contracts. The results of the research
concern the “impact of the firm’s creditworthiness on the value of the renegotiation option”
(Berlin and Mester 1992). When firm’s creditworthiness is low, its increase will positively affect
the value of the renegotiation option. However, beyond some minimum level further
creditworthiness increase will only reduce the option value. Their model also suggests that bank
loans include covenants because they allow the bank to restrict the set of actions borrower can
take, not because of their loan call back option as Gorton & Kahn argue. The following finding is
quite intuitive: the possibility of renegotiation should make covenants stricter. The overall results
are consistent with empirical and anecdotal evidence. Although authors believe, that their
findings are quite intriguing, they do not answer any questions about the optimal contract design
with different types of debt (public and private), thereby setting a path for further research.
Unfortunately, this “fruitful avenue of future research” was left undeveloped. Some authors like
Bulow and Shoven (1978); Gertner and Scharfstein (1991) discuss the issue, but fail to provide
the optimal contract design in the face of debtholder conflicts.
Garleanu and Zwiebel in their 2006 “Design and Renegotiation of Debt Covenants” also
consider the effect of design and renegotiation of covenants in debt contracts as a “specific
example of the contractual assignment of property rights under asymmetric information”
(Garleanu and Zwiebel 2006). In particular, they explore the situation where managers have
more information than the bank on future investment transfers from debt to equity. The findings
23
of their research indicate that this adverse selection problem results in allocation of greater exante decision rights to the less informed party – which is the creditor – that would be under
symmetric information condition. The practical implication would be tighter covenants upon
interception and their further waive after the negotiation (creditor is giving away the excessive
rights). The results of Garleanu and Zwiebel research contradict the previous incomplete
contracting models – they were assuming that management is the one gaining excessive rights ex
ante. The authors rely on 2005 work by Dessein (2005) “Information and Control in Ventures
and Alliances” when they consider control rights allocation under asymmetric information.
Dessein, however, did not explore renegotiations and was more focused on private entrepreneurinvestor relations. His findings indicate that the entrepreneur (informed party) ceases control
rights to the venture capitalist (the uninformed party) in order to signal congruent preferences.
(Dessein 2005)
To sum up, incomplete contracts approach mainly focuses on information asymmetries
between lenders and borrowers in debt contracts and suggests covenants are a way to reduce
them – while debt renegotiation makes the contract more effective, covenants are actually
embedded options to renegotiate.
We observe that main three approaches to covenants include agency theory, incomplete
contracts approach and various accounting theories on the subject. The agency approach treats
covenants as means to resolve the agency conflict between the debtholders and shareholders and
to reduce agency costs of debt. Accounting theories focus on their impact on accounting
decisions and figures, mostly via the covenant breaches. Research on incomplete contracts treat
covenants as means to reduce information asymmetries.
The three approaches to covenants more or less exhaustively construct the theoretical
background of the covenant research. Belyaev (2015) also emphasizes on results of resourcebased theory of strategic management as being important for covenant studies. As this
perspective is not specifically relevant for the research, it is still worth mentioning to provide the
full picture on the subject.
Belyaev claims, that resource-based theory stems the success of the company in its
possession of the specific assets that give the company competitive advantages. For instance,
Williamson distinguishes organizational routines and core competences that are possible due to
allocation of specific resources into clusters (Williamson 1999). Belyaev suggests, that credit
risk is directly related to unique company resources. When company loses its own core
competences, its ability to conduct business smoothly is jeopardized. That might be followed by
future problems with the ability to make interest payments, thus the credit risk is increased. The
ideas of strategic management resource-based theory suggest that creditors should be interested
24
in borrowers retaining their core competences and unique resources. Belyaev argues that lenders
can use specific positive covenants for that reason. For example, if the unique management
competence is the competitive advantage of the organization, then loan covenant can demand
from the borrower to retain the key managers, keep the managerial structure or distribution of
responsibilities. (Belyaev, 2015)
Belyaev refers to work of Bishara and Orozco while discussing the implications of the
resource-based theory. In their 2012 article, “Using Resource-Based Theory to Determine
Covenant Not to Compete Legitimacy” authors explored the nature of modern knowledge-based
economy and concluded that today main assets of the company are not comprised of production
facilities, inventory and other tangibles. Instead, leaders of the industry like Apple, Google or
Facebook constantly generate new ideas that develop into innovations. (Bishara & Orozco, 2012)
Aside from new ideas and commercialized innovations, Belyaev selects managerial practices,
policies regarding the health, safety, and environment (HSE) as new kind of company assets that
are becoming a crucial factor of the credit risk of the borrower. The idea is to incorporate the
new reality into the covenants that are stuck in world of financial ratios and company reporting.
Banks can start to include covenants that protect strategic intangible assets of the firm in the loan
agreements to maintain the credit risk of the borrower. Moreover, covenants can be a mechanism
that controls the borrower’s compliance of ecological and social norms and rules. (Belyaev 2015)
1.5 The effect of covenant loan structure on borrowing terms
We have discussed the main theoretical approaches to covenants and specifically focused
on the ways they can resolve certain problems. It turned out, that various theories isolate
different benefits that covenants provide while included in the debt contracts: from resolving the
agency conflict and minimizing the agency costs to making debt contracts more effective via
renegotiation and establishing the oversight mechanisms. We must admit that all approaches
have their practical point. K. Paglia and D. Mullineaux in their 2006 article “An Empirical
Exploration of Financial Covenants in Large Bank Loans” studied the reasons for inclusion of
covenants in the contracts and found out that their use is affected by potential agency problems,
information asymmetries, incentives to monitor, and growth opportunities alike.
(Paglia and Mullineaux, 2006) All theories, however, conclude that the effect is double-sided:
both lender and borrower should be voting for the inclusion of covenants in the contract.
In order to proceed with the research we have to estimate the practical effects that
covenants have on the loan agreement. Most of them are already mentioned as a part of
theoretical background. We will continue with a short recap of the main practical consequences
of inclusion of covenants on the loan agreements from previously discussed theoretical
25
approaches and then list the findings from the empirical academic literature. These articles
commonly test the relations between covenants and other variables inside the specific
environment. They are mostly not related to aforementioned theoretical frameworks – the
research is primarily focused on relations and covariance.
As we discussed earlier, according to research on Agency Theory of Covenants, both
lender and borrower enjoy the decreased agency costs when special provisions are included in
the debt contract. However, reducing agency costs comes at a price that both the borrower and
lender have to pay. At first, covenants reduce the managerial flexibility that may be both a good
and a bad thing at the same time. On the one hand, reducing the decision-making power of the
management may protect the debtholders from detrimental company strategies. On the other
hand, less flexibility may result in lower firm value as company would have to pass out on
possible profitable transactions. Moreover, bank can establish a better monitoring system via
covenants. In any case, cost that borrower pays is still a gain for the lender. The bank has to pay
the price as well – by reducing the cost of debt for the lender.
The accounting perspective on covenant research mainly studies the covenant violations
and managers’ decisions when violations are close or already happened. It turned out that
accounting choices of managers clearly depend on covenant thresholds. This is also a cost for the
company – by including covenants in the contract, managers would make certain choices to
maintain the appropriate accounting figures rather than thinking about company value instead.
The bank, on the other hand, will probably divide the breaches into technical and material and
will waive the violation if the breach is not related to financial distress.
Finally, the incomplete contracts approach sees covenants as a way to make debt
contracts more effective – they are perceived as an option to renegotiate the terms. Renegotiation
allows reducing the information asymmetries between parties and allocating control rights better.
As a result, we get a more efficient contract where banks do not charge extra cost of debt for
uncertainties and borrower’s decision-making is more transparent.
While different perspectives on covenants may contradict each other on specifics – like
would companies with significant growth opportunities rely on covenants – they agree on basics:
covenants are beneficial for both lender and borrower (for various reasons), both parties pay the
price for including them in the contract but also enjoy the benefits. The gains of the company are
most material: theory suggests that contract with provisions would be cheaper for the borrower
than agreement without them.
The idea, that presence of restrictive covenants reduces the yield on debt is generally
obvious in theory, but needs empirical confirmation. We find support for this statement in an
article by Russian researcher Anna Zadorozhnaya “The Influence of Covenant Protection on
26
Yield of Corporate Bonds” (2015). The article presents the results of empirical study of how
covenant protection affects the cost of debt. Covenants mitigate problems of risk transfer in
agency conflicts, and influence the corporate financial policy with cost of debt reduction.
Research results demonstrate that a negative relation between bond spread and the presence of
covenants exists, which is consistent with the costly contracting hypothesis (CCH), but
registered only in Eurobond market. So, if the bond covenants on domestic market in Russia are
not affecting cost of debt, then valuation of loan covenants with instruments of real option
analysis is becoming more significant for all market participants. (Zadorozhnaya 2015)
Article by Deng et. al. “The Role of Debt Covenants in the Investment Grade Bond
Market – The REIT Experiment” (2015) examines the investment grade bonds and covenant
protection. Investment grade bonds usually do not have covenants in debt terms. The rationale is
that investment grade firms are financially stable and covenants do not provide extra benefits.
Authors, however, claim that in the case of Real Estate Investment Trusts (REITs), investment
grade REITs tend to hve a covenant protection with leverage limits and requirements to maintain
particular interest coverage ratios and fixed charges. This unique environment of REIT bonds
gave a perfect opportunity for the researchers to examine the importance of debt covenants on
debt markets. The findings of the article is somewhat surprising: debt covenants are indeed
common among investment grade REITs in the REIT market and solid evidence exists of their
higher use compared to non-investment grade REITs. Authors show that debt covenants are
rarely a mandatory term in the market, as investment grade REITs choose covenant provisions
based on accounting ratios for which they have enough slack. Finally, article provides evidence
for lower cost of debt when these investment grade REIT bonds are issued with covenants. The
last finding supports the idea of negative relation between yield and presence of covenants,
suggested and tested by other researchers. (Deng et. al., 2015)
Jin Yu (2010) in the work “The Value of Debt Covenants: A Quantitative Framework”
was the first to incorporate debt covenants into dynamic structural model with endogenous
choice of real investment and capital structure in the presence of agency conflicts. This research
framework is crucial for any research on covenants as it includes calibrated simulations – with
current unavailability of data, simulated numbers become a good substitution. Findings by Yu
indicate that coverage ratio covenants dominate book leverage ratio covenants. With optimal
interest coverage (book leverage) ratio covenants, market leverage ratio increases from 26% to
39% (37%) and overall firm value increases by 3.0% (2.5%) and the market leverage ratio
increases from 26% to 39% (37%). Yu concludes that covenants and their violations are
important determinants of firm’s investment and financing policies and provides model for
optimal covenant tightness. (Yu 2010)
27
Finally, N. Reisel in her 2004 article “On the Value of Restrictive Covenants: An
Empirical Investigation of Public Bond Issues” contributes to the covenant theory by
investigating the costs of including covenants in the bond issues. The results of her research
again prove that lender pays the borrower with decreased costs of debt for covenant protection.
Reisel finds out that on average firms can reduce the costs of debt by astonishing 311 basis
points by including restrictive financial covenants in the contract. She also tested the same
relationship with asset sale, payout and investment covenants, but did not find the significant
relationship. Reisel suggests that results of the research imply that high growth firms should find
it too costly to include covenants that restrict investments, asset sale and payoff. (Reisel, 2004)
This is contradictory to results of 2007 empirical work “Growth Opportunities and the Choice of
Leverage, Debt Maturity, and Covenants” by M. Billet, T. King and D. Mauer, who find a
positive relation between restrictive covenants and growth opportunities, debt maturity and
leverage.
The recent empirical works on the subject again prove that covenants in the loan
contracts have a certain value and lender pays its part by lowering the costs of debt for the
borrower.
2. Methodology and Data
2.1 Propositions
As we mentioned in the introduction, the aim of this research is to calculate the potential
of an interest rate decrease that is available for loan with a specific covenant. By “potential for
an interest rate decrease” we understand the ability of borrower to obtain lower interest rate on
loan through negotiation by referring to the extra option value that covenant provides the bank.
As we have observed in the literature review section, various theoretical studies imply that debt
covenants are beneficial for both parties (lender and borrower) and both parties have to pay for
them. The lender pays via decreased costs of debt. If so, we are going to estimate the value of the
specific covenant in the contract using the real options approach. If the value is more than zero, it
means that all else equal, contract with the covenant is more beneficial for the bank than light
covenant loan structure (covenant-free debt). However, we assume that bank should not offer
loans of different cost to the same borrower at a specific point in time, thus covenant of a
specific tightness can be used by a borrower to justify interest rate discount.
In order to achieve the research goal, several research objectives should be met:
Define the term “loan covenant”, classify existing covenants and determine their
roles in loan process and debt contracts;
Identify theoretical approaches to covenants and related borrower and lender
activity;
28
Evaluate the effects of covenants on both lender and borrower;
Create a model of a firm that has a loan with a covenant in its debt structure, and
estimate the value of real options this covenant provides;
Calculate how much the interest rate on loan with covenant can be reduced
compared to the covenant-free loan
Several obstacles stand in the path of the research. We need to overcome them to achieve
the research goal successfully:
At first, every company is unique and has a different set of characteristics. The oil
extraction company from the UK with low share of debt and high-levered Russian
telecommunications industry player are different in terms of cash flows, revenues, margins, etc.
Therefore, each company will receive individual set of covenants of different tightness. Even
within the same country and industry, companies have different accounting ratios that need to be
maintained. Thus, each possible effect of covenants on loan terms would be different. In order to
overcome this obstacle we would focus on one specific industry, which is Russian oil industry.
We are going to use the industry data on distribution of company characteristics to simulate the
accounting variables of the generic company. This is going to be a dataset for the research.
Second, there is a degree of uncertainty of how to incorporate management decisions.
Managers create a strategy for the company and decide, whether the firm should borrow cash,
invest in projects, etc. Simulating some of the variables can be invalid, unless management
decisions are incorporated into the model or specific assumptions are made. Moreover, as we
have observed in the accounting approach to covenants, managers tend to make revenuegenerating accounting choices in order to prevent the covenants from breach. In our model, we
are not going to incorporate the preventive managerial activity before the breach, as it is a part of
a separate research and a good area of future development. At this point in time, it is crucial to
build a stable working model without overcomplicating the results. As for the variables not prone
to simulation – we are going to make industry-specific and variable-specific assumptions
concerning each of them, which will be presented later on.
Third, many covenants exist in the contract at the same time and each probably has its
own impact on the loan. As a part of the research, we decided to focus on debt service coverage
ratio (DSCR) and stick to the threshold of this covenant. In Corporate Finance DSCR is a
measure of the cash flow available to pay current debt obligations. The current debt obligations
include interest, principal and lease payments. The ratio is calculated as (Rehmann 2013):
DSCR=
Net operating income
Total Debt service
(1)
The higher the ratio is, the easier is to obtain the loan. On practice, Earnings before
Interest and Taxes (EBIT) are used as a proxy of the available cash flow. During the simulation
29
procedure, at each iteration the actual values of DSCR will be compared to the threshold value at
each period. In case of a breach, lender’s actions will be incorporated in the model and value of a
covenant will be calculated. Later, the distribution of values of a covenant will be then presented
with a choice available for the company (managerial implications). We are going to assume that
DSCR is the only covenant in the loan contract of the firm.
Fourth, what are the consequences of covenant breach? The violation of DSCR covenant
means that the borrower failed to maintain the minimum DSCR ratio (threshold). If the bank sets
the DSCR covenant ratio to 3, then the borrower would violate it if the actual ratio goes lower –
like 2, or 1, or 0, or even -1 (in case EBIT is negative). From different theoretical approaches to
covenants we have observed that lender can interact with the borrower in many different ways
after the covenant violation. At first, bank can announce the technical default of the borrower and
demand an early repayment of the loan. Second, bank can impose sanctions on the borrower and
increase the interest rate of the debt, add more covenants, take control over some of the
operations, and many more. Finally, lender can give a waiver and leave it “as is” with hope that
situation is going to improve in the future.
We are going to incorporate the approach of Borgonovo & Gatti (2013) to covenant
violations in our model. They divide the breaches into material (possible debt services default
due to inability of the borrower tp pay debt-related payments) and technical (credit risk is not
changing, no risk of future default – accidental covenant violation).
In case of the technical breach, we assume that bank utilizes the cash sweep strategy. The
cash sweep is the mandatory use of excess free cash flow to pay down outstanding debt rather
than distribute it to shareholders (Ventureline 2015). In other words, lender restricts the dividend
payments to shareholders in the period when the breach occurred. This is not the only way to
deal with the technical covenant violation – as we mentioned in the theory section banks may
waive the violation and let the company continue its operations without any sanctions. (Chen and
Wei 1993; Beneish and Press 1995) However, it is hard to determine the value of this option to
engage as the decision to waive should be supported by careful evaluation of company’s risk
profile. What if the borrower is clearly off track and waive is guaranteed only if management do
some revenue-generating decisions? Cash sweep is the solution as the outcome is clear, decision
it is certainly in the interest of debtholders and it eliminates the detrimental strategy of
unauthorized distribution of assets (agency conflict).
In case of the material breach, lenders will demand an early repayment of the outstanding
debt at its present value. The cash is taken from asset sales and free cash flow. The material
breach is not exactly a default – but a signal for a future one. If the company is unable to make
30
debt service payments in full out of cash flow and current assets in this period, then it is in
default. While material breach happens in Y n the following default happens in period Yn+1.
Simulation is essential part of our analysis as it allows us to model the behavior of the
firm in the long run, see how specific covenant is breached and what consequences that will
have. Researchers have used simulation before in their works devoted to covenants. For
example, Gamba and Triantis (2013) in their article “How Effectively Can Debt Covenants
Alleviate Financial Agency Problems?” examine how effective debt covenants can be in moving
shareholders’ investment and financing policies closer towards first-best policies, and the extent
to which agency costs can thus be mitigated. Methodology of their research includes modelling
the investment and financing decisions in an infinite-horizon discrete-time dynamic and
stochastic framework. The control variables are the book value of assets in place and the face
value of outstanding debt. At first, authors model the finite set of heterogeneous firms in the
economy, each driven by an independent company specific risk, but then describe the behavior of
the individual firm.
Another methodological step is the application of the real option analysis to the simulated
data. After we have understood how company’s accounting variables are going to act over time
and what is the effect of covenant breach, it is necessary to estimate the additional value the
specific covenant provides the bank. We are going to apply the real options methodology, which
is quite similar to financial options framework. Some authors included covenants into the real
options analysis frameworks; however, their research topics are not relevant for this Master
Thesis.
In the next sections, we are going to describe the process of model creation, simulation of
the outcomes and option value of DSCR covenant calculation.
2.2 Model description
There is a firm that belongs to shareholders and equityholders alike. A firm comes from
oil industry in Russia. The firm starts at time zero (t=0) and its condition would be observed in
the next ten periods (t{0,1,2…10})
There are several reasons why we decided to use the industry data of oil companies. At
first, there is a sufficient number of similar firms in the business and they operate under the same
market conditions. Second, the accounting data we used is prepared under the same IFRS
standards and similar approach to asset valuation, revenue recognition and debt valuation is used.
Third, this is one of the oldest industries in modern Russia, thus the accounting data for more
than 10 years is available. Forth, Russian oil businesses have long relations with the banks; they
are familiar with the debt financing and can borrow at comparable interest rates. Finally, oil
31
industry is considered a stable one, even in the period of low oil prices, with equity beta being
close to one for all companies in the list. (Investfunds.ru, 2016) A stable industry may expect
stable cash flows in the next ten years.
The firm enters time zero with a certain amount of assets, total debt/total equity ratio,
dividend payout rate, and amount of existing debt service payments. It borrows the amount of N
US dollars at the interest rate i and has to repay the debt in equal installments during the next 10
periods. Principal payments are made during the whole duration of the loan. Bank sets the
threshold for the Debt Service Coverage Ratio covenant at the amount of T. In case of the
threshold violation in any given period, bank can call the technical default on debt and demand
the early repayment of the outstanding debt at PV. However, as we mentioned earlier, the bank
would divide the breaches into material and technical and utilize the “cash sweep” strategy if the
breach is not related to possible default.
It is necessary to mention the meaning of “covenant threshold” here. For DSCR covenant,
the threshold value is the minimum value the company should maintain throughout the duration
of a loan. If the value of DSCR falls beyond the threshold, that means that company is either not
maintaining the sufficient levels of operating income, or is paying excessive debt payments, or
both. The value of the threshold is determined by bank in our case; however, in reality both
lender and borrower have to agree on the threshold.
The future states of the company should be forecasted. You may find two main
approaches to forecasting the value of the parameters in literature: qualitative and quantitative.
Qualitative forecasting is used when no reliable historical or comparable data exists and it is
based on expert opinion approach, management expectations, polling, surveys, etc. These
methods are not applicable to our research. Quantitative forecasting requires reliable numerical
historical or comparable data that can be divided into time-series, cross-sectional or panel. Here
are some well-known methods of quantitative forecasting:
ARIMA and auto-ARIMA (autoregressive integrated moving average) takes the
historical time-series data to perform the back-fitting optimization routine to
account for autocorrelation. The model’s advantage as it corrects the
nonstationary characteristics of the data for the sake of stability and learns over
time by correcting the forecasting errors.
Multivariate regression models the relationship structure of the dependent variable
on the exogenous variables. The model itself can be linear and non-linear.
Maximum likelihood estimation is used to forecast the probability of the event
with the independent variables. The approach can be used on Logit, Probit and
Tobit regression models and is effective when traditional regression models fail
(for instance, predict the probability more than one)
32
ARCH and GARCH (generalized autoregressive conditional heteroscedasticity)
models are used to characterize and model time series. In the financial modelling
they are particularly effective in estimating the historical volatility and predicting
the future volatility of a marketable security. Many variations of the original
techniques exist now.
Stochastic process forecasting is used when variables can’t be predicted using the
traditional means of regression analysis. A stochastic process is a sequence of
events of probabilistic nature. The main stochastic processes include Random
Walk, Geometric Brownian motion, Jump-Diffusion, etc.
Markov chains, non-linear extrapolation, etc.
In our research, we are going to us the Geometric Brownian Motion to forecast future
values of some parameters of the company and Maximum Likelihood models with GARCH (1,1)
to obtain the historical volatilities of these parameters.
The data for the oil industry was used from the 2016 work by Igor Sennikov
“Implementation of Simultaneous Equations Model to Forecasting Residual Income of Russian
Oil Companies” and verified with Thomson Reuters Eikon.
We took the largest Russian
companies in the industry and collected their balance sheet and income statement figures to
estimate the amount of sales revenue (SR), operating expenses (OPEX), total assets (TA),
Debt/Equity ratio (D/E), share of debt service payments in total debt outstanding (P), portion of
quick assets in total assets (Q), and dividend payout ratio (Div) for the period from 2004 to
2014. The full list of the firms can be found the Appendix 1.
After the panel data on variables was obtained, we calculated the industry averages of the
revenue, OPEX and total assets for each year to eliminate the discrepancies coming from poor
management, size of the company and internal factors. It turned out that for ten years firms on
average gained net operating profit and experienced asset growth.
The future revenues, costs and asset values were simulated for the next 10 periods using
the geometric Brownian motion (GBM).
GBM is certainly not the best approach to use when forecasting the future amounts of
company revenues or operating costs; however, it is intuitive, applicable to the simulation
procedure and incorporates uncertainty of outcomes that is required for option analysis. The
process itself takes the form of:
S
=( t ) +σξ √ dt
S
;
(2)
Where
S
is the previous value of the variable,
S
is the change in the variable’s value in the next step,
33
is the annualized volatility,
σ
is the annualized growth or drift rate
By using this model, we assume that each period the variables will “drift up” by a
specific amount, but also will experience a shock which will be a standard deviation ( σ)
multiplied by a random number (ξ). (Brewer et al. 2012) We are not applying the GBM directly
to the cost or revenue values, but instead we are using the log returns of the variables as a proxy.
The log returns have a sound meaning when they are applied to asset prices – they represent a
convenient mathematical proxy of the rate of return investor in the asset gets after the prices
change. Log returns have a following formula:
R t =log (
Pt
) ;
P t−1
(3)
Where,
Rt is the log return of the asset
Pt is the price of asset in period t
Pt-1 is the price of asset in period t-1
In our case, log returns help us to convert the non-stationary data of revenues, costs and
asset values to stationary log differences thus allowing us to apply the Geometric Brownian
Motion. The drift of the stochastic process was calculated as the average value of the log
differences and the volatility of the returns (sigma) was obtained by building the GARCH (1;1)
model.
Standard deviation (sigma) can be obtained by using several methods: historical variance;
applied variance and maximum likelihood method. The historical variance method is the easiest
one to calculate but the results it produces are the most unreliable. (Hull, 2012) In order to
evaluate the applied sigma the underlying asset should be actively traded on the market which is
not the case of our research. Therefore we used the maximum likelihood method in order to
estimate the parameters of GARCH (1;1) and find the long-term volatility.
The equation for GARCH (1;1) sigma is the following:
2
2
2
σ n =γ V L +a un−1 +β σ n−1 ;
(4)
γ +β +a=1∧ω=γ V L ;
(5)
where
VL
is the long-term volatility, that was required for the GBM. The first step was to
calculate the historical sample variance and assume that it is V L , when we set some random
ω ,
β
and find a
as 1-
β
– ω / V L . The likelihood measure is:
2
– ln ( v i )−ui /2 . ;
(6)
34
Its sum has to be maximized by choosing ω ,
β
and a
via the iterative search
procedure, with the following restrictions:
ω ≥ 0 , ≤ 1,a ≥ 0, b ≤ 1∧b ≥ 0. ;
(7)
After finding the optimal values of the parameters, we can find the long-term variance as
V L=ω /(1−α−β ) ;
(8)
After the volatility of differences of industry revenues, operating expenses and total
assets was estimated, we used the methods of VBA programming to simulate the future value of
assets, revenues and operating expenses. The number of iterations was set as 50 and then the
average outcome was taken for a reason to control for the managerial actions. We suggest that
although the revenues and operating expenses of the firm may follow a random walk on the time
period of ten years, managers still have the influence to use accounting to move the figures to
average values. Moreover, by averaging out the random walk we lower the amount of future
simulations we would have to take in order to estimate the distribution of covenant violations.
Several assumptions had to be made in order to proceed with the simulation. At first, we
were interested in the amount of interest payments the firm makes for the debt that was issued
before time zero. We have analyzed the industry time-series data of interest payments in the
Thomson Reuters Eikon (Thomson Reuters 2016) and concluded that annual interest payments
of the companies constitute a fixed share in the total debt outstanding. For instance, in the
Russian oil industry interest payments are usually 6% of the Total Debt – which makes sense,
that on average companies simply pay the annual cost of debt as interest and accounting
discrepancies are ruled out. Therefore, for the sake of simplicity we make an assumption 1:
annual interest payments, other than newly borrowed debt, represent the fixed share of value of
liabilities.
Second, dividend payments vary from period to period. Again, for the sake of simplicity
we make an assumption 2: each year the company has the same fixed dividend payout ratio (if
the income attributable to shareholders is more than 0). The ratios were taken from Thomson
Reuters Eikon industry averages section and then checked on the basis of the academic research
papers. For instance, average dividend payout ratio in oil industry is 20% according to Thomson
Reuters and D. Romanov independently proved that in his 2014 “Dividend Policy of Russian
Oil&Gas Public Companies”
Third, we forecasted the future values of the D/E ratio. We considered the ratio as the
amount of total company liabilities divided by total value of equity, so that D+E=Total assets.
Therefore, the values of both company debt and equity were derived from the simulated value of
assets. We further make an assumption 3: the values of D/E ratio are normally distributed among
the companies with a mean equal to industry mean and variance equal to industry variance. We
35
calculated the industry mean and variance of the D/E ratio on the basis of its historical ratio.
Then we used the Monte-Carlo simulation method to generate random D/E ratios for the
company in each time period under the standard normal distribution. We did not consider
including the ratio values in the GBM process as the proportions of debt and equity do not follow
a random walk of any kind, but are rather a product of the CFO activity. There is a certain body
of literature that studies company leverage: some suggest that managers tend to stick to a target
ratio (Korteweg 2010), other argue that leverage is a product of managers trying to benefit on the
signals they send to market (Ross 1977) or that it represents how much managers rely on the
equity as a last resort. (Myers and Majluf 1984) Although it is a topic of a separate research, we
have observed the evidence that supports all of the findings within different firms. Therefore, we
can suggest that on average, the strategies of managers will be ruled out and our firm would have
a randomly distributed leverage ratio for the 10 year horizon, with the mean value being “target”
leverage. The borrower strives to achieve the target capital structure and makes certain
adjustments each period, but cannot hold the target leverage in the long term. As a result, certain
discrepancies exist each period (that will be determined by the standard deviation in the MonteCarlo simulation procedure)
Forth, in our model we have to control for the income taxes. As companies tend to defer
tax payments or pay taxes in advance for the following periods, tax payments won’t nessesary
represent the net income multiplied by income tax rate. However, taxes play secondary role in
our model and are primarily used to calculate the amount of income attributable to shareholders
and excess cash flow to be paid out in a form of dividends. Thus, we make the assumption 4 that
each period the company would pay income taxes in the amount Tx equal to:
Tx=¿∗TR ;
(9)
Where NI is net income and TR is income tax rate that is equal to 20% in Russia.
It is worth mentioning, that Russian oil industry, which was used as the source of base
data of the analysis, has a different tax structure, than other sectors. However, some works have
proved that still on average the tax for the upstream operations is 20% (Sennikov, 2016)
Fifth, in our research we suggest that one of the possible outcomes of the simulation may
be the default of the borrower (firm) at any given period. In our model, we call a default the
inability to pay the interest and principal payments in full due to lack of cash flow and quick
assets in place. In other words, if the amount of all interest payments and principal payments to
the bank is higher than company’s EBIT plus quick assets, then we can call it a default. By quick
assets we assume assets that are equal to cash or can be quickly converted to cash without
substantial loss of value (liquid inventories, marketable securities, etc.) (Encyclopedia of
Finance). After careful analysis of the historical data, we concluded that it is possible to make
36
assumption 5 that share of the quick assets in total assets is a variable that is normally distributed
among the companies with a mean equal to industry mean and variance equal to industry
variance. Again, we used the Monte-Carlo method to generate random share of quick assets for
our company in each period.
Finally, when all the exogenous parameters were simulated and necessary industryspecific assumptions were made we were able to calculate the EBIT (earnings before interest and
taxes) value as difference between revenue and operating costs and Net Income as EBIT minus
interest payments and tax payments. After that, we were able to calculate the value of DCSR
covenant in each period and compare it with the threshold to identify the violations.
There were also other options for the model creation: we could use the specific company
data for the simulation, set the main parameters randomly, or determine them at our own
discretion. However, these choices would make the model either too specific or too indefinite for
the research. If we chose the company data on revenues, costs and assets, we would have
incorporated its specific strategy and management style into the model, thus making the results
irrelevant in a broad sense. If the parameters were set randomly, then the practical implications
of the outcome would be questionable. If the parameters of the simulation were picked by
ourselves, then results cannot be truly reliable.
To sum up, we have created a model of a company in a certain industry with the main
parameters being revenue, operating costs, EBIT, interest payments, tax payments, net income
dividend payments, assets, equity and liabilities value. Revenues, costs and assets were
forecasted with their log differences being taken as stochastic process and applied to Geometric
Brownian Motion. The values of other parameters were obtained under specific assumptions. By
creating this model, we can observe the possible covenant violations for the generic company on
a ten period time horizon.
2.3. Covenant violations
The simulation procedure allow us to get possible infinite outcomes of future company
performance. The borrower has a choice now: it can either give a loan without any covenants or
include the DSCR provision in the terms. If the loan terms are covenant-free, then the borrower
can still observe the probable “violations,” but can do nothing about them. With the DSCR
covenant included, however, the bank can engage when the breaches occur. We run the
simulation procedure to obtain the distribution and probability of both technical and material
covenant breaches for a specific threshold.
Consider a loan agreement with covenants. If the CB is the event “covenant is violated,”
let us consider (10) as the probability of this event:
37
p cb=PR(CB) ;
(10)
According to Borgonovo and Gatti (2013), the material breach is associated only with a
subset of states in which a violation occurs. Then, we can assume that (11) and (12) denote the
probabilities of technical breach and material breach respectively:
pTB =PR(TB) ;
(11)
p MB =PR( MB) ;
(12)
The two events are considered disjointed event though the state of the world with
technical violation is equal or included to the state of the world with material violation.
According to some research, the default probability can be calculated as the relative
frequency of events of default across scenarios. (Elsinger et al. 2006) We are able to observe the
events of default in our model and with substantially large number of simulations can get the
default probability for our firm. By running the simulation, we can also get the probabilities of
technical and material breaches, which are equal to:
pTB = lim
N →∞
nTB (N )
,
N
p MB= lim
N →∞
n MB (N )
N
(13)
If we introduce a Boolean variable U Tb(n), that counts 1 when the breach is technical and
0 if its not, then:
nTB =∑ U Tb (n) and
^
ptb =∑ U Tb ( n ) / N ;
where N is the number of simulations and
^
ptb
(14)
is our estimate for the probability of
technical violation. Same applies to material breach.
This is the approach to division of breaches into technical and material originally
proposed by E. Borgonovo and S. Gatti (2013) and we are happy to partly employ it in the
model.
Let us consider a scenario, when the event of default happens. In our case, the default of
the borrower in period t is confirmed when:
Dp t >CF t + LA t ;
(15)
Where:
Dpt is the debt payments in the period t,
CFt is the Cash flow to firm in the period t and
LAt is the amount of liquid assets in place that can be used immediately for debt
payments
In other cases, the company is still able to pay its obligations in period t and default is not
considered, even if borrower is unlikely to make debt payments in the next period t+1.
38
The event of default is accompanied by DSCR covenant violation, but we are not
considering it as a material breach. At this scenario, the borrower will not be able to utilize any
of the strategies, as company will not have sufficient funds for that.
Borgonovo and Gatti separate the technical and material violations on the basis of the
numerical value of the threshold. In their words:
“…contract usually includes two sets of values for this financial covenant. The first
indicates the minimum threshold below which lenders can require an accelerated repayment of
the loan; the second, which is lower than the first, triggers the resolution of the credit agreement
and represents the material breach of covenant. The joint consideration of the two sets of values
generates a range of values for DSCR associated with a technical default that is frequently
waived by means of a renegotiation of the loan repayment terms.” (Borgonovo and Gatti 2013).
We used a different criterion for the separation. Imagine that in period t, the company
fails to maintain minimum level of DSCR ratio, thus violating its covenant. If there is a default in
the period t+1, we suggest that covenant breach was a predictor of a financial distress in a next
period and thus the breach is material. However, if the company is still able to make its debt
payments in t+1, then the breach is considered technical (the violation happened due to
accounting reasons and does not represent the future default on payments). This does not mean,
however, that technical and material breaches are not connected – the borrower may experience a
streak of covenant violations, with n of them being technical, but final one before default being
material. If the breach is technical, the bank uses the “cash sweep” strategy and restricts the
dividend payments in the period. Dividends are considered an excess cash flow and used to
repay some of the outstanding debt. The debt service payments are recalculated for the following
periods. The cash sweep strategy is an adequate answer from borrower – if excess cash flow is
used to repay the outstanding debt, then interest payments decrease in the following periods and
the probability that a material breach happens in future decreases. In addition, academic research
states that credit risk rises right before the violation (Fargher et. al. 2001), so bank tries to lower
it back. If the breach is material bank sees it as a threat to future debt payments and demand the
early repayment of the outstanding principal. We assume, that lender can positively identify both
technical breaches and material breaches (for instance, by using the presented model). In other
words, the uncertainty in bank’s actions is ruled out. The lender will not miss any violations and
will positively identify which breach is technical and which is material. In order to price the
option well we need to use backward induction for the bank actions – the lender should have all
the information regarding the future states and be able to make payoff-maximizing decisions.
Finally, the borrower may be unable to make debt payments in period t even if previously
there were no violations of covenants. The lender is not protected from this sudden default with
39
the DSCR covenant. We assume that in case of default in period t the bank will receive a
payment that represent the portion of the loan value (50-100%) in period t+1.
For the research we have run 4000 simulations of future states of the company, obtained
the distribution of covenant violations and probable states of default. Then we incorporated the
bank actions given the covenant violations and obtained the loan value playoffs for the lender at
each run. The detailed findings are presented in the third part of the Thesis.
2.4. Real options analysis
After we have modelled the company characteristics, observed the situations when the
covenant is breached, we calculated possible value of the options that covenant provides to
lender. What real options exist for the bank in this case?
In general, the covenant can be considered as an option “to exercise control” or simply
“to engage” for the lender (incomplete contracts approach focuses on “option to renegotiate”).
When the borrower violates the covenant, the bank can make payoff-maximizing or riskdecreasing decisions, which are unavailable if covenants are not present. In our case, the two
strategies that are available for the bank are cash sweep and early repayment.
The general “option to engage” can be decomposed into more specific opportunities for
the lender. At first, when the bank observes the material breach, it demands early repayment of
the principal. The material breach is a predictor of a company default in the next period and
thereby the bank is able to avoid the uncertainties associated with borrower default and losses
related to time value of money – the default payment will represent only a portion of the
outstanding principal and will happen after some time as borrower settles the amounts with other
creditors and obtains the financing (by selling assets, issuing stock, etc.). Technical default
payment is more valuable as it happens instantly after the material breach and covers all the
remaining principal. Thus, in our model DSCR covenant provides the bank an option to abandon
(or terminate) the loan when the PV of future debt payments falls below the required amount.
This act is effectively exercising of a put option.
Second, when the bank observes the technical violation of the covenant, it uses cash
sweep strategy to repay part of the outstanding principal. A technical breach signals the lender
that borrower may currently be off track with its operations. Here DSCR covenant gives the bank
a flexibility option on the debt payments – lender can vary them to help borrower avoid the
default.
Third, there are several options that we are not discussing in this work due to complexity
of the analysis involved. For example, by setting the tightness of covenants the borrower may
vary the tenure of the loan. If the threshold is too high for the DSCR provision, then company is
40
very likely to violate it even in the first years. Due to technical nature of these breaches (as
covenant strictness is not affecting the probability of default), loan will have an effective tenure
lower than 10 years (bank will constantly repay parts of the principal with cash sweeps). Another
option is a follow-up loan option: when the bank is terminating the loan (by calling the technical
default on the borrower), it can instantly lend the obtained money to another borrower with
better financial health. This act essentially increases the flexibility of bank portfolio
management.
If a covenant is a set of real options, then is should have a specific value, and according
to options theory, lender should pay the premium to borrower. The value paid to the borrower is
not simply monetary – it can be paid in form of lower costs of debt (in our case lower interest
rate). Is covenant truly a real option? According to Mun (2012), real options analysis is
applicable to situations when:
A financial model must exist. Real options analysis requires the use of an existing
DCF model, as real options build on the existing financial modeling techniques. If
a model does not exist, it means that strategic decisions have already been made
and no financial justifications are required, and hence, there is no need for
financial modeling or real options analysis. We are going to use the financial
model within our analysis and observe the future states of the company
Uncertainties must exist. Without uncertainty, the option value is worthless. If
everything is known in advance, then a DCF model is sufficient. Uncertainties are
observable here, as future state of the accounting variables within the company is
unknown and lender is risking the money if company cannot make debt service
payments.
Uncertainties must affect decisions when the firm is actively managing the project
and these uncertainties must affect the results of the financial model . These
uncertainties will become risks, and real options can be used to hedge the
downside risk and benefit on the upside. The uncertainty within the borrower
affects the decisions of the lender. Credit risks of the borrower for the bank is
obvious.
Management must have strategic flexibility or options to make midcourse
corrections when actively managing the projects. Otherwise, do not apply real
options analysis when there are no options or management flexibility to value.
With covenants in place banks have a flexibility to engage in the operations of the
borrower, or may let it go.
Management must be smart enough and credible enough to execute the options
when it becomes optimal to do so. Otherwise, real options are useless unless they
41
are correctly executed – both at the right time and under right conditions. We are
going to assume that lenders are behaving rationally and decide to choose the best
course of action when the covenants are violated. If the violation is not associated
with default risk, then banks will not demand immediate repayment as it will not
be beneficial for anybody. On the other hand, a clear default-type covenant
violation is regarded as an early repayment event. (Mun 2012)
The first application of option analysis for debt valuation happened when Robert Merton
in 1970s pioneered a different approach to credit assessment while interpreting the yield spread
on bond as being the option premium to the debtholders for selling a put option on the assets of
the company to equity holders. If the issuing company assets have a good performance,
debtholders simply receive their principal and stay away from participation (unlike
equityholders) If assets perform poorly, debtholders will receive a claim on those assets during
liquidation for whatever recovery value they can get, rather than the full principal value.
Debtholders lose an amount equal to the difference between what they lent to the company and
the recovery value of those assets in such event (the option’s “intrinsic value”), net of the “option
premium” which is the bond’s yield spread. (Merton 1974)
We could employ the Merton model to estimate the value of debt without and with
covenants and find the value of the option as a difference between these two. However, there are
specific drawbacks of the model that prevent us from doing so (assumed zero-coupon debt
structure is not realistic, stock market prices should be derived to reflect the true intrinsic value
of equity – and we can’t simulate those prices, etc.)
Another way to value the real option is to use the instruments of financial options, like
Black-Scholes option pricing model. S. Benninga characterized the use of Black-Scholes formula
for real options as “…no: Black-Scholes is not the appropriate tool. However, the Black-Scholes
model is by far the most numerically tractable (i.e., easiest) model we have for valuing options
of any kind. In valuing real options we often use the Black-Scholes model, realizing that at best it
can give an approximation to the actual option value. Such is life.” (Benninga 2014, p. 496)
Again, certain complications make the B-S approach ineffective in our case. First, the
option terms for the model are ambiguous – the intuition for the volatility of returns of the
underlying, time to maturity, maturity itself, etc. is either not clear or don’t make economic and
financial sense. Second, the diversity of outcomes and uncertainties may produce faulty results.
Other option pricing models – like binominal trees, Heath-Jarrow-Morton framework, finite
difference models and others – may be very effective for financial options but produce
complications or ambiguities in our case.
42
In order to estimate the value of our option, we will use the Monte Carlo method for
option pricing (Boyle 1977; Broadie and Glasserman 1996). It requires producing the large
number of possible paths for the underlying (which we have already done by now with our
simulation procedures), then calculating the payoffs (discounted debt payments) and finally
averaging them. We will calculate the payoffs for a loan with and without covenants, discount
the payoffs to obtain debt value and derive the option value from formula:
DVC= DV 0+OV
(16)
Where DVC is the debt value with a covenant,
DV0 is the value of covenant-free debt and OV is the option value
The loan value will be calculated as follows:
DV o =
P 1 + I 1+O 1 P 2 + I 2 +O 2
P + I +O
¿
+
+…+ n n n n + n+1n+2
1
2
(1+r)
(1+r )
( 1+r )
( 1+r )
(17)
Where Pn is principal payment in period n,
In is interest payment in period n (I+P is equal in each period if no changes are made),
On is other payments (cash sweep or early redemption due to technical default),
Defn is a payment the lender gets in case of borrower’s default (default payment), which
represents part of outstanding principal and paid in a year following the default
Finally, after the option price is estimated, we can compare the covenant-free debt
contract and contract with covenants of the same agreement terms to find out whether the
borrower can lower the interest rate to the lender to secure the same overall value of two
agreements. Moreover, we will provide the distribution of covenant values for different
thresholds, dividend payout ratios, industries, etc. In the end, practical implications of the results
will be discussed.
43
3. Results
3.1 Industry
For the reasons previously mentioned, the simulation procedure of the model uses the
data from the Russian oil industry. Therefore, the outcomes of the simulation are industry
specific. Nevertheless, with certain adjustments the model can be used with parameters of a
specific company, another industry averages, or even randomly generated parameters. The data
of Russian oil companies was taken for its reliability and accessibility and in order to
demonstrate the possible managerial implications under normal conditions. The results are
presented as follows: we are going to provide the short industry description first together with
results of simulation. Then we will present findings on the DCSR covenant violations and real
option value. Finally, we will analyze abnormal scenarios (extreme volatility and leverage) to
investigate whether results hold.
By Russian oil industry, we understand companies that have production and export of
crude oil and oil products as their main business. The oil companies are typically large integrated
players that benefit from economies of scale. The market value in 2015 was 71,1 billion USD
and is believed to grow by a compound rate of 5% annually by 2020 (Marketline 2016). The
market is quite competitive with seven giants (Rosneft, Lukoil, Bashneft, Surgutneftegaz,
Slavneft, Tatneft and Gazpromneft) occupying a share of more than 80% (Sennikov, 2016).
During 2004-2014 industry players experienced stable growth in revenues and assets due to high
commodity prices cycle and continuous increase in production volumes. According to our
estimates, the drift rates for assets values, revenues, and operating expenses were equal to 12%,
16% and 18% respectively and good financial health of the companies was only spoiled during
the financial crisis. However, aggressive two-digit growth is not expected in the following
periods due to various negative factors: from sharp decrease of oil prices in 2015 to revised
investment plans and economic sanctions by EU and US. (Henderson 2015) Therefore, the
further drift rates of the variables were adjusted in order to match the economic outlook more
accurately to 5%, 9% and 11% respectively. The Russian company output estimates are currently
seen as moderately optimistic. Four of the largest oil producers (Rosneft, Lukoil, Surgutneftegas,
and Tatneft) have all claimed that production in 2015 will remain flat compared to 2014 in the
worst-case scenario. We believe, that throughout the duration of the loan (10 years), the players
would be able to exploit their capability to attract stable financing to restructure the business and
invest in revenue-generating projects, as they all insist on in their investor presentations. The
volatility of changes of asset values, revenues and operating expenses estimated via GARCH
(1,1) remained low at 0,18, 0,21, and 0,22 respectively.
44
Russian oil companies use debt to finance investment projects and current operation, but
not too aggressively – during the period of 2004-2014 the average Debt to Equity ratios rarely
exceeded the value of 1. (Thomson Reuters 2016). We expect that future capital structure
policies will not change radically and industry will maintain the mean ratio of 0,81 with a
standard deviation of 0,21.
It is hard to tell how Russian oil companies use covenants in their bank loans, as this data
is usually a confidentiality issue. Thus, we did not incorporate existing covenants in the analysis
and simply introduced a new debt with DSCR condition to the model.
The full data upon which the simulation was built upon can be found in the Appendix 2.
Figure 1 illustrates the several possible runs of the model and represents future states of
revenues, operating costs and value of assets:
45
Object 75
Value
Simulation run 2
$180,000.00
$160,000.00
$140,000.00
$120,000.00
$100,000.00
$80,000.00
$60,000.00
$40,000.00
$20,000.00
$0.00
04 05 0 6 07 0 8 09 10 11 12 13 1 4
20 20 20 20 20 20 20 20 20 20 20
Y1
Y2
Y3
Y4
Y5
Y6
Y7
Y8
Y9 Y1 0
Period
Assets
Revenues
Costs
46
Simulation run 3
$140,000.00
$120,000.00
$100,000.00
Value
$80,000.00
$60,000.00
$40,000.00
$20,000.00
$0.00
04 0 5 06 07 08 09 10 11 1 2 1 3 1 4
20 20 20 20 20 20 20 20 20 20 20
Y1
Y2
Y3
Y4
Y5
Y6
Y7
Y8
Y9 Y10
Y4
Y5
Y6
Y7
Y8
Y9 Y10
Period
Assets
Revenues
Costs
Simulation run 4
$140,000.00
$120,000.00
$100,000.00
Value
$80,000.00
$60,000.00
$40,000.00
$20,000.00
$0.00
04 0 5 06 07 08 09 10 11 1 2 1 3 1 4
20 20 20 20 20 20 20 20 20 20 20
Y1
Y2
Y3
Period
Assets
Revenues
Costs
Figure 1. Possible future states of the modelled company.
Each run represents the possible (but not necessarily most possible) company
performance in the next ten periods. We have conducted 4000 runs of a simulation and obtained
a full range of future states of the abstract borrower. Each run produces different outcomes and
each outcome can be explained from the economic viewpoint. For instance: in the first run, we
can observe sharp rise in operating income and expenses with a more moderate growth of asset
value. This might be a result of company increasing the extraction rates at existing drilling sites
and hiring new workers in order to capture a bigger share of the market. That strategy pushed
both revenues and costs up, but low profit and conservative financing strategy resulted in lower
47
growth rates for assets. In the second run, we observe sharp increase in company revenues, but
moderate or no growth in assets or costs. The reason might be a positive shift in commodity
prices with devaluation of some company assets. Third run shows the situation, when both
revenues, costs and asset value experienced stable growth which might be the result of some
beneficial investments done by company management. Finally, in the fourth run of the
simulation procedure we witnessed the operating expenses exceeding operating income and
some gains in asset value. This can be a consequence of poor growth strategy, when management
fueled the expansion with raising new financing (debt, equity or both) while being unable to
maintain the profitability levels.
Table 1 represent the loan terms of the borrower that were used for the model:
Table 1. Loan dataXLoan data (mln USD)
Principal amount
Loan term (years)
Annual interest rate
Payments per year
Payment amount (Interest+Principal)
Discount rate for bank
$3 000,00; $7000,00; $15000;
10
7,00%; 9,00%
1
$427,13; $996,64; $2 135,66 for 7% loan
$467,46; $1 090,74; $2 337,30 for 9% loan
5,87%
Different principal amounts represent different cases – either the borrower is asking for a
relatively small debt (6% of the total assets – 3000 mln. USD), medium one (15% of the total
assets – 7000 mln. USD) or needs substantial amount of financing (30% of the total assets –
15000 USD). The discount rate the lender would use to estimate the PV of the loan is equal to
risk-free rate of USD debt (4,87% – yield on 10Y sovereign Eurobonds of Russia) (Cbonds
2016) plus market risk premium of 1% (Thomson Reuters 2016). Two different interest rates
represent the ability of the lender win higher or lower margins.
3.2 Covenant-free debt
First, we will consider the situation when the bank gives a loan without any covenants.
The simulation allow us to observe the probabilities of the borrower to default at certain time
periods under different loan conditions. Moreover, the analysis provides us with the expected
loan value for each case. Figure 2 represent the individual and cumulative default probabilities
for different loan agreements:
48
Object 84
Figure 2. Default probabilities for the borrower under 7% loan with various principal values.
The histogram for the 9% loan can be found in the Appendix 3. The results indicate that
the company is more likely to default in the later periods (Y8-Y10) and that individual default
probabilities are not sensitive to the principal amount of the loan. This has a logical explanation
– the more the time passes for the borrower, the higher the discrepancy can be between costs and
revenues (result of random walk specification and features). In addition, according to the
assumption 3, the firm sets its D/E ratio for each period and thus if the amount of new debt is
excessive, then company would offset it with equity issue or early redemption of other debt
obligations. The different interest rates do not have a major impact on the default rates of the
borrower, probably because the increase in annual payment is relatively low (9,5%)
The cumulative probability for the whole duration of the loan barely exceeds 14% for
various loan terms. That means that in the rest 85% of the cases the lender will receive all the
debt payments in full.
In case of the covenant-free debt, the only issue that bothers the bank is the possible
borrower default. The borrower can observe potential covenant breaches but can do nothing
about them. After we run the simulation procedure for loan without covenants, we get the
distribution of expected loan values, which are equal to a sum of payments discounted at bank’s
discount rate. The empirical distribution of loan values ($7000,00 principal; 7% rate) is
presented in Figure 3:
49
Object 86
Object 88
Figure 3. Distribution of loan values in case of the covenant-free debt.
The empirical distribution of loan values of covenant-free debt indicates several things.
At first, there is a high chance the bank will get the maximum value of debt as probability of
default (and thus earning lower PV) is not high – around 13%. As we have discussed earlier, this
default rates holds for various loan terms (Figure 2). Second, the distribution of loan values in
the default scenarios is left-skewed with the mass of the figure concentrating on the right side.
Even if the default strikes, the lender has a good chance of getting relatively high loan value.
However, low PV values in the left tail indicate that extreme losses are probable in no-covenant
scenario, even though their relative frequency is extremely low (less than 0,1%).
50
Table 2 represents the expected debt values for different loan conditions with the
volatility of payoffs in percent. The difference in percent between the expected value and
maximum possible value is less than 5% through all agreement terms. It seems that in case of our
borrower the bank is capable of obtaining substantial amounts of debt value without usage of
covenants. Figure 2 showed that majority of defaults happens right before the loan maturity. With
larger part of interest paid in the first periods, the low amount of outstanding debt is easier to
cover if the borrower defaults.
Table 2. The expected loan value and related parameters for the covenant-free debt
Volatility of
loan values
(%)
Expected loan
value
Maximum loan
value
Difference
(%)
N=$3000,00; i=7%
$3 132,78
$3 163,19
0,97%
3,1304%
N=$3000,00; i=9%
$3 422,54
$3 461,84
1,14%
3,5273%
N=$7000,00; i=7%
$7 303,12
$7 380,77
1,06%
3,3122%
N=$7000,00; i=9%
$7 987,37
$8 077,62
1,13%
3,5854%
N=$15000,00; i=7%
$15 632,27
$15 815,93
1,17%
3,3656%
N=$15000,00; i=9%
$17 093,78
$17 309,19
1,26%
3,7454%
Agreement terms
The expected debt values will be later compared to their counterparts from loans with
DSCR covenant in place.
In the covenant-free loan contracts, the lender still has an option to observe possible
violations of the covenant like if there were any. This may be a valuable experience even if the
lender has no power in case of the violations, as management of the borrower will act normally
and will not make any decisions that affect the accounting ratios of the firm. Thus, the bank can
compare the obtained frequencies of different violation types, distributions of violations, etc.
with the same data of the loan with DSCR covenant to evaluate the effectiveness of its strategy.
In the next sections, we are going to use the term “violation probability” regularly.
According to the frequentist interpretation of probability, probability of a certain event is the
limit of its relative frequency in a large number of trials. (Friedman 1999) This definition forms
the foundation of modern statistics and supports its need for experimentation. In our case, we
conduct runs of the simulation, collect the results after each run and then estimate the frequency
of certain events (defaults, covenant violations, etc.) According to the definition above, the limit
of the frequency is statistical probability. In our case, we will not estimate the limit, but we
suggest, that number of simulation runs is high enough to estimate the statistical probability with
sufficient accuracy. Further, we will use both words “frequency” and “probability” in the same
meaning.
51
Figure 4 represents the possible distributions of technical and material breaches over time
for DSCR threshold of 2.
Object 90
Figure 4. Distribution of possible covenant violations for different agreement terms of covenantfree debt.
The threshold is quite low so, as we expected, violations become common only at the end
of the loan and combined frequency barely exceed 15%. The share of material breaches in total
breaches rises the closer the contract is to maturity – for $7000,00 9% loan it is 0% in Y1-2, but
19% in Y9. There could not be any material breaches in Y10 as the only two possible options are
default (which is not a material breach for Y10) or technical violation. If the borrower is able to
make debt payments in Y10, no matter if there is a breach, or not, the lender will get the debt
payments on schedule and in full amount. We see, that for the contract terms with higher interest
rate and principal amount the frequency of violations increases.
We test the sensitivity of violations frequency and distribution to various parameters by
running simulation with different loan agreement terms. It turned out that threshold value has the
biggest effect on probability of a covenant breach at certain period. Figure 5 represents the
distribution of possible covenant violations for different DSCR thresholds (N=$7000,00; i=7%)
52
Object 93
Figure 5. Distribution of potential covenant violations over time for DSCR thresholds
from 1 to 4,5 for covenant-free debt (principal equal to $7000,00 and interest rate 7%)
The evidence presented on the graph (Figure 5) indicates that rising the threshold of
DSCR ratio even by 0,5 can drastically increase the frequency (and thus, the probability) of
covenant breaches at certain time periods – sometimes even more than 15%. These violations are
purely hypothetical as lender can only observe them, but does not have any right to engage.
Interestingly, extremely high thresholds (4-4,5) guarantee that breaches will be more common for
earlier periods. The explanation, however, is simple – if the threshold does not match the
empirical value of the DSCR ratio in period Y1-1, then the borrower has limited capability to
produce good results right from the start. The probability of the violation is even lower for 4,5
threshold in the later periods, as it takes time for borrower to reach the necessary profitability
levels. According to some high-level corporate banking executives, in real world lenders use the
rule of thumb for threshold determination, being:
Threshold t =Value t−1−25
(18)
We also see that even with a very loose threshold the violations happen significantly more
often than defaults of the borrower. Thus, we make a hypothesis that even a low DSCR threshold
can help the lender capture extra debt value and help negotiate lower cost of debt.
The maximum probability of a violation at certain period slightly exceeds 50% for a very
tight covenant (4,5). Of course, the lender can theoretically offer debt with even more tight
covenants; however, the borrower is very unlikely to agree on a contract terms he is going to
violate with >50% chance in the first period. Another hypothesis that we make at this point is
that there are optimal threshold values that maximize the positive effect for the lender and they
are not too punishing for the borrower. The issue of optimal threshold is regularly raised in
academic literature and our hypothesis is not contradicting the findings (e.g. Yu 2010).
53
Interest rate and principal amount of loan both have a minor positive impact on the
frequency of covenant violations. The rise of interest payment amount is not drastically affecting
the DSCR ratio and effect of high loan principal is offset by the company’s ability to repay other
debt obligations and issue equity to secure the D/E ratio in the period. Figure 6 represents the
probabilities of covenant breach for loan with N=$7000,00, covenant threshold of 2 and variable
interest rate.
Object 98
Figure 6. Distribution of potential covenant violations over time for interest rates from 5% to
12% covenant-free debt (principal equal to $7000,00 and DSCR threshold equal to 2)
At first, interest rate increase does not affect the skewness of distribution of covenant
breaches over time – periods that are more distant are still having higher probability of a
violation. Second, loans with higher interest rate generally showed higher frequencies of
violations (which is partly seen on the graph); however, for almost 2x increase in interest the
differences more than 3% were extremely rare. We do not have any solid proof that the overall
effect is significant and not a result of randomness and imperfections in the simulation
procedure.
The same graph for the loan with changing principal value can be found in the
Appendix 4. The analysis of sensitivity of covenant violations to changes in principal amount
produces almost the same outcome as its interest rate analogue: even significant changes in N
result in relatively small changes in violation frequencies, although graph shows a positive
relation between principal amount and covenant breach probability. The effect, again, is
ambiguous and relies more on the common sense and assumption 3 of the model.
The evidence suggests that DSCR threshold has a major impact on the covenant violation
probability, while the interest rate and principal play a minor role. We argue that under
assumption 3 of our model the effect of interest rate and principal is the same fundamentally –
54
while company adjusts the capital structure after the loan was taken, rising N and i basically
increases the annual debt payment. If so, the volatility of revenues and costs in our model is
offsetting the magnitude of change in interest payment. The positive effect of increase in
threshold is observable and quite substantial. While firm’s cash flow is volatile, it needs time to
rise above the certain threshold and that result in high violation rates in early periods for loans
with tight thresholds. When thresholds are more loose, then most of the violations happen in later
periods, which matches the borrower default probability distribution curve. As the strategy of the
lender is to maximize loan value with certain strategies for each violation type, then covenants
that are too loose or too tight may be destructive for the bank. We will observe the empirical
proof for this proposition in the next sections.
If the threshold is the main factor affecting the violation frequencies, then it is useful to
observe the distribution of technical and material violations for different thresholds. Figure 7
shows this distribution for loan with N=$7000,00 and i=7%
Object 100
Figure 7. Distribution of hypothetical covenant breaches for various thresholds in covenant-free
debt, N=$7000,00; i=7%
Results of the analysis show that when thresholds reach values higher than 3, material
breaches become the very precise predictor of the future default. When the threshold is high
enough, the amount of material breaches is equal to almost 99% of the amount of defaults. The
discrepancies on the graph for the material breaches are caused by the uncertainties of the
outcome that are incorporated in the simulation procedure. The amount of technical violations,
however, increases dramatically. In real world, the bank would be swarmed with irrelevant data
on constant breaches and will suffer losses from nonstop demand for decision-making and
negotiations with the borrower. While these costs are not incorporated in our model, the lender
would still suffer losses, but in a different way. As the bank uses cash sweep strategy in case of a
55
technical violation, high volume of these violations will stimulate artificial acceleration of
principal repayment. Thus, lender will lose part of the interest when no default was in sight.
Again, quite intuitively, the outcome suggests that there should be an optimal threshold
ratio. It will enable the lender capture greatest value of loan. If the threshold is too low, we see
that material breaches only capture part of future defaults (and do not protect the lender from
sudden borrower nonpayment). If the threshold is too high, then defaults are less surprising, but
numerous technical violations of covenants become a problem and prevent the borrower from
capturing the optimal value of debt.
We have analyzed the covenant-free loan, obtained the expected values for several
different loan terms, and observed the hypothetical covenant breaches. The results suggest that
threshold value is the main factor of violation probabilities and that there should be an optimal
threshold value that would maximize the expected value of loan with covenants for the lender.
3.3 Loan with DSCR covenant
If the borrower is taking out a loan with DSCR provision, then bank will do cash sweeps
if the violation is technical and will demand early repayment of the loan if the breach is material.
It is worth mentioning that our model allows to see the possible violations even if the
loan has been fully repaid and we will include them in the analysis. If, on average, the loan with
covenant is fully repaid by Y9 and we do not consider possible breaches at Y10, then sharp drop
of probability will be falsely attributed to bank’s option strategy.
In order to test the effectiveness of DSCR provision incorporation, we will observe how
default rates are different in the new model. Figure 8 shows cumulative and snapshot default
probabilities for loans with and without covenants. (N=$7000,00; i=7%). The same analysis for
loans with different interest payments can be found in the Appendix 5.
Object 103
56
Figure 8. Default probabilities for loan with and without covenant (DSCR threshold=2,
N=$7000,00; i=7%)
The default probabilities for no covenant debt are higher than for loan with DSCR
provision. The higher is the annual interest payment, the bigger will be the discrepancy. If the
debt payment is low (for instance, N=$3000,00, i=7%), the difference between the cumulative
default probabilities at Y10 will be less than simulation error (>1%). However, for N=$15000,00
the cumulative default probability at Y10 decrease is 3,5% (as seen in the appendix 5). This
outcome has an intuitive explanation: if the debt payment is high and lender protects from
default by demanding early repayment at material breach, then potential decrease debt payments
in subsequent periods will result in lower chance of default (cash flow and quick assets will be
the same but debt service will be lower, than for the no covenant case).
It does not mean that lender’s strategy is beneficial for the borrower. In fact, to make a
default payment in case of the material breach, the borrower would have to rise liquidity through
short-term debt, equity or asset sale. All three of these may have a detrimental effect on financial
health of the borrower. However, we can’t incorporate all specifics in the theoretical model so in
our case the default rate is lower for DSCR covenant debt and management is qualified to make
the early repayment of the loan in case of material breach with no negative consequences for
firm’s operations. However, if the firm is in default already, then obtaining the financing is
becoming almost impossible.
Figure 9 shows the distribution of technical and material breach frequencies with various
DSCR thresholds for loan with and without covenants. The effect of covenant on amount of
breaches is unidentifiable in our model. The discrepancies were tested and results showed that
they are attributable mostly to randomness in the simulation procedure.
Object 105
57
Figure 9. Distribution of covenant breaches for various thresholds in DSCR covenant and
covenant-free debt, N=$7000,00; i=7%
We suggest that due to assumption 3 of the model and its specification the effect of
covenant on the amount of breaches is unobservable. If the company is focusing on the leverage
ratio, then the option of the early loan repayment will slightly lower the default rates but the
amount of breaches will be unaffected – managers will simply adjust the capital structure in the
next period so that more loans are taken/redeemed or more equity is issued/bought back.
Moreover, following this logic, accelerated redemption of our loan may trigger issuance of new
debt securities with higher interest rate and longer maturities. In real life, however, accounting
theories of covenants teach us that managers will make revenue-generating, cost-cutting
accounting decisions in order to avoid breaches and we omit this proposition in our model.
We have observed the distribution of loan values for covenant-free debt. In a case where
DSCR covenant is present, the distribution of values will depend on the threshold. Figure 10
shows the value distribution of covenant debt with DSCR threshold of 2.
Object 108
Object 110
58
Figure 10. Distribution of loan values in case of the DSCR covenant threshold equal to 2
The empirical distribution of loan values of debt with covenants indicates several things.
At first, there is a lower chance the lender will obtain maximum value of debt comparing to no
covenant distribution, as the probability of value being lower due to covenant violations is higher
than default rate at Figure 3. Second, the distribution of loan values in the covenant breach
scenarios is left-skewed with the mass of the figure concentrating on the right side. Moreover,
the potential values are very close to the actual maximum value, comparing to the no-covenant
scenario. The analysis indicates that for the loan with covenant it is likely to obtain the value that
is 95% or 99% of the maximum value. The PV values on the left tail of the distribution are also
higher than their counterparts from no-covenant debt (Figure 3); they are also quite high with
lowest being around 90% of the maximum value.
The results of the analysis show that distribution of values for covenant and no-covenant
debt are different. For no-covenant loan, the lender is more likely to obtain maximum loan value;
however, there is also a chance to get a significantly lower value represented with the left tail of
the distribution. If there is a DSCR provision, then the chances of getting maximum is lower;
however, the risk of extreme loss is also lower. Two debt types belong to different risk profiles;
so, they should be compared based on expected debt value and payoff volatility.
Tables 3 and 4 compare the expected debt values of DSCR covenant and no-covenant
debt for two cases (N=$7000,00; i=7% and N=$15000,00; i=9%). Comparison for other cases
can be found in the Appendix 6.
Table 3. The expected loan value and volatility for covenant and covenant-free debt
(N=$7000,00; i=7%). Maximum loan value is $7380,77
Threshold
Expected loan
value with
covenant
1,50
2,00
2,50
3,00
3,50
4,00
4,50
$7 365,60
$7 368,93
$7 363,42
$7 350,82
$7 328,51
$7 294,21
$7 255,59
Volatility (%)
with covenant
1,3788%
0,7688%
0,5637%
0,5249%
0,7245%
0,8935%
0,9385%
Expected
loan value
without
covenant
$7 303,12
Volatility (%)
no covenant
3,3122%
Difference
$62,48
$65,81
$60,30
$47,69
$25,39
-$8,91
-$47,53
59
Table 4. The expected loan value and volatility for covenant and covenant-free debt
(N=$15000,00; i=9%). Maximum loan value is $17309,19
Threshold
Expected loan
value with
covenant
Volatility (%)
with covenant
Expected
loan value
without
covenant
Volatility (%)
no covenant
Difference
1,50
$17 254,65
1,2397%
$160,87
2,00
$17 233,74
0,8443%
$139,96
2,50
$17 157,46
1,0643%
$63,68
$17 093,78
3,7454%
3,00
$17 016,12
1,3658%
-$77,66
3,50
$16 827,63
1,4850%
-$266,15
4,00
$16 650,42
1,3566%
-$443,36
4,50
$16 533,60
1,1670%
-$560,18
The outcome clearly indicates that the expected value of loan with covenants is higher for
thresholds from 1,5 to 2,5. Then the results show ambiguity – for some loan terms the value
decreases, for others it increases. At first, it provides strong supporting evidence that a certain
threshold will help to maximize the loan value for the lender that uses covenants. In our case,
this optimal threshold varies from 1,5 to 2,5, where the positive difference between expected
loan values is the highest. Second, the positive difference itself is the value of options embedded
in the covenant, according to the Monte-Carlo method of option pricing. Thus, combination of
options that we analyze (option to terminate the loan and flexibility option) have positive value
for some thresholds. When the covenant threshold is too high, then the difference is negative –
indicating that lender will lose money if DSCR provision is used. Nevertheless, we argue that
options embedded in the covenant are still “in the money” in that case. Why? The model that
we’ve built is limited in terms of lender’s actions – we do not consider him increase the interest
rate, waive the violation, ask for violation fee, take control of the operations, etc. in case of a
breach. However, if the threshold is too high, giving a waiver or asking for an extra fee would
probably be the beneficial strategy for a technical breach, not the cash sweep. As we’ve
discussed earlier, if the lender is doing cash sweeps regularly, then instead of lowering the risk of
default he would artificially accelerate the loan repayment for no good reason and fail to capture
a large portion of interest. This is displayed in tables 3-4, when expected loan values with
covenants decrease while the threshold rises. Therefore, if the option is still in the money, while
the threshold is high, the negative difference between expected payoffs should be caused by
suboptimal option execution and rising costs of dealing with frequent violations.
The value distribution of covenant loan is also different from covenant-free distribution
in terms of payoff volatility. As seen in tables 3 and 4, the volatility of covenant loan values is
2x-3x times lower than volatility of no-covenant debt. Figure 11 shows the loan value volatility
distribution over different thresholds for various loan terms:
60
Object 112
Figure 11. Volatility of loan values for loan with DSCR covenant
The volatilities of covenant-free debt all exceed 3%; we observe much lover numbers for
loan with DSCR covenant, exclusive of the loan terms. There is a range of thresholds from 2 to 3
that provide minimum value volatility; then it usually starts to rise with threshold tightness. Very
loose thresholds have a value volatility higher than minimum as they fail to capture some of the
unexpected defaults that produce low value outcomes. Very tight thresholds sometimes have an
increase in volatility; there is no exact explanation for that but we suppose that it is a result of
increased payoff uncertainties due to cash sweeps. When the thresholds are set too tight, the
borrower is doing cash sweeps regularly. Amount of cash collected is not stable – in fact, cash
sweeps are related to potential dividends and they are as volatile as revenues and earnings.
The evidence is solid: by using the DSCR provision, the lender not only can shift the
payoff distribution structure to more “loss-averse” (and less risky, one can argue), but also
increase the expected payoff from the borrower and decrease the volatility of loan values.
If the execution of options that are implemented in the covenant is giving these benefits
to the borrower, then there should be a premium for that in the loan contract. As we suggested,
the borrower pays the lender with the decreased costs of debt. If the option price is the difference
between the expected values of two loan types, then interest rate for covenant-free debt should
be higher in order to match the increased value of loan with DSCR provision. We estimated the
interest rate discount the lender has to make for covenant debt. The values we have obtained can
be referred as minimum: in fact, they can be higher in real life as some option premiums were
omitted in the model.
At first, bank actions in the model are limited. We argue that given full range of
strategies, the lender will further increase the expected payoff from the borrower thus making
61
options more valuable and interest rate discount higher. Second, covenant drastically decreases
the loan payoff volatility and thus makes the risk profile of the borrower more favorable for the
bank. In that case, the lender should obtain cheaper financing by definition – as its debt becomes
less risky. Third, the covenant has other embedded options that we do not consider – for instance,
a follow-up loan option. In our case, if the loan is repaid earlier due to cash sweep or material
breach happening, the lender is simply left with the money. However, he would probably reinvest
them into less risky securities or offer another loan to a different client until the end of the 10Y
period. Thereby option is in the money as additional value is captured, option price is higher, and
debt cost for the borrower is lower. The introduction of these elements to the analysis is a good
topic for future research.
Final step of the analysis was the estimation of interest rate reduction potential for debt
with DSCR covenant. In order to obtain the values for different loan terms we ran the
simulations for the covenant-free loan, got its expected value and then ran the simulations for
covenant debt using brute-force – we tried every possible interest rate variation to get the
expected values of loans match. As a result, we got the loans with different interest rates but all
else being equal. The difference in interest rates is the reduction potential for a certain threshold.
For instance, if the expected loan value of covenant-free loan is equal to E, and the value of the
same loan with DSCR provision and threshold of 2,5 is equal to C, we simulated the outcomes of
this loan for different interest rates i n to find such i when C=E. The difference between interest
rates for two loans is the reduction potential (given the nominal value, term and DSCR
threshold).
Figure 12 shows, what is the minimum amount of interest rate reduction the borrower
should be able to negotiate for debt with various terms and DSCR thresholds.
Object 114
62
Figure 12. Interest rate reduction potential for different loan terms & thresholds
Results show that in case of our standard, “generic” borrower, the interest rate can be
lowered at least by 0,19%-0,22% for optimal thresholds. By setting the threshold very loose (11,5), the lender will fail to exploit some the volatility decrease potential. If the threshold is too
tight, then potential cost of debt decrease is ambiguous, as lender needs to have more decisionmaking flexibility that we fail to provide in our model.
The research goal of the paper was to calculate the potential of interest rate decrease for
Russian borrowers that is achieved by implementing covenants in the contract. We have obtained
evidence that helps us answer this question positively: indeed, the company can ask for a lower
interest rate if there is a DSCR covenant in the contract. If the threshold for DSCR is set
optimally (captures future default and not triggering technical violations too frequently), then the
reduction amount can be at least 0,2% for a normally operating borrower.
3.4 Abnormal leverage and volatility cases
In previous sections, we have modelled the generic company that has attributes of the
Russian oil industry players. The results of the analysis indicated that DSCR provision allows the
borrower to ask for lower costs of debt under normal conditions. In this section we are going to
make one step further and analyze the situation, when conditions are not “normal” or “stable”
anymore. Imagine, the bank gave a loan to the company under principal amount N and interest
rate i. The borrower’s risk level was assessed based on set of parameters. However, soon after
the loan was given the borrower decided to increase its credit risk – by either increasing the
leverage or volatility. Will the DSCR provision help the bank capture more loan value?
For simplicity we are going to analyze only one case in this part, where N=$7000,00 and
i=7%. We will consider two stress scenarios – where the volatilities for revenues, costs, and
assets are increased by 20% and where the mean leverage ratio increases by 1. We will also test
the scenario when the volatility decreases by 10% – assuming that the company decided to
become less risky. Drift rates for this scenario will be also eliminated (company doesn’t grow)
For simplicity, only the thresholds of 2; 2,5 and 3 will be considered (as they proved to be
most effective for normal scenario).
Let us consider the low-volatility scenario first. The volatility rates were lowered by 10%,
making the revenue, cost and asset volatility equal to 11%, 12% and 8%. The simulation
procedure of the no-covenant model showed only 45 breaches out of 40000 possible for DSCR
threshold of 2, no defaults and expected payoff equal to maximum possible payoff. It turned out
that serious volatility decrease together with “no-growth” strategy eliminated all possible default
risk for the company and turned it into a “perfect” borrower – statistically speaking, the bank has
63
zero chance of not getting the maximum loan value if the company parameters remain. The use
of covenants here was not beneficial at all – while breaches still counted and cash sweeps were
made, the expected loan value for covenant loan represented only 99% of the maximum value
for all thresholds. The profitability levels of the borrower remained high (as they were taken
from Russian oil industry) and low volatility with no growth ensured the profit margin holds.
Thus, the company was always able to make debt payments. The whole case, obviously, is
almost impossible in real life – the world is changing, management makes investment decisions
that are not always good, financial crises occur and volatility skyrockets, etc.
The results of the analysis of this scenario show that a low-volatility borrower can agree
on a wide range of accounting covenants with tight thresholds. If the violation risk is too low,
then bank is most likely to continuously “waive” the breaches, as it will be in its own interest.
Table 5 shows results of the stress scenarios modelling.
Table 5. The expected loan value and volatility for covenant and covenant-free debt in
stress scenarios 1 and 2. N=$7000,00; i=7%). Maximum loan value is $7380,78
Threshold
Expected loan
value with
covenant
2
2,5
3
$7 244,43
$7 275,55
$7 293,79
2
2,5
3
$7 362,06
$7 341,89
$7 296,76
Expected
loan value Volatility (%)
with
no covenant
covenant
Stress scenario 1 (extreme volatility)
4,88%
$6 924,46
9,55%
3,97%
2,68%
Stress scenario 2 (extreme leverage)
0,54%
$7 282,42
1,92%
0,61%
0,89%
Volatility (%)
with covenant
Interest rate
reduction
potential
-1,00%
-1,10%
-1,20%
-0,25%
-0,20%
-0,08%
The results of scenario 1 show the true power of covenant use for borrower with high risk
of default. We assumed that initial valuation of the borrower was incorrect and volatility of
assets, costs and revenues was underestimated. When the true volatility figures turned out to be
higher by 20%, the default probability on a 10-year horizon was calculated to be around 43%.
High default risk made the expected payoff of the covenant-free loan less than initial ($7000,00)
principal, thus making the current loan not profitable – it was not generating any value. The
DSCR provision helped to avoid the destructive conditions of poor evaluation of the borrower.
As the bank was able to demand cash sweeps and early repayment, the expected loan value
exceeded principal, so the bank at least did not lose the money. The conditions held even for
loose threshold of 2. In line with the previous findings, the volatility of payoffs for covenant-free
debt was almost 2 or 3 times higher than volatility for covenant loan. The interest rate reduction
potential was calculated differently here, as reducing the covenant loan to a value below
64
principal is pointless – no money is generated for the bank. Thus, we looked for the interest rate
increase that was sufficient for covenant free-loan to match the expected value. It turned out that
covenant-free loan required at least 1% interest rate increase to get its value equal to covenant
loan. This is a substantial number for a USD debt in Russia. However, this figure is purely
fictional – no bank would agree to negotiate any potential cost of debt decrease for such a risky
client.
Stress scenario 2 turned out to be not dangerous for the lender, as the potential increase in
default probability was minimal – around 2%-3%. The difference between expected values was
slightly higher than for a normal case. The potential decrease in interest rates was also
comparable to normal conditions. We can say that either our leverage increase was not sufficient
or proportion of debt is not affecting the value of DSCR covenant in a severe way in our model.
The results of analysis of the three abnormal scenarios show that volatility is the main
determining factor of the option value for the DSCR covenant. When volatility of the firm
parameters was decreased to the negligible levels, covenant was not providing any benefits to the
lender – the default risk was almost equal to zero. If the company can sustain stable profitability
levels throughout the duration of the loan (apparently, by sacrificing its growth), then it can agree
on any covenants the bank is offering. When volatility of the firm is extremely low, then
covenant breaches would likely be waived by the bank. If the volatility of the parameters is
extremely high, then even loose thresholds can capture significant loan values for the bank. The
importance of volatility in our model is similar to importance of volatility in financial options
theory. Historical volatility of the underlying is crucial for pricing of options on stocks, indices,
etc.
While abnormal volatility of the firm’s parameters proved to be significant factor of the
value of options embedded in the DSCR covenant, we could not estimate meaningful effect of
abnormal leverage. Overall, the results of this section confirm the findings that we got when
analyzing the regular scenario. The lender can capture extra loan value by including the DSCR
covenant in the loan contract. The borrower can also negotiate lower costs of debt if the covenant
is included, as we assume that lender offers same loan value for the same borrower. In our model
if the values of the firm’s parameters are normal (or represent the real-life situation), then the
interest rate reduction potential can be up to around 0,2%. If the volatility of the borrower
increases dramatically, then option value and reduction potential increases as well. However,
when volatility is unrealistically low, then covenant does not capture extra value. Options
become more expensive when the uncertainties rise and lose their value when the uncertainties
are low. From the viewpoint of financial theory, this makes perfect sense – if the default risk of
the borrower is negligible, the bank does not need any additional provisions to protect the debt
65
payments. However, if the company’s future performance is highly uncertain, then default risk is
massive and lender needs certain provisions incorporated in the contract to protect the loan
value. Our model captures this relation. We proved that the results of our analysis hold under
both normal and irregular conditions given the assumptions we have made earlier.
3.5 Managerial implications
In previous sections, we determined that loan covenants actually have real options
embedded in them. These options allow the bank to capture additional loan value in the face of
potential borrower’s default. In fact, value of the loan contract increases when DSCR covenant is
included. We assume, that bank should only offer the contracts of the same value to the same
borrower, so it pays the price of including the covenant by agreeing to lower the costs of debt
(interest rate in our case). In this section, we are going to discuss the implications of our
theoretical model that can be used in real world.
At first, we ascertained that when DSCR covenant is included in a loan agreement, a
reduction potential for the interest rate exists. Moreover, we have estimated this potential – in the
normal scenario, it can be up to around 0,2%. If the company decides to increase its risk level by
boosting the volatility of its parameters, then this number can increase further to more than 1%.
The first direct managerial implication of our work is the following: the company should be able
to negotiate lower interest rate for the debt contract with covenants than it will have for
covenant-free debt with the same terms. In part one of the thesis we have already discussed how
it is justified from the theoretical viewpoint and how empirical works confirm the connection
using the sample of real loan contracts. In this Thesis we used modelling, simulation and real
options approach to come up to results that uphold the same claim.
A more specific conclusion that we came to: the generic borrower from Russian oil
industry should be able to negotiate at least 0,2% interest rate decrease if the Debt Service
Covenant Ratio is included and optimal threshold is set up. We used the industry data of Russian
oil producers for our model and analyzed the effects of DSCR covenant. The average optimal
reduction potential we obtained was around 0,2% if the covenant is included (number heavily
depends on the loan terms and threshold). We say that this is the smallest discount borrower
should get, as in our model only part of the bank’s actions were captured and not all real options
analyzed. We assume that interest rate discount for loan with DSCR covenant should be higher in
real life, but at this point, this is the most we can get.
Another real-life implication is: company with stable profitability levels and low
volatility of main parameters should agree on DSCR covenant even with tight thresholds. As we
observed in the case of abnormally low volatility, the default risk becomes negligible and it is
66
costly for the bank to apply particular strategies. Thus, lender is likely to waive any potential
violations. The borrower may get certain benefits (like possible lower interest rate) without
suffering consequences of covenant breaches.
Banks (lenders) can also use the results of the paper for their own benefit. As the interest
rates have increased in the past two years and foreign financing is limited for Russian companies
for various reasons, local lenders should stay both profitable and supportive for their clients. The
Russian interbank lending rate in RUB for the period up to 1 year is currently around 11% (Bank
of Russia 2016) and average commercial bank lending rate is even higher, so borrowers will be
open for any opportunity to lower their costs of debt. At the same time, results of our research
imply that banks can offer these interest rate discounts without loss in loan value or taking
additional credit risk. The main implication for the lenders is the following: by including certain
covenants in the loan contract, Russian banks can offer lower costs of debt for particular clients
without losing loan value; thus, banks can become more competitive among other players and
provide support for their clients.
Another finding of the paper that is relevant for the lender is the potential ability to
decrease the volatility of loan values when DSCR covenant is introduced to the contract, even if
the threshold is not tight. We have estimated that while the difference in loan values can be small
for covenant and covenant-free loan, potential volatility of loan values, that bank gets, can be
two or even three times lower in our model. As covenants protect the lender from negative
consequences of borrower default, the distribution of loan values becomes denser and left tail of
this distribution becomes less “fat.”
Finally, if the banks know they can provide an interest rate discount, they should be able
to calculate the interest reduction potential for a specific loan. In our thesis, we analyzed only
one covenant (DSCR) and a limited set of bank’s actions. However, it is clear that each client
needs unique set of assumptions and set of covenants for the loan contract. Bank may use the
structure of our model to create a particular framework for each client, simulate future
performance of the borrower, and estimate probabilities of default, material and technical
breaches. After the simulation procedure is done, the lender can determine optimal set of
responses for violation of each covenant in the set. Thereby the real options approach can be
employed to estimate the covenant value and interest rate reduction potential. According to the
Citibank St. Petersburg Branch Manager R. Belyaev, banks in Russia work really well with the
risk analysis and modelling when they assess their clients, but real options approach is not
widely employed. The example of this paper may motivate some to use this approach and price
the options inside particular covenants to calculate extra loan value.
67
4. Conclusion and Areas for Future Research
In this Master Thesis, we applied the real options framework to explore the connection
between presence of covenants in the loan contract and potential costs of debt for the borrower.
On the basis of a set of assumptions we created a model of a borrower, incorporated a loan with
Debt Service Coverage Ratio covenant in it and estimated the potential decrease in interest rate
that borrower can negotiate by referring to the option value that covenant provides.
Our work is based on the wide range of academic research on the subject. During the
literature analysis, we discovered several relevant lines of research. The Agency Theory of
Covenants primarily focuses on covenants’ ability to mitigate agency conflict between
shareholders and debtholders inside the firm. If a covenant is included in debt terms, both parties
enjoy decreased agency costs and associated benefits, like more transparent monitoring process,
but have to pay the price for that. While management activity will be more limited, creditors will
need to offer lower spread on debt securities. Various accounting theories on the subject focus
primarily on the impact of covenants on firms’ reporting. The highest effect both ex post and ex
ante is achieved via covenant violations. Accounting research provides substantial proof that
managers make earnings-increasing and cost-cutting accounting choices around violations while
both systematic and unsystematic risk increases at that point. Banks are often eager to waive the
violation and set tight covenants to act as “trip wires” and provide effective monitoring. Hence,
covenant breaches can be divided into material (effective predictor of future default) and
technical (accidental violation). Finally, the incomplete contracts theory treats covenants as
renegotiation options in debt contracts that help to increase contract effectiveness in the future.
We adopted this vision in our work but broadened the understanding of options embedded in the
covenant – we claim that covenants contain a set of real options the lender benefit on. While
these options are “in the money” and have certain option value, lender needs to pay the premium
for their use by providing a lower interest rate for covenant loan.
In the paper, we have estimated the value of some real options that DSCR covenant
provides the lender according to our model. The results of the simulation procedure indicate that,
given a coherent bank strategy for different violation types, even the covenant with loose
threshold brings extra loan value as lender becomes more protected against borrower default. We
also find empirical proof that given the set of assumptions an optimal threshold exists for the
loan with particular terms. If the DSCR threshold is set below the optimal levels, then covenant
fails to capture some the violations that indicate serious credit risk increase and hence fails to
capture some of the option value. If the threshold is set too tight, then coherent bank strategy is
not effective anymore and lender is swarmed with information on covenant breaches that is not
relevant.
68
We also evaluated the volatility of loan values that lender gets for both covenant and
covenant-free debt. There is significant evidence that payoff volatility becomes from two to three
times lower if the DSCR provision is introduced in the contract. Covenant changes the loan
value distribution – in light covenant debt there are usually more chances to capture maximum
loan value, while in covenant loan the value distribution is much denser near the maximum.
Moreover, the left tail of the distribution (low loan values) is much “fatter” for covenant free
loan, indicating that bank can obtain the payoff even smaller than initial principal amount.
Finally, we calculated the potential of interest rate decrease when DSCR covenant is
introduced to loan terms. According to our model, the reduction potential for normally operating
borrower given the consistent bank strategy (cash sweeps for technical violations, early
repayment for material violations) and optimal threshold value can be around 0,2% for USD
loan. This value may not be impressive, but for current economic conditions, even such small
discount can be relevant for borrower. Moreover, we refer to this value as minimal, as some of
the real options were not considered in the analysis, limited lender strategy was taken, and only
one covenant analyzed. If the whole set of different covenants is included, then reduction
potential can be higher.
We tested our model for abnormal scenarios with extreme volatility and high leverage.
The results generally hold. The reduction potential in abnormal scenarios can be up to 1,2%, but
obviously this value is not feasible in reality – the relations between a bank and such anomalous
client are too complicated to assume that such big interest discount will simply be given.
Our model is not a perfect representation of real world, it has its own drawbacks, and
future research on the subject should develop it in order to create a more serious contribution in
field of real options and covenants.
At first, we omit some important findings of the accounting research – particularly,
accounting decisions of managers near the breach. While we build the model upon accounting
data, it is necessary to incorporate at least some of the managerial activity, either through
creating algorithms or introducing investment function in the analysis. We decided to leave it for
future research in order to obtain results that are more reliable at the current moment.
Second, model is built with a set of assumptions (can be found in the methodology part).
As assumptions do not truly represent the real world, future research on the subject may
substitute the assumptions with equation models, functions, and other more reliable methods of
analysis.
Third, Geometric Brownian Motion has its advantages as a way of simulating future
values, but also has its drawbacks. Being one of the simplest stochastic processes, GBM is not
intuitively applicable in this case. Future research may test other simulation procedures to check
69
the results of this Master Thesis, provide adjustments where required in order to make findings
more relevant and reliable.
Fourth, the classification of violations into technical and material requires the optimal
behavior of the lender at given point in time, and that means the banks knows exactly what type
of breach it is. We stepped away from classifying the violations through comparing current
DSCR ratio values with two thresholds. We argue that the approach utilized in this Thesis is
better, but again it omits some of the uncertainties of the real world. In our opinion,
implementation of control theory methods will greatly improve the model and results.
Finally, future research can incorporate more real options in the analysis and create a
better lender strategy for violations. At the moment, the results become ambiguous when the
covenant threshold is set too tight. Also, our procedures were too “heavy” for the methods of
VBA programming that we used, so outcome generation was taking its time. Some other coding
language can be used by future research (like R or Python) to speed up the process.
Overall, findings of this Master Thesis are relevant for both the modern business and
financial theory. Managerial implications exist for borrowers – they can refer to option value to
negotiate lower interest on debt, and for banks – they can become more competitive if they use
the real options embedded in loan covenants. The results are reliable, but can be further
improved by future research on the subject.
70
5. References
Achleitner, Ann K., Carolin Bock, and Florian Tappeiner. "Financial covenants and their
restrictiveness in European LBOs - an assessment in the aftermath of the financial
crisis." IJEV 4, no. 3 (2012): 214. doi:10.1504/ijev.2012.048600.
Aghion, Philippe, and Patrick Bolton. "An Incomplete Contracts Approach to Financial
Contracting." The Review of Economic Studies 59, no. 3 (1992): 473.
doi:10.2307/2297860.
Altman, Edward I. "Financial Ratios, Discriminant Analysis and the Prediction of Corporate
Bankruptcy." The Journal of Finance 23, no. 4 (1968): 589. doi:10.2307/2978933.
Barth, Mary E., Donald P. Cram, and Karen K. Nelson. "Accruals and the Prediction of
Future Cash Flows." The Accounting Review 76, no. 1 (2001): 27-58.
doi:10.2308/accr.2001.76.1.27.
Bazzana, Flavio, and Eleonora Broccardo. "The Role of Covenants in Public and Private
Debt." SSRN Electronic Journal, October 2009. doi:10.2139/ssrn.1496352.
Begley, Joy. "Restrictive Covenants Included in Public Debt Agreements: An Empirical
Investigation." January 1994.
University of British Columbia
Begley, Joy, and Sandra L. Chamberlain. "The Use of Debt Covenants in Public Debt: The
Role of Accounting Quality and Reputation." SSRN Electronic Journal (n.d.).
doi:10.2139/ssrn.930848.
Begley, Joy, and Gerald A. Feltham. "An empirical examination of the relation between debt
contracts and management incentives." Journal of Accounting and Economics 27, no.
2 (1999): 229-259. doi:10.1016/s0165-4101(99)00006-3.
Belyaev, Ruslan V. "Covenants as Risk Management tool in Credit Relation." Vestnik of St.
Petersburg University, no. 1 (2015): 91-124.
Beneish, Messod D., and Eric Press. "Costs of Technical Violation of Accounting-Based
Debt Covenants." The Accounting Review 68, no. 2 (April 1993): 233-257.
http://www.jstor.org/stable/248399.
Benninga, Simon. Financial Modeling, 4th ed. The MIT Press, 2014.
Berger, Philip G., and Eli Ofek. "Diversification's effect on firm value." Journal of Financial
Economics 37, no. 1 (1995): 39-65. doi:10.1016/0304-405x(94)00798-6.
Berlin, Mitchell, and Loretta J. Mester. "Debt covenants and renegotiation." Journal of
Financial Intermediation 2, no. 2 (1992): 95-133. doi:10.1016/1042-9573(92)90005x.
Besanko, David, David Dranove, and Mark Shanley. Economics of Strategy. New York:
Wiley, 2000.
Billet, Matthew T., Tao-Hsein D. King, and David C. Mauer. "Growth Opportunities and the
Choice of Leverage, Debt Maturity, and Covenants." The Journal of Finance 62, no.
2 (2007): 697-730. doi:10.1111/j.1540-6261.2007.01221.x.
Bishara, Norman D., and David Orozco. "Using the Resource-Based Theory to Determine
Covenant Not to Compete Legitimacy." Ross School of Business Working Paper,
April 2015, 979-1031. http://hdl.handle.net/2027.42/85367.
Black, Fischer. "The Dividend Puzzle." The Journal of Portfolio Management 2, no. 2
(1976): 5-8. doi:10.3905/jpm.1976.408558.
71
Bloomberg. "Stocks - Bloomberg." Bloomberg.com. Accessed May 25, 2016.
http://www.bloomberg.com/markets/stocks.
Bolton, Patrick, and David S. Scharfstein. "A Theory of Predation Based on Agency
Problems in Financial Contracting." The American Economic Review 80, no. 1
(March 1990): 93-106. http://www.jstor.org/stable/2006736.
Borgonovo, Emanuele, and Stefano Gatti. "Risk analysis with contractual default. Does
covenant breach matter?" European Journal of Operational Research 230, no. 2
(2013): 431-443. doi:10.1016/j.ejor.2013.04.047.
Boyle, Phelim P. "Options: A Monte Carlo approach." Journal of Financial Economics 4, no.
3 (1977): 323-338. doi:10.1016/0304-405x(77)90005-8.
Bradley, Michael, and Michael R. Roberts. "The Structure and Pricing of Corporate Debt
Covenants." SSRN Electronic Journal, May 2004. doi:10.2139/ssrn.585882.
Bratton, William W. "Bond Covenants and Creditor Protection: Economics and Law, Theory
and Practice, Substance and Process." The Law and Economics of Creditor
Protection, 2008, 37-85. doi:10.1007/978-90-6704-633-6_3.
Brewer, Kevin D., Yi Feng, and Clarence Kwan. "Geometric Brownian Motion, Option
Pricing and Simulation: Some Spreadsheet-Based Exercises in Financial Modelling."
Spreadsheets in Education 5, no. 3 (November 2012).
http://epublications.bond.edu.au/cgi/viewcontent.cgi?article=1131&context=ejsie.
Broadie, Mark, and Paul Glasserman. "Estimating Security Price Derivatives Using
Simulation." Management Science 42, no. 2 (1996): 269-285.
doi:10.1287/mnsc.42.2.269.
Bulow, Jeremy I., and John B. Shoven. "The Bankruptcy Decision." The Bell Journal of
Economics 9, no. 2 (1978): 437. doi:10.2307/3003592.
Cbonds (Financial News Agency). Cbonds. Bond Market in Russia. Russian Bonds and
Fixed Income. Accessed May 25, 2016. http://ru.cbonds.com/.
Chava, Sudheer, and Michael R. Roberts. "How Does Financing Impact Investment? The
Role of Debt Covenants." The Journal of Finance 63, no. 5 (2008): 2085-2121.
doi:10.1111/j.1540-6261.2008.01391.x.
Chen, Kevin, and John Wei. "Creditors' Decisions to Waive Violations of Accounting-Based
Debt Covenants." The Accounting Review 68, no. 2 (March 1993): 218-232.
http://www.jstor.org/stable/248398.
Christensen, Hans B., and Valeri V. Nikolaev. "Capital Versus Performance Covenants in
Debt Contracts." Journal of Accounting Research 50, no. 1 (2011): 75-116.
doi:10.1111/j.1475-679x.2011.00432.x.
Cremers, K. J., and Vinay B. Nair. "Governance Mechanisms and Equity Prices." The
Journal of Finance 60, no. 6 (2005): 2859-2894. doi:10.1111/j.15406261.2005.00819.x.
DeAngelo, Harry, Linda DeAngelo, and Douglas J. Skinner. "Accounting choice in troubled
companies." Journal of Accounting and Economics 17, no. 1-2 (1994): 113-143.
doi:10.1016/0165-4101(94)90007-8.
Dechow, Patricia M., S.P. Kothari, and Ross L. Watts. "The relation between earnings and
cash flows." Journal of Accounting and Economics 25, no. 2 (1998): 133-168.
doi:10.1016/s0165-4101(98)00020-2.
72
DeFond, Mark L., and James Jiambalvo. "Debt covenant violation and manipulation of
accruals." Journal of Accounting and Economics 17, no. 1-2 (1994): 145-176.
doi:10.1016/0165-4101(94)90008-6.
Demerjian, Peter R. "Financial Ratios and Credit Risk: The Selection of Financial Ratio
Covenants in Debt Contracts." SSRN Electronic Journal, 2007.
doi:10.2139/ssrn.929907.
Demerjian, Peter R., and Edward L. Owens. "Measuring Financial Covenant Strictness in
Private Debt Contracts." SSRN Electronic Journal, 2014. doi:10.2139/ssrn.2232880.
Demiroglu, Cem, and Christopher M. James. "The Information Content of Bank Loan
Covenants." SSRN Electronic Journal, 2010. doi:10.2139/ssrn.959393.
Deng, Yongheng, Erik Devos, Shofiqur Rahman, and Desmond Tsang. "The Role of Debt
Covenants in the Investment Grade Bond Market – The REIT Experiment." The
Journal of Real Estate Finance and Economics 52, no. 4 (July 2015): 428-448.
doi:10.1007/s11146-015-9511-6.
Dessein, Wouter. "Information and Control in Ventures and Alliances." The Journal of
Finance 60, no. 5 (2005): 2513-2549. doi:10.1111/j.1540-6261.2005.00806.x.
Diamond, Douglas W. "Financial Intermediation and Delegated Monitoring." The Review of
Economic Studies 51, no. 3 (1984): 393. doi:10.2307/2297430.
Diamond, Douglas W. "Monitoring and Reputation: The Choice between Bank Loans and
Directly Placed Debt." Journal of Political Economy 99, no. 4 (1991): 689-721.
doi:10.1086/261775.
Dichev, Ilia D., and Douglas J. Skinner. "Large-Sample Evidence on the Debt Covenant
Hypothesis." J Accounting Res 40, no. 4 (2002): 1091-1123. doi:10.1111/1475679x.00083.
Elsinger, Helmut, Alfred Lehar, and Martin Summer. "Risk Assessment for Banking
Systems." Management Science 52, no. 9 (2006): 1301-1314.
doi:10.1287/mnsc.1060.0531.
Fargher, Neil L., Michael S. Wilkins, and Lori M. Holder-Webb. "Initial Technical Violations
of Debt Covenants and Changes in Firm Risk." J Bus Fin & Acc 28, no. 3-4 (2001):
465-480. doi:10.1111/1468-5957.00381.
Freixas, Xavier, and Jean-Charles Rochet. Microeconomics of Banking. Second edition.
London, England: The MIT Press, 2006.
Friedman, Charles. "The Frequency Interpretation in Probability." Advances in Applied
Mathematics 23, no. 3 (1999): 234-254. doi:10.1006/aama.1999.0653.
Gamba, Andrea, and Alexander J. Triantis. "How Effectively Can Debt Covenants Reduce
Financial Agency Costs?" SSRN Electronic Journal, 2013. doi:10.2139/ssrn.2101645.
Gertner, Robert, and David Scharfstein. "A Theory of Workouts and the Effects of
Reorganization Law." The Journal of Finance 46, no. 4 (1991): 1189.
doi:10.2307/2328856.
Gorton, Gary, and James Kahn. "The Design of Bank Loan Contracts, Collateral, and
Renegotiation." 1993. doi:10.3386/w4273.
Gorton, Gary, and Andrew Winton. "8. Financial Intermediation." Handbook of the
Economics of Finance 1, no. 1 (2003): 431-552. doi:10.1515/9781400845422-012.
Greenbaum, Stuart I., and Anjan V. Thakor. Contemporary Financial Intermediation
(Academic Press advanced finance series). Academic Press, 2007.
73
Gârleanu, Nicolae, and Jeffrey Zwiebel. "Design and Renegotiation of Debt Covenants."
Review of Financial Studies 22, no. 2 (2008): 749-781. doi:10.1093/rfs/hhn017.
Hart, Oliver, and John Moore. "Incomplete Contracts and Renegotiation." Econometrica 56,
no. 4 (1988): 755. doi:10.2307/1912698.
Henderson, James. Key determinants for the future of Russian oil production and exports.
Oxford: Oxford Institute for Energy Studies, 2015. http://www.worldcat.org/title/keydeterminants-for-the-future-of-russian-oil-production-and-exports/oclc/910847028.
Holthausen, Robert W., and Richard W. Leftwich. "The economic consequences of
accounting choice implications of costly contracting and monitoring." Journal of
Accounting and Economics 5 (1983): 77-117. doi:10.1016/0165-4101(83)90007-1.
Hoshi, Takeo, Anil Kashyap, and David Scharfstein. "Bank Monitoring and Investment:
Evidence from the Changing Structure of Japanese Corporate Banking
Relationships." 1989. doi:10.3386/w3079.
Hoshi, Takeo, Anil Kashyap, and David Scharfstein. "The Role of Banks in Reducing the
Costs of Financial Distress in Japan." 1990. doi:10.3386/w3435.
Huberman, Gur, and Charles M. Kahn. "Limited Contract Enforcement and Strategic
Renegotiation." American Economic Review 78, no. 3 (1988): 471-484.
http://EconPapers.repec.org/RePEc:aea:aecrev:v:78:y:1988:i:3:p:471-84.
Hull, John. Options, Futures, and Other Derivatives. Boston: Prentice Hall, 2012.
"Interbank Money Market from 01.08.2000." Bank of Russia. Accessed May 25, 2016.
http://www.cbr.ru/eng/hd_base/?Prtid=mkr_base.
Investfunds. Eng.Investfunds.Ru - Mutual Funds and Asset Management in Russia,
Investment Funds of Russia, Russian Funds, Fund Quotes and Performance. Accessed
May 25, 2016. http://eng.investfunds.ru/.
Jensen, Michael C., and William H. Meckling. "Theory of the firm: Managerial behavior,
agency costs and ownership structure." Journal of Financial Economics 3, no. 4
(1976): 305-360. doi:10.1016/0304-405x(76)90026-x.
Jha, A. "Earnings Management Around Debt-Covenant Violations - An Empirical
Investigation Using a Large Sample of Quarterly Data." Journal of Accounting,
Auditing & Finance 28, no. 4 (2013): 369-396. doi:10.1177/0148558x13505597.
John, Kose, and Simi Kedia. "Institutions, Markets and Growth: A Theory of Comparative
Corporate Governance." SSRN Electronic Journal, 2006. doi:10.2139/ssrn.686824.
Karapetov, Artem G. "Pravovaya priroda i posledstviya narusheniya kovenantov v
finansovykh sdelkakh." Chastnoe pravo i finansovyj rynok: Sb. statej (Legal Nature
and Consequences of the Violation of Covenants in Financial Transactions. Private
Law and Financial Market: The Collection of Articles) 1 (2011): 106-168.
Korteweg, Arthur. "The Net Benefits to Leverage." The Journal of Finance 65, no. 6 (2010):
2137-2170. doi:10.1111/j.1540-6261.2010.01612.x.
Lancett, Peter. "Types of Debt Covenants." EHow. Accessed May 24, 2016.
http://www.ehow.com/list_6464801_types-debt-covenants.html.
Malitz, Ileen. "On Financial Contracting: The Determinants of Bond Covenants." Financial
Management 15, no. 2 (1986): 18. doi:10.2307/3664974.
Marketline. Oil & Gas in Russia. WWW.MARKETLINE.COM, 2016. Reference Code:
0153-2116.
74
Merton, Robert C. "On the Pricing of Corporate Debt: The Risk Structure of Interest Rates."
The Journal of Finance 29, no. 2 (1974): 449. doi:10.2307/2978814.
Metelitsa, Ekaterina, Siranush Sharoyan, and Zhanna Nemtsova. "Если завтра BB ::
Общество :: Газета РБК." Газета РБК. Last modified January 12, 2015.
http://www.rbc.ru/newspaper/2015/01/12/56bd45059a7947299f72c4fa.
Minnis, Michael, and Andrew G. Sutherland. "Financial Statements as Monitoring
Mechanisms: Evidence from Small Commercial Loans." SSRN Electronic Journal,
2016. doi:10.2139/ssrn.2327378.
Mun, Johnathan. Advanced Analytical Models: Over 800 Models and 300 Applications from
the Basel II Accord to Wall Street and Beyond. Hoboken, N.J.: Wiley, 2008.
Mun, Johnathan. Modelling Risk - Applying Monte Carlo Risk Simulation, Strategic Real
Options, Stochastic Forecasting, and Portfolio Optimization, 2nd ed. New Jersey:
John Wiley & Sons, Inc, 2012.
Murfin, Justin. "The Supply-Side Determinants of Loan Contract Strictness." The Journal of
Finance 67, no. 5 (2012): 1565-1601. doi:10.1111/j.1540-6261.2012.01767.x.
Myers, Stewart C. "Determinants of corporate borrowing." Journal of Financial Economics
5, no. 2 (1977): 147-175. doi:10.1016/0304-405x(77)90015-0.
Myers, Stewart, and Nicholas Majluf. "Corporate Financing and Investment Decisions When
Firms Have Information That Investors Do Not Have." Journal of Financial
Economics 13, no. 2 (1984): 187-221. doi:10.1016/0304-405X(84)90023-0.
Nash, Robert C., Jeffry M. Netter, and Annette B. Poulsen. "Determinants of contractual
relations between shareholders and bondholders: investment opportunities and
restrictive covenants." Journal of Corporate Finance 9, no. 2 (2003): 201-232.
doi:10.1016/s0929-1199(02)00007-x.
Paglia, John K., and Donald J. Mullineaux. "An empirical exploration of financial covenants
in large bank loans." Banks and Bank Systems 1, no. 2 (2006): 103-122.
Petlevoy, Vitaliy. "ВЕДОМОСТИ - Группа «Кокс» нарушила ковенанты по кредиту
Газпромбанка на 4,5 млрд рублей." ВЕДОМОСТИ - новости бизнеса и
финансов, аналитика, смарт-версия газеты. Last modified May 5, 2016.
https://www.vedomosti.ru/business/articles/2016/05/06/640165-gruppa-koksnarushil-kovenanti-kreditu-gazprombanka-45-mlrd-rublei.
"Quick Assets." Encyclopedia of Finance (n.d.): 224-224. doi:10.1007/0-387-26336-5_1657.
Rajan, Raghuram, and Andrew Winton. "Covenants and Collateral as Incentives to Monitor."
The Journal of Finance 50, no. 4 (1995): 1113. doi:10.2307/2329346.
Rehmann. Analyzing Cash Flows. 2013. http://www.mibankers.com/files/Analyzing
%20Cash%20Flows-%20Rehmann.pdf.
Reisel, Natalia. "On the Value of Restrictive Covenants: Empirical Investigation of Public
Bond Issues." SSRN Electronic Journal (n.d.). doi:10.2139/ssrn.644522.
Roberts, Michael R., and Amir Sufi. "Control Rights and Capital Structure: An Empirical
Investigation." The Journal of Finance 64, no. 4 (2009): 1657-1695.
doi:10.1111/j.1540-6261.2009.01476.x.
Romanov, Daniil. "Dividend Policy of Russian Oil&Gas Public Companies." Master's thesis,
Saint Petersburg State University, 2014.
Ross, Stephen A. "The Determination of Financial Structure: The Incentive-Signalling
Approach." The Bell Journal of Economics 8, no. 1 (1977): 23. doi:10.2307/3003485.
75
Saavedra, Daniel. "Renegotiation and the Choice of Covenants in Debt Contracts." SSRN
Electronic Journal, 2014. doi:10.2139/ssrn.2626942.
Sennikov, Igor. "Implementation of Simultaneous Equations Model to Forecasting Residual
Income of Russian Oil Companies." Master's thesis, Graduate School of
Management, Saint Petersburg State University, 2016.
Shleifer, Andrei, and Robert W. Vishny. "Management entrenchment." Journal of Financial
Economics 25, no. 1 (1989): 123-139. doi:10.1016/0304-405x(89)90099-8.
Smith, Clifford W. "A Perspective on Accounting-Based Debt Covenant Violations." The
Accounting Review 68, no. 2 (April 1993): 289-303.
http://www.jstor.org/stable/248402.
Smith, Clifford W., and Jerold B. Warner. "On financial contracting." Journal of Financial
Economics 7, no. 2 (1979): 117-161. doi:10.1016/0304-405x(79)90011-4.
Sweeney, Amy P. "Debt-covenant violations and managers' accounting responses." Journal
of Accounting and Economics 17, no. 3 (1994): 281-308. doi:10.1016/01654101(94)90030-2.
Thomson Reuters. "Eikon Financial Analysis | Thomson Reuters." Financial Software, Data,
Analytics and News | Thomson Reuters. Accessed May 25, 2016.
http://financial.thomsonreuters.com/en/products/tools-applications/tradinginvestment-tools/eikon-trading-software.html.
Tirole, Jean. "Cognition and Incomplete Contracts." American Economic Review 99, no. 1
(2009): 265-294. doi:10.1257/aer.99.1.265.
Ventureline. "Cash Sweep Definition." Financial Analysis - Ratio Analysis - Industry Ratios.
Accessed May 24, 2016. https://www.ventureline.com/accounting-glossary/C/cashsweep-definition/.
Watts, Ross L., and Jerold L. Zimmerman. Positive Accounting Theory. Englewood Cliffs,
N.J.: Prentice-Hall, 1986.
Wilkins, Michael S. "Additional evidence on the consequences of debt covenant violation."
PhD diss., University of Arizona, 1994. http://hdl.handle.net/10150/186963.
Williamson, Oliver E. "Strategy research: governance and competence perspectives."
Strategic Management Journal 20, no. 12 (1999): 1087-1108. doi:10.1002/(sici)10970266(199912)20:12<1087::aid-smj71>3.3.co;2-q.
Yu, Jin. "The Value of Debt Covenants: A Quantitative Framework." PhD diss., Vienna
Graduate School of Finance, 2010.
Zadorozhnaya, Anna. "The influence of covenant protection on yield of corporate bond."
Journal Finance and credit 7, no. 631 (February 2015): 34-44.
Zinbarg, Edward D. "The Private Placement Loan Agreement." Financial Analysts Journal
31, no. 4 (1975): 33-35. doi:10.2469/faj.v31.n4.33.
"Сбербанк планирует встретиться с новым руководством АВТОВАЗа." TLT.ru Новости Тольятти. Last modified May 22, 2016. http://tlt.ru/articles.php?
n=1975964.
76
Appendix 1. Russian oil industry players
FIRM NAME (Bloomberg 2016)
Public Joint Stock Oil Company Bashneft
Open Joint Stock Company Surgutneftegas
Open Joint-Stock Oil and Gas Company Slavneft
Open Joint Stock Company Rosneft Oil Company
Public Joint Stock Company Oil Company LUKOIL
PJSC Tatneft
PJSC Gazprom Neft
77
Appendix 2. Simulation Assumptions
CONSOLIDATED (mln USD)
Revenues
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Gneft
8796
14332
19931
22428
33205
23773
32176
35032
39694
39789
36500
Tatneft
7180
10619
11714
13936
17860
12004
15411
14205
14301
14281
12347
Lukoil
34058
56215
67684
81891
107680
81083
104956
133650
139171
141452
144167
Rosnef Slavnef Surgut Bashnef Average
t
t
t
4628
5938 10699
1877
10454
18332
6331 15780
2462
17724
24004
6348 19382
3503
21795
41727
5937 24146
3777
27692
51113
5717 23175
4710
34780
46422
3505 16608
6433
27118
63057
4023 20274
11707
35943
92461
5329 27951
16544
46453
99472
6398 27358
17148
49077
147340
6059 26296
17681
56128
142632
5117 23083
16517
54338
Opex
Gneft
Tatneft Lukoil
Rosneft Slavneft Surgut Bashneft Average
2004
5645
5851
21976
4979
3026
7734
2159
7339
2005
11127
8848
39748
15143
2735 11294
2803
13100
2006 11249
10088
49023
27799
2818 15471
3670
17160
2007 13187
11450
57761
32986
2997 19407
3906
20242
2008 19720
16267
79463
42858
3726 19840
4854
26675
2009 16413
9685
58536
32577
2274 13852
3748
19584
2010 21371
13244
78818
43102
2889 16611
9936
26567
2011 27863
8361 103124
68715
5329
8874
13869
33734
2012 32396
8115 109237
77704
5413
9573
14507
36706
2013 32528
8171 111636 117080
5210
9644
14940
42744
2014 30745
6824 118175 114562
4685
8975
14383
42621
Assets
Gneft
Tatneft Lukoil
Rosneft Slavneft Surgut Bashneft Average
2004 10262
11171
29761
25093
5224 26469
2169
15736
2005 10639
9817
40345
30658
5505 31201
2483
18664
2006 14102
12136
48237
34191
5694 36006
2962
21904
2007 16609
15068
59632
57351
6806 42539
3433
28777
2008 20205
12874
71461
83931
7455 39579
3140
34092
2009 29912
16355
79019
89900
7518 43401
12664
39824
2010 31251
18267
84017
98621
7713 52110
14944
43846
2011 34732
18867
91192 107568
7441 58895
13573
47467
2012 42553
20640
98961 129975
8821 67784
15171
54844
2013 47540
20495 109439 228972
8581 72513
13614
71593
2014 36137
12626
81638 150495
5184 56954
9021
50294
CONSOLIDATED (proportion)
78
D/E
Gneft
Tatneft Lukoil
Rosneft Slavneft Surgut Bashneft Average
2004
0,43
0,82
0,43
3,88
1,08
0,06
0,11
0,97
2005
0,39
0,42
0,51
1,88
0,96
0,06
0,16
0,63
2006
0,42
0,38
0,47
1,77
0,89
0,06
0,16
0,59
2007
0,59
0,40
0,45
2,88
1,09
0,07
0,10
0,80
2008
0,45
0,51
0,42
0,89
1,13
0,05
0,18
0,52
2009
0,85
0,62
0,41
0,83
1,10
0,06
1,39
0,75
2010
0,76
0,73
0,42
0,71
1,58
0,12
1,84
0,88
2011
0,63
0,62
0,35
0,67
2,08
0,13
1,42
0,85
2012
0,56
0,47
0,35
0,74
2,30
0,14
0,91
0,78
2013
0,64
0,40
0,39
1,41
2,11
0,16
0,90
0,86
2014
0,97
0,32
0,56
2,04
3,34
0,14
1,67
1,29
Mean: 0,81; Standard Deviation: 0,21
Quick assets Gneft Tatneft Lukoil Rosneft Slavneft Surgut Bashneft Average
2004
0,11
0,08
0,05
0,07
0,02
0,09
0,02
0,06
2005
0,03
0,08
0,04
0,07
0,01
0,13
0,01
0,05
2006
0,10
0,07
0,02
0,05
0,02
0,21
0,10
0,08
2007
0,04
0,07
0,01
0,04
0,02
0,15
0,07
0,06
2008
0,11
0,06
0,04
0,04
0,04
0,33
0,14
0,11
2009
0,03
0,05
0,03
0,05
0,04
0,31
0,11
0,09
2010
0,04
0,05
0,03
0,11
0,02
0,19
0,07
0,07
2011
0,03
0,06
0,03
0,09
0,03
0,21
0,06
0,07
2012
0,07
0,04
0,03
0,09
0,12
0,18
0,04
0,08
2013
0,09
0,06
0,02
0,04
0,10
0,20
0,04
0,08
2014
0,06
0,08
0,03
0,04
0,05
0,23
0,10
0,08
Mean: 0,08; Standard Deviation: 0,02
Dividend payout ratio (average)
Interest as part of debt
20,00%
9,00%
GEOMETRIC BROWNIAN MOTION PARAMETERS
Parameters:
Sigma
Drift
Time step
Risk-free rate
Revenue Costs Assets
0,21
0,22
0,18
0,09
0,11
0,05
1,00
1,00
1,00
4,87%
79
Appendix 3. Default probabilities of the firm (i=9%, various principal)
Object 117
80
Appendix 4. Distribution of covenant violations over time (various principal)
Object 119
81
Appendix 5. Default probabilities for loan with and without covenant
Object 121
Object 123
82
Appendix 6. Expected values and volatilities for covenant and covenant free
loan
N=$3000, i=7%
Threshold
Expected loan
value with
covenant
1,50
2,00
2,50
3,00
3,50
4,00
4,50
$3 156,95
$3 157,35
$3 155,03
$3 150,87
$3 142,11
$3 129,93
$3 111,58
Volatility (%)
with covenant
1,2967%
0,8217%
0,8764%
0,6005%
0,8746%
1,1342%
1,3434%
Expected
loan value
without
covenant
Volatility (%)
no covenant
$3 132,78
3,1304%
Expected
loan value
without
covenant
Volatility (%)
no covenant
Difference
$24,17
$24,57
$22,25
$18,09
$9,33
-$2,85
-$21,20
N=$3000, i=9%
Threshold
Expected loan
value with
covenant
1,50
2,00
2,50
3,00
3,50
4,00
4,50
$3 452,08
$3 449,19
$3 440,52
$3 424,13
$3 400,09
$3 358,16
$3 307,40
Volatility (%)
with covenant
1,1326%
1,0108%
1,2455%
1,7038%
2,3311%
3,1011%
3,6023%
$3 422,54
3,5273%
Expected
loan value
without
covenant
Volatility (%)
no covenant
Difference
$29,54
$26,65
$17,97
$1,59
-$22,46
-$64,38
-$115,14
N=$7000, i=9%
Threshold
Expected loan
value with
covenant
1,50
2,00
2,50
3,00
3,50
4,00
4,50
$8 051,28
$8 048,22
$8 028,23
$7 987,93
$7 913,55
$7 809,97
$7 696,30
Volatility (%)
with covenant
1,4561%
1,0590%
1,0799%
1,4542%
2,0028%
2,4393%
2,4895%
$7 987,37
3,5854%
Difference
$63,90
$60,84
$40,86
$0,56
-$73,83
-$177,40
-$291,08
83
N=$15000, i=7%
Threshold
Expected loan
value with
covenant
1,50
2,00
2,50
3,00
3,50
4,00
4,50
$15 790,92
$15 791,73
$15 769,85
$15 726,73
$15 661,88
$15 597,22
$15 551,53
Volatility (%)
with covenant
0,9201%
0,3679%
0,4107%
0,4957%
0,5755%
0,5329%
0,4620%
Expected
loan value
without
covenant
$15 632,27
Volatility (%)
no covenant
3,3656%
Difference
$158,65
$159,45
$137,57
$94,45
$29,61
-$35,06
-$80,74
84
Отзывы:
Авторизуйтесь, чтобы оставить отзыв