master thesis maurits kruithof

63
Basel III Capital Requirements Impact of Higher Capital Requirements on Bank Funding Costs Maurits J. H. Kruithof Master’s Thesis Business Economics, Finance Track Date: March 14, 2013 Student Number: 5603404 Supervisor: Professor Arnoud W. A. Boot Second Examiner: Dr. Jeroen E. Ligterink University of Amsterdam, Faculty of Economics and Business

Upload: mauritskruithof

Post on 01-Dec-2015

111 views

Category:

Documents


3 download

DESCRIPTION

Basel III Capital Requirements: Impact of Higher Capital Requirements on Bank Funding CostsMaster Business Economics, Finance (University of Amsterdam)

TRANSCRIPT

 

Basel III Capital Requirements

Impact of Higher Capital Requirements on Bank Funding Costs

Maurits J. H. Kruithof Master’s Thesis

Business Economics, Finance Track

Date: March 14, 2013 Student Number: 5603404 Supervisor: Professor Arnoud W. A. Boot Second Examiner: Dr. Jeroen E. Ligterink

University of Amsterdam, Faculty of Economics and Business

 

[this page intentionally left blank]

 

Basel III Capital Requirements

Impact of Higher Capital Requirements on Bank Funding Costs

Abstract

This thesis analyzes the impact of higher capital requirements on bank funding costs.

Often is claimed that capital is an expensive form of funding. An extensive literature

review of theoretical insights points out that this is not necessarily the case. The

foundation of capital structure research is the Modigliani-Miller theorem. The question is

whether this theorem holds in practice and is applicable to banks. It proves to be a

useful theorem to identify relevant frictions and distortive policies. This thesis scrutinizes

the public policies that favor debt financing over equity financing, because the corporate

tax system and implicit government guarantees create a significant lower cost of debt

funding. The Basel III capital requirements should therefore be complemented with tax

policy reforms and recapitalization of the banking system with the use of a contingent

capital (CoCo) requirement.

JEL classifications: G21, G28, G32, G38, H25

Keywords: Banking Regulation, Basel III, Capital Requirements, Capital Structure,

Funding Costs, Government Guarantee, Lending Spread, Leverage, Modigliani-Miller

Theorem, Tax Shield.

 

Table of Contents

1. Introduction.…………………………………………………………………………………………… 7 PART I Theory and Empirics

2. Higher Capital Requirements: Theoretical Insights.……………………… 10 2.1 Modigliani and Miller (1958)..…………………………………………………………………… 10 2.2 Modigliani-Miller Theorem versus CAPM.…………………………………………………… 12 2.3 Having More Equity Capital: Steady State.……………………………………………… 15 2.3.1 Cost of Capital Fallacy.……………………………………………………………………………… 15 2.3.2 The Role of Subsidies on Debt in New Equilibrium……………………………………… 16 2.4 Raising More Equity Capital: Transition Phase………………………………………… 17 2.4.1 Information Asymmetry.…………………………………………………………………………… 17 2.4.2 Debt Overhang………………………………………………………………………………………… 18 2.5 Conclusion.………………………………………………………………………………………………… 20

3. Empirical Studies on Higher Capital Requirements..……………………… 21 3.1 Kashyap, Stein and Hanson (2010).………………………………………………………… 21 3.2 King (2010).……………………………………………………………………………………………… 22 3.3 Angelini et al. (2011)..……………………………………………………………………………… 24 3.4 Cosimano and Hakura (2011)..………………………………………………………………… 27 3.5 Santos and Elliott (2012).………………………………………………………………………… 29 3.6 Conclusion.………………………………………………………………………………………………… 30 PART II Tax Shield, Government Guarantee and Policy Reforms 4. Tax Shield on Debt………………………………………………………………………………… 32 4.1 The Methodology of Debt Tax Shield Calculation.…………………………………… 32 4.2 The Size of the Dutch Bank Tax Shield.…………………………………………………… 33 4.3 The Future of the Tax Shield.…………………………………………………………………… 37 5. Government Guarantees and Recapitalization..……………………………… 39 5.1 Impact and Consequences of Government Guarantees.………………………… 39 5.2 The Size of the Dutch Government Guarantee..……………………………………… 40 5.3 Recapitalization of the Banking System..………………………………………………… 44 6. Summary and Conclusion.…………………………………………………………………… 46 List of Abbreviations.…………………………………………………………………………………………………… 48 Bibliography…………………………………………………………………………………………………………………… 49 Other References, Sources and Data..……………………………………………………………………… 53 Appendices..…………………………………………………………………………………………………………………… 54

 

List of boxes, figures and tables Box 2.1: Roles of Capital.………………………………………………………………………………………… 10 Figure 1.1: Process Display of the Statement.…………………………………………………………… 7 Figure 2.1: Alternative Responses to Increased Capital Requirements.…………………… 19 Figure 3.1: Alternative Responses to Increased Capital Requirements.…………………… 26 Figure 4.1: Leverage (Equity Multipliers) of Three Largest Dutch Banks.………………… 34 Figure 4.2: Key Interest Rates.…………………………………………………………………………………… 35 Figure 4.3: Tax Shield on Debt (in EUR millions).……………………………………………………… 36 Figure 4.4: Tax Shield as a Percentage of Total Assets.…………………………………………… 36 Figure 4.5: Fee Income as a Percentage of Total Interest and Fee Income..…………… 37 Figure 5.1: Notches Between “Stand-Alone” and “Supported” Credit Ratings.………… 41 Table 4.1: Profit and Loss Account, Tax Shield………………………………………………………… 32 Table 4.2: Corporate Tax Rate in the Netherlands…………………………………………………… 33 Table 4.3: Leverage (Equity Multipliers) of Three Largest Dutch Banks.………………… 34 Table 5.1: Implicit Subsidy High (In Millions) 1999-2012..……………………………………… 42 Table 5.2: Implicit Subsidy Low (In Millions) 1999-2012………………………………………… 42 Appendices 1. Data Leverage Calculation…………………………………………………………………………………… 54 2. Key Interest Rates..……………………………………………………………………………………………… 55 3. Data Tax Shield.…………………………………………………………………………………………………… 56 4. Net Interest and Fee Income.……………………………………………………………………………… 57 5. Data and Calculations Implicit Government Guarantee……………………………………… 58 6. Capital Ratios Graphs…………………………………………………………………………………………… 62 7. Bank Lending Spreads Graph.……………………………………………………………………………… 63

 

Preface

After a difficult start, where I’ve spent several months reading all sorts of papers that

were very interesting but totally not relevant, I finally got the spirit during my first

extensive meeting with professor Boot in his office. I’m very thankful for his support,

advice and time he spent with me on my thesis. I also think that this is the right place to

express my gratitude to him for all he has done to help Room for Discussion achieve

success. It happens rarely these days that a mentor-student relationship is possible

when so many students are pursuing their ambitions. I was privileged to have such a

great mentor.

I also want to thank my parents for their ongoing support and interest during my studies

and work for Room for Discussion. They’ve helped me creating the circumstances in

which I could do all the things I needed to do, for that I’m very grateful.

After all the interviews and debates I did for Room for Discussion, I certainly knew that

my thesis had to be about banks and the financial sector. During one of my preparations

for a debate I found a speech by Thomas Huertas, which was written before the collapse

of Lehman Brothers. It contained the following quote1: ‘Capital is the cornerstone of

banking. Capital is the foundation on which banks take risks and achieve rewards, and

capital is ultimately what protects deposits.’ If capital really is the cornerstone of banking,

why were banks so poorly capitalized that the banking crisis of 2007-2009 could happen?

Well, I’ve tried to find an answer and that ultimately resulted in this Master’s thesis.

Maurits Kruithof

Amsterdam, March 2013

                                                                                                                         1  Derived  from  a  speech  by  Thomas  Huertas,  FSA  United  Kingdom,  June  26th  2008,  via  www.fsa.gov.uk.  *  I  thank  my  friends  Richard  Evers  and  Gerben  Smit  for  their  useful  comments  and  suggestions.  I  also  thank  the  banks,  accountancy  and  consultancy  firms  with  whom  I’ve  had  several  informal  meetings  for  their  time,  advices,  comments,  criticism  and  willingness  to  spar  with  me.  

7    

1. Introduction

The recent financial crisis of 2007-2009 shows that the ability of the banking sector to

deal with major shocks must be strengthened. The assets that are held by the banking

sector are too risky or high priced compared to the amount of capital on the liability side

of the balance sheet. The sector is not capable to absorb losses on their own positions,

portfolios and loans provided to companies and households. In 2008 and 2009

governments across the world had to recapitalize banks and guarantee bank debt.

To prevent recurrence of such problems in the long run the Basel Committee presents

reforms to strengthen global capital and liquidity rules, henceforth Basel III (BCBS,

2010a). This should improve the resilience of individual institutions, but also contribute

to greater stability of the financial system as a whole. The measures of Basel III

intervene in the capital structure of banks. The banking sector has to acquire more

capital and of higher quality. Banks’ balance sheets and capital structures will be

different after the implementation of Basel III. Many bankers argue that equity capital is

expensive and higher capital requirements increase the total cost of funding and the

price of a bank loan. Therefore, this thesis examines the following statement: equity

capital is the most expensive form of funding compared to debt and depositors’ money,

therefore raising capital requirements increases the total costs of bank funding (figure

1.1 and equation (1)).

Figure 1.1: Process Display of the Statement

𝑟!"#$%&'%  <  𝑟!"#$!!!"#$  <  𝑟!"#$  !"#$  !"#$  <  𝑟!"#$%& (1)2

This master thesis explains the relationship between equity capital held by a bank and

total costs of bank funding. It tries to set out what impact higher capital requirements,

such as the Basel Committee’s (BCBS, 2010a) capital requirements, have on the bank’s

funding costs of debt and equity and the price of a bank loan. The research question of                                                                                                                          2  Source:  King  (2010),  𝑟  is  the  (required)  rate  of  return.      

Higher  capital  requirements  

Higher  cost  of  equity  funding  

Higher  total  cost  of  funding  

8    

this thesis is: What impact have higher capital requirements on the cost of equity capital

and total funding costs? This study reviews important theoretical literature that is

available on the impact of higher capital requirements. It also analyzes recent papers

that empirically tested the effects of higher capital requirements on total funding costs

and prices of bank loans. The assumptions of these empirical studies are compared to

the fundamental and theoretical insights, including those from Admati et al.’s (2011) key

paper. Analyzing higher capital requirements demands a clear distinction between having

more equity capital (steady state) and raising more equity capital (transition phase). The

dynamics of a steady state and transition phase analysis are rather different.

Based on the Modigliani-Miller capital structure theory (1958), Admati et al. (2011) state

that bank capital is not expensive and that many arguments are fallacious (steady state).

Though, not a few other authors question the applicability of the Modigliani-Miller

theorem on banking. Banks have a certain “special” role, which is to provide liquidity to

the economy and debt is the instrument that banks need to fulfill this role. However, it

turns out that public policy creates distortions in the funding costs of debt and equity,

mainly due to fiscal incentives and government guarantees. The high leverage ratio of

the financial sector is partly explained by these policies that subsidize debt. Higher

capital requirements mean that banks can make less use of these implicit subsidies.

These distortions can be resolved, since they are part of public policy.

This thesis is relevant to financial policy makers, people working in the financial sector

and those engaged in scientific research into capital regulation. It should lead to a better

understanding and awareness of the importance of correct and sufficient capital

regulation in relationship with capital structure. It provides a discussion of “state of art”

theories and concepts written in fundamental papers about capital regulation, capital

structures and bank lending. This thesis explains that public policy favors debt financing

and gives proposals of how policies on taxation and government guarantees can be

reformed, or at least which direction new policies should have to reduce the incentives

for debt financing. It also provides a notion of recapitalization, which reduces the

government guarantee and debt overhang problem. The system has to be robust and

able to absorb losses, while less dependent on government guarantees.

This thesis proceeds as follows. The first part of this thesis focuses on theoretical and

empirical studies of higher capital requirements and funding costs. Starting with chapter

two, it discusses the theoretical insights related to higher capital requirements, the

difference between raising and having more equity capital and what kind of impact this

9    

has on banks’ funding costs. Chapter three compares assumptions and results of

empirical studies with the theoretical insights of chapter two. The result of this analysis

emerges two main distortive policies in the discussion of higher capital requirements,

namely corporate tax rules and governments guarantees that implicitly favor debt

financing. These two distortions will be at the center of the second part of this thesis,

respectively chapters four and five. Chapter four elaborates the tax shield methodology,

provides a calculation of the Dutch size of the tax shield and proposes tax policy reforms.

Chapter five sets forth the implicit government guarantee, estimates the size of the

Dutch guarantee and proposes efficient recapitalization in relationship with government

guarantees. This thesis ends with a summary and conclusion in chapter six.

10    

PART I Theory and Empirics 2. Higher Capital Requirements: Theoretical Insights

This chapter discusses theoretical concepts and insights related to the capital structure

of banks, banks’ funding costs and higher capital requirements. The basic theory of

capital structure composed by Modigliani and Miller (1958) is the starting point of this

chapter. Many arguments that are put forward by Admati, DeMarzo, Hellwig and

Pfleiderer (2011) are based on this theory of corporate finance. Admati et al. (2011)

explain why bank equity is not expensive and refute many fallacious, irrelevant and/or

very weak arguments. But the academic literature is ambivalent in thinking about the

Modigliani-Miller theorem and its applicability to banks. This chapter also explains

important distinctions between raising more equity capital (transition phase) and having

more equity capital (steady state). The dynamics of raising and having more equity

capital are quite different. Raising more equity capital implies debt overhang problems

and creates information asymmetry problems. Having a higher equity capital ratio in a

new, steady state, equilibrium affects e.g. the benefits of the tax shield and implicit or

explicit government guarantees. This chapter ends with a brief summary and conclusion.

2.1 Modigliani and Miller (1958)

Modigliani and Miller (1958), who wrote a fundamental paper about capital structure,

state that under certain assumptions a firm’s capital structure is irrelevant for I) its value

and II) weighted average cost of capital (WACC). These are known as the Modigliani and

Miller Propositions I and II. The four assumptions made in the frictionless Modigliani and

Miller world are severe; they include no information asymmetry, taxes, financial distress

costs and transaction costs.

Box 2.1 Roles of Capital

Capital is one of the most fundamental concepts in economics. Wherever there is entrepreneurial

activity, investments made and clients served, capital plays an essential role in businesses. It

provides funding, receives profits and is the only mechanism on the balance sheet to absorb

losses. The role of capital can roughly been split into three main characteristics3: 1) as a technical

instrument on balance sheets, 2) as a governance tool and 3) as a systemic buffer.

The first role of capital is technical. The amount of capital relative to the amount of debt is crucial

in this wide debate on the role of capital. The extent to which capital is risky or costly relies on the

leverage of a firm (Modigliani and Miller, 1958) and the riskiness of the assets. The second                                                                                                                          3  Inspired  by  Modigliani  and  Miller  (1958,  p.  261)  who  viewed  their  main  question  “what  is  the  cost  of  capital?”  through  three  perspectives,  the  one  of  the  corporate  finance  specialist  (technical),  the  manager  (governance)  and  macroeconomist  (systemic  level).  

11    

important and fundamental function of capital is that it monitors the management and distributes

risk among its shareholders. Shareholders are more or less the owners of the company and they

use their voting rights to control or influence management decisions. The agency theory explains

the moral hazard and adverse selection problems (Jensen, 1986). Third, capital provides as a

buffer for the system as a whole to absorb losses. The first and second roles of capital are

important for individual firms, while the systemic role of capital matters the economy as a whole.

Frequently, the systemic role of capital is wrongly separated from the individual interests of a firm

(Admati et al., 2011). Some argue that capital is too expensive and that cheaper debt finance is

preferable (Gorton, 2010). However, this may not be the case if welfare costs of high leverage

ratios are included (Admati et al., 2011). Berger (1995) states that regulators use capital

requirements to create safety nets and to protect the economy from negative externalities. In this

way, the systemic role of capital is taken into account for individual financial institutions.

A second misunderstanding that is important to mention here is that capital is not something to

keep idle or that must be set aside (Admati et al., 2011). Cochrane (2013) states: “capital is a

source of money, not a use of money.” There is a difference between capital requirements and

liquidity or reserve requirements. Capital requirements prescribe banks how to fund themselves

with debt or equity (leverage ratio), while liquidity or reserve requirements relate to the type of

assets and asset mix banks must hold (Admati et al., 2011). Capital requirements address the

right-hand side of the balance sheet and liquidity or reserve requirements the left-hand side.

However, there is a link between capital requirements and assets, because of the Risk Weighted

Assets (RWA) rule that is included in all Basel Accords.4 Nevertheless, once a bank meets reserve or

liquidity requirements, all capital can be used for new loans and investments.5

An important proposition to discuss is whether the Modigliani-Miller theorem is applicable

to banks, under the same assumptions mentioned above. Bank balance sheets and

operations are fundamentally different compared to non-bank firms. For example, banks

produce financial debt instruments such as deposits, short-term commercial paper and

repurchase agreements to provide liquidity to the economy, while non-bank firms do not

fulfill this function. Admati et al. (2011) conclude that, based on the framework of

Modigliani and Miller, higher capital requirements have no significant, long-term,

negative consequences for the economy that offset the benefits. This only concerns the

new equilibrium (steady state), thus after the equity capital is acquired. Miller (1995)

states that the Modigliani-Miller theorem is only applicable ex ante, when equity capital

ratios can be fully anticipated in an equilibrium (at t=0 or t=1). The theorem is not

applicable during the transition phase, the time between t=0 and t=1. The interest rates

on debt do not reflect the new equity capital infusion between t=0 and t=1, simply

                                                                                                                         4  Each  asset  that  a  bank  holds  is  risk-­‐adjusted,  which  means  that  high  risk  assets  require  a  higher  capital  ratio.  5  More  about  capital  and  bank  lending,  see  Cebenoyan  Strahan  (2004)  Fabi  et  al.  (2005)  Gambacorta,  Mistrulli  (2004)  Inderst,  Mueller  (2008)  Thakor  (1996)  Elliott  (2010a).  

12    

because most of the debt is already in place and terms and conditions cannot be

renegotiated. However, there is an extensive collection of literature available that

discusses the applicability and relevance of the Modigliani-Miller theorem on banking in

the steady state (the equilibrium of today at t=0). If the Modigliani-Miller theorem does

not hold on banking in its pure form, could increasing capital requirements have

significant consequences for bank’s overall cost of capital and eventually their lending

spreads in the new equilibrium (steady state at t=1)?

The applicability of the Modigliani-Miller theorem is questioned in a paper by Gorton,

Lewellen and Metrick (2011). They argue that bank debt is information-insensitive6

similar to government debt. According to Gorton et al. (2011) and Gorton (2010), bank

debt is immune to adverse selection in trading because agents do not want to acquire

private information about the current health of the bank, since acquiring or generating

information is costly. Gorton et al. (2011) regress the fraction of financial liabilities in the

economy against the fraction of government liabilities in the economy and find that

government and financial liabilities are viewed as acceptable substitutes by investors.

Gorton et al. (2011) argue that bank debt therefore may contain a convenience yield,

like government debt. A convenience yield is a yield below what might be expected

according to standard fixed income calculations7. In other words, investors in bank debt

are willing to accept a lower rate of return due to the implicit or explicit government

guarantee. Gorton et al. (2011) use an average convenience yield of 70 basis points. In

a world with such implicit government guarantees the Modigliani-Miller theorem no

longer holds. Implicit and explicit government guarantees are therefore a distortion in

the pricing of banks’ funding costs. Paragraph 2.3 and chapter five will elaborate on the

distortive effects of implicit government guarantees.

2.2 Modigliani-Miller Theorem versus CAPM

A widely debated consequence of higher capital requirements is that more equity capital

should lower the Return On Equity (ROE). Although many bankers claim that equity

capital is expensive and consider the ROE as fixed, basic corporate finance theory shows

that these propositions are inconsistent. The ROE increases both by more asset risk

and/or more leverage and vice versa. The proposition that more equity capital decreases

the ROE can be interpreted on the basis of two theories. First, the Modigliani-Miller

theorem state that the distribution of total asset risk among more shareholders lowers

                                                                                                                         6  Note  from  the  author:  the  definition  ‘information-­‐insensitive’  seems  a  contradictio  in  terminis,  like  ‘risk-­‐free’  is  too.  7  For  example,  the  Dutch  State  has  recently  issued  short-­‐term  debt  with  negative  interest  rates.  

13    

the ROE, while total funding costs of the bank remain unchanged (proposition II). Admati

et al. (2011) rely heavily on this theorem and proposition. Second, the Capital Asset

Pricing Model (CAPM) calculates the required rate of return of a security (in this case a

bank stock, thus the return on equity) in relation to its risk (𝛽!"#$%&  ). The risk (𝛽!"#$%&  ) is

dependent on leverage (!!!!

). The CAPM formula states that the ROE (𝑅!"#$%&  ) equals the

risk-free rate (𝑅!) plus a risk premium (𝑅!) multiplied by the risk factor (𝛽!"#$%&  ):

𝑅!"#$%&   = 𝑅! + 𝛽!"#$%&  𝑅! (2)

This model is independent of the Modigliani-Miller theorem and is using different

assumptions. But, since both the Modigliani-Miller theorem and CAPM calculate the ROE,

Gorton et al. (2011) and Miles, Yang, Marcheggiano (2012) find it useful to examine

their relationship. Does the Modigliani-Miller theorem holds simultaneously with CAPM?

Gorton et al. (2011) state that if a bank satisfies the minimum capital requirements it

can produce information-insensitive debt with a convenience yield. Banks that do not

satisfy the minimum capital requirements are considered insolvent: their debt will

become information-sensitive. The existence of a convenience yield on debt breaks the

basic corporate finance theory on capital structure, risk and return. Holding asset risk

and return on assets unchanged, existing shareholders benefit from cheaper debt at the

expense of debt holders (and at the expense of taxpayers when a bail-out is needed).

The implicit government guarantee enables shareholders to receive a higher return. If

banks would follow the corporate finance theory, the advantage of the convenience yield

must result in lower interest rates charged on loans. Because banks can obtain cheaper

debt, they are able to offer loans with lower interest rates. This would imply that the

relationship between the Modigliani-Miller theorem and CAPM does not hold. According to

Gorton et al. (2011), either one of the following two statements can be true:

I. The ROE of banks exceeds their cost of capital under the CAPM. In this case, the

Modigliani-Miller theorem still holds, but CAPM no longer holds (= higher return

on equity with no significant change of risk).

II. Because of the existence of a convenience yield, banks can lower their return

on assets (interest rates charged on loans), leaving the returns on equity and

debt unchanged. Now, the CAPM holds, but Modigliani-Miller Proposition II no

longer holds.8

Gorton et al. (2011) regress bank equity returns against the market portfolio to test

whether statement I is true and banks earn a significant higher equity return given their

                                                                                                                         8  The  required  return  on  assets  is  independent  of  the  firm’s  capital  structure.  

14    

level of risk. They find no abnormal equity return relative to the CAPM, which means

statement II is most likely the case and statement I is not true (Gorton et al., 2011).

Because the counterfactual of statement II is not directly observable, Gorton et al. (2011)

cannot test this statement. However, they assume that the convenience yield influences

the interest rate that banks charge on loans. Banks are driven by competition and will

therefore lower their interest rates charged on loans to gain as much clients as possible.

Gorton et al. (2011) argue that non-bank firms don’t issue similar debt with a

convenience yield, because these firms are able to obtain the same gain as a borrower

by taking out a bank loan with a lower interest rate.

Miles et al. (2012) state that the Modigliani-Miller theorem is unlikely to hold exactly and

use the theorem to assess its relevance for measuring the social costs of more equity

financed lending by banks. They refer to key questions such as how the probability of

crisis falls when banks hold more capital and to what extent the ROE lowers when banks

hold more capital and reduce the risk of that capital. Miles et al. (2012) mention the tax

and guarantee distortions as important factors that influence financial structure. As

mentioned earlier, these distortions ensure that Modigliani-Miller theorem does not hold

completely. Miles et al. (2012) use data on UK banks to test this empirically (chapter 3).

Similar to Gorton et al. (2011), Miles et al. (2012) use the CAPM to test if bank leverage

and risk/return are correlated. The risk of bank assets (𝛽!""#$") is distributed among debt

and equity holders. Therefore, 𝛽!""#$" can be written as follows:

𝛽!""#$"   = 𝛽𝑒𝑞𝑢𝑖𝑡𝑦!

!!!+ 𝛽!"#$

!!!!

(3)

(D=debt, E=equity and 𝛽!"#$=risk of debt)

Assuming that debt is riskless (𝛽!"#$ = 0), this equation implies:

𝛽!"#$%&   =!!!!𝛽!""#$ (4)

Equation (4) shows the similarity of the CAPM and Modigliani-Miller theorem (Miles et al.,

2012), namely a linear relationship between risk and leverage. Under the assumption of

riskless debt, which is more or less the same as a convenience yield, the ROE depends on

leverage. More equity capital results in a decrease of risk and return on equity.

Miles et al. (2012) regress equation (4) to test the linear relationship between risk and

leverage. Will the CAPM and Modigliani-Miller theorem hold if banks halve their leverage?

This implies that equity risk (𝛽!"#$%&  ) is reduced by 50%. Their results show that the

relationship between the Modigliani-Miller theorem and the CAPM does not hold.9 The

                                                                                                                         9  Gorton  et  al.  (2011)  draw  the  same  conclusion.  

15    

equity risk is not linearly related to leverage because externalities influence the return

on equity.10 However, they use the test results (the coefficients that Miles et al. (2012)

found of the CAPM formula 𝑅!"#$%&   = 𝑅! + (𝑎 + 𝑏  leverage)𝑅!) to estimate the weighted

average cost of capital (WACC) assuming that the cost of debt is fixed (risk-free rate)

while leverage halves. As mentioned earlier, the second proposition of the Modigliani-

Miller theorem states that the WACC is irrelevant to the capital structure, therefore the

WACC should not change. Miles et al. (2012) find an increase of the WACC and estimate

that the rise in WACC is only about 55% of what it would be in the absence of the

Modigliani-Miller theorem. In other words, there is a Modigliani-Miller effect and the

theorem holds for approximately 45% of the full extent.

2.3 Having More Equity Capital: Steady State

As the previous paragraphs have shown, the theoretical consequences of higher capital

requirements are ambiguous. This paragraph discusses the dynamics of having more

equity capital on the balance sheet in a new steady state. How can poorly capitalized

banks of today be compared with banks that have a low financial leverage in the new

equilibrium when all banks are better capitalized? Important to mention here is that, ex

ante, the new equilibrium is hard to predict. Today it is unknown how banks’ assets or

liabilities must be priced in the future. However, could bank capital be an attractive and

safe asset class with lower required returns on equity in the new equilibrium?

2.3.1 Cost of Capital Fallacy

The cost of capital fallacy, namely that equity capital is expensive and the return on

equity is fixed at a high level, creates a vicious circle. Once a bank has a capital surplus,

i.e. any “available” equity capital above the minimum capital requirement, it tends to

economize on capital to increase ROE by engaging in certain activities.11 According to

Boot (2013), “putting capital to use” increases the cost of this capital and may not create

value at all. Boot (2013) states that shareholders and other market participants foresee

that banks will economize on capital and thus raise their required return on equity. This

confirms the belief of banks that equity capital is expensive and that the best response

to higher capital requirements is to increase risk on the short-term to realize the

required return in the future. Stating that the cost of capital is also expensive in a new

equilibrium and that the ROE will be (or must be) fixed in a new equilibrium is

fundamentally flawed and misleading, since they do not adjust for risk (Admati et al.,

2011). As a caveat, this belief of banks suggests that the new equilibrium consists of

                                                                                                                         10  E.g.  corporate  tax  system,  government  guarantees,  capital  regulation  and  market  sentiment.    11  E.g.  proprietary  trading.  

16    

many more risky assets and activities, which would be the opposite of what higher

capital requirements are meant for.

Despite of this self-fulfilling belief (or vicious circle), Admati et al. (2011) advocate a

banking system with more equity capital. They argue that if the asset risk remains

constant, i.e. no significant change on the left side of the balance sheet, an increase in

capital requirements lowers the ROE due to less leverage. Note that this applies to the

new equilibrium (steady state at t=1). By assuming that asset risk remains constant,

Admati et al.’s (2011) statement is correct: more equity capital distributes risk.

Substantial more equity capital reduces the total per unit risk that is borne by the equity

holder, which should result in lower required rates of return. Thus, holding a bank share

could be an attractive and safe asset class in the new equilibrium when all banks have

more equity capital on their balance sheets and asset risk remains constant.

2.3.2 The Role of Subsidies on Debt in New Equilibrium

The impact of having more equity capital in a new equilibrium is that today’s subsidies

on debt can be less used in the future. On the one hand, more equity capital distributes

risk and lowers the required returns on equity. On the other hand, the implicit

government guarantee and tax shield play a smaller role because there is less debt on

the balance sheet. This could increase the cost of debt. As mentioned earlier, Gorton et

al.’s (2011) analysis indicates that when banks have less information-insensitive debt on

their balance sheet, the benefit from the convenience yield decreases in the new

equilibrium. This also applies to the benefits of the tax shield. The corporate tax system

gives a fiscal incentive to finance with debt, because interest payments on debt are tax

deductible. When there is less debt on the balance sheet, this fiscal advantage of debt

disappears. These reductions of debt-financing advantages increase the total cost of

funding, which according to Gorton et al. (2011) results in higher prices for a bank loan.

Chapters four and five elaborate the tax shield and implicit government guarantees in

more detail and propose some reforms that can be applicable to the new equilibrium and

transition phase.

Following the theory of Admati et al. (2011), Gorton et al.’s (2011) assumptions suffer

from a neglect of external costs and misaligned incentives. Gorton et al. (2011) and

Gorton (2010) argue that banks produce debt, which distinguishes banks from other

companies and makes them “special”. They also state that the economy needs debt and

it is socially desirable that banks produce liquid securities, e.g. securitization of individual

mortgages or short-term commercial paper. Although debt is a useful instrument for the

17    

economy, this observation does not imply that banks should be highly leveraged. Admati

et al. (2011) state that investors do not always need those liquid securities in the form

of short-term debt. Bank capital can be a safe and attractive asset class (in a new

equilibrium) for long-term investors that are now holding long-term and senior debt.

Admati et al. (2011) argue that the attractiveness of short-term debt is enhanced when

banks are better capitalized, while investors with longer time horizons hold more equity

capital.

Given the huge costs of the system’s breakdown in the 2007-2009 financial crisis,

Admati et al. (2011) see strong reasons to question the social value of much of this debt

creation that Gorton et al. (2011) advocate.12 The call for more equity capital and the

use of less debt suggest a long-term transition to a banking landscape that is much

different than that of today and hard to predict upfront.

2.4 Raising More Equity Capital: Transition Phase

The road leading to higher capital levels entails other sorts of issues, such as debt

overhang, information asymmetry and stigmatization problems (“new-issuance costs” or

flow costs). Kashyap, Stein and Hanson (2010) explain that there is a crucial distinction

to make when discussing costs of capital in relation to acquiring more equity capital.

First, if a poorly capitalized bank is trying to attract more equity capital from the market,

it could face debt overhang problems while better-capitalized banks do not. Second,

costs associated with information asymmetry also play a bigger role when the bank in

dispute is highly leveraged. Kashyap et al. (2010) state that the frictions of raising more

equity capital are more severe than the “ongoing costs” of holding more equity capital.

2.4.1 Information Asymmetry

An important contribution to the information asymmetry discussion is the flow-cost

theory, which is set up by Myers and Majluf (1984). They explain the difference between

more and less information available for respectively firm management and outside

investors. Assuming that management acts on behalf of existing shareholders, then an

equity issue will be taken as a negative signal, since management prefers to sell shares

when they think shares are overvalued. This is also known as the signaling effect (or

stigmatization) and share issues will tend to be associated with negative share-price

impacts. Because management knows that there is a negative impact of this

                                                                                                                         12  Admati  et  al.  (2011)  state  that  the  financial  crisis  is  due  to  high  leverage  ratios  and  that  if  the  equity  cushion  was  big  enough,  the  crisis  did  not  occur.  On  the  other  hand,  Gorton  et  al.  (2011)  argue  that  the  conversion  of  information-­‐insensitivity  debt  into  information-­‐sensitivity  debt  (e.g.  repo)  is  the  cause  of  the  financial  crisis  and  not  necessarily  the  level  of  debt  (leverage).  

18    

stigmatization, they will postpone or not propose an equity capital issuance. This

disturbs the leverage reduction during the transition phase.

If a bank faces higher capital requirements, it might not be raising new external equity

and instead prefers to shrink its assets and stop lending. Kashyap et al. (2010) conclude

that, in the sense of the Myers-Majluf model and empirical work they have surveyed,

new capital requirements should be phased-in sufficiently, in order to reduce the

information asymmetry problem and to give banks time to generate the necessary

additional capital largely out of retained earnings and maintain lending activities

normally.

On the other hand, Admati et al. (2011) argue that if the share issue decision is not

taken by the management, but required by the regulator, the negative signaling effect

can be neutralized. They refer to the Troubled Asset Relief Program (TARP) in 2009,

where banks didn’t have a choice whether to accept government intervention or not, and

the information asymmetry was not an issue. If new capital requirements are

accompanied by regulation mandating all banks to issue new shares at a pre-specified

scheme, the negative signaling effect would be removed, and banks have no reason to

reduce lending in order to meet the new capital requirements during the transition phase

(Admati et al., 2011). Also Admati et al. (2011) recommend regulators to postpone

dividend payments by banks for a period of time, and use the retained earnings to build

up bank capital. Again, if done under force of regulation, this will not lead to a negative

signaling effect on the health of any particular bank (Admati et al., 2011).

2.4.2 Debt Overhang

In addition to the information asymmetry problem, poorly capitalized banks face debt

overhang problems. Myers (1977) was the first to describe the problem of debt overhang.

For a firm with outstanding debt, equity capital issuance reduces leverage. Leverage

reduction of these firms benefits existing debt holders and providers of debt guarantees.

For each unit of equity capital that is added to the balance sheet, debt becomes safer

and a transfer of value takes place from shareholders to existing debt holders (“dilution”)

during the transition phase. This transfer of value leads to underinvestment; new (partial

equity financed) projects are not carried out, because dilution will occur (Myers, 1977).

In a paper about debt overhang in relation to banks, Admati, DeMarzo, Hellwig and

Pfleiderer (2012) state that shareholders do not want to reduce the leverage even if the

reduction would not change the total value of the bank. In some cases, new equity

19    

capital that is invested in good assets (loans with positive NPV) might increase the total

value of the bank. Due to debt overhang and the “addiction” to leverage new loans are

not provided (Admati et al., 2012). During financial crises, when the probability of

default is significantly higher, debt overhang problems partly explain the credit rationing.

Repayments of existing loans are used to strengthen the banks’ balance sheets.

Figure 2.1 shows three options that are possible to reduce leverage in response to higher

capital requirements:

Initial Balance Sheet Revised Balance Sheet with Increased Capital Requirements to 20%

New assets:

12.5 Equity:

22.5

Loans: 100

Equity: 10

Loans: 100

Equity: 20

Loans: 100 Deposits

and Debt:

90

Deposits

and Debt:

90

Deposits

and Debt:

80

Loans: 50

Equity: 10

Deposits and

Debt: 40

10% capital requirement 1) Asset Liquidation 2) Recapitalization 3) Asset Expansion

Figure 2.1: Alternative Responses to Increased Capital Requirements, source: Admati et al. (2012)

Assume the initial capital requirement is set at 10% and suppose that the bank has €100

worth of assets (loans). The bank is financed with €10 of equity capital and €90 of

deposits, debt and other liabilities. Now assume that capital requirements are raised to

20%. Following figure 2.1, the first option is asset liquidation, where the bank “delevers”

its balance sheet by liquidating €50 in assets and using the proceeds to reduce liabilities

from €90 to €40. Option two is a pure recapitalization, where issuing €10 of additional

equity capital and buying back €10 of debt satisfy the new capital requirement. The third

option is a balance sheet expansion. Raising equity by €12.5 and using the proceeds to

acquire new assets or provide new loans also satisfy the 20% capital requirement

(Admati et al., 2012).

Admati et al. (2012) analyze shareholders’ incentives to find out if shareholders have a

preferred option. They find that, from shareholders’ perspective, all three options are

equally undesirable because of the debt overhang problem (Admati et al. 2012). This is

important to know, because if the asset liquidation was preferred and the transition

period is accompanied by a significant number of assets sales, it could spark a fire sale

and have a destabilizing effect in the midst of financial crisis. In any case, more equity

20    

capital increases debt holders’ safety and value. Admati et al. (2012) emphasize that

these problems would be less significant when the banking system is better capitalized.

2.5 Conclusion

This chapter discussed important insights and concepts related to capital structure of

banks, banks’ cost of funding and higher capital requirements. In conclusion, banks are

“special” and have some unique dynamics on their balance sheets (e.g. deposits).

However, some important corporate finance insights are applicable to banks and should

be taken into account when discussing the capital structure of banks. The Modigliani-

Miller theorem holds partially and is independent of CAPM. The theorem is also useful to

expose and identify frictions and distortions. The impact of higher capital requirements

on the total cost of funding is negatively affected mainly by two externalities. First,

implicit and explicit government guarantees affect the banks’ cost of funding due to a

discount on the interest rates on debt: the convenience yield. Second, the tax shield is

also subsidizing debt and makes the total cost of funding cheaper. These problems, and

many more issues, play a minor role when the banking system is better capitalized;

hence higher capital requirements are necessary to reduce frictions and distortions.

The theoretical analysis of discussing and implementing higher capital requirements

must be segregated in two ways, namely having more equity capital and raising more

equity capital. The self-fulfilling beliefs of banks that having more equity capital is

expensive and the ROE is fixed are fundamentally flawed. Having more equity capital

reduces the required return on equity, since risk is distributed among more shareholders.

It is therefore misleading if banks engage in risky activities to remain their ROE constant.

In the new equilibrium (steady state, when all banks are better capitalized), bank capital

can be an attractive and safe asset class to hold in the portfolio, with a low risk profile

and a reduced required return on equity.

Raising more equity capital could transfer value from shareholders to existing debt

holders. This so-called debt overhang problem makes the decision to acquire more

equity capital difficult for banks’ shareholders and managers. Along with information

asymmetry problems, high leverage ratios are hard to breach by raising more equity

capital. The combination of these two problems demands that higher capital

requirements should be phased in gradually. However, these problems can be alleviated

if the regulator requires banks to postpone dividend payouts and to issue new equity

capital on short notice under force of that same regulator.

21    

3. Empirical Studies on Higher Capital Requirements

This chapter surveys five recently published, empirical-based papers and publications on

the effect of higher capital requirements on loan growth and bank lending spreads. In

chronological order of publication it discusses Kashyap, Stein and Hanson (2010), King

(2010), Angelini et al. (2011), Cosimano and Hakura (2011) and Santos and Elliot

(2012). The assumptions and results of these studies are compared with the theoretical

concepts and insights discussed in chapter two. These five studies empirically test the

long-run impact of higher capital requirements. Their results relate to a new, steady

state, equilibrium. Some of these papers mention the transition phase briefly by

explaining debt overhang and asymmetry information problems. These problems are not

involved in their empirical parts, with the exception of Santos and Elliott (2012). In some

cases, chapter two will be complemented with other insights and arguments. A summary

and conclusion form the end of the chapter.

3.1 Kashyap, Stein and Hanson (2010)

Kashyap, Stein and Hanson (2010) examine the impact of “substantially heightened”

capital requirements on large financial institutions, and on their customers. They begin

their empirical study by validating the Modigliani-Miller theorem. A large sample of banks

is used to test if the 𝛽!"#$%&   halves when the equity capital ratio is doubled (similar to

Miles et al. (2012)). The regression results are roughly in line with what is predicted

upfront. There is some empirical evidence that justifies the use of the Modigliani-Miller

theorem for further calibrations. Note that Kashyap et al. (2010) assume, for simplicity

matters, that debt is risk-free (𝛽!"#$), which implies the existence of a convenience yield.

Their baseline regression results are not corrected for the loss of subsidized debt when

the equity capital ratio is doubled. Thus, the Modigliani-Miller effect must be dampened.

The second conclusion Kashyap, Stein and Hanson (2010) draw is that if the minimum

capital ratio is raised by ten percentage points, the loan rates will increase by 25-45

basis points13 according to their methodology. This applies to the new equilibrium

(steady state). They qualify this as a minor change in loan rates and small in absolute

terms. The outcomes are only as good as the model that underlies them and the main

assumption of the model is the loss of the tax shield when debt is replaced with equity.

They assume the cost of long-term debt is 7% and the corporate tax rate is 35%. Thus a

ten percentage points increase of the capital ratio would raise the lending spread with 25

basis points (=10% x 7% x 35%). Kashyap et al. (2010) correct for the loss of                                                                                                                          13  100  basis  points  =  1  percent.  

22    

subsidized debt in an aggressive case (violation of the Modigliani-Miller theorem) and

find an increase of the lending spread by 45 basis points.

There is an incentive for banks to be highly leveraged, because of these benefits

provided by a convenience yield and the tax shield. Admati et al. (2011) explain that

when debt has indeed a tax advantage over equity, this assumption is correct, but

irrelevant to capital regulation. Both capital regulation and tax rules are matters of public

policy. Tax policy should aim at discouraging behavior that generates negative

externalities, such as increases in leverage ratios. High leverage ratios raise the

probability of bank failures and weaken the financial system. The probability of

government intervention, using public funds, is also increased (Admati et al., 2011).

The final conclusion of Kashyap, Stein and Hanson (2010) is that intense competition

drives the banks in the direction of high leverage. The most competitive advantage that

banks have is the ability to fund themselves cheaply (i.e. short-term debt or “repo”14).

Even the smallest increase in cost of funding relative to direct competitors can lead to

the loss of much business (Kashyap, Stein and Hanson, 2010). They also argue that

substantially heightened capital requirements will lead to greater banking activity within

the so-called “shadow banking” sector due to these competition forces. This

phenomenon is also known as regulatory arbitrage. Kashyap et al. (2010) find empirical

evidence that large banks in particular tend to hold less capital and are able to exploit

regulatory arbitrage. Admati et al. (2011) point out that most activities and entities in

the “shadow banking system” relied on commitments made by regulated entities, and

thus were within regulators’ reach. They believe it is unhelpful in the context of the

capital regulation discussion to refer to the “shadow banking system” like that. Capital

regulation is focused on reducing excessive leverage and regulators should be able to

assess the true leverage of banks. This includes banks’ contribution to the entities within

“shadow banking system” that are being used to hide leverage and exposures (Admati et

al., 2011). Obligations to the shadow banking system could be higher than expected.

3.2 King (2010)

The second paper, a BIS working paper by King (2010), outlines a methodology for

mapping the increases in capital and liquidity requirements proposed under Basel III to

bank lending spreads. He finds that a one-percentage point increase (steady state) in

the capital ratio can be recovered by increasing lending spreads by 15 basis points. This

is a bigger change in the lending spread compared to the figures Kashyap, Stein and                                                                                                                          14  For  the  role  of  repo  financing,  see  Gorton  (2010)  and  Gorton  and  Metrick  (2010).  

23    

Hanson (2010) estimated with their methodology. King’s (2010) most important

assumption is that the return on equity (ROE) and the cost of debt are unchanged when

more equity capital is acquired. He argues that theoretically both the cost of debt and

the cost of equity should decline as leverage decreases and the risk of default becomes

smaller, but it is not evident that these theories hold in practice (King, 2010). According

to King, this is due to implicit government guarantees on bank debt, which reduce the

risk of default, leading shareholders to expect a lower ROE. At the same time King (2010)

mentions the implicit subsidy on cost of deposits due to the deposit insurance schemes,

lowering the cost of wholesale funding compared to firms with similar leverage ratios. As

mentioned in chapter two, bankers argue that higher capital requirements will increase

funding costs, since indeed more equity capital will reduce banks’ ability to benefit from

these guarantees and subsidies. Following this reasoning, capital is indeed expensive.

Admati et al. (2011) argue that this is not a legitimate reason for regulators not to

propose new capital requirements. The existence of these subsidies cannot be neglected,

but that does not justify high leverage ratios. Admati et al. (2011) find it paradoxical

that the government subsidizes the leverage of banks at the same time that it

recognizes that this leverage is socially very costly and considers imposing higher capital

requirements to prevent the banks from taking advantage of this subsidy.

Admati et al. (2011) make a clear distinction between private costs and social costs,

which is important to do when empirically testing higher capital requirements. King

(2010) seems to neglect this. Similar to the case of the tax advantage of debt,

government guarantees on debt concern private costs of bank capital. Admati et al.

(2011) take into account the default risks borne by the taxpayer and the costs of these

risks to taxpayers as social costs. Once these costs are included, there is a strong case

for requiring banks to have more equity capital. Equity cushions are valuable, as they

reduce the likelihood and cost of the guarantees (Admati et al., 2011). Note that this

refers to the new equilibrium (steady state).

King (2010) holds the ROE and the cost of debt constant while calculating the effects of

new capital requirements. This is contrary of what should happen according to the

Modigliani-Miller theorem, as extensively stated in chapter two. Raising the amount of

capital should reduce risk per unit of capital and thus lower the ROE. King (2010)

mentions that it is possible to empirically identify an inverse relationship between bank

24    

capital ratios and historical ROEs, with lower returns for more highly capitalized banks.15

Because there is a lack of data on secondary market prices for bank debt, the empirical

relationship between bank capital ratios and the cost of wholesale funding is less clear

(King, 2010). Therefore, King (2010) argues that it is reasonable to assume that ROE and

cost of debt are unchanged despite new higher capital levels. This is false, since ROE does

not adjust for risk. King’s (2010) reasoning shows a misunderstanding of the way in

which risks must be taken into account when calculating the cost of funding. Referring to

chapter two, the required return on equity is higher than the required return on debt and

this difference reflects the greater riskiness of equity relative to debt. Reducing the

amount of capital (increasing leverage) has an effect on the riskiness of debt and equity

and, therefore, on the required expected return on equity.

Modigliani and Miller (1958) state that, with or without tax advantages and public

subsidies to debt and deposits, increasing the amount of equity simply re-distributes the

total risk that is borne by investors in the bank, the right side of the balance sheet. The

total risk of the bank is given by the risks that are inherent in the bank’s asset return,

the left side of the balance sheet (Admati et al., 2011). According to the Modigliani-Miller

theorem, changing the capital structure must affect the return on equity and cost of debt,

therefore King’s (2010) assumption cannot hold. King’s (2010) calculations and test

results are incomplete, since previous mentioned arguments are not taken into account.

3.3 Angelini et al. (2011)

The third paper, a NY Fed Staff Report by Angelini et al. (2011), assesses the long-term

economic impact of the new regulatory standards (the Basel III reform). In line with

Kashyap et al. (2010) and King (2010), this third study also examines the steady state

(new equilibrium). However, Angelini et al.’s (2011) method is a completely different

approach compared to Kashyap et al. (2010) and King (2010), which have studied the

new capital requirements at the level of banks’ balance sheets and used partial

equilibrium models.16 Angelini et al. (2011) address the impact of the new capital

requirements on economic performance and fluctuations. They also discuss the adaption

of countercyclical capital buffers on economic fluctuations. When the economy is

booming (shrinking), capital ratios should be increasing (decreasing). Angelini et al.

(2011) use different general equilibrium models to calculate output17, welfare18 and

consumption. The general equilibrium theory assumes that investments and savings are                                                                                                                          15  This  is  similar  to  the  Modigliani-­‐Miller  effect  mentioned  in  chapter  two.  16  Assuming  other  sectors  are  not  affected  due  to  the  change  in  the  banking  sector,  hence  ceteris  paribus.  17  Output  is  the  volatility  of  macroeconomic  variables.  18  The  welfare-­‐model  of  Van  den  Heuvel  (2008)  is  a  well-­‐known  example.  

25    

in equilibrium and equal, therefore savings are needed when capital investments

increase across different sectors. A conversion of savings into investments in the

financial sector changes the equilibrium of welfare, consumption and economic output for

all sectors. Angelini et al.’s (2011) focus is on the costs of the new regulation and how

these costs affect the behavior of supply and demand in the whole economy. A highly

stylized version of the new scenario (higher capital requirements, conversion of savings

into investments) is translated into model inputs and different variables. The results, or

the model output, are steady state values and volatility of key macroeconomic variables,

which determine the new general (macro) equilibrium (Angelini et al., 2011).

Angelini et al. (2011) derive three results about long-term economic performance,

fluctuations and countercyclical capital buffers. The first result is that a one-percentage

point increase in the capital ratio translates into a 0.09 percent output loss relative to

the level that would have prevailed in the absence of capital tightening. Their

interpretation of this figure is that the impact on long-term economic performance is

modest, which is in line with results obtained in similar studies19 (Angelini et al., 2011).

The second estimate is about the impact of higher capital requirements on economic

fluctuations. According to Angelini et al. (2011), higher capital requirements should

dampen output volatility (the magnitude of economic shocks). Their used models

estimate that a one-percentage point increase in the capital-to-asset ratio reduces the

standard deviation of output by 1.0 per cent, which they opine as a modest result.

Angelini et al. (2011) also find that a one per cent increase in capital raises the lending

spread with 13 basis points20. The final result of Angelini et al. (2011) is that a

countercyclical capital buffer could have a more sizeable dampening effect on output

volatility. The equity capital buffers that are accumulated in good times reduce the

downward impact of an economy in recession.

A modest loss of welfare, as Angelini et al. (2011) estimated, could suggest that

increasing capital requirements reduces the ability of banks to provide loans or hold

deposits, which can be consumed. Admati et al. (2011) claim that increasing capital

requirements do not have to lead to a decline of welfare. Figure 2.1 from chapter two

provides three options that are possible to reduce leverage. The third response, asset

expansion, gives a bank the opportunity to increase the equity capital ratio, while at the

same time providing new loans to the economy.

                                                                                                                         19  MAG  (2010b),  BCBS  (2010b).  20  King  (2010)  estimates  a  comparable  increase.  

26    

Initial Balance Sheet Revised Balance Sheet with Increased Capital Requirements to 20%

New assets:

12.5 Equity:

22.5

Loans: 100

Equity: 10

Loans: 100

Equity: 20

Loans: 100 Deposits

and Debt:

90

Deposits

and Debt:

90

Deposits

and Debt:

80

Loans: 50

Equity: 10

Deposits and

Debt: 40

10% capital requirement 1) Asset Liquidation 2) Recapitalization 3) Asset Expansion

Figure 3.1: Alternative Responses to Increased Capital Requirements, source: Admati et al. (2011)

Admati et al. (2011) argue that this example of a single bank is just as pertinent when

analyzing the banking sector as a whole or even the overall economy, like Angelini et al.

(2011) do with the general equilibrium theory. The assumptions made for most of the

models that are used by Angelini et al. (2011) exclude an adjustment to higher capital

requirements concerning the third option, according to Admati et al. (2011).

Theoretically, if all banks use the asset expansion option to satisfy the new capital

requirements, the whole economy would expand. Since savings and investments must

be equal in the general equilibrium model, it is not realistic that massive asset expansion

by banks is an obvious option.21 As an example, a combination of asset liquidation and

asset expansion, where some banks become smaller and other larger, financed with new

equity by a conversion of savings into investments is a more realistic option.

In the particular model of Van den Heuvel (2008), used by Angelini et al. (2011), banks

are financed only with equity and deposits, thus increased capital requirements are at

the expense of deposits, resulting in a welfare loss under the model’s assumption that

consumers derive utility from holding deposits (Admati et al., 2011). As option three

suggested, banks can satisfy higher capital requirements without reducing their deposit

base, therefore Admati et al. (2011) find it highly suspect if not meaningless to apply

this model to assess the welfare costs of capital requirements. However, Admati et al.

(2011) seem to forget that the new investments in bank capital must come from savings,

since they must be equal. In order to expand the assets, like option three, savings must

be converted into investments. In general equilibrium models this is seen as a loss of

consumption and welfare (Angelini et al., 2011).

                                                                                                                         21  A  conversion  of  deposits  into  equity  does  not  expand  the  balance  sheet  (both  liabilities  of  a  bank).  

27    

Concluding, Angelini et al. (2011) seem to neglect the social costs that arose from

misalignments and distortions underlying the system’s breakdown in the crisis. Angelini

et al. (2011) focuses on costs in terms of the loss of welfare and consumption. In 2008,

Van den Heuvel concluded that capital requirements were too high and he estimated that

one upper bound for the cost of a one-percentage point increase in capital requirements

is $1.8 billion per year. Given these facts, Admati et al. (2011) find it remarkable that

Van den Heuvel’s (2008) welfare-model is used by Angelini et al. (2011). The loss of

output, consumption or welfare due to higher capital requirements is significantly smaller

than the costs of the financial crisis (Admati et al., 2011). If more equity capital reduces

the costs of financial crises, than equity capital should be taken into account as a benefit.

3.4 Cosimano and Hakura (2011)

The fourth paper that is discussed, an IMF Working Paper by Cosimano and Hakura

(2011), investigates the impact of the new capital requirements of Basel III on bank

lending rates and loan growth (steady state, new equilibrium). The method used by

Cosimano and Hakura (2011) models three variables simultaneously; the generalized

method of moments (GMM). The first variable that Cosimano and Hakura (2011) regress

is the choice of capital, depending on the capital requirement, interest rate on deposits,

noninterest costs of loans and total assets (Cosimano and Hakura, 2011). The second

regression variable is the loan rate, which is dependent of the first variable plus interest

rate on deposits, costs of loans and economic activity. The last step they examine is the

elasticity of bank loans, for which they use the loan rate from the second regression. The

elasticity of bank loans indicates the effect of higher capital requirements and loan rates

on loan growth. Cosimano and Hakura (2011) assume that higher capital requirements

raise banks’ marginal cost of funding, which leads to higher lending rates. They also

assume that the ROE is fixed, which means that all costs of increasing capital are

reflected by a higher loan rate. This is a violation of the Modigliani-Miller theorem, as

mentioned earlier. The last assumption is that bank liabilities consist only of equity and

deposits (Cosimano and Hakura, 2011). Three different groupings of banks are (cross-

country) analyzed: 1) the 100 largest banks worldwide; 2) commercial banks or bank

holding companies (BHC’s) in advanced economies that experienced the 2007-2009 crisis;

and 3) commercial banks or BHC’s that did not experience the 2007-2009 crisis.

The first finding of Cosimano and Hakura (2011) is that a one percent increase in the

capital requirement (equity-to-asset ratio) raises the loan rate for the 100 largest banks

with 12 basis points. For the second group, banks that faced the 2007-2009 crisis, a one

percent increase is associated with a 9 basis points average increase in the loan rate.

28    

The banks that did not experience the 2007-2009 crisis have a 13 basis point average

increase. Cosimano and Hakura (2011) also find a 12 basis point increase in marginal

cost of equity relative to the marginal cost of deposits, which is evidence against the

Modigliani-Miller theorem. Chapter two explained that there is a Modigliani-Miller effect,

thus not all assumptions hold in their pure form. A higher level of equity would reduce

the riskiness of the bank equity such that the ROE declines. However, Cosimano and

Hakura (2011) refer to the government guarantees and subsidies as a possible source of

the higher cost of equity. Since there is less room for subsidized debt, as stated in

chapter two, total cost of funding becomes higher and thus raising equity capital is

expensive. It is therefore not surprising that they find these increases.

Cosimano and Hakura (2011) use the increases in loan rates to estimate the loan

demand or elasticity for the three groups and different countries. The 100 largest banks

estimations imply a reduction in the volume of loans by on average 1.3 percent in the

long run when Basel III is in force (1.3 percent increase of equity-to-asset ratio).

Cosimano and Hakura (2011) use 2007 data as baseline scenario, which is outdated (see

paragraph 2.5). For banks in countries that experienced the 2007-2009 crisis,

implementing Basel III would reduce loan growth with 4.6 percent on average and 14.8

percent for banks in countries that did not experience the 2007-2009 crisis (Cosimano

and Hakura, 2011). According to Cosimano and Hakura (2011), the wide variance in the

results of loan rate increases and loan demand decreases reflects the differences

between countries’ interest elasticity of loan demand and bank’s net cost of raising

equity.22

Admati et al. (2011) state that highly leveraged banks are generally subject to

distortions in their lending decisions, such as frictions23 associated with governance and

information. This may lead to worse lending decisions compared to a better-capitalized

bank. If shareholders and management of a highly leveraged bank work on the basis of

ROE, they have incentives to make excessively risky investments, especially when

governments guarantee debt. The upward potential, a high ROE, is intended for

shareholders and managers, while the downward potential is shifted to taxpayers.

Cosimano and Hakura (2011) show that higher capital requirements raise lending rates

and reduce loan growth. Admati et al. (2011) would see this as a social benefit, since

excessive lending is reduced. The reduce in loan growth is not a necessity as is shown in

                                                                                                                         22  Differences  in  cost  of  capital  are  due  to  different  tax  policies  and  ex-­‐  and/or  implicit  government  guarantees  on  debt  and  deposits  across  countries.  23  i.e.  agency  theory,  moral  hazard,  asymmetric  information,  debt  overhang.  

29    

figure 3.1, thus there should be no concern with any negative impact on the economy of

increased equity capital requirements (Admati et al., 2011). However, this argument is

questionable when a general equilibrium method from Angelini et al.’s (2011) paper is

used where savings and investments are assumed to be equal.

3.5 Santos and Elliott (2012)

As stated in the introduction of this thesis, the first reason for higher capital

requirements is to strengthen the resilience of banks and the banking sector (BCBS,

2010). The previous four studies showed that, with or without correct and justified

assumptions, higher capital requirements result in higher cost of funding and ultimately

higher loan rates in a new equilibrium. In addition to higher loan rates due to the loss of

tax advantage and government guarantees, Santos and Elliott (2012) compose three

extensions of methodologies used by Kashyap et al. (2010), King (2010), Angelini et al.

(2011) and Cosimano and Hakura (2011). These extensions lead to substantially lower

net economic costs and are more in line with arguments of Admati et al. (2011). Santos

and Elliott (2012) state that financial reform comes at a price and that higher capital

requirements do add operating costs for banks that result in higher loan rates. However,

Santos and Elliott (2012) estimate in their study that loan rate increases will likely be

significantly smaller compared to King (2010) and Angelini et al. (2011).24

The first extension of Santos and Elliott (2012) is that market forces demand banks to

have greater safety margins above the minimum capital requirement. They state that

simply comparing the new Basel capital requirements with the old misses the crucial

point that banks hold capital on top of the minimum requirements, as a result of their

own desire to operate safely and because of pressure from the markets and rating

agencies (Santos and Elliott, 2012). Therefore, Santos and Elliott (2012) use the end-

2010 levels as baseline for their estimates, which are higher than the Basel II capital

requirements. The distance between end-2010 levels and Basel III is smaller.

The second extension of Santos and Elliott (2012) assumes that banks will cut costs and

take other measures to reduce the effect on loan rates and remain competitive. This

accounts for an average reduction of 14 basis points on the lending rate (end-2010

                                                                                                                         24  In  the  IMF  Staff  Discussion  Note  by  Santos  and  Elliott  (2012),  the  estimates  are  mainly  compared  to  official  BIS,  IIF  and  OECD  studies.  

30    

levels versus Basel III capital requirements). Santos and Elliott (2012) mention eight

different bank responses to cost increases which they have included in their study.25

The final extension is more in line with Admati et al. (2011), namely investors will lower

their required rate of return on bank equity when the bank reduces its leverage and

improves safety (Santos and Elliott, 2012). Holding the ROE fixed at a high level is

misleading, as explained in chapter two. With these three extensions taken into account,

Santos and Elliott (2012) estimate that average loan rates increase by 28 basis points in

the United States, 17 basis point in Europe and 8 basis points in Japan in the long term

because of the new capital requirements. Santos and Elliott (2012) mention by

comparison that the smallest step by which central banks change the interest rate is 25

basis points, which has no dramatic effect on the economy. Note that Santos and Elliott

(2012) assume that during the transition phase many permanent cost-saving measures

are implemented.

3.6 Conclusion

This chapter analyzed five empirical-based studies that researched the impact of higher

capital requirements on bank funding costs. The assumptions made in these studies are

compared with the theory of chapter two. The first four studies mainly examined the cost

side of higher capital requirements in a new equilibrium, the steady state. Overall, some

benefits of higher capital requirements are mentioned, but generally not taken into

account in the empirical tests and calculations. King (2010), Angelini et al. (2011) and

Cosimano and Hakura (2012) find substantial increases of funding costs when higher

capital requirements are implemented. The figures are in the range of 12 to 15 basis

points for each percentage point that equity capital increases. However, under the

assumption of fixed and high-level ROE’s and neglecting social benefits of a better-

capitalized banking system, these conclusions are misleading, flawed and/or incomplete.

Admati et al. (2011) and Miles et al. (2012) advocate an empirical analysis of the impact

of higher capital requirements that considers not only costs, but also benefits of

increased equity capital ratios. Such an analysis requires a clear distinction between

costs and benefits to individual banks (private costs and benefits) and overall economic

or social costs and benefits. The empirical studies discussed in this chapter show that

private costs of banks may rise. The loss of subsidized debt on the balance sheet when

more equity capital is acquired cannot be ignored. On the other hand, private benefits

                                                                                                                         25  See  Santos  and  Elliott  (2012,  p.  8).  Many  costs  savings  can  be  realized  due  to  increased  safety  and  lower  volatility.  Admati  et  al.  (2011)  mention  these  effects  as  the  largest  benefit  of  increased  capital  requirements.    

31    

are difficult to quantify when the ROE is assumed to be fixed. Therefore the outcomes are

biased to the cost side of higher capital requirements. For a more balanced empirical test

of a new equilibrium, the following four factors should be included (Miles et al., (2012)):

1) Changes in required return on debt and equity as capital structure changes.

2) Changes in weighted average cost of capital (WACC) due to a different capital

structure and tax treatments of debt and equity. 26

3) A lower probability of banking problems as equity buffers rise (safety net).

4) Economic costs generated when banking sector problems arise (bailouts).

The empirical parts of the studies by Santos and Elliott (2012) and Miles et al. (2012) do

take into account these beneficial factors and find significantly lower increases of steady

state funding costs. Furthermore, Miles et al. (2012) even find that the optimal level of

bank capital relative to the proportion of GDP is between 8 and 10 percent of total bank

assets.27 This estimation is twice the capital requirement of Basel III.

The five papers discussed in this chapter generally agree that public policies, such as the

corporate tax system and implicit government guarantees, create subsidized debt. The

main conclusion that can be derived from the empirical studies is that most of the

increase of funding costs due to higher capital requirements is caused by the loss of this

subsidized debt on the balance sheet (steady state). This subsidized debt creates an

incentive for banks to prefer debt financing and to be highly leveraged. The transition

towards a better-capitalized banking system is easier when these negative externalities

are removed. Therefore, higher capital requirements should be complemented with

reforms of policies concerning the corporate tax system and implicit government

guarantees.

                                                                                                                         26  Note  that  all  analyzes  discussed  in  this  chapter  assume  that  asset  risk  and  return  are  unchanged  in  a  new  equilibrium  (for  simplicity  matters).  As  stated  in  chapter  two,  the  self-­‐fulfilling  belief  that  equity  capital  is  expensive  may  cause  risk-­‐seeking  bank  managers  during  the  transition  phase.  Santos  and  Elliott  (2012)  add  to  this  that  banks  will  cut  operational  costs  to  increase  equity  capital  buffers  and  remain  competitive.    27  The  capital  must  be  explicitly  loss  absorbing.  

32    

PART II Tax Shield, Government Guarantee and Policy Reforms 4. Tax Shield on Debt

The first part of this thesis, chapters two and three, eventually showed that tax policies

and government guarantees are serious distortions in the discussion about higher capital

requirements and bank capital structure. The second part, chapters four and five,

respectively scrutinizes the corporate tax policy and implicit government guarantee. This

chapter investigates, as an example, the Dutch situation of how the tax shield on debt

works and what the relative size is. The theory and empirics discussed in part I clarify

that high-leveraged banks benefit from the tax shield and that deleveraging creates a

loss of tax shield in a new equilibrium. Since interest payments on debt are fiscally

deductible, banks have an (implicit) incentive to prefer debt when financing new loans.

For example, if a bank pays 7% interest rate on their debt, it actually costs 7% x (1 –

corporate tax rate). This fiscal benefit, the tax shield, is part of many funding cost

calculations. Banks’ earnings are significantly higher due to lower tax payments. This

negative externality encourages banks to be more leveraged. Therefore, this chapter

ends with a possible reform of the corporate tax policy.

4.1 The Methodology of Debt Tax Shield Calculation

In a straightforward cost of capital or funding cost calculation the interest rate on debt,

determined by 𝑟!"#$, is corrected for the corporate tax rate, 𝑇!:

𝑊𝐴𝐶𝐶 =   !!!!

×𝑟!"#$%& +!

!!!×𝑟!"#$× 1 − 𝑇! (5)28

An annual report of a bank does not provide a calculation for the size and value of its tax

shield. A very simplified example shows the methodology how it is accounted. Assume

bank A having €100 interest income (e.g. loans and mortgages) and €72 interest

expenses (e.g. debt and deposits). Also assume that bank B is a 100%-equity financed

bank (no debt or deposits, therefore no interest expenses) with similar assets providing

€100 interest income. The profit and loss accounts of both banks state the following:

Table 4.1: Profit and Loss Account, Tax Shield, source: author compilation

                                                                                                                         28  Berk  and  DeMarzo  (2008)  Corporate  Finance,  Pearson  Education  Ltd.  

Bank A Bank B

Interest income €100 Interest income €100

Interest expenses -72 Interest expenses 0

Profit 28 Profit 100

Taxes (tax rate 25%) 7 Taxes (tax rate 25%) 25

Net profit €21 Net profit €75

33    

The tax shield of bank A is interest expenses x 25% = €18, which is the difference

between the tax expenses of bank A and B (€25 - €7 = €18). The key insight of this

example is that investors of the leveraged bank, debt and equity holders of bank A,

receive €93 (€72 + €21), while those of bank B only obtain €75. Note that the asset-

sides of the balance sheets of banks A and B are 100% comparable. The difference of

income that investors of bank A and B obtain, is the tax shield of €18 (€93 - €75). Thus,

investors of a (leveraged) bank can capture more income if leverage increases, while the

government misses tax revenues.29 Hence, leverage is implicitly subsidized.

Many Western countries allow tax deductions for expenses made to generate revenue,

mostly by taxing gross profits. As mentioned in chapter two, banks are “special” and use

debt to provide liquid securities to the economy, such as deposits and short-term debt.

This debt includes many interest expenses, whereby tax deductibility plays a bigger role

for banks compared to non-financial firms. Admati et al. (2011) state that the current

tax shield on debt induces a distortion in the allocation of public funds between firms

that can borrow extensively (e.g. banks) and firms that use more equity (non-financial

companies). Given the “special” role that banks have and given the high levels of bank

debt, the impact of the tax shield grew significantly to an undesirable extent.

4.2 The Size of the Dutch Bank Tax Shield

As an example and illustration of the distortion, this paragraph quantifies the total size of

the Dutch tax shield on bank debt over the period 1999-2012 for the largest three banks

ING, ABN AMRO and Rabobank. The main factors that determine the relative size of the tax

shield are leverage, interest rates on debt and corporate tax rates. 30 31 These three

factors are all positively correlated with the size of the tax shield. The first two factors

are applicable to individual banks and can be determined by banks themselves. Policy

makers set the third factor, the corporate tax rate. The corporate tax rate in the

Netherlands has declined from 35% in 1999 to 25% since the beginning of 2011.

1999-2001 2002-2004 2005 2006 2007-2010 2011-…

35% 34,5% 31,5% 29,6% 25,5% 25%

Table 4.2: Corporate Tax Rate in the Netherlands 1999-2012, source: Dutch Ministry of Finance

                                                                                                                         29  Note  that  from  a  government’s  perspective  the  tax  shield  (debt  subsidy)  is  not  an  actual  expense.        30  Examples  of  interest  rates  are  LIBOR,  EURIBOR,  ECB  interest  rates,  Federal  Funds  Rate  and  deposit  rates.  31  The  last  part  of  equation  (5)   !

!!!×𝑟!"#$× 1 − 𝑇!  shows  that  leverage   !

!!!,  total  assets  (𝐷 + 𝐸),  interest  

rates  𝑟!"#$  and  corporate  tax  rates  1 − 𝑇!  are  correlated  with  the  tax  shield  on  debt.    

34    

The leverage of ING, ABN AMRO and Rabobank is quite different over the period 1999-2012.

While Rabobank is most stable (total assets between 17,1 and 20,9 times their equity),

ING and ABN AMRO have chosen relatively more debt to finance their assets. ING’s balance

sheet only contained 2,05% of loss-absorbing equity capital at the end of 2008. Table

4.3 and figure 4.1 show all ratios from 1999 to 2012:

Table 4.3: Leverage (Equity Multipliers) 1999-2012 of Three Largest Dutch Banks, source: annual

reports and author calculation

Figure 4.1: Leverage (Equity Multipliers) 1999-2012 of Three Largest Dutch Banks, source: annual

reports and author calculation (appendix 1)

0  

10  

20  

30  

40  

50  

60  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

Leverage  (E

quity

 Mul1p

lier)  

ABN  AMRO  ING  Rabobank  

1999 2000 2001 2002 2003 2004 2005

ING 13,6x 22,4x 27,6x 32,0x 31,3x 29,8x 30,1x ABN AMRO 27,0x 30,5x 36,5x 37,3x 33,4x 43,9x 36,4x Rabobank 18,7x 19,0x 19,7x 17,6x 17,1x 19,2x 20,9x

(Cont…) 2006 2007 2008 2009 2010 2011 2012 ING 29,7x 33,2x 48,7x 29,9x 29,0x 28,0x 21,5x ABN AMRO 38,1x 33,3x 38,9x 24,8x 31,1x 35,4x 28,1x

Rabobank 20,3x 19,9x 20,4x 19,2x 18,9x 20,4x 20,9x

35    

Some key interest rates that are applicable or relevant to debt financing are given in

figure 4.2. The EURIBOR / LIBOR and ECB deposit rate are most frequently used in banking

and have peaks in 2000-2001 and 2007-2008, during bull markets.

Figure 4.2: Key Interest Rates 1999-2012, source: DNB and author compilation (appendix 2) The combination of these three factors, respectively the corporate tax rate, leverage and

interest rates, displays that the tax shields are at their largest in 2000-2001 and 2007-

2008. As leverage and interest rates increases, so does the tax shield. The extent to

which the relative size of the tax shield increases when leverage increases depends on

the type of debt instruments that are issued by the individual bank and in what volume.

Some debt securities are more volatile (e.g. when linked to the EURIBOR or LIBOR), while

deposits have a more stable interest rate. ING stands out in 2008 with an all time high

leverage, while at the same time interest rates are at their highest points.

Figures 4.3 and 4.4 (see next page) indicate that ING indeed deviates from ABN AMRO and

Rabobank. The absolute sizes of the tax shields on debt of ING, ABN AMRO and Rabobank

show similar movements up to 2004, with ABN AMRO’s peak of €9.7 billion at the turn of

the century (figure 4.3). From 2004 to 2008 the size of ING’s tax shield grows rapidly to

a maximum of €22.2 billion. This number indicates the impact and magnitude of the

subsidized debt distortion. The tax shields of ABN AMRO and Rabobank decline as a portion

of total assets, contrary to ING’s tax shield prior to the financial crisis (see figure 4.4).

This is not due to external factors, but inherent to ING’s strategy and management

decisions. Why ING differs so much from the other banks is not certain, based on these

figures. It could be that large quantities of ING’s new debt are linked to interest rates

that rise sharply, like EURIBOR or LIBOR. The growth of ING Direct and its exposure to US

mortgages is also a possible explanation for this striking deviation.

0,00%  

1,00%  

2,00%  

3,00%  

4,00%  

5,00%  

6,00%  

7,00%  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

Interest  Rate  

Eurozone  Bonds  10-­‐y  

Avg.  3-­‐month    Saving  Deposit  Rate    EURIBOR  12-­‐month  

LIBOR  ($)  12-­‐month  

ECB  Deposit  Rate  

36    

Figure 4.3: Tax Shield on Debt (In Millions), source: annual reports, Dutch Ministry of Finance and

author calculation (appendix 3)

Figure 4.4: Tax Shield as a Percentage of Total Assets, source: annual reports, Dutch Ministry of

Finance and author calculation (appendix 3)

 0    

 5.000    

 10.000    

 15.000    

 20.000    

 25.000    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ING  

Rabobank  

ABN  AMRO  

0,00%  

0,20%  

0,40%  

0,60%  

0,80%  

1,00%  

1,20%  

1,40%  

1,60%  

1,80%  

2,00%  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ING  

ABN  AMRO  

Rabobank  

37    

Banks are not solely relying on interest income and expenses.  They also have significant

fee revenues (or “commission”) that are part of their profits. Depending on which

strategy they choose, banks mutually differ in the relationship between interest income

and fee income. This is a fourth factor that influences the tax shield. Figure 4.5

expresses the contribution of fee income as a percentage of the combined revenues.

Figure 4.5: Fee Income as a Percentage of Total Interest and Fee Income, source: annual reports

and author compilation (appendix 4)

This figure and the graphs of appendix 4 show that ABN AMRO was focusing more on fee

income compared to their most direct competitors. It also demonstrates that fee income

is of less importance after the financial crisis of 2007-2009, presumably because

investment-banking activities are reduced and these banks “return to basics”. This

increases the relative weight of the tax shield and the distortion of subsidized debt. As is

clear from the preceding figures, the tax shield on debt is large and creates a significant

distortion in favoring debt finance over equity finance in the Netherlands. Debt financing

is made cheap. This distortion can be taken away by reforming the corporate tax system

and establish equal treatment of debt and equity.

4.3 The Future of the Tax Shield

Higher capital requirements reduce the ability to benefit from the tax shield on debt,

because less leverage results in a smaller tax shield. Increasing capital requirements

does not take away the distortion of tax advantages that subsidize leverage. According

to Admati et al. (2011), tax policies should discourage behavior that generates negative

externalities (high leverage and high risk of failure) and encourage behavior that

generates positive externalities (deleveraging and lower risk of failure). As paragraphs

4.1 and 4.2 pointed out, the tax shield on debt significantly lowers the tax expenses if

banks are more leveraged. Investors of high-leveraged banks capture more income

0%  5%  10%  15%  20%  25%  30%  35%  40%  45%  50%  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ABN  AMRO  ING  Rabobank  

38    

compared to their better-capitalized competitors, due to the tax shield on debt. A reform

of the tax shield should be designed in such a way that the fiscal incentive shifts to

equity financing, while total tax expenses remain more or less unchanged. Therefore,

phasing out the tax shield on debt could be done stepwise and during a long and orderly

transition period.32

The most simple and obvious solution is to remove the tax deductibility of interest

payments on debt, whereby tax is paid over gross interest income instead of net interest

income. Debt and equity are now treated equally from a fiscal perspective and the fiscal

incentive to finance the activities with debt is removed. New investments, e.g. loans or

acquisitions, no longer have a tax benefit on debt (or equity) and high leverage ratios

aren’t fiscally favored. However, this means that banks have to pay enormous amounts

of tax, namely several times their profit. This will increase the price of a loan (correction

for a higher tax base) and give an unbalanced incentive to focus more on fee income. On

the other hand, a full tax exemption on interest income means de facto that banks do

not have to pay taxes. The current banking tax could be expanded to correct for this, but

that is highly sensitive for subjective measurements.

A less cumbersome solution is to set a maximum on the tax deductibility of interest

expenses in order to remove the negative externality (incentive to prefer debt financing

and high leverage). At the same time, the operating profit that banks obtain through net

interest income could be less taxed, either by exemption or a lower corporate tax rate

for that specific profit.33 The challenge is to hold the reduction of the deductibility similar

to the exemption in absolute numbers, so that total tax payments remain about equal.

Such a reform will remove the incentive for banks to finance their activities with debt,

since it is significantly less subsidized. It is important to notice that such radical tax

reforms will only have a chance of success if they are set internationally in order to

establish a level playing field. Finally, if the current tax policies will not be reformed,

banks will benefit less from the tax shield on debt when higher capital requirements

(Basel III) are implemented. On the one hand, public policy makers require banks to

reduce leverage, while on the other hand they conserve tax policies that incentivize

excessive leverage. This is quite paradoxical and inconsistent. Thus, tax policy reforms

are of great importance.

                                                                                                                         32  Van  Dijkhuizen  (2012)  proposed  similar  reforms  for  mortgage  interest  deduction  and  income  taxes  in  the  Netherlands.  33  The  tax  rules  that  apply  to  fee  income  are  excluded  of  these  reforms,  since  the  fiscal  treatment  of  fee  income  does  not  create  negative  externalities  that  encourage  banks  to  be  highly  leveraged.  

39    

5. Government Guarantees and Recapitalization

This chapter analyzes implicit government guarantees, the second large distortion that

affects the banks’ cost of debt and equity. Large and complex banks are labeled as

Systemically Important Financial Institutions (SIFI’s), too-big-to-fail (TBTF) or too-

important-to-fail (TITF). Their debt contains a convenience yield, as explained in chapter

two. Investors in SIFI’s know that bankruptcy will not occur, since governments

(implicitly) guarantee that they will step in to prevent this. Therefore, debt investors are

willing to accept a lower required rate of return. Hence, the cost of debt does not fully

reflect the inefficiencies of excessive leverage caused by implicit government guarantees.

If this debt is replaced by equity capital, due to higher capital requirements, banks will

lose a share of this subsidy.

This chapter begins with explaining how implicit government guarantees affect the

banking system. The second paragraph shows, as an example, calculations of the size of

the implicit government guarantee and the total implicit subsidy in the Netherlands. This

chapter concludes with a possible solution that lowers the distortive effects of the implicit

government guarantee.

5.1 Impact and Consequences of Government Guarantees

The recapitalizations of many banks all over the world in 2008 and 2009 provide

examples of implicit government guarantees that became explicit. Governments had to

bail out large, complex and high-leveraged financial institutions to avoid a breakdown of

the financial system. As stated in chapter two, there is an important distinction between

raising more equity capital and having more equity capital. High-leveraged banks find it

difficult to increase the equity capital ratio due to debt overhang problems and

information asymmetry. As to debt overhang problems, for each unit of capital that is

acquired, debt holders’ risk is reduced. The formal characteristics of debt do not change,

but since risk is distributed among more shareholders, this debt becomes safer and thus

more valuable. This transfer of value creates an incentive for managers and shareholders

to maintain excessive leverage and to postpone equity capital issuances to prevent

dilution.

This debt overhang problem is exacerbated by the implicit government guarantee. In the

absence of the implicit government guarantee, debt holders have a disciplining effect on

management. They monitor the company and its management’s strategic choices to

ensure that the company will pay back the debt instead of going bankrupt due to

40    

excessive risk-taking. Since debt holders know that the government will intervene to

prevent a potential default, their market discipline reduces. This may result in a

preference for high leverage and excessive risk-taking incentives by banks. Once banks

are better capitalized under pressure of higher capital requirements the debt overhang

problem is solved or at least plays a minor role depending on the amount of new equity

capital. If there is sufficient equity capital in the steady state, banks can internalize

losses and depreciations using their own buffer. In this case, implicit government

guarantees are less important and the likelihood of a bailout is significantly reduced.

There are more distortive effects of implicit government guarantees. For example, the

convenience yield on debt is only obtainable by SIFI’s. Small banks do not have the

implicit guarantee and pay a relatively higher required return on debt. Therefore, their

funding costs are higher and competitiveness is reduced. This shifts business to large

banks and enhances the too-big-to-fail problem. Noss and Sowerbutts (2012) state that

the implicit government guarantee also attracts more resources from other sectors of the

economy to the financial sector. Thus, the guaranteed banking sector as a whole has a

competitive advantage over those sectors that are not or less guaranteed.

Besides implicit government guarantees, there are explicit guarantees. The most explicit

form of a government guarantee is the deposit insurance, which is partially financed by

the sector itself. The Federal Deposit Insurance Corporation (FDIC) and the Dutch Deposit

Guarantee Scheme (DGS) are two examples of explicit guarantees, which work as an

insurance for deposit holders. This system protects small deposits holders from losing

their money due to insolvency of the bank. The insurance is also established to prevent

bank runs. For an extensive discussion of deposit insurance, see Diamond and Dybvig

(1983, 2000) and Pennacchi (2009).

5.2 The Size of the Dutch Government Guarantee

This paragraph estimates the size of the implicit government guarantee of the three

largest Dutch banks over the period 1999-2012, similar to the tax shield calculation

made in paragraph 4.2. The impact of the implicit government guarantee on bank

funding costs is indirectly observable. Therefore, estimating the total size of the implicit

government guarantee is subject to a degree of judgment and some severe assumptions.

According to Gorton et al. (2011), bank debt contains a convenience yield. One method

to calculate the size of the implicit government guarantee is to multiply all outstanding,

interest rate-sensitive debt (minus deposits, which have an explicit guarantee) with this

convenience yield. The credit spread between bonds issued by small (non-SIFI) and large

41    

(SIFI) financial institutions is another example of a convenience yield. Small banks do not

enjoy the benefits of the government guarantee and thus issue bonds with a higher

credit spread (or risk premium) compared to large banks. This method is based on the

size of banks, where large banks have a government guarantee and a lower probability

of default. Gorton et al. (2011) use an average convenience yield of 72 basis points that

is found by Krishnamurthy and Vissing-Jorgensen (2010) over the period 1926-2008.

Baker and McArthur (2009) find a lower funding cost advantage using the size-based

method, ranging from 9 to 49 basis points for US banks.

A second method, used in this paragraph, is based on credit ratings provided by

Standard and Poor’s, Moody’s or Fitch. These credit rating agencies often issue two

ratings: the “supported” credit rating and the “stand-alone” credit rating. The first rating

reflects the actual costs of funding that are observed in the market. This is the normal

rating that is used by the market and reported in annual reports. The second rating is

based on an estimate of funding costs that banks would pay without the government

guarantee, in a stand-alone situation (Noss and Sowerbutts, 2012). The distance

(number of credit rating steps or “notches”) between these two ratings is dependent on

macroeconomic events and the likelihood of government support. Figure 5.1 shows how

Moody’s estimated this over the period 1999-2012:

Figure 5.1: Notches Between “Stand-Alone” and “Supported” Credit Ratings, source: Moody’s

Moody’s (2011) state that impact of the implicit government guarantee is equal to zero

notches from 2002 to 2006. This assumption does not neglect the existence of an

implicit government guarantee, but it implies that the market estimates the likelihood of

government intervention to be small, whereby the market is indifferent between

supported and non-supported banks. The difference between stand-alone and supported

credit ratings is assumed to be two notches since the beginning of the financial crisis in

0  

0,5  

1  

1,5  

2  

2,5  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

Notches  

42    

2007.34 Recently, Moody’s rescaled it to one notch in 2012 for US banks and two notches

for Dutch banks (Moody’s, 2012).

The credit spread between the stand-alone and supported rating (notches) captures the

margin that is needed to calculate the size of the implicit government guarantee. One

way to obtain the margin is to use the average credit spread between corporate bonds

and government bonds.35 All outstanding, interest rate-sensitive debt (minus deposits)

multiplied by this margin gives an estimation of the total size of the implicit government

guarantee. Assuming that all outstanding, interest rate-sensitive bank debt is affected by

the implicit government guarantee could be considered as a broad approach. A more

conservative approach is to use “issued bonds” only. Table 5.1 shows the total size of

the implicit government guarantee over the period 1999-2012, using Moody’s credit

ratings and the method described above. Table 5.2 shows the lower, conservative

variant.

1999 2000 2001 2002 2003 2004 2005

ING 214 217 191 0 0 0 0 ABN AMRO 148 164 323 0 0 0 0 Rabobank 69 356 208 0 0 0 0

(Cont…) 2006 2007 2008 2009 2010 2011 2012 ING 0 900 938 3.735 2.892 12.474 7.884 ABN AMRO 0 797 705 2.106 794 4.150 3.435 Rabobank 0 146 584 1.274 383 512 573 Table 5.1: Implicit Subsidy High (In Millions) 1999-2012, sources: annual reports, Moody’s,

Bloomberg and author calculations (appendix 5)

1999 2000 2001 2002 2003 2004 2005

ING 46 33 30 0 0 0 0 ABN AMRO 38 42 87 0 0 0 0 Rabobank 22 100 70 0 0 0 0

(Cont…) 2006 2007 2008 2009 2010 2011 2012 ING 0 80 116 684 570 2.294 1.721 ABN AMRO 0 210 178 794 442 2.162 1.965 Rabobank 0 71 285 756 236 299 335 Table 5.2: Implicit Subsidy Low (In Millions) 1999-2012, sources: annual reports, Moody’s,

Bloomberg and author calculations (appendix 5)

                                                                                                                         34  Ueda  and  Weder  di  Mauro  (2012)  assume  three  to  four  notches  based  on  Fitch  credit  ratings  35  For  example,  the  spread  between  a  US  government  bond  and  an  Aaa-­‐rated  bond  issued  by  a  US  bank  is  31  basis  points  in  1999.  If  the  spread  of  a  Baa1-­‐rated  bond  issued  by  a  bank  is  143  basis  points,  then  the  margin  between  the  Aaa-­‐rate  and  Baa1-­‐rate  is  112  basis  points  in  1999  (see  appendix  5,  source:  Bloomberg).    

43    

The results of the calculations show that the implicit subsidy of ING and ABN AMRO

increased substantially after the financial crisis, while Rabobank demonstrates a more

stable impact of the implicit subsidy. As stated in chapter four, the leverage ratios of ING

and ABN AMRO are higher than Rabobank’s, which have an influence on the credit ratings

of these banks. The increase of the implicit subsidy is caused by reduced confidence that

markets have in banks. This significantly lowers the credit ratings during and after the

financial crisis. Second, the number of notches between “stand-alone” and “supported”

has increased, whereby the margin and spreads grow exponentially. Rabobank is the

least leveraged bank and has a relative small and strong balance sheet. Due to

Rabobank’s relatively low leverage and high credit ratings, the margins between

“supported” and “stand-alone” are smaller. Hence, the implicit subsidy of Rabobank is

the smallest of these three banks. Still, the most conservative approach of this method

yields an annual average implicit subsidy of €330 million over the past six years for

Rabobank. In other words, implicit government guarantees ensure that not all

inefficiencies of high leverage are reflected in the costs of bank debt funding. In absence

of the government guarantee Rabobank would have paid €330 million per year more

interest on their issued bonds, which is approximately 15% of their total annual profit.

As stated before, these calculations are subject to a degree of judgment. The credit

ratings and the number of notches between “supported” and “stand-alone” credit ratings

are based on Moody’s interpretation.36 Applying the same credit spread or margin to all

different types of debt instruments on the balance sheet is also objectionable. Therefore,

the conservative approach (issued bonds only) is included and considered to be more

accurate. Other authors using the credit rating-based method find different margins. For

example, Ueda and Weder di Mauro (2012) find a margin of 60 to 80 basis points over

the period 2007-2009 for US banks. Using the same method and different credit spreads,

Van Tilburg (2012) estimated that the implicit subsidy of the three largest Dutch banks

in 2011 is between €3,8 and €11,4 billion. It is clear that the exact benefit of implicit

government guarantees is hard to quantify. Despite of the bandwidth of the estimations,

these methods indicate that the problem has a significant impact on bank funding costs.

They also expose differences between better-capitalized and high-leveraged banks.

                                                                                                                         36  Note  that  up  until  the  financial  crisis  credit  rating  agencies  made  mistakes  in  their  judgments.  This  calculation  is  based  on  credit  ratings,  thus  the  results  are  not  entirely  objective.  Most  important  is  that  the  differences  between  ING,  ABN  AMRO  and  Rabobank  can  be  made,  since  all  three  banks  are  subjected  to  the  same  method.  

44    

5.3 Recapitalization of the Banking System

Paragraphs 4.2 and 5.2 show that the distortive effects of the tax shield and implicit

government guarantees are significantly reduced when a bank is better capitalized. This

also implies that high-leveraged banks exploit the implicit subsidy relatively more. Their

incentive to acquire more equity capital is negatively influenced by these distortions,

which enlarge the debt overhang problem (transition phase). Besides the loss of

subsidized debt, banks do not have importance in raising equity capital because dilution

may occur. Therefore, it seems inevitable that governments and/or regulators should

intervene to recapitalize the banking system.37 Higher capital requirements are one form

of government intervention, but can be complemented with other solutions. For example,

Admati et al. (2011) advocate a regulatory rule that forces all large banks to issue equity

capital according to a fixed schedule. This may help to reach the higher capital

requirement faster and to avert the stigmatization of an equity issuance, but it does not

mitigate the other problem of the transition phase, debt overhang. Admati et al. (2011)

also advocate that dividend payments should be suspended during the transition phase,

because retained earnings should be used to increase the equity cushion. Although these

proposals create better-capitalized banks more efficiently, the existing debt holders

benefit from this situation.

The challenge is to find an instrument that deals with the lack of incentives of high-

leveraged banks to acquire more equity capital timely, hence solves the debt overhang

problem. Calomiris and Herring (2011) describe such an instrument and propose a

contingent convertible (CoCo) capital requirement in addition to higher capital

requirements. Contingent convertible capital is a debt instrument that converts to equity

when the equity capital ratio falls below a certain threshold. This mandatory conversion

of debt to equity is a direct form of recapitalization for which the term “bail-in” is used

frequently. The automatic conversion ensures that banks can avoid the debt overhang

problem described earlier. Note that the risk and probability of conversion lead to a

higher required rate of interest by investors in CoCos compared to normal debt (based

on theoretical insights of chapter two).

Calomiris and Herring (2011) state that if banks have a choice between issuing equity

capital or CoCos, they should prefer CoCos. The dilutive effect of a forced equity capital

issuance immediately takes place. As to CoCos, dilution only occurs when debt is

converted into equity capital depending on the conversion rate. Calomiris and Herring

                                                                                                                         37  See  e.g.  Scharfstein  and  Coates  (2009),  Admati  et  al.  (2011,  2012  and  2013),  and  Philippon  and  Schnabl  (2012)  for  a  more  detailed  discussion  about  the  need  of  government  intervention.    

45    

(2011) state that the primary aim of a CoCo should be “to incentivize the voluntary, pre-

emptive, and timely issuance of equity into the market as a means of avoiding highly

dilutive CoCo conversion”. CoCos also facilitate bail-ins and signal bank risk, but the

encouragement of timely equity capital issuances is far more important. According to

Calomiris and Herring (2011), the design of a CoCo requirement should include at least:

a large size of CoCos, a credible and observable moment of conversion (the trigger) and

a conversion rate that is dilutive of existing shareholders. If a bank faces significant

losses that will “trigger” the automatic conversion of debt into equity capital soon, it will

avoid this by issuing new equity capital timely. The large amount of CoCos being

converted, combined with the dilutive conversion rate, must be an unattractive option

compared to issuing new equity capital timely. Existing shareholders and management

anticipate the possibility of a conversion and will have strong incentives to be adequately

capitalized and have accurate risk management (Calomiris and Herring, 2011).

Lastly, the role of implicit government guarantees is significantly reduced if such a CoCo

requirement is implemented. As explained in paragraph 5.1, debt holders have a less

disciplining effect on banks’ management and strategic decisions in the presence of

implicit government guarantees and in the knowledge that the bank is too-big-to-fail.

CoCo holders are keen to prevent conversion, as well as existing shareholders that fear

heavy dilution, therefore the disciplining effect will return. Second, if a bank faces

significant losses of equity and is not able to issue new capital, the conversion of CoCos

reduces the likelihood and magnitude of a government bailout. The higher the CoCo

requirement, the smaller is the role of the implicit government guarantee.

46    

6. Summary and Conclusion

This thesis answers the question if higher capital requirements increase total bank

funding costs. The first part provides an analysis of theoretical and empirical studies. The

starting point of this analysis is the Modigliani-Miller theorem that is explained in chapter

two. Discussing the impact of higher capital requirements on funding costs and capital

structures requires a clear distinction between the steady state and the transition phase,

since their dynamics are different. Following the Modigliani-Miller theorem, higher capital

requirements will not change (steady state) total funding costs under some severe

assumptions. From the discussion of theoretical papers can be concluded that this

theorem does not hold on banking in its pure form, although it is useful to identify

frictions and distortions. The theoretical analysis emerges two distortions, namely the

tax shield and implicit government guarantee. These public policies have a significant

damping effect on the funding costs of bank debt and implicitly subsidize debt.

Given the fact that in the current situation bank debt is subsidized, then replacing this

debt with more equity capital reduces the ability of banks to exploit the implicit subsidies

in the new equilibrium. Second, the self-fulfilling beliefs of banks that capital is

expensive and the ROE is fixed in a new steady state are fallacious and incorrect.

Theoretical insights show that having more equity capital distributes risk and must lower

the required return on equity. Holding a bank share in the new equilibrium could be an

attractive asset class if risk is distributed and all banks are better capitalized so that they

internalize losses and depreciations. Meantime, raising more equity capital entails debt

overhang and information asymmetry problems, especially if banks are poorly capitalized.

The dilutive effect of an equity capital issuance creates incentives for bank management

and existing shareholders to resist reductions in leverage that make existing debt safer.

Banks have no importance in higher equity capital ratios, due to this debt overhang.

The empirical studies examined in chapter three mainly focus on the cost side of higher

capital requirements and hold the ROE fixed. This is not consistent with the theoretical

insights discussed in chapter two. The outcomes of these studies over-estimate the

increases of funding costs and neglect some beneficial consequences of higher capital

requirements, such as lower required returns on equity and the reduction of distortions

and inefficiencies. These studies also focus on private costs and do not take into account

the social benefits of a better-capitalized banking system. Most importantly, a large part

of the estimated rise in funding costs is caused by the loss of subsidized debt. Having

more equity capital reduces the ability of banks to obtain the benefits of the tax shield

and implicit government guarantees. Without these distortions, the transition towards a

47    

better-capitalized banking system would be easier. Reforms of the corporate tax system

and the government guarantee policy should complement higher capital requirements.

The second part of this thesis focuses on these two distortions and reforms. Chapter four

shows that the implicit subsidy that is created by the deductibility of interest expenses

favors debt financing significantly. As leverage and interest rates increases, so does the

distortive effect of the tax shield. This allows investors of leveraged banks to capture

more revenue due to lower tax expenses. Calculations of the Dutch situation confirm

that high-leveraged banks indeed exploit this implicit subsidy relatively more than their

better-capitalized competitors. This negative externality incentivizes banks to prefer debt

financing, while on the other hand more equity capital needs to be acquired. The

challenge of reforming the corporate tax system is to remain total tax expenses more or

less unchanged and to shift the incentive from debt financing to equity financing.

Therefore, the deductibility of interest expenses should be maximized or capped, while

net interest income could be less taxed. Such a reform needs to be set internationally.

The distortive effect of the implicit government guarantee lowers funding costs of bank

debt, as explained in chapter five. In the presence of implicit government guarantees,

not all inefficiencies of high leverage are reflected in the costs of bank debt funding. The

guarantee also exacerbates the debt overhang problem, because debt holders have

fewer incentives to address excessive risk-taking and high leverage ratios. The implicit

government guarantees can be significantly reduced if the banking system is better

capitalized and able to internalize losses and depreciations. Estimations of the Dutch

implicit subsidy show that the impact of the guarantee significantly increased during and

after the financial crisis. The calculations demonstrate that due to a lower market

confidence in banks and higher levels of uncertainty, the advantages of the implicit

government guarantee on funding costs increased, especially for high-leveraged banks

with lower credit ratings. To reduce the need and impact of an implicit government

guarantee, the banking system must be recapitalized through intervention by the

government and/or regulator. This could be done by imposing a contingent capital (CoCo)

requirement. CoCo is a form of debt that automatically converts to equity if a bank faces

too many losses. If the amount of CoCos is large and the conversion rate dilutive, banks

will have incentives to prevent conversion. Hence, banks will have interest to acquire

new equity capital timely. Due to the large size of CoCos, the likelihood of a bail out

reduces and the implicit government guarantee is of less importance.

Concluding, equity capital is not expensive, but bank debt is made cheap. Therefore,

Basel III should be complemented with tax policy reforms and recapitalization of the

banking system with the use of a contingent capital (CoCo) requirement.

48    

List of Abbreviations

BCBS Basel Committee on Banking Supervision BHC Bank Holding Company BIS Bank for International Settlements CAPM Capital Asset Pricing Model CEO Chief Executive Officer CoCo Contingent Convertible DNB De Nederlandsche Bank (NL) ECB European Central Bank (EU) EURIBOR EURo InterBank Offered Rate ESM European Stability Mechanism FDIC Federal Deposit Insurance Corporation Fed Federal Reserve System (US) FSA Financial Services Authority (UK) GDP Gross Domestic Product IIF Institute of International Finance IMF International Monetary Fund LIBOR London Interbank Offered Rate MAG Macroeconomic Assessment Group MMMF Money Market Mutual Fund NPV Net Present Value OECD Organization for Economic Co-operation and Development Repo Repurchase agreement ROE Return On Equity RWA Risk Weighted Assets SIFI Systemically Important Financial Institution TARP Troubled Asset Relief Program TBTF Too Big To Fail TITF Too Important To Fail WACC Weighted Average Cost of Capital Note: capital = equity capital, unless explicitly mentioned

49    

Bibliography Acharya, V., Mehran, H., Schuermann, T., Thakor, A. (2011) Robust Capital Regulation Federal Reserve Bank of New York, Staff Report no. 490 Acharya, V., Mehran, H., Thakor, A. (2010) Caught Between Scylla and Charybdis? Regulating Bank Leverage When There is Rent-Seeking and Risk-Shifting, working paper Admati, A., DeMarzo, P., Hellwig, M., Pfleiderer, P. (2011) Fallacies, Irrelevant Facts, and Myths in the Discussion of Capital Regulation: Why Bank Equity is Not Expensive, Stanford GSB Research Paper No. 2063 Admati, A., DeMarzo, P., Hellwig, M., Pfleiderer, P. (2012) Debt Overhang and Capital Regulation, Working Paper, no. 114, Rock Center of Corporate Governance Admati, A., Hellwig, M. (2013) The Bankers’ New Clothes: What’s Wrong with Banking and What to Do about It, Princeton University Press Angelini, P., Clerc, L., Cúrdia, V., Gambacorta, L., Gerali, A., Locarno, A., Motto, R., Roeger, W., Van den Heuvel, S., Vlček, J. (2011) Basel III: Long Term Impact on Economic Performance and Fluctuation, Federal Reserve Bank of New York, Staff Report no. 485 Baker, D., McArthur, T. (2009) The Value of the “Too Big to Fail” Big Bank Subsidy, 09-2009, Center for Economic and Policy Research Bank for International Settlements (2010, revised 2011) Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems, Basel Committee on Banking Supervision Berger, A., Herring, R., Szegö, G. (1995) The Role of Capital in Financial Institutions, Journal of Banking & Finance, vol. 19, pp. 393-430 Berk, J., DeMarzo, P. (2008) Corporate Finance, Pearson Education Limited Blum, J. (1998) Do Capital Adequacy Requirements Reduce Risks in Banking?, Journal of Banking and Finance, vol. 23, pp. 755-771 Boot, A.W.A. (2011) Banking at the Cross Roads: How to Deal with Marketability and Complexity?, Amsterdam Center for Law & Economics, working paper No. 07 Boot, A.W.A. (2013) Financial Sector in Flux, Journal of Money, Credit and Banking, forthcoming

50    

Brei, M., Gambacorta, L., Von Peter, G. (2011) Rescue Packages and Bank Lending, BIS Working Paper, no. 357 Calomiris, C., Herring, R. (2011) Why and How to Design a Contingent Convertible Debt Requirement, Financial Institutions Center Working Paper No. 11-41, University of Pennsylvania Wharton School Cosimano, T., Hakura, D. (2011) Bank Behavior in Response to Basel III: A Cross-Country Analysis, IMF Working Paper, no.119 Diamond, D., Dybvig, P. (1983) Bank Runs, Deposit Insurance and Liquidity, Journal of Political Economy, no. 91, pp. 401-419 De Nicolo, G., Gamba, A., Lucchetta, M. (2012) Capital Regulation, Liquidity Requirements and Taxation in a Dynamic Model of Banking, Discussion Paper no. 10, Deutsche Bundesbank Fatica, S., Hemmelgarn, T., Nicodème, G. (2012) The Debt-Equity Tax Bias: Consequences and Solutions, European Commission Working Paper no. 33 Frenkel, M., Rudolf, M. (2010) The Implications of Introducing an Additional Regulatory Constraint on Banks’ Business Activities in the Form of a Leverage Ratio, working paper Gambacorta, L., Mistrulli, P. (2004) Does Bank Capital Affect Lending Behavior? Journal of Financial Intermediation, vol. 13, pp. 436-457 Gorton, G. (2010) Slapped By the Invisible Hand: The Panic of 2007, Oxford University Press Gorton, G., Metrick, A. (2010) Securitized Banking and the Run on Repo, forthcoming, Journal of Financial Economics Gorton, G., Metrick, A. (2010) Regulating the Shadow Banking System, Brooking Papers on Economic Activity, Q3-2010, pp. 261-312 Gorton, G., Lewellen, S., Metrick, A. (2011) The Cost of Bank Capital: Thinking Beyond Modigliani and Miller, working paper Gorton, G., Ordonez, G. (2012) Collateral Crises, NBER Working Paper, No. 17771 Hellwig, M. (2010) Capital Regulation after the Crisis: Business as Usual? Working paper

51    

Ivashina, V., Scharfstein, D. (2010) Bank Lending During the Financial Crisis of 2008, Journal of Financial Economics, forthcoming Jensen, M. (1986) Agency Cost of Free Cash Flow, Corporate Finance, and Takeovers, American Economic Review, vol. 76, no. 2, pp. 323-329 Kashyap, A., Stein, J., Hanson, S. (2010) An Analysis of the Impact of ‘Substantially Heightened’ Capital Requirements on Large Financial Institutions, Working Paper King, M. (2010) Mapping Capital and Liquidity Requirements to Bank Lending Spreads, BIS Working Papers, no. 324 Krishnamurthy, A., Vissing-Jorgensen, A. (2010) The Aggregate Demand for Treasury Debt, working paper Miles, D., Yang, J., Marcheggiano, G. (2012) Optimal Bank Capital, The Economic Journal, vol. 122, no. 563 Miller, M. (1995) Do the M&M Propositions Apply to Banks?, Journal of Banking and Finance, vol. 19, pp. 483-489 Modigliani, F., Miller, M. (1958) The Cost of Capital, Corporation Finance, and the Theory of Investment, American Economic Review, vol. 48, pp. 261-297 Moody’s Analytics (2011) Quantifying the Value of Implicit Government Guarantees for Large Financial Institutions, Quantitative Research Group Moody’s Investors Service (2012) Key Drivers of Dutch Bank Ratings Actions Myers, S. (1977) Determinants of Corporate Borrowing, Journal of Financial Economics, vol. 5, pp. 145-175 Myers, S., Majluf, N. (1984) Corporate Financing and Investment Decisions when Firms Have Information that Investors Do Not Have, Journal of Financial Economics, vol. 13, pp. 187-222 Noss, J., Sowerbutts, R. (2012) The Implicit Subsidy of Banks, Financial Stability Paper, no. 15, Bank of England Pennacchi, G. (2009) Deposit Insurance, working paper

52    

Pfleiderer, P (2010) On the Relevancy of Modigliani and Miller to Banking: A Parable and Some Observations, Stanford University, Rock Center for Corporate Governance, Working Paper no. 93 Philippon, T., Schnabl, P. (2012) Efficient Recapitalization, Journal of Finance, vol. 68, nr. 1, February 2013 Repullo, R., Saurina, J. (2011) The Countercyclical Capital Buffer of Basel III: A Critical Assessment, CEMFI Working Paper, no. 1102 Santos, A., Elliott, D. (2012) Estimating the Costs of Financial Regulation, IMF Staff Discussion Note, SDN/12/11 Scharfstein, D., Coates, J. (2009) Lowering the Cost of Bank Recapitalization, Yale Journal of Regulation, 07-2009 Schich, S., Lindh, S. (2012) Implicit Guarantee for Bank Debt: Where Do We Stand?, OECD Journal, no. 1, Financial Market Trends Thakor, A. (1996) Capital Requirements, Monetary Policy and Aggregate Bank Lending: Theory and Empirical Evidence, Journal of Finance, vol. 51, pp. 279-324 Ueda, K., Weder di Mauro, B. (2012) Quantifying Structural Subsidy Values for Systemically Important Financial Institutions, IMF Working Paper No. 128 Van den Heuvel, S. (2008) The Welfare Cost of Bank Capital Requirements, Journal of Monetary Economics, vol. 55, pp. 298-320 Van Tilburg, R. (2012) The Financial Overweight of the Netherlands, SOMO Paper, November 2012

53    

Other References, Sources and Data Annual reports and form 20-F of ABN AMRO 1999-2012 via www.abnamro.nl, www.shareholder.com and www.sec.gov Annual reports and form 20-F of ING 1996-2012 via www.ing.com and www.sec.gov Annual reports of Rabobank 2000-2012 via www.rabobank.nl and Rabobank Investor Relations via [email protected]  Annual report DNB (2011) and www.dnb.nl Bloomberg Fixed Income Database Interim report Committee Van Dijkhuizen (2012) Lecture by Miles, D., Optimal Bank Capital, Stanford Finance Forum, June 2011 via www.youtube.com Lecture by Admati et al., Fallacies, Irrelevant Facts, and Myths in the Discussion of Capital Regulation: Why Bank Equity is Not Expensive, Stanford Finance Forum, June 2011 via www.youtube.com Ministry of Finance, Corporate tax rates 1996-2012, via www.rijksoverheid.nl The Economist, Strength in Numbers: How Much Capital Did Banks Opt to Hold When They Had the Choice?, November 10th 2012 The Financial Times, More Capital Will Not Stop the Next Crisis, R. Rajan, October 1st 2009 The Region, Federal Reserve Bank of Minneapolis, Interview with Gary Gorton, December 2010, pp. 14-29 The Wall Street Journal, Running on Empty, J. H. Cochrane, March 2nd 2013 ** Several informal meetings with banks, accountancy and consultancy firms **

54    

Appendices 1. Data Leverage Calculation 38

                                                                                                                         38  Sources:  annual  reports,  20-­‐F  form  (SEC)  

55    

2. Key Interest Rates39

3. Data Tax Shield40

[EXCEL INPUT ON THE NEXT PAGE]

                                                                                                                         39  Source:  LIBOR  via  Fed  www.research.stlouisfed.org  and  others  via  DNB  www.statistics.dnb.nl  40  The  different  corporate  tax  rates  among  countries  and  jurisdictions  are  neglected  because  the  data  that  is  

needed  for  allocation  of  interest  payments  is  confidential  and  not  publically  available.    

56    

57    

4. Net Interest and Fee Income41

                                                                                                                         41  Sources:  annual  reports,  20-­‐F  form  (SEC)  

 0    

 5.000    

 10.000    

 15.000    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ABN  AMRO  

Net  fee  income  

Net  interest  income  

 0    

 5.000    

 10.000    

 15.000    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

Rabobank  

Net  fee  income  

Net  interest  income  

 0    

 5.000    

 10.000    

 15.000    

 20.000    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ING  

Net  fee  income  

Net  interest  income  

58    

5. Data and Calculations Implicit Government Guarantee42

 Aaa   Aa1   Aa2   Aa3   A1   A2   A3   Baa1   Baa2   Baa3  

1999   31   36   41   48   55   65   89   143   172   210  

2000   68   88   94   99   106   113   137   182   241   282  2001   65   77   82   86   98   112   134   203   400   491  

2002   22   36   39   43   55   66   86   149   297   394  2003   39   46   51   57   70   79   94   146   243   326  

2004   20   29   34   46   56   67   81   145   303   383  2005   14   20   25   34   38   46   54   110   229   345  

2006   15   19   23   28   34   41   48   98   143   261  

2007   21   23   26   32   38   44   53   100   121   243  2008   126   135   147   156   163   168   176   217   265   364  

2009   65   82   109   130   156   182   213   345   411   508  2010   47   53   59   65   74   91   116   267   333   431  

2011   41   45   50   57   64   75   99   239   306   403  

2012   18   22   27   32   42   51   73   193   260   358  

                      Source: Bloomberg (Average Credit Spreads of US Corporate Bonds 1-year versus US Treasury Bills 1-year in Basis Points)

Credit Rating Notches (difference between “stand-alone” and “supported”), source: Moody’s

                                                                                                                         42  Sources:  annual  reports,  form  20-­‐f,  Moody’s  (2011,  2012),  Bloomberg  

0  

0,5  

1  

1,5  

2  

2,5  

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

Notches  

59    

Implicit Subsidy (High) 1999-2012, sources: annual reports, Moody’s and Bloomberg

Implicit Subsidy (Low) 1999-2012, sources: annual reports, Moody’s and Bloomberg

 0    

 2.000    

 4.000    

 6.000    

 8.000    

 10.000    

 12.000    

 14.000    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ING  

ABN  AMRO  

Rabobank  

 0    

 500    

 1.000    

 1.500    

 2.000    

 2.500    

1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  

ABN  AMRO  

ING  

Rabobank  

60    

61    

62    

6. Capital Ratios Graphs

Source: Kashyap, Stein, Hanson (2010)

63    

7. Bank Lending Spreads Graph

Bank Lending Spreads, USA and EU 1982-2009, source: King, M. (2010) BIS Working Paper 324