The JOURNAL OF FINANCE:

The Case For Intervening In Bankers’ Pay JOHN THANASSOULIS∗

July 30, 2011

Abstract

This paper studies the default risk of banks generated by investment and remuneration pressures. Competing banks prefer to pay their banking staff in bonuses and not in fixed wages as risk sharing on the remuneration bill is valuable. Competition for bankers generates a negative externality driving up market levels of banker remuneration and so rival banks’ default risk. Optimal financial regulation involves an appropriately structured limit on the proportion of the balance sheet used for bonuses. However stringent bonus caps are value destroying, default risk enhancing and cannot be optimal for regulators who control only a small number of banks.

Keywords: Bonuses, default risk, competition for bankers, financial regulation.

JEL Classification: G21, G34.

∗John Thanassoulis is at the University of Oxford. I would first like to thank the editor, Campbell Harvey, the anonymous associate editor and the reviewer. Their comments have substantially improved this paper. I am grateful to the Economic and Social Research Council of the UK for financial support (grant RES-000-22-3468). I am also grateful to seminar participants at Oxford University, the Bank of England, the New York Federal Reserve and the Centre for Competition Policy at UEA. Finally special thanks to Howard Davies, Alex Edmans, Alex Guembel and Alan Morrison for detailed and helpful comments. I have no employment or consulting link with any bank or financial institution nor any regulator of such institutions.“We will not go back to the days of reckless behavior and unchecked excess at the heart of this crisis,” President Obama, September 2009.

“What is unfair is if you are an analyst who takes zero risk for the bank, and yet you can still be harmed by somebody who has been acting like they work for a mortgage hedge fund rather than an investment bank,” former Lehman employee, October 2008.

Is competition between banks for trader and executive talent destabilizing? Does the competition for bankers raise the risk of bank default? If so is there a case for financial regulation which intervenes in bankers’ pay? Many policy makers in the EU, US, and G20 feel that something is not right with bankers’ pay – that the high bonuses cannot just be the necessary incentives bankers require to work. Consequently the EU, US and individual countries within the EU have implemented or proposed policies which amount to the regulation of bankers’ pay.

This paper presents a model of banker remuneration in a competitive market for banker talent. This allows us to study, and then calibrate, the default risk of the banks generated by investments and remuneration pressures. We will show that competition by banks for bankers generates an empirically relevant negative externality driving up rival banks’ default risk. Op- timal financial regulation will involve some intervention in the competitive market for bankers’ labour. In particular it will involve weak caps on the proportion of the balance sheet which can be used for bonus payments.

To appropriately frame a financial regulator’s concern, let us explore what problem interven- tion in the competitive market for bankers’ labour could hope to solve. At the level of the CEO and the other most senior executives in an organization the answer is immediate. Senior officers of a bank select the level of bank risk. It is important to ensure that these senior officers would not wish to take on more risk than their shareholders (Bebchuk 2009, Davies 2010) or more risk even than their debt holders (Bolton, Mehran and Shapiro 2010, Edmans and Liu 2011). However the argument for intervening in the labour market for all bankers is much less clear. Individual bankers work under a risk control regime overseen by the CEO and the Board. These senior executives can control bank risk through their policies on hedging, diversification and as- set allocation. Financial regulation exists to make sure that CEOs and Boards properly exercise their duties to understand and then build structures allowing them to manage the risks taken by their employees. Corporate governance failures may well exist in reality rendering the CEO

1

and Board unwilling/unable to manage the risks (Thanassoulis 2009) – but if this is the problem then tackling corporate governance directly would be advisable. As Corporate Governance and risk supervision is improved by more active regulation, ill-founded interventions in the labour market for bankers will soon become constricting and outlive their usefulness. Rather, if there is a case for intervening in bankers’ pay in general it must be that competition to hire the best bankers forces banks to, knowingly, pay and risk too much.

I offer a model of multiple banks of different sizes competing to hire a team of bankers from a population of bankers who are differentiated in terms of their skill, which is publicly observable. I model skill as affecting the expected return on assets. I assume that there are no corporate governance failings and so assume that the banks control the aggregate level of risk taken by their bankers so as to maximize bank value. The bankers can be paid in fixed wages or bonuses or both - the choice is endogenous. If the net realization of the bankers’ investments is to lower the bank’s assets to below some given level, then a default event (or run) occurs which results in some extra cost for the bank. For example, the costs of premature liquidation of some of the assets to meet the higher than usual demands from debt holders, or the increased costs of capital which a weaker financial institution faces. Thus the banks compete to hire bankers and the remuneration and risk which the banks run are decided endogenously.

In the competitive market equilibrium I find that the banks will opt to pay their bankers entirely in bonus. This is optimal for them not because of any incentive effects, instead a different mechanism is at work. Bonuses have much better insurance properties than wages do as the remuneration payment is better connected with the realized state of investments. If investment returns are low, the required compensation is also low – just when the danger of a default event is present. Thus banker bonuses, by facilitating risk sharing, allow banks to deliver utility to all their bankers in the least value reducing way.

Competition between banks for bankers creates a negative externality between the banks which drives banks’ default risk upwards. Each bank would like to hire the best team of bankers they can. For a bank to ultimately hire a given team of bankers they must meet the compe- tition for those bankers from the marginal competitor. The negative externality is exerted by the marginal competing bank as by driving up remuneration they are imposing costs on the employing bank which they do not have to bear. The costs arise as the increased remuneration

2

raises bank default risk, as a default event occurs even at smaller investment losses. Financial regulators would care about any increase in default risk as, firstly, the default of a financial institution has negative repercussions on all counterparties which are not internalised by the failing firm; and secondly, these negative repercussions may be so widespread that they cause other institutions to default resulting in a systemic crisis and so even greater costs for society as a whole.

Can remuneration payments by a bank ever appreciably add to the risk of a run or of default and so merit policy-maker attention? Just before the financial crisis the total remuneration bill in one year at Merrill Lynch and at Morgan Stanley was running at 50% of their entire stock of shareholder equity.1 Looking at data on the banks and financial institutions traded on the NYSE over the last 10 years, in about 10% of cases the remuneration bill was worth more than 80% of total shareholder equity.2 These are huge payments which can potentially make the difference between investors having and losing confidence in a bank.3 Thus it is imperative that a study of financial regulation considers remuneration.4

In the competitive equilibrium bonuses depend not only on the distribution of banker talent, but also on the size distribution of the banks competing to hire these bankers. We are able to solve for the equilibrium remuneration in closed form allowing an exploration of how financial regulation should adjust to properly reflect remuneration effects. This analysis is conducted in Section III and the results can be summarised as follows:

Bonus Limits. A modest cap on the proportion of the balance sheet which can be used for bonuses – modest in that it is close to the current equilibrium rate of bonuses – lowers default risk and raises the value of the largest banks. If a maximally weak bonus cap is implemented then these benefits are maximised for almost all banks. The maximally weak bonus cap is bank specific. It is the most stringent cap on the proportion of the balance sheet which can be used for bonuses which just remains generous enough to allow the bank to pay their staff in bonuses without recourse to increased fixed wages. This cap is strictly lower than the bonus paid without a cap as the bonus cap dampens the negative externality inherent in remuneration by impacting the ability of rival banks to poach banking staff more than it impacts the equilibrium employer of a banker. Thus compensation levels for all banks fall and in addition the insurance problems of fixed wages are avoided. Such

3

a regulation is of value even to a regulator who only controls a subset of the financial institutions. Features of this approach can be seen in the new Basel III restrictions on remuneration if tier 1 capital should fall below 7%.

However very stringent bonus caps, or a requirement to use wages rather than bonuses increase bank default risk. The EU have legislated to introduce per person caps on bonuses. US Federal Agencies are consulting on an implementation of the Dodd-Frank Act which would see them introducing sign off requirements at a per banker level on the structure and level of compensation.5 Banks under such a regime are forced (in the case of Europe) or encouraged (in the case of the US) to use fixed wages. Competition for bankers keeps wages high and results in banks having high fixed costs they must pay whatever the outcome of their bankers’ investments. As remuneration is such a large fraction of shareholder equity, the increase in risk is substantial. (See the calibration in Section IV). Appropriately designed bonus caps can therefore lower bank default risk even with well functioning corporate governance. A rationale behind the approach of per person caps is to correct for poor corporate governance and so impose a government limit on the incentive individual bankers have to take risks. However corporate governance and risk monitoring are being explicitly improved in financial regulation thus weakening this rationale for per person regulation over time. Appropriately designed bonus caps at the balance sheet level would not become constricting or outlive their usefulness once other corporate governance focused measures take effect.

Taxation. A financial regulator might consider taxing the bonus pools of the banks in the hope of dampening the negative externality created by remuneration. Such a policy was followed in the UK (“the 50% super-tax on bonuses”) and in France. The taxation policy does lower the amount delivered to bankers, but it leaves bank default risks unaffected. In a competitive equilibrium, the amount a bank is willing to pay to hire a better banking team depends upon the quality of the alternative bankers it can hire. The tax does not alter this calculation. Though default risk is not altered, money is diverted from bankers’ pay to the government.

Increase capital adequacy ratios. Financial regulators limit the risks that a bank can take on by requiring the bank to maintain a certain proportion of its risk weighted assets as

4

equity (tier 1 capital). Changes in this capital adequacy requirement, or in the risk weight- ings, have the effect of altering the investment freedom of the bank for given equity levels. A reduction of the proportion of the balance sheet which can be put at risk does lower the default risk of the banks. It will not lower the bonus rate which the bankers receive, but as they are restricted in the activities they can pursue with the balance sheet, their total pay declines. Thus if banks respond to stricter capital adequacy regulations by lowering the risk profile of their assets, then the negative externality inherent in remuneration is lowered, though so too is the value of the banks.

Choices as to the level of equity and the size of the balance sheet are outside the scope of this model. Admati et al. (2010) have argued that substantially higher equity requirements are called for which would force the banks to raise equity rather than lower the risk profile or size of their assets. They argue that, absent the tax deductibility of interest, such a rebalancing towards equity would be (close to) costless for the banks for the reasons described in the seminal work of Modigliani and Miller (1958). If banks do respond to stricter capital adequacy requirements by raising more equity then they would become safer. Remuneration would still make up a part of the default risk in the manner this model describes – but the absolute level of risk would be reduced by the enhanced equity cushion.

Too big to fail? Limit bank size. There have been calls for a limit on bank size, or a tax which would have the same effect. An arbitrary cap on balance sheet size would result in a number of banks being of equal size at the bank size limit. These banks would compete with each other intensely for their bankers. The end result would be that banker bonus rates amongst the best bankers would rise and the default risk of all the banks affected by the size cap would be pushed up. (It is possible that a smaller bank will not be as important systemically and so the increased risk of default is acceptable.)

Of the possible regulatory tools analysed only two act to reduce the negative externality cre- ated by remuneration and so have the desired effect on default risk. These are capital adequacy and bonus caps. However only one of these, appropriately structured, can both lower default risk and raise bank values. The promising intervention is, as described above, a modest cap on the proportion of the balance sheet which can be used for bonuses. Optimal bonus caps would

5

be at least as strict as the maximally weak bonus cap. Conversely, a more stringent bonus cap can be shown to not be optimal for regulators who do not control all of the banks in the market. A calibration exercise suggests that remuneration can increase the risk of default of financial institutions by a quarter (Table I). A further calibration yields that for the bonuses awarded by a sample of major household name banks to have been limited to the level of the maximally weak bonus cap would have required over $3 billion to be removed from each of their annual bonus pools (Figure 4). Thus remuneration contributes substantially to risk, and optimal interventions are significant. Hence understanding how financial regulation should adapt to the impact of

remuneration pressures is a first order issue.

Related Literature

To be able to assess financial regulation in the light of bankers’ pay I offer a model of a competitive labour market by banks for teams of bankers. Such an endeavour builds on the seminal contributions of Gabaix and Landier (2008) and of Edmans, Gabaix and Landier (2009). In this series of papers the authors offer a model of a competitive labour market for CEOs. My contribution is to reformulate the models to consider competition between banks which requires the introduction of the possibility of a bank default event into the labour market model. This addition is key. First it drives the risk sharing properties of bonuses as compared to wages. Secondly with no default possibility there can be no assessment as to how pay policy impacts default risk. The latter is of key interest in the case of banking. In a similar vein, Bulow and Levin (2006) consider a competitive labour market in which pay cannot be made worker specific and they demonstrate that wages are compressed. Here I allow banks to make offers conditional on who they are employing and so such remuneration compression does not result.

To capture the nature of a bank and the implications for default I have built on the model of banking offered by Wagner (2009). As in Wagner if the size of a bank’s balance sheet should fall below some level (for example its level of liabilities) a default event or run occurs which results in an extra cost to the bank. Wagner however does not investigate the supply side competition for bankers and so is silent on banker pay in general.

The aim of this paper is to understand how intervention in the labour market for bankers would alter bank risk – widely considered policies go much broader than purely the regulation of CEO pay. Other researchers have used market equilibrium models to assess the effect of

6

CEO pay regulation. Recently Dittmann, Maug and Zhang (2010) argue that capping CEO pay would lower firms’ value in general; while Llense (2010) argues that pure pay for performance cannot explain the full increase in French CEO pay. Edmans and Gabaix (2011) argue that hiring decisions are more significant for wealth creation than optimising the CEO remuneration contract. The analysis I offer here makes at least two contributions. First it focuses on the market equilibrium of pay in a whole sector rather than just the CEO. Secondly it studies the risk of a run which banks have, which allows us to focus on the key policy question of the default risk accepted in equilibrium by the banking sector.

The work presented here identifies how competition between banks for bankers can create a negative externality which operates by driving up the remuneration rival banks must pay their bankers and so increases their default risk and expected costs of default. This new insight is related to a nascent literature which explores the externalities created for corporate governance by competition on the labour market. See Acharya and Volpin (2010), Acharya, Gabarro and Volpin (2010) and Dicks (2010).

An important insight from the model is that bonuses have a risk shifting role totally separate from incentive effects. Banks in my model have a concavity in their objective function which arises from the fact that, at sufficiently small investment realizations, a default event occurs which causes an extra cost for the bank. This insight has, to my knowledge, not been noted as being relevant to bankers’ pay. That a risk averse entity would like to share risk with her contractees is known and parallels exist, for example, in the share-cropping literature which stemmed from Newberry and Stiglitz (1979). That contracts should be structured to share risk appropriately is known in the contracting literature as the Borch rule.6 In a similar vein to this paper Nocke and Thanassoulis (2009) show that firms would seek to share risk with their suppliers if bad market outcomes would result in less investment than the first best in the future.

My model has focused on the aggregate level of risk which a bank would knowingly allow their team of bankers to take on rather than the risk choices of individual bankers. Others have focused on how competition between banks affects individual bankers’ moral hazard in a parallel stream in the literature (Axelson and Bond 2009, Bijlsma Boone and Zwart 2010, Inderst and Pfeil 2009).

Plan Of The Paper

7

The model is given in Section I and is solved for the competitive equilibrium in Section II. Financial regulation in the light of bankers’ pay is analyzed in Section III. Some empirical evidence and calibrations are provided in Section IV while the conclusion follows in Section V. Technical proofs are provided in Appendix A. Further data and calibration details are provided in the Internet Appendix B.