“Too‐Small‐To‐Survive” versus “Too‐Big‐To‐Fail” banks: The two sides of the same coin

• Date: 01 August 2018
• Author: Nikolaos I. Papanikolaou,Theoharry Grammatikos
• Published date: 01 August 2018
• DOI: http://doi.org/10.1111/fmii.12094

DOI: 10.1111/fmii.12094

ORIGINAL ARTICLE

“Too-Small-To-Survive” versus “Too-Big-To-Fail”

banks: The two sides of the same coin

Theoharry Grammatikos1Nikolaos I. Papanikolaou2

1LuxembourgSchool of Finance, University of

Luxembourg,6, rue Coudenhove-Kalergi, L-1359,

Luxembourg

2BournemouthUniversity, Department of

Accounting,Finance and Economics, BH8 8EB,

UK

Correspondence

NikolaosI. Papanikolaou, Bournemouth

University,Department of Accounting, Finance

andEconomics, BH8 8EB, UK.

Email:npapanikolaou@bournemouth.ac.uk

Abstract

In the recent crisis, the U.S. authorities bailed out numerous banks

through TARP, whilst let many others to fail as going concern enti-

ties. Even though both interventions fully protect depositors, a bail

out represents an implied subsidy to shareholders, which is not yet

the case with closures where creditors are not subsidised. We inves-

tigate this non-uniform policy, demonstrating that size and not per-

formance is the decision variable that endogenously determines one

threshold below which banks are treated as TSTS by regulators and

another one above which are considered to be TBTF. We, hence,

provide a pair of economic rather than regulatory cut-offs for TBTF

and TSTS banks. The shareholders and the other uninsured creditors

of a distressed bank are not bailed out if the bank is considered to

be TSTS. We further document that the less complex a bank is, the

less likely is to be bailed out and, hence, to have all of its creditors

protected.

KEYWORDS

bank creditors, complexity, size threshold, Too-Big-To-Fail,

Too-Small-To-Survive

JEL CLASSIFICATION

D02, G01, G21, G28

1INTRODUCTION

In October 2008, the U.S. Congress passed the Emergency Economic Stabilization Act (EESA) and authorised the

Department of the Treasuryto launch the Troubled Asset Relief Program (TARP)with the purpose to offer emergency

financial aid to corporate firms but, most importantly,to bolster the resiliency of banking institutions. Not surprisingly,

many banks that had been largely affected by the turmoil in credit markets triggered by the U.S.subprime mortgage

crisis received TARP funds via the Capital Purchase Program(CPP), which was the key component of TARP.Through

CPP,the U.S. Treasury invested up to $250 billion in the preferred equity of banks to enhance their capital ratios.The

c

2018 New YorkUniversity Salomon Center and Wiley Periodicals, Inc.

Financial Markets,Inst. & Inst. 2018;27:89–121. wileyonlinelibrary.com/journal/fmii 89

90 GRAMMATIKOSAND PA PANIKOL AOU

primary aim of this rescue package was the prevention of the sudden and simultaneous collapse of a large number of

distressed banks, which would have had destructiveeffects on the entire financial system.

Nonetheless, every coin has two sides: on 28 September 2007, NetBank was the first banking firm to fail as a going

concern entity in the U.S.in the latest financial crisis. The Federal Deposit Insurance Corporation (FDIC) took receiver-

ship of NetBank and all the insured deposit accounts were transferred to an assuming institution. Some days later,on

4 October 2007, Miami ValleyBank was also shut down by the authorities. The collapse of Miami Valley Bank was fol-

lowed by those of Douglas National Bank and Hume Bank in early 2008. Importantly,the number of failures increased

rapidlyfrom 2008 onwards. In total, more than 500 collapses were recorded during the crisis. The FDIC was appointed

receiver of the bankrupt institutions and this inflicted a total loss of $76 billion on the system.

Accordingly,the U.S. federal authorities -like European and other national authorities worldwide- providedsubstan-

tial financial aid to a number of troubled banking organisations during the crisis while, at the same time, allowed many

others to go bankrupt as going concern entities. The FDIC-backed resolution mechanism is designed to cope with the

insolvency of distressed banks under either normal or unstable economic conditions, whereas TARP is viewed as an

emergency mechanism designed to cope with bank fragility.Even though under both interventions small bank deposi-

tors remain fully protected, TARPgovernment bailouts represent an implicit subsidy to the bank's shareholders, which

is not yet the case with FDIC-backed failures where shareholders are not subsidised. As documented in Gandhi and

Lustig (2015), government guarantees in the form of bailouts protect the shareholders of large banks, but not those

of small banks in a financial disaster.In a similar vein, Veronesi and Zingales (2010) calculate the costs and benefits of

TARPfrom the perspective of big banks’ shareholders and conclude that these firms receive large subsidies. Hence, the

question should not be bailout vs. failure, but rather subsidy for stockholders (via TARP)vs. only protecting depositors

(via FDIC-backed failures) with severe consequences for the uninsured creditors and the stockholders of the failed

banks.

Such a differential treatment of distressed banks raises some important questions: do regulators consider some

institutions as being very important for the system -or alternatively ‘Too-Big-To-Fail’ (TBTF)- inthe sense that a col-

lapse of any of them has to be deterred for not to trigger contagious defaults in the entire banking network, whereas

some others are perceived as being ‘Too-Small-To-Survive’(TSTS) in that their failure as going concern entities has no

material impact on their counterparts, let alone on the system as a whole? Is it the size of financial institutions the

key determinant that makes the authorities to treat distressed banks differently,or it is that failed banks lag behind

in terms of performance compared to those that the authorities decide to financially support via TARP? To put it dif-

ferently,is it that regulators are reluctant to support the uninsured creditors and the shareholders of some distressed

banks because they consider these banks as being TSTS? Crucially, is there anyspecific threshold size below which a

banking firm is viewed as TSTS by regulatory authorities? And, if so, is there a relevant threshold size for TBTF banks?

And, finally, what is the role of bank complexity in regulators’ interventionsin the case of financial distress and how

complexity interactswith size?

Admittedly,size lies in the very centre of banking research. Banks of different sizes follow diverse business models

which are related to various levelsof risk, increased or reduced earnings, higher or lower failure probabilities and carry

a different weight for the financial system. Notwithstanding the fact that these variations havebeen well-documented

in the extant literature,the still growing crisis literature, within which our study falls, has not paid the necessary atten-

tion to the role that size and complexity playin the decision of regulatory authorities of how to treat a distressed bank

and its creditors. Hence, in this paper,we focus on the recent financial crisis aiming to provide concrete answers to the

aforementioned questions.

We collect data for the entire population of U.S. commercial and savings banks and distinguish our sample banks

into two key groups: the non-distressed banks, and the distressed banks composed bythe TARP and the FDIC-backed

(failed) banks. We first conduct a univariate analysis to compare the averagesize and the performance of each banking

group in the years prior to the onset of the crisis, accounting also for the organisational and operational complexity of

banks given the systemic implications of complexity in resolving distressed banks and the relativeimportance of com-

plexity vs. size. The measurement of bank performance relies on the CAMELS rating system, which has been utilised

by the U.S. regulators for more than two decades to monitor the safety and soundness of banks. We, then, employ a

GRAMMATIKOSAND PA PANIKOL AOU 91

multivariate technique which can endogenously define one or more threshold levels of an observed variable to exam-

ine whether we can specify one threshold size below which banks are considered by the authorities as being TSTS and

a second one above which banks are considered to be TBTF.To put otherwise, what we are modelling is whether FDIC

funds or TARP funds are used to support the distressed banks in the recent crisis, and whether the bailout extends

beyond bank debts.

The univariate analysis suggests that the decision of regulatory authorities to choose between a TARP-assisted

bailout and an FDIC-backed failure is more influenced by bank's size and less influenced by bank's performance, com-

plexity, and risk-taking profile. The regression results of the multivariate threshold analysis reveal that the failure

and TARP probabilities are, in essence, determined by size. In this context, our threshold technique endogenously

specifies two cut-off points for size: one for the TSTS banks and another one for the TBTF banks, which are con-

sidered to be the two sides of the same coin. Regulators choose not to bailout the shareholders and the uninsured

creditors of a distressed bank if the bank is considered to be TSTS. From a market equilibrium viewpoint, the exis-

tence of these size regimes means that the free market outcome cannot be reached because it is the size and not

the performance of banks which is the key decision variable for the failure and TARP probabilities. The complex-

ity of banks are found to be negatively related to the probability of failure across the different size regimes, imply-

ing that the more complex a bank is the more likely is to receive TARP assistance and, hence, to have all of its

creditors protected. The impact of complexity on failure is considerably stronger in the TBTF regime and the esti-

mates are highly statistically significant in this regime, showing that banks which are perceived as TBTF are also too-

complex-to-fail.

The paper proceeds as follows. Section 2 reviews the importance of size and complexity in the banking literature.

The aim of this section is not to provide an extensive review of the past and the current literature;rather, its key pur-

pose is to present how bank size is intertwined with performance, and discuss the relevance of size and complexity

through the lenses of the most influential studies. Section 3 presents our data set and outlines how it is constructed;

the relevant key variables and the univariate analysis we conduct are also presented in this section. The multivari-

ate threshold econometric technique we employ is fully described in Section 4. The regression results are presented

and discussed in Section 5; a set of interesting policy implications drawn from the results are also offered. Section 6

is devoted to sensitivity analysis, whereas Section 7 provides a brief summary of our main findings and offers some

concluding remarks.

2BANK SIZE AND COMPLEXITY

In September 1984, the Office of the Comptroller of the Currency (OCC) made, for the first time, a public distinction

between TBTF and non-TBTFbanking institutions. It, specifically, announced that the biggest 11 from a total of approx-

imately 14,000 banks that were in operation at that time were considered as being TBTF and as such they would be

offered a full deposit insurance, whereas all the other banks would remain only partially covered.After that announce-

ment, the spotlight of the relevant literature turned to shine large banking organisations and to examine the impor-

tance of size for the smooth functioning of the whole financial system. Of the most prominent studies in the early TBTF

banking literature were those of Boyd and Runkle (1993), Demsetz and Strahan(1997), O'Hara and Shaw (1990), and

Galloway,Winson, and Roden (1997).

A significant part of the current banking literature, which has been sparkedby the emergence of the late 2000s cri-

sis, explores the relevanceof TBTF banks in the propagation of the crisis and its subsequent dissemination throughout

the global economy.For instance, Huang, Zhou, and Zhu (2009) construct a framework for measuring and stress testing

the systemic risk of 12 U.S.major commercial and investment banks. Adrian and Shin (2010) examine the procyclicality

in leverage of the 5 biggest U.S. investmentbanks prior to the crisis. Patro, Qi, and Sun (2013) employ the 22 largest

commercial and investment banks in U.S.to analyse the relevance of stock return correlations in assessing the level of

systemic risk. Papanikolaou and Wolff (2014) use a sample of 20 U.S.systemically important banks to study how off-

balance-sheet activities affect the overall risk profile of banks as well as the levelof systemic risk before and after the