The determinants of banks' AT1 CoCo spreads

• Published date: 01 March 2022
• Author: Axel Kind,Philippe Oster,Franziska J. Peter
• Date: 01 March 2022
• DOI: http://doi.org/10.1111/eufm.12314

Eur Financ Manag. 2022;28:567–604. wileyonlinelibrary.com/journal/eufm © 2021 John Wiley & Sons Ltd.

|

567

DOI: 10.1111/eufm.12314

ORIGINAL ARTICLE

The determinants of banks' AT1 CoCo spreads

Axel Kind

1

|Philippe Oster

2

|Franziska J. Peter

3

1

Department of Economics, University of

Konstanz, Konstanz, Germany

2

Lucerne School of Business, Institute of

Financial Services Zug IFZ, Lucerne

University of Applied Sciences and Arts,

Rotkreuz, Switzerland

3

Department of Corporate Management

and Economics, Zeppelin University

Friedrichshafen, Friedrichshafen,

Germany

Correspondence

Philippe Oster, Lucerne University of

Applied Sciences and Arts, Lucerne

School of Business, Institute of Financial

Services Zug IFZ, Suurstoffi 1, CH‐6343

Rotkreuz, Switzerland.

Email: philippe.oster@hslu.ch

Abstract

We conduct a comprehensive pricing study of addi-

tional tier 1 (AT1) contingent convertible (CoCo) bonds

issued by Eurozone banks. By accounting for an ex-

tensive set of pricing determinants related to the reg-

ulatory framework, the security design and key market

variables, we show that the regulatory concept of the

maximum distributable amount (MDA) introduced in

2016 has a significant and economically meaningful

impact on CoCo spreads. Furthermore, we examine

whether the market stress induced by the COVID‐19

pandemic influences the determinants of CoCo

spreads. Our results show that the pricing factors re-

main stable throughout tranquil and volatile periods.

KEYWORDS

additional tier 1 (AT1), contingent convertible(CoCo), COVID‐19

pandemic, maximum distributable amount (MDA), point of

nonviability (PoNV), regulation, security design, systemically

important banks (SIB)

JEL CLASSIFICATION

G12, G13, G21, G28

EUROPEAN

FINANCIAL MANAGEMENT

CreditSights, 2017 offered free access to their worldwide CoCo database in December 2017, for

which we would like to express our gratitude. Also, we appreciate the helpful comments on

binding MDA criteria provided by Scope Ratings (Lambert & Gurbuz, 2017) and the highly

appreciated feedback from various reviewers and discussants, in particular René Marian

Flacke, at the 26th Annual Meeting of the German Finance Association (DGF) in September

2019. Furthermore, we are especially grateful for the constructive comments and suggestions of

the anonymous reviewers and John A. Doukas (the editor).

1|INTRODUCTION

Under the current Basel III framework, banks are required to issue, depending on their na-

tional and international importance, a certain ratio of bail‐in‐able securities to risk‐weighted

assets (RWA). Due to this regulation, the new hybrid asset class of additional tier 1 (AT1)

contingent convertible (CoCo) bonds has emerged with 252 issuances between 2014 and

November 2020, leading to a market valuation of €231 billion as measured by Barclays Global

Contingent Capital Index.

1

The literature on CoCos has focused on the following aspects

2

: The advantages and the

limitations of CoCos (Benink, 2018; Flannery, 2016), the optimal capital structure with CoCos

(Attaoui & Poncet, 2015; Barucci & Del Viva, 2012), the security design and its effects (Allen &

Tang, 2016; Cai et al., 2018; Chen et al., 2017; Davis & Prescott, 2017; Furstenberg, 2017), CoCo

issuance motives (Araten & Turner, 2013; Bancel & Mittoo, 2004; Dutordoir et al., 2014; Fajardo

& Mendes, 2020), its announcement effects on credit default swap (CDS) spreads (Ammann

et al., 2017) and stock performances (de Spiegeleer et al., 2016; Liao et al., 2017), a banks’risk‐

taking (Martynova & Perotti, 2018; Walther & Klein, 2015) and, specifically, the emergence and

mitigation of moral hazard risks (Berg & Kaserer, 2015; Díaz et al., 2018; Hilscher &

Raviv, 2014).

Promising research has been carried out on their valuation (Abed Masror Khah et al., 2019;

Barucci & Del Viva, 2012; Chang & Yu, 2018; Chen et al., 2017; Chung & Kwok, 2016; Corcuera

et al., 2014; Davis et al., 2014; de Spiegeleer et al., 2017; Leung & Kwok, 2017; Pennacchi &

Tchistyi, 2019a,2019b; Pennacchi et al., 2014; Turfus & Shubert, 2017). However, theoretical

contingent claim models, which are typically based on firm value (structural approach), credit

default swaps (credit derivative approach), or dynamic replication with shares (equity deriva-

tive approach), often have difficulties taking contextual parameters into account. In particular,

this applies to the multifaceted regulatory requirements imposed on banks.

We contribute to the current literature by offering empirical rather than theoretically

motivated insights into the determinants of CoCo spreads. This paper presents a comprehen-

sive pricing study of AT1 CoCo bonds, that is, CoCos that are triggered before default and

before bank resolution mechanisms come into play.

3

In detail, our study offers several con-

tributions to the ongoing debate on CoCos. First, related to the regulatory framework, our

paper is, to the best of our knowledge, the first to show that the distance‐to‐maximum dis-

tributable amount (MDA)‐threshold has a negative, economically relevant and statistically

significant impact on CoCo spreads. The distance‐to‐MDA‐threshold constitutes the major

proxy for regulatory reserves that prevent restrictions of banks’distributions. Our results in-

dicate that the new MDA concept is effective and MDAs are reflected in AT1 CoCo spreads. In

this respect, our article contributes in explaining the scope and the consequences of the in-

troduction of the Supervisory Review and Evaluation Process (SREP) guidelines in January

2016 (European Banking Authority [EBA], 2017b) on CoCo pricing—an issue that has re-

mained largely unexplored. In this respect, our paper directly relates to recent studies that

1

Barclays Global Contingent Capital Total Return Index Value Unhedged EUR.

2

For a comprehensive literature survey, see Oster (2020).

3

Although CoCos include both going‐concern (AT1) and gone‐concern (tier 2 and other) instruments, we focus on the former because for our pricing study we

require a homogeneous sample of instruments. In fact, tier 2 CoCos deviate considerably from AT1 CoCos in terms of their regulatory framework and security

design (e.g., tier 2 CoCos have mandatory coupons that are not subject to MDA restrictions).

568

|

EUROPEAN

FINANCIAL MANAGEMENT

KIND ET AL.

measure the influence of regulatory reforms on the pricing of financial instruments in the

banking sector (e.g., Ahnert et al., 2020; Crespi et al., 2019).

Second, building on the results of Goncharenko & Rauf, 2016, who find lower spreads for

CoCos issued by banks with a global systemically important bank (G‐SIB) status, we extend the

issuers’regulatory status to domestic systemically important banks (D‐SIBs), respectively, other

systemically important institutions (O‐SIIs), as set by national regulators.

4

Though the banks’

G‐SIB status has a negative and statistically significant impact on spreads, we show that the

D‐SIB status, respectively, the O‐SII status, has no statistically significant influence on spreads.

This is remarkable, as national capital surcharges for D‐SIBs may exceed those under the G‐SIB

framework, notably in smaller countries. Moreover, our results indicate that issuers with an

affiliation to Southern Europe trade at significantly higher spreads, which can be explained by

the larger regulatory leeway.

Third, in the field of security design, we contribute to the literature by analysing, whether

the loss‐absorbing mechanism of principal write‐down offers significantly higher spreads than

equity‐conversion AT1 securities, as suggested on theoretical grounds by Martynova & Perotti

(2018) and empirically by Hesse (2016,2018). In doing so, we distinguish between equity‐

conversion and the temporary write‐down loss‐absorbing mechanism and account for the trigge

level. We show that AT1 CoCos with equity‐conversion and low‐trigger trade at significantly

lower spreads compared to their equivalents with permanent write‐down or high‐trigger.

5

Our

results confirm and reinforce anecdotal evidence by Avdjiev et al. (2013) and the results of

Hesse (2016,2018). Furthermore, we find that the Macaulay duration to next call date is a

statistically significant driver of CoCo spreads.

Fourth, with respect to the market‐related determinants of CoCo pricing, our baseline

regression model demonstrates that the credit default swap (CDS) spread (a proxy for the

issuers’probability of default) is positively and significantly related to AT1 CoCo spreads.

Similarly, also the coefficients of the 3‐month total return and the volatility of the issuers’

shares over 90 days are positive and significant.

Fifth, we complement the empirical results initially gained in a low‐volatility environment

by a panel analysis conducted during a high‐volatility period. Our results indicate that relevant

pricing factors are consistent during tranquil and volatile times. Reliability of the results is

obtained by focusing on a homogeneous sample of securities that comply and adhere to the

very same regulatory framework with regard to the implementation of Basel III. The analysis

achieves the highest data quality through manual data preparation based on original security

prospectuses and issuer reports.

The paper is organised as follows. Section 2introduces the regulatory framework, describes

AT1 CoCos and develops suitable pricing hypotheses. Section 3discloses our model and ex-

plains the data set. Section 4presents the results and compares them to previous studies.

Section 5examines panel data in the context of the COVID‐19 crisis and thus considers

different volatility regimes. Section 6concludes.

4

Financial institutions are classified on the basis of cross‐jurisdictional activity, size, interconnectedness, substitutability and complexity (Basel Committee on

Banking Supervision [BIS], 2013).

5

It is worth noting that the issue of multiple equilibria (or no equilibrium) raised by Sundaresan and Wang (2015) and Glasserman and Nouri (2016) does not

apply to the sample of contingent convertibles studied in this paper. As noted by Pennacchi and Tchistyi (2019b), the problem of multiple equilibria is only

relevant under very specific conditions and CoCo provisions. Most importantly, CoCos (and the related stocks) might be plagued by this problem only if they

possess a market trigger, which is never the case under the EU regulatory framework that, in CRD IV Article 54, explicitly requests for AT1 CoCos an

accounting trigger (of at least CET 5.125%). This is probably also the reason why Demertzisand Zenios (2019) correctly note that ‘international institutions do

not seem to be overly concerned about such a problem’(p. 84).

KIND ET AL.EUROPEAN

FINANCIAL MANAGEMENT

|

569