Investor behaviour and reaching for yield: Evidence from the sterling corporate bond market

• Author: Robert Czech,Matt Roberts‐Sklar
• DOI: http://doi.org/10.1111/fmii.12122
• Published date: 01 December 2019
• Date: 01 December 2019

DOI: 10.1111/fmii.12122

ORIGINAL ARTICLE

Investor behaviour and reaching for yield:

Evidence from the sterling corporate bond market∗

Robert Czech Matt Roberts-Sklar

Bank of England

Correspondence

RobertCzech, Bank of England, Threadneedle

Street,London, EC2R 8AH, UK.

Email:Robert.Czech@bankofengland.co.uk

∗Theviews expressed in this paper are those

ofthe authors, and not necessarily those of the

Bankof England or its committees. We wish to

thankan anonymous referee, our two discussants

NicolaBranzoli and Joao Santos, Franklin Allen,

MatteoAquilina, Andrea Buraschi, Christopher

Hansman,Harrison Hong, Raj Iyer, Petri Jylha,

MarcinKacperczyk, Andrei Kirilenko, Alex

Michaelides,David Miles, Jon Relleen, Savitar

Sundaresan,Felix Suntheim, Nick Vause and

seminarparticipants at the Bank of England, the

FinancialConduct Authority, Imperial College,

the13th Annual Central Bank Conference

onthe Microstructure of Financial Markets,

the2017 FSB Workshop on Systemic Stress,

InvestorBehaviour and Market Liquidity and the

68thAnnual Meeting of the French Economic

Associationfor their helpful comments.

Abstract

We provide evidence on how corporate bond investors react to a

change in yields, and how this behaviour differs in times of market-

wide stress. We also investigate ‘reaching for yield’ across investor

types, as well as providing insights into the structure of the corpo-

rate bond market. Using proprietary sterling corporate bond trans-

action data, we show that insurance companies, hedge funds and

asset managers are typically net buyers when corporate bond yields

rise. Dealer banks clear the market bybeing net sellers. However, we

find evidence for this behaviour reversingin times of stress for some

investors. During the 2013 ‘taper tantrum’,asset managers were net

sellers of corporatebonds in response to a sharp rise i nyields, poten-

tially amplifying price changes. At the same time, dealer banks were

net buyers. Finally, we provide evidence that insurers, hedge funds

and asset managers tilt their portfolios towards higher risk bonds,

consistent with ‘reaching for yield’ behaviour.

KEYWORDS

banks, corporate bonds, cyclicality, financial stability, insurers,

investment decisions, non-bank financial institutions, trading

volume

JEL CLASSIFICATION

G11, G12, G15, G21, G22, G23

1INTRODUCTION

The market for corporate debt playsa crucial role in the global financial system by providing funding to the real econ-

omy. However, little is known about the investment behaviour in the secondary corporate bond market. Who buys

corporate bonds when yields are rising and who is on the other side of the trade? And how do different investortypes

c

2019 Bank of England. Financial Markets,Institutions & Instruments c

2019 New YorkUniversity Salomon Center and Wiley Peri-

odicals, Inc.

Financial Markets,Inst. & Inst. 2019;28:347–379. wileyonlinelibrary.com/journal/fmii 347

348 CZECH ANDROBERTS-SKLAR

react in times of market-widestress? For example, if large investors were to behave in a ‘procyclical’ manner - i.e. selling

bonds as prices fall and buying as prices rise - theymight amplify yield moves, potentially causing markets to overshoot.

Sharp and sustained falls in corporate bond prices may reduce the ability of some companies to raise finance, which

could affect their investment decisions and potentially threaten their solvency. Understanding the behaviourof end

investors has become even more pressing since the financial crisis. Dealer banks have decreased their inventoriesof

corporatedebt by at least 50% (Dick-Nielsen & Rossi, 2019), leading to an increased cost of immediacy and heightened

potential for the investment decisions of large investorsto move market prices.

Agrowing theoreticalliterature analyses the trading behaviour of financial institutions in fixed-income assets. Rajan

(2006) shows that the compensation structure of delegated investment managers generatesstrong incentives to take

additional hidden risks. Importantly for our paper,these perverse incentives equally apply to investment managers of

mutual funds, hedge funds or insurance companies. Rajan (2006) also shows that the underlying incentives to ‘reach

for yield’ are even stronger in a low interest rate environment. Feroli,Kashyap, Schoenholtz, and Shin (2014) suggest

that this return chasing can reverse sharply during market‘tantrums’ (such as the ‘taper tantrum’ in 2013), resulting in

a bond sell-off by delegated investors. By contrast,banks become ‘illiquidity seekers’ in stress periods, as proposed by

Diamond and Rajan (2011). Hanson, Shleifer,Stein, and Vishny (2015) show that banks have a comparative advantage

at holding illiquid fixed-incomeassets with substantial price volatility. ‘Sticky deposits’,protected by deposit insurance,

serve as a relatively stable source of funding compared to shadow banks. Funding fragility is particularly pronounced

for asset managers, because common redemption policies create first mover advantages and strategiccomplementar-

ities for fund investors (Chen,Goldstein, & Jiang, 2010).

We provide novel empirical evidence for these theoretical predictions. Thus far,empirical research has been con-

strainedby a lack of comprehensive trade data in many markets. We are able to fill this gap in the literature with propri-

etary transaction-level data from the Financial Conduct Authority’s Zen database. The unique feature of the dataset

is that it includes the identity of both counterparties, allowing us to analyse differences in trading behaviour across

institutions. The data cover all sterling corporate bond trades for firms regulated in the UK, or branches of UK firms

regulated in the EEA. We combine the transactiondata with a hand-collected classification of firms into broad investor

types. Finally, we match our data to publicly-available information on the corresponding bonds such as the date of

issuance or the remaining time-to-maturity.

Our main contribution to the literature is to analyse the heterogeneity in investmentbehaviour on a relatively high

frequency and across different investor types: dealer banks, non-dealer banks, insurance companies, hedge funds and

asset managers. We examine investorbehaviour in the sterling corporate bond market by regressing the logarithm of

buy and sell volume on the lagged change in bond yields, controlling for a rangeof factors. Over the period 2011-2016,

we find that insurance companies, hedge funds and asset managers, on average, significantly increase purchases and

reduce sales of sterling corporate bonds following an increase in corporate bond yields. Dealer banks clear the mar-

ket byreducing purchases and increasing sales. This result is robust across various regression specifications, including

changing the dependent variable to number of trades, and using the change in corporatebond spreads rather than the

change in yields.

However, we find that the average behaviour reverses in times of stress. For example, during the 2013 ‘taper

tantrum’,asset managers sold more and bought less in response to a rise in yields, potentially amplifying yield changes.

Asset managers mayface incentives to behave procyclically in stress periods given their exposure to short-term bench-

marks and redemption risk (see, e.g., Buffa, Vayanos,& Woolley, 2017; Feroli et al., 2014; Goldstein, Jiang, & Ng, 2017).

At the same time, dealer banks were net buyers, thereby clearing the market. These findings are consistent with the

theoretical predictions of Diamond and Rajan (2011) and Hanson et al. (2015) that banks behave countercyclically

during stress periods in fixed-income markets. More generally,we find that when the VIX is high (in the top decile),

asset managers respond much less to an increase in yields, only slightly increasing their purchases.

Wealso perform a cross-sectional analysis of the relationship between bond characteristics and investor behaviour.

For dealer banks, insurers and asset managers, the coefficients on the change in yields increase with the bond residual

CZECH ANDROBERTS-SKLAR 349

maturity. With regard to credit quality,we find the largest magnitudes for high yield bonds, followed by investment

grade and unratedbonds.

What drivesthe heterogeneous behaviour across different investor types? In the US corporate bond market, ‘reach-

ing for yield’ has been cited as a key driver of investor behaviour (see, e.g., Becker & Ivashina, 2015; Choi & Kron-

lund, 2017; Kacperczyk & Schnabl, 2013). Therefore, we explorethe heterogeneity in the ‘reaching for yield’ behaviour

across investor types in the sterling corporate bond market. We construct three ‘reaching for yield’ factors to cap-

ture buying or selling of i) higher yielding lower rated bonds, ii) higher yielding longer maturity bonds, or iii) higher

yielding bonds within the same credit-quality and maturity bucket (Choi& Kronlund, 2017). We regress the logarithm

of buy and sell volume on these factors and find statistically significant coefficients on all three factors for hedge

funds and asset managers. Therefore, both investor types tilt their portfolios to higher yielding bonds, either within

a given credit risk and maturity bucket, or by pushing further along the credit and duration risk spectra. Consistent

with Becker and Ivashina (2015), we find that insurance companies primarily ‘reach for yield’ by favouring higher

yielding bonds within the same credit quality and maturity bucket, and to a lesser extent through longer maturity

bonds.

For robustness, we perform additional tests to account for different measures of trading activity or bond

performance.1First, we alter the dependent variable by using the number of trades or the net trading amount. Sec-

ond, we also change the main independent variable by using end-of-dayyields or yield spreads. Third, we obtain results

for a reduced sample, excludingsmaller retail sized trades (<£100,000). Our results remain statistically significant and

economically meaningful throughout these different specifications. Moreover,the economic magnitude is even larger

when we filter out retail-sized trades, suggesting that our headline results are relatively conservative due to the con-

sideration of these smaller trades.

Our results have important financial stability implications. First, the procyclical behaviour of asset managers in

times of stress could amplify asset price moves and increase volatility.Such disruption to the corporate bond market

could have an adverse impact on real economic activity given the critical function this market performs in directly

providing lending to the real economy. Second, ‘reaching for yield’ behaviour could reflect responses to regulatory

factors and may also provide evidence of an excessivecompression in risk premia, possibly creating vulnerability to a

correction.

2RELATED LITERATURE

There is a growing theoretical literature on the trading behaviour of financial institutions in fixed-income markets.A

significant part of this literature focuses on banks (see, e.g., Diamond & Rajan, 2011; Greenwood, Landier,& Thesmar,

2015; Shleifer & Vishny,2010). Hanson et al. (2015) show that traditional banks hold illiquid bonds during stress peri-

ods because of their relatively stable source of funding when compared to shadow banks. Diamond and Rajan (2011)

confirmthat banks load up on liquidity risk to profit from high returns when the depressed value of illiquid assets recov-

ers. Importantly,the incentives and strategies of bank managers deviate significantly from those of delegated portfolio

managers. Rajan (2006) shows that the compensation structures of investment managers (across different investor

types) generate perverse incentivesto add hidden risks. Guerrieri and Kondor (2012) confirm that delegated portfolio

managers tend to underestimate tail events and therefore often shift to riskier assets. However, as investorsadjust

their risk assessment, they engage in a flight-to-safety and markets become increasingly fragile (Gennaioli,Shleifer, &

Vishny, 2012). In their model of the corporate bond market,Baranova, Coen, Lowe, Noss, and Silvestri (2017) show

that redemptions from open-ended investment funds and subsequent corporate bond sales can materially increase

spreads.

On the empirical front, our approach is closely related to two papers that provide insights into the investment

behaviour of different investor types in debt securities for the German market (Abassi, Iyer,Peydró, & Tous, 2016;

Timmer,2018). Using quarterly holdings data of financial institutions in Germany, Timmer (2018) shows that banks and