European Bond ETFs: Tracking Errors and the Sovereign Debt Crisis

• Date: 01 November 2014
• DOI: http://doi.org/10.1111/j.1468-036X.2012.00649.x
• Published date: 01 November 2014

European Financial Management, Vol. 20, No. 5, 2014, 958–944

doi: 10.1111/j.1468-036X.2012.00649.x

European Bond ETFs: Tracking Errors

and the Sovereign Debt Crisis

Mikica Drenovak

Faculty of Economics, University of Kragujevac, Serbia

E-mail: mdrenovak@kg.ac.rs.

Branko Uroˇ

sevi´

c

Faculty of Economics, University of Belgrade, Serbia and National Bank of Serbia

E-mail: branko.urosevic@cif.co.rs.

Ranko Jelic

University of Birmingham, Business School, Birmingham, B15 2TT,UK

E-mail: r.jelic@bham.ac.uk.

Abstract

This study examines the tracking performance of 31 eurozone sovereign debt

exchange traded index funds (ETFs) during 2007–2010. The tracking performance

is assessed by four different tracking error models. Overall, funds underperform

their respective benchmarks. Active returns (net of fees) vary substantially (from

+46.74 to −30.36 basis points) and are of considerable economic interest.

The significant differences in the performance of swap-based and in-kind funds

highlight the importance of appropriate (e.g. correlation vs. cointegration based)

metrics requiredfor the assessment of funds adopting different replication methods.

We also document important changes in the tracking performance due to the

changing characteristics of EU sovereign bonds since the start of the sovereign

debt crisis.

Keywords: exchange-traded funds,tracking errors,fixed-income,sovereign debt

crisis

JEL classification codes: D81, E43, G15, G24

The authors are grateful to three anonymous referees and John Doukas (the editor) for

valuable comments. We thank Sergey Gelman, Drago Indjic, Miles Kumaresan, Carsten

Sprenger, FelixGoltz, Chien-Hui Liao, Jose Garcia-Zarate, and Florian Bitsch, participants at

the European Financial Management Symposium on Alternative Investments,Toronto (April,

2011), and seminars at the ICEF- Moscow (May, 2011) and National Bank of Serbia (June,

2011) for their valued input. Mikica Drenovak and Branko Uroˇ

sevi´

c gratefully acknowledge

the support of the Serbian Ministry of Science and Technology, Grant OH 179005. The views

presented in this paper are those of the authors and do not necessarily represent the views of

our respective institutions.

C

2012 Blackwell Publishing Ltd

© 2012 John Wiley & Sons Ltd

European Bond ETFs 959

© 2012 John Wiley & Sons Ltd

1. Introduction

In times of market distress and high uncertainty, investors typically rebalance their

portfolios towards less risky and more liquid assets (Longstaff, 2004). The ‘flight

to quality’ associated with the recent financial crisis spurred the development of the

eurozone sovereign debt exchange traded funds (ETFs), which today represent one of

the fastest growing segments of the overall European ETF market. For example, in 2009

these products gathered $6.2bn in new cash, which represented 7.5% of total inflows in

all European ETFs.1The sheer number of such funds, their liquidity, and the significant

total assets under management make them one of the most important new investment

classes.

Despite their increasing importance, there is a paucity of research on the European

EFTs and the analysis of ETF tracking errors remains a widely misunderstood and

frustrating process for investors (Flood, 2010a).2Drenovak and Uroˇ

sevi´

c (2010) is the

only previous study on the performance of EU sovereign ETFs. The authors discuss

institutional issues related to the eurozone sovereign debt ETF market and perform a

preliminary ranking of funds based on the standard correlation-based tracking error

method. In this paper, we examine the tracking performance of 31 eurozone sovereign

debt exchange traded index funds between January 2007 and December 2010. We

contribute to the existing literature by examining the tracking performance using four

different models, including cointegration analysis. We further present novel evidence on

the determinants of tracking errors using panel data analysis and consider the impact

of the tracking method (i.e. in-kind vs. synthetic replication) and the financial crisis on

tracking errors.

The new evidence on the tracking performance is particularly important in the

European context for the following reasons. First,since the collapse of Lehman Brothers,

several EU governments have implemented various measures aiming to stabilise their

troubled financial systems, which were further followed by the divergence in credit

spreads of EU countries.3Before the crisis, the risk associated with euro sovereign bond

indexes was lowand almost entirely attributable to the yield curve risk. During the crisis,

the risk of euro sovereign indexes rose sharply. Notably, sovereign bonds from so-called

peripheral EU countries such as Belgium, Greece, Italy,Ireland, Portugal and Spain have

become more akin to corporate bonds.4The above changes in the European bond market

are expected to have an important impact on the tracking performance of the sovereign

ETFs.

1By mid 2009, there were 753 registered European exchange traded funds (ETFs) with assets

under management (AUM) of overUS$ 183 bn. The propor tion of fixed income ETFs in the

total ETF assets grew from 5% in 2003 to more than 25% in 2009 (calculations made by the

authors, based on data from Barclays Global Investors (2009), Flood (2010b) and iShares

(2010)).

2Edhec (2009) reports that only 50% of European investors see ETFs’ tracking quality as

fairly good.

3We take the end of August 2008 (i.e. the collapse of Lehman Brothers) as the beginning of

the sovereign debt crisis.

4The changes are sometimes described as the new risk and return paradigm in the bond

market (see Nomura, 2011; p.80).

Mikica Drenovak, Branko Uroševic´ and Ranko Jelic

960

© 2012 John Wiley & Sons Ltd

Second, a new debate has recently emerged on the merits of synthetically structured

(i.e. swap-based) ETFs compared to those that use the underlying assets to track the

benchmarks (i.e. physically-based or in-kind funds). The proponents of physically-

based funds highlight the counterparty risk that exists with synthetically structured

funds, while the proponents of synthetically structured funds would argue that swap-

based funds deliver the lowest tracking errors.5Importantly, the largest European ETF

providersf avourdifferent structures. For example, the UK market predominantly consists

of physically-based while German and French markets are dominated by swap-based

ETFs. Given this segmentation of the European market and substantial differences

within the individual product structures, it is important to examine the various types of

ETFs and to compare their tracking performance.

Bond indices are expected to be less volatile than major equity indices. They also tend

to have far fewer constituents than leading equity indices (e.g. S&P). It is, therefore,

expected that passively managed ETFs tracking sovereign bond indices should exhibit

consistently low tracking errors similar to their total expense ratio (TER). Our results,

however, show substantial levels and variations in tracking errors for passively managed

European sovereign bond ETFs. The tracking errors are higher than those reported for

US Treasury bond ETFs (Houweling, 2011) and US passive equity ETFs (Blume and

Edelen, 2004). In some cases tracking errors are more than three times higher than their

TERs and are, therefore, more in line with results for corporate bond ETFs reported in

the previous literature. Importantly, some funds (e.g. Lyxor) consistently outperformed

benchmarks. Both univariateand our panel data analysis reveal a significant deterioration

of the sample ETFs’ tracking performance during the crisis period.

We further document that, as a result of the sovereign debt crisis, credit risk

considerations (i.e. credit default swaps (CDS) of index constituents) are increasingly

important for the tracking performance of sovereign bond ETFs. We also find strong

evidence for the importance of volatility of underlying indexes, duration, the replication

method, bid-ask spreads, TER, and ETFs’ size for the tracking performance.6Overall,

our results suggest that producing ETFs which genuinely match performance of

European sovereign bond indexes has become more complex in recent years. Each

ETF should, therefore, be assessed on its own merits based on error metrics suited to

fund characteristics and target indexes.

The remainder of the paper is organised as follows. In Section 2 we present important

characteristics of euro sovereign debt ETFs and their underlying indexes. Section 3

discusses relevant literature and motivates the hypotheses. Section 4 describes the

data and provides sample descriptive statistics. Section 5 presents the methodology

for assessment of tracking performance. The results of our sample ETFs’ performance

analysis are discussed in section 6. Section 7 presents the results of our panel data model

for determinants of the tracking performance. In Section 8, we test for robustness of our

results. Section 9 concludes.

5Controversially, the synthetically structured ETFs have been mentioned as a possible new

source of systematic risk by the Financial Stability Board, the International Monetary Fund,

the Bank for International Settlement and the Securities and Exchange Commission (SEC).

See Ramaswamy (2011), International Monetary Fund (2011), Financial Stability Board

(2011), Aboulian (2011) and The Wall Street Journal (2011).

6Our results are robust to use of data from different providers, alternative proxies, and

alternative panel models specifications.