See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/284113894

ETFs Performance Europe - A Good Start or Not?

Article  in  Procedia Economics and Finance · December 2015

DOI: 10.1016/S2212-5671(15)01346-5

CITATIONS READS

2 224

1 author:

Simona Mihai Yiannaki

European University Cyprus
32 PUBLICATIONS   93 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Environmental medicine: social and medical aspects View project

ICT ENTREPRENEUR View project

All content following this page was uploaded by Simona Mihai Yiannaki on 04 January 2016.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com

ScienceDirect
Procedia Economics and Finance 30 (2015) 955 – 966

3rd Economics & Finance Conference, Rome, Italy, April 14-17, 2015 and 4th Economics &
Finance Conference, London, UK, August 25-28, 2015

ETFs performance Europe - a good start or not?

Simona Mihai Yiannaki*
Associate Professor, European University Cyprus, Diogenes 6 str,2404 Engomi, Nicosia, Cyprus

Abstract

Under the premises that the U.S. Exchange Traded Funds (ETFs) hold over 70% of the (7)V¶World market, it seems that the
European ones have been either under-researched or less demanded. This study provides some insights into the performance of two
ETFs hubs, holding over 80% of the European ETFs activity, namely those operating in Luxembourg and Ireland, due also to their
tax similarities. Following an updated literature review on the topic, the paper compares these two ETFs hubs by using secondary
data publicly available, interpreted under a framework of previously identified performance methods: Tracking EUURU-HQVHQ¶V
alpha and Modigliani- M2 measure of performance. This methodology completes the descriptive statistics analysis, while aiming
at answering two hypotheses. The first hypothesis states that the Tracking Error of ETFs compared to their benchmark or market
indexes equals zero, which is confirmed by the study. The second hypothesis suggests that these particular ETFs do not present
significant alphas, which is partially confirmed. Moreover, the second hypothesis is tested not only against various features of these
IXQGV µbenchmarks, but also from risk measurement perspectives, while employing correlation significance between the two
countries ETFs. Overall, it appears that from the risk adjusted performance perspective, the ETFs domiciled in Luxembourg
outperform the Irish ones, leading also to potential M&As in this industry.

© 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
‹ 2015 The Authors. Published by Elsevier B.V.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-reviewunder
Peer-review under responsibility
responsibility of IISES-International
of IISES-International Institute
Institute for Social
for Social and Economics
and Economics Sciences.
Sciences.

Keywords: ETFs, -HQVHQ¶VDOSKD0RGLJOLDQLPHDVXUH M2, performance, risk adjusted performance, Tracking Error

* Corresponding author. Tel.: +357 22713197;

E-mail address: S.Mihai@euc.ac.cy

2212-5671 © 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of IISES-International Institute for Social and Economics Sciences.
doi:10.1016/S2212-5671(15)01346-5
956 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

1. Introduction

As the euro crisis is followed swiftly by economic convergence for a European fiscal union in order to balance
monetary policies which seem to have damaged several areas of previous economic growth in the Eurozone, a good
escape for investors would be to look not only in emerging economies, but also in alternative investments.
Several investment strategies of High Growth Markets (HGM), namely from the BIRC countries, followed by
Mexico, Singapore and South Korea, still seem shook up by financial PDUNHWV¶ high volatility, thus for European
investors searching for high returns and diversification, financial products alternatives become a must.
Considering that the ETFs looks under-researched (Blitz et al. 2012), with few studies on European ETFs suffering
from underperformance compared to their American competitors which hold over 70% of the World ETFs market,
there is still high interest in these funds, as alternative to traditional investments.
Since their introduction in 1993, as financial innovation, ETFs popularity has surged, while stock, bonds,
commodities and currency index tracking ETFs offered acceptable diversification in terms of asset classes, region and
maturities. Innovations on ETFs include leverage vs. non leveraged, synthetic, reverse funds with double or triple
exposure to an index.
The European ETFs market, especially through Undertakings for Collective Investments in Transferable Securities
(UCITS) became the second largest after the American one (Deutsche Bank, 2014), and it is now governed by the EU
directives for UCITS.
Meanwhile, Luxembourg and Ireland hold over 80% of European ETFs in terms of domiciliation
(PricewaterhouseCoopers, 2013), and follow the regulations of the Directives for UCITS, the Markets in Financial
Instruments Directive (MiFID) and the Alternative Investment Fund Managers Directive (AIFMD). Both countries
tax regime is considered similar, with no tax imposed on the fund level and no tax for nonresidents (Carne, 2012),
while also double-taxation treaties make these countries attractive for international investors in terms of dividends
repatriation.
Both countries have English speaking business environment, attractive for international investors, and at the same
time the ETFs are exempt from the proposed Financial Transaction Tax (FTT) suggested by the European Commission
in 2011 (European Commission, 2015).
However, of interest for this paper is the performance of the European ETFs domiciled in these two largest fund
hubs, and a comparative approach is taken looking at tKHDSSURSULDWHYDOXDWLRQPHWKRGVRI(7)V¶SHUIRUPDQFH, by
testing two hypotheses:
x H1: The Tracking Error of ETFs in Luxembourg and Ireland is equal to zero; and
x H2: Luxembourgish and Irish ETFs do not display significant alphas.
The methodology used includes on one hand the pervasive quantitative methods of Tracking Error (TE) and risk-
adjusted performance evaluation DQG RQ WKH RWKHU KDQG WKH -HQVHQ¶V alpha coefficient and expense ratio, as
management performance offering insights of qualitative nature. The sources employed are: Reuters, Bloomberg,
Morningstar, and Yahoo Finance websites, as databases (daily fund returns at closing values and daily index levels)
collected for the period 2008-2014 IRU  IXQGV LQ HDFK RI WKH WZR MXULVGLFWLRQV DQG XVLQJ 6WDWD Š DV VWDWLVWLFDO
software. As part of the research framework, the paper evolves around the following figure:

Fig. 1.Research Framework- Author sources

 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966 957

2. Literature Review

Most of the literature is focused on the ETFs of the United States, where performance research tends to eye the
ETFs tracking only by the major indexes VXFKDV'RZ-RQHV,QGXVWULDO$YHUDJH6WDQGDUGVDQG3RRUV¶DQGIHZ
on EURO STOXX50.
The choice of analyzing WKHVHWZRFRXQWULHV(7)V¶SHUIRUPDQFHGHULYHVIURPWKHIDFWWKDWXQGHUDJOREDOL]HGPDUNHW
threaten by increased debt proliferation, many investors explore under an apparently safer but of higher yield
investment, more diversified portfolios, beyond those of only stocks and mutual funds. At the same time Irish and
Luxembourgish ETFs benefit of almost identical tax regimes (Carne, 2012).
Investors have certain strategies in mind and are normally aware that over a long term period of time, risk adjusted
annual returns on hedged foreign stock portfolios have shown volatilities similar to risk adjusted returns to unhedged
foreign stock portfolios. Thus, it becomes a challenge to invest in high liquidity funds offered by the Irish and
Luxembourgish ETFs, which are listed on various stock exchanges, such as the London Stock Exchange.

ETFs by definition represent a basket of securities, which together provide the investors an immediate
diversification and exposure to many indexes. They are managed as any fund on a daily basis for the purpose of
tracking of an index, their portfolio is maximized under existing and forecasted market opportunities, and hence their
performance depends on the securities selection and risk reduction through diversification (Markovitz, 1952, and
Jensen, 1968).
Based on various identified literature results, the performance valuations of funds are separated in three areas of
concern applicable to ETFs:
x Fund return and performance persistence
x Index replication and Tracking Error
x Risk Adjusted Portfolio Performance

Overall, the literature provides mixed results in the sense that the European ETFs tend to underperform their
benchmark indexes in comparison to the U.S. hosted funds, either because of regulatory differences or the use of
expense ratios and dividend withholding taxes (Blitz et al. 2012). In other empirical research the Tracking Error seems
one of the main performance measures to test the liquidity of the underlying securities in a given index (Buetow and
Henderson, 2012).

2.1 Fund Returns and Performance Persistence

According to Blitz et al. (2012, p. 650), index-WUDFNLQJIXQGVLQWKH86ILQGUHWXUQV³SUHGLFWDEOHZLWKDhigh degree

RIDFFXUDF\´. Elton et al. (2004) confirms such findings due to the fact that these funds hold similar assets with same
proportions, though performance differences persist in the case of funds which track the same index, based on the past
values and expenses.
The opposite results occur in the regional comparisons via index-tracking (Blitz et al. 2012), where expense ratio
cannot easily justify the fund performance differences among European, U.S. and Japanese ETFs, where European
underperformed the U.S. ones, and the U.S. ones underperformed the Japanese ones. The reason resides in examining
dividend taxes and reinvestment policies, beyond the area of expense ratio. For example, the withholding tax in the
U.S. exceeded the Japanese one by 23%, indicating the U.S. underperformance vs. Japan, where dividends can be
reinvested after taxes. Noticing the underperformance of the European funds vis-j-vis the U.S. ones regarding the
dividend taxation, one can question the use of synthetic vs. direct replication of indexes. In brief, derivative products
can achieve similar exposures and help avoid dividend payments, including any associated withholding taxes.
Other studies (Adjei, 2009) found no significant differences in the performance attributed to expense ratios
variation over longer periods of time, while a study on over 34 mutual funds (Sharpe, 1966) found no substantial proof
that such trends persisted. A two-period return consistency research related to the size of ETFs found that large size
ETFs exhibited a higher performance persistency than smaller ones, while on a multi-period, a medium size ETFs
combination offered better results. In a study by Soe (2014) on S&P Dow Jones Index on a sample of 687 U.S. equity
mutual funds, over a five-year horizon, it appeared both likely that top-performing funds became worst performing,
958 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

and vice versa. To summarize on these aspects, historical data needs to be combined with additional predictors apart
from region, size, expense ratio and overall risk profile in such evaluations.

2.2 Index replication and Tracking Error

In a research by Buetow and Henderson (2012), the optimal portfolio allocation tends to consider risk and return
characteristics of assets included in the portfolio, but some assets may not be available to all investors due to large
direct holding costs. Thus, ETFs, as substitutes to regular mutual funds (Agapova, 2011) can provide investors with a
broader diversification and inclusion of previously unavailable assets.
Regular investment funds are priced on the basis of their NAV (net asset value), thus for an index-tracked fund,
the NAV represents the proportional value of securities in the corresponding index, despite the complicated process
of the daily net asset valuation, resulting in a unit price. However, for ETFs, their share price depends on the value of
the IXQG¶VDVVHWZKLFKVLJQLILFDQWO\IOXFWXDWHVDURXQGLW$VSHU3HWDMLVWRLQWKH86(7)V¶DQDO\VLVLWVHHPV
that price differences would not persist over time due to the arbitrage mechanism, which relates to both :
x The increase difficulty in portfolio duplication at reasonable costs and
x The fact that the NAV valuation may not reflect the current value of ETFs portfolio, especially in those
funds holding international and illiquid equities.
At its turn the NAV calculation can provide prices that diverge due to the closing times of the international
exchanges and the intra-day exchange rate modifications.
Therefore, these price differences give room for calculating the tracking error (TE) of any ETF, and by definition
(Vardharaj et al. 2004), the TE are reported as standard deviation percentage difference between the price behaviour
of a position or a portfolio and the price behaviour of a benchmark. Hwang and Satchell (2001) found that ex-post TE
tends to be larger than predicted TE, while TE should not be considered by itself as performance indicator (Buetow
DQG+HQGHUVRQ%HVLGHVWKH(7)¶VUHWXUQVDUHDOVRLQIOXHQFHGE\VXSSO\DQGGHPDQGRIWKHLUVKDUHVWUDGLQJ
volumes, and systematic risk. In this case, risk performance measures such as Treynor, Sharpe, and Modigliani
measure M2 are widely used (Arugaslan and Samant, 2014<HWLQWKH&$30PRGHOLQWKHHYDOXDWLRQRIWKHIXQG¶V
management performance, these may tend to be overly pessimistic if the impact of taxation is not taken into
consideration (Blitz et al. 2012).
The Tracking Error of an ETF stems from two sides:
x One is the NAV- TEPDSSLQJWKHHIILFLHQF\RIWKHIXQG¶VPDQDJHPHQW and
x The second is represented by the price fluctuations around the NAV, which incorporates the demand for the
fund (Buetow and Henderson, 2012).
Therefore, ETF prices can generate returns that differ from the underlying index. A similar situation is observable
from liquidity point of view, when ETF confirms the link between the liquidity of a fund and that of the underlying
securities (Petajisto, 2013), while ETFs that track indexes with less liquid securities present significantly lower
correlations with the benchmark ( Buetowand Henderson, 2012).
According to Charupat and Miu (2013), the ETFs¶ TE depends on five factors:
x The management fees that tend to be lower than those of mutual funds, but affect the expense ratio, thus
the larger the TE the less performant the ETF compared to the benchmark (index);
x The transaction costs are not part of the expense ratio mentioned above; these costs tend to be higher the
more volatile the indexes, hence the higher the TE;
x Indirect replication and representative sample of the underlying securities usually with a lower transaction
costs can increase the TE, while direct replication tends to lower it, since the securities in an index and in
an ETF are the same;
x Investors in ETFs are entitled to dividends, hence they have to be considerHG DV SDUW RI WKH IXQG¶V
performance;
x For leveraged or inverse ETFs the compounding of daily returns lead to TE if exposure is not adjusted by
end of the day.
Generally, the TE is a result of many components such as those presented above, therefore it becomes difficult to
dissociate among their importance in performance interpretation.
 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966 959

2.3 Risk Adjusted Portfolio Performance

In studies of behavioral finance, risk and performance are correlated, while the modern portfolio theory provides
several measurement tools of quantitative nature for portfolio optimization.
Treynor (1965) suggested that the performance measure adjusted WKHIXQG¶VH[FHVVUHWXUQIRUV\VWHPDWLFULVN, thus
investors can rank the funds based on the compensation obtained at every risk appetite level they prefer.
Sharpe (1966) studied 34 mutual funds and replaced the market risk with beta coefficient, utilizing overall portfolio
volatility, thus helping investors settle for the reward provided for bearing the risk.
Later on Brian M. Rom and Sortino (2010) identified the reward-to-variability ratio or the Soretino ratio, looking
this time at the downside of standard deviation of the portfolio, which penalizes manages only for the returns falling
below a specified required rate of return.
Michael C. Jensen (1968) has built on the work of Treynor and Sharpe and identified the JHQVHQ¶s alpha coefficient
as an absolute measure of performance, yet studies indicate negative alphas for ETFs, irrespective of their management
strategies (Rompotis, 2009).
Modigliani and Modigliani (1997) measure based on CAPM allows investors to easily estimate by how much any
given fund outperforms or underperforms its benchmark, when additional risk of the portfolio is incorporated.
More recently, Robet Shiller (2015) has looked into the element of cyclical adjusted price earnings ratio (CAPE)
for S&SP 500 index for potential long term investments on the U.S. equity sector. Professor Shiller extended the
CAPE measure to equity sectors and together with Barclays, designed the Shiller Barclays CAPE Europe Sector Value
Net TR Index, the basis of the new ETF (Smith , 2015).
In terms of derivatives the VIX indicator, introducHG LQ  E\ 'XNH 8QLYHUVLW\¶V 3URIHVVRU 5REHUW :KDOH\
(www.cboe.com) is a key measure of implied volatility, market expectations of near-term volatility derived from S&P
500 stock index option prices. Since VIX is meant to be forward looking, VIX futures were introduced in 2004, and
VIX options were introduced in 2006, in order to search the use of instruments with potential to diversify portfolios
in times of market stress. The short term, medium term and leveraged VIX are ETFs which stand out immediately in
the volatility space. They can eliminate the credit risk that many exchange traded notes (ETNs) carry, but at the same
time will make it vulnerable to TE, which ETNs do not exhibit.
To summarize the literature review findings, it appears there are notable gaps in the European ETFs industry, some
related to their underperformance compared to other international ETFs, yet the World ETFs studies are not conclusive
either on the performance persistence during longer periods, past returns, expense ratios and dividend taxes. Several
identified widespread methods in these funds performance are:
x index-tracking via TE, both important for managers and investors, yet a method which is strongly affected
by a multitude of factors, and which can vary greatly with the fund replication techniques and applied
strategies;
x risk adjusted performance based on various ration, such as Treynor, Sharpe, Sortino and the Modigliani
performance measure;
x management value performance, supported by Jensen.
In the following part of the paper, the research presents the methodology used in evaluating the two countries¶
ETFs environments, concentrating on the index replication via Tracking Error method and on risk adjusted
performance measures. Thereafter, testing the research two hypotheses related to TE and risk adjusted performance
will be discussed under the results analysis.

3. Methodology

The first part of the methodology is introducing the TE and Index replication strategies, and the second part is the
risk adjusted performance measure selected for this study. The sample selection utilized in this research is based on a
pool of 12 equity ETFs from Luxembourg and 12 ETFs from Ireland, which are traded on three major stock exchanges:
London Stock Exchange, Frankfurt Stock Exchange and Euronext, of which about 80% are traded on a daily basis on
two of these markets.
960 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

Their prices represent historical data and are publicly available from various databases such as Bloomberg,
Mornigstar, Reuters, Google finance and Yahoo finance. The convenience sample is collected on a period of 7 years
(2008-2014), so that the results include the activity before and after the events of the financial crisis. Instead of the
NAV, the actual µFORVHRIWKHGD\¶SULFHVKDYH been used to calculate returns.
Additional data includes risk free rates and underlying index prices used in determining the TE, the return
correlations and the risk-adjusted performance from the Bloomberg and Morningstar databases. However, the
examined funds are benchmarked against a variety of indexes, due to the fact that one market proxy does not reflect
technically the correctness of the choice for a single index.
Following the research methodology, the results interpretation will focus on the empirical research utilizing
descriptive statistics, TE , -HQVHQ¶VDOSKD and Modigliani measure- M2 analysis. 7KHVRIWZDUHXVHGLV6WDWDŠ

3.1 The Tracking Error and the Index replication strategies

The Tracking Error (TE) measure provides a proper way to assess passive funds capability to minimize it
(Vardharaj et al. 2004).
Frino and Gallagher (2001) use three methods of tracking error: the absolute difference between portfolio and
benchmark returns, the standard deviation of the differences between the same returns and the standard error of returns
regression analysis. The first two measures are more widely used (Rompotis, 2011, and Scozzari et al. 2013), while
last method tends to overstate the TE if portfolio and benchmark returns are not linearly related.
For the purpose of this research the second method will be used, as per Frino and Gallagher (2001), as it is widely
used in the definition of standard error for ex-post TE, allowing for interpretable calculation on any TE data frequency.

1 2
TE
n 1
¦
n
t 1

a ETF a ETF  (1)

where ƒ୉୘୊ is the difference between fund and index returns, called active return, defined as:

a ETF R ETF  R b (2)

Thereafter, besides calculating the TE deviations, the correlations between the index and the ETFs are interpreted
in order to find how good of a proxy the ETF is in replicating the benchmark return.

3.2 The Risk Adjusted Performance Measure

In line with these parameters the research makes use of two methods: RQHUHO\LQJRQ-HQVHQ¶Valpha and the other
on Modigliani measure- M2. To lead to the two measures, the methodology employs the intermediary formula of
Capital Asset Pricing Model (CAPM) (Lintner, 1965).
6HOHFWLQJWKH-HQVHQ¶VDOSKDPRGHODQGXVLQJWKHUHWXUQVFDOFXODWHGLQWKH&$30PRGHOWKHUHVXOWVREWDLQHGZLOO
determine the active return of the fund and if the fund managers outperformed the relative benchmark, by seeking
positive alphas+RZHYHUDFFRUGLQJWR5RPSRWLVLIWKHPDUNHWLVHIILFLHQWDQGWKHSRUWIROLRRI(7)¶VLVSULFHG
in a proper manner, the expected alpha should not be different than zero, whereas ex-post alpha is:

D >
R p  R f  E p Rm  R f  @ (3)

Calculating each alpha value is needed in ranking the funds management performance, yet some limitations apply,
such as:
x alpha does not adjust for the nonsystematic risk, an element that investors may be searching for while
following various levels of risk strategies (Aragon and Ferson, 2006);
 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966 961

x well diversified ETFs are not rewarded by alpha;

x an assumption is made for funds with beta coefficient higher than one, that the funds borrow at risk free rate,
fact which is not realistic;
x further verification would be needed via Modigliani measure M2, as a new method applied in portfolio theory.

Modigliani measure M2 has the advantage of considering the inherent risk of the portfolio apart from systematic
risk measured by the beta coefficient.
Vm
M2 R  R   R f (4)
Vp m f
where:
ım- standard deviation of the returns of the market, as a proxy;
ıp- standard deviation of the returns of the ETF.

Modigliani¶V performance measure M2 offers interpretation for the relative performance of the funds by indicating
that the highest percentage represents the best risk adjusted performance. Nevertheless, M2 matches his research when
dealing with stand-alone ETFs as well, it is based on standard deviation as a risk measure and assumes normal
distribution. However, skewed return distributions may misinterpret M2 values by either understating or overstating
the results, but the aim of the paper is to look for results with normality test of returns.
Following the research methodology, and the literature review preceding it, the research leads to testing two
possible hypotheses:
H1: The Tracking Error of ETFs in Luxembourg and Ireland is equal to zero; and
H2: Luxembourgish and Irish ETFs do not display significant alphas.

4. Results Interpretation

In testing the two hypotheses, the study presents the results of running the calculations for several descriptive
statistics, and then the findings are compared to the results of the TE, the -HQVHQ¶VDOSKDDQG02.
While the selected 12 ETFs from each of the two countries track the same indexes, the replication strategies and
fund currencies can differ. The majority of funds use full replication strategy, while others employ indirect replication
or a mix of the two.

4.1 General Statistics interpretation

The data presented in Appendix A. shows the holding period return (HPR), the mean and standard deviation of the
convenience sample selected. Over the period 2008-2014, the results show that the observations of each ETF and
benchmark index which were monthly returns annualized, have produced positive returns, with best performance the
ETFs domiciled in Luxembourg, tracking FTSE 250, while the second best performant was the Irish fund tracking the
same index.
During the same 7 year analysis period, the benchmark index had a 73% HPR, thus both ETFs performed better
over the same period. Out of the top performers, most were from Luxembourg, while in the worst SHUIRUPHUV¶
category, most were Irish. Among the worst performers, European and Latin American indexes obtained negative
+35ZKLOHWKHVDPHIXQGV¶EHQFKPDUNVSURGXFHGQHJDWLYHUHWXUQVRYHUWKHVDPHSHULRGHJ(XUR672;;'LYLGHQG
30- 32.4%, MSCI Brazil index -51.4%).
Comparing the 1 year (2014) timeframe versus the 3 year period timeframe, while FTSE2150 index obtained
outstanding returns, the best performers on the long term remained the same, but on short term slight changes were
observed. Although the Latin American ETFs remained the least performant, followed by the European funds, the
ETFs tracking Japanese equity index had obtained the highest returns, while emerging markets index funds were
placed on the second position. The opposite was experienced in the worst performers category, with the European
equity funds stagnating and the Brazilian ETFs barely producing positive returns.
962 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

In terms of mean, minimum, maximum and standard deviation statistics over a period of 3 years and last year
(2014) the findings were:
x The Mean: there are large gaps within the last year and less on the 3-year basis, while all-time annual
Means range from -3.45% to 11.6%, with the lowest values occurring during the market crash of 2008,
depressing returns until 2009.
x The highest Max percentage of 35.2% was in Luxembourg, and the lowest Min of -29.4% in Ireland.
x The standard deviation over all-time period was relatively high, considering the underlying indexes track
international securities, usually highly volatile and collected over an unstable period of time. It appears
that funds with negative mean all time have the highest risk.
These statistics depict the Luxembourgish ETFs performing better on HPR, bearing similar high risks as the Irish
ones have, with limits on the Irish funds going for higher risk and higher losses, regardless the point in time. Based
on these statistics results, further evaluation of ETFs is needed to reflect risk appetite and replication performance.

4.2 Index Replication and Tracking Error

As a measure for quality, passive index tracking by ETFs is their ability to reproduce benchmark returns. The TE
is presented in Appendix B. By examining the data, it appears that on a shorter term, the 3±year tracking error fell
below 3% mark for half of the ETFs, while on the annualized TE over the 7-year research period Luxembourg ETFs
averaged 5%, while Irish around 4%. In over 80% of the cases, the TE for Luxembourg ETFs exceeded the Irish one.
Over a 3-year period the Irish funds averaged 2.7% and the ones domiciled in Luxembourg exceeded 3.5%. Over
the past year 2014, TE increased up to 2.9% in Ireland and over 3.9% in Luxembourg, while on the T-test, TE
differences between the funds in each of the two countries were not statistically significant (p>0.05). In calculating
the correlation, all funds indicate a strong positive correlation ranging from r=0.94 till 0.99, with a relatively lower
correlation between the ETFs and their benchmarks in international and emerging markets. The calculated correlation
coefficients post a strong positive linear relationship between the risk of the underlying benchmark/index and the TE
of the ETF, meaning that when index volatility raises so does the TE. Based on these findings it appears that hypothesis
H1 is rejected.

4.3 The ETF’s Jensens’s alpha and the Modigliani measure M2

In Appendix C. the data presents the basis for the interpretation of Jensen¶VDOSKDZKLOHLQ$SSHQGL['for the
M2 over a period of 3 and 7 years respectively, for the sample observations.
5HJDUGLQJWKH-HQVHQ¶VDOSKDLWFDQEHLQIHUUHGWKDWWKHH2 hypothesis is rejected only for one fund (p<0.01), DBXA
(Luxembourg), tracking the MSCI Europe Index, which outperformed the index by 0.26% on a monthly basis. Also
over a 3-year period, the H2 is rejected for other three funds DBXA and XMCX (Luxembourg) and MIDD (Ireland),
whereas the last two funds which track FTSE 2510 index managed to outperform the U.K. benchmark by a significant
0.8% per month. Yet, considering that most funds show insignificant alphas, it may be correct to assume that over a
longer period of time, alpha should average to zero value. Overall, the funds tracking indexes with equities in the
BRIC markets showed lowest alphas, failing to offer expected returns, best performing being the U.K. markets with
abnormally higher returns and statistically significant alphas.
Over a long time periodERWKFRXQWULHV¶(7)¶VDOSKDVDYHUDJHFORVHUWR]HURwith -0.4% for Ireland and 1.4% for
Luxembourg, while the 3-year alphas were -1.16% and 0.09% respectively. Yet, regardless the time period in 90% of
the cases, the Luxembourg ETFs exhibit higher alphas, but without a statistically significant difference between the
ETFs¶ alphas in both countries.
It has emerged that the funds which were underperforming in terms of TE and overall returns revealed high negative
alphas, though they were not statistically different from zero. This method seems to consistently help in identifying
underperforming ETFs.

Validating these findings against the results of M2 is the next final step in this research. According to M2 calculations
data analysis indicates that on a 3-year all European ETFs outperformed their benchmarks, those of Far East
outperformed, too, but with Korean equities tracking in line with the market. About 4 ETFs underperformed in their
 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966 963

own markets. The Irish ETFs were still outperformed by the Luxembourgish ones in 9 out of 12 cases. The same
appears to show in the 7-year Modigliani measure.
Over the 7-year the M2 measure calculations shows that about 80% of the ETFs did better than their benchmarks,
with some even over the 10% value, with the exception of the Brazilian and the European dividend paying equities.
Irish funds tracking Latin America were slightly under the zero value, and both two Irish and Luxembourgish ETFs
outperformed the U.K. market by 9.17% and 9.93% respectively. Within the European Monetary Union market, the
EURO STOXX 50 index funds from Ireland and Luxembourg beat the market by 2.62% and 3.38% respectively,
while the EURO STOXX Select Dividend 30 fund fell behind the benchmark with 2.4%. In 11 out of 12 cases in the
7-year timeframe the ETFs in Luxembourg outperformed their Irish counterparts.

Overall the findings suggest that over a period of 7 years and during the last 3 years, only one fund rejected
hypothesis H2, three other funds show statistically significant alphas, other funds partly confirming the H2, meeting
thus the findings of (Rompotis, 2011). From Modigliani measure perspective, the findings suggest that still
Luxembourg ETFs outperformed the Irish ones.

Despite some certain limitations of the research, such as: the selection of a convenience sample, the relative limited
timeframe, the data availability, frequency and variation, the selection of only equity ETFs and passive indexes, during
a financial distress period, with hiccups regarding liquidity and the IXQGV¶ access to borrowing and fluctuating interest
rates, still the research can bring some conclusive remarks on the European ETFs performance.
These are all important in further research regarding the risk and performance measures, alternative sources of
investments, whereas ETFs start picking up more than other mutual funds in the market.

5. Conclusions

The study provides insights into the two countries differences in terms of investment performance, despite the
financial crisis pertaining at the time of analysis, while serving as a guiding tool for European investors looking into
ETFs as alternative sources for obtaining financial benefits.
Innovation in the investment industry is a constant paradigm bridging the appetite for high returns with the level
of safety for investors. ETFs seem to have succeeded to grasp the LQYHVWRUV¶attention increasingly in the last years
also in the European markets, despite limited research in this region. However, based on the analysis extended to two
E.U. markets of Luxembourg and Ireland, where over 80% of ETFs are domiciled, it seems that:
x Regardless of the timeframe, the Tracking Errors are significantly different than zero, though no statistically
significant differences are observed between the two countries (Charupat and Miu, 2013).
x As per Buetow and Henderson (2012), it appears that the tracking of European indexes exhibits statistically
smaller Tracking Error than funds tracking international equity indexes. Also, a strong correlation is found
between TE and volatility, implying that ETFs with high volatility exhibit increased TE, but there is a weak
correlation between TE and trading volumes, meaning the less liquid the funds the higher the TE.
x Risk±adjusted performance measure shows ETFs with domicile in Luxembourg perform better than those
with domicile in Ireland, the H2 hypothesis is partly confirmed (Rompotis, 2011);
x Since Luxembourg performance seems better off, there could be possibility for further M&As also at ETFs
levels between the two countries (Mihai Yiannaki, 2013).
Overall, in conclusion, several portfolios allocations from ETFs domiciled in Luxembourg exhibit higher return
under equal weight conditions and a better risk management compared to the ETFs domiciled in Ireland.
964 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

Appendix A. Descriptive statistics of the selected ETFs domiciled in Luxembourg and Ireland

Ticker Country Benchm ark Holding Period Return Mean Std. Deviation
1Y 3Y All tim e 1Y 3Y All tim e Min Max 1Y 3Y All tim e
MIDD IE FTSE 250 Index 5.60% 60.80% 81.20% 6.00% 17.80% 10.90% -12.80% 10.20% 9.00% 10.70% 18.50%
IJPN IE MSCI Japan Index 30.30% 46.50% 39.70% 31.50% 14.50% 6.40% -15.60% 16.50% 14.20% 13.00% 15.20%
ISF IE FTSE 100 Index 4.70% 29.80% 31.70% 5.00% 9.60% 4.80% -21.50% 18.70% 8.00% 10.10% 15.10%
IEMM IE MSCI Emerging Markets Index 27.20% 9.10% 27.70% 27.90% 3.70% 5.00% -22.30% 20.10% 11.20% 12.20% 20.20%
IDWR IE MSCI World Index 5.80% 36.90% 22.50% 6.10% 11.70% 4.60% -16.50% 13.50% 7.50% 11.00% 18.40%
IMEU IE MSCI Europe Index 2.50% 26.60% 14.40% 2.90% 8.90% 3.90% -19.30% 18.00% 9.80% 12.00% 18.50%
EUE IE EURO STOXX 50 Index 18.10% 58.80% 10.50% 19.00% 17.70% 2.70% -12.30% 9.20% 13.00% 13.60% 19.40%
IKOR IE MSCI Korea Index -7.30% -3.40% 5.80% -6.70% 0.00% 6.60% -13.00% 8.40% 11.60% 15.40% 32.90%
IDTW IE MSCI Taiw an Index 15.00% 16.20% 3.30% 15.90% 5.90% 6.40% -14.30% 12.10% 12.70% 12.50% 25.90%
IBZL IE MSCI Brazil Index -14.20% -48.10% -51.40% -10.70% -16.90% -3.40% -29.40% 20.50% 28.60% 25.40% 33.90%
LTAM IE MSCI Emerging Markets Latin America 10/40 Index 11.40% -25.50% -23.40% 13.50% -8.00% -1.00% -22.40% 16.20% 20.00% 17.40% 24.00%
IDVY IE EURO STOXX Select Dividend 30 Index 2.20% 15.60% -31.80% 2.80% 6.10% -2.30% -21.50% 29.80% 11.30% 14.80% 24.00%
XMCX LU FTSE 250 Index 5.30% 61.60% 90.40% 5.80% 18.00% 11.60% -12.70% 10.20% 9.50% 10.90% 18.10%
DBXA LU MSCI Europe Index 19.10% 59.50% 55.50% 19.70% 17.40% 7.10% -19.90% 18.90% 10.80% 10.10% 16.40%
XJP LU MSCI Japan Index 32.40% 52.50% 53.70% 33.70% 16.10% 7.90% -18.30% 20.10% 14.80% 13.30% 15.00%
XEM LU MSCI Emerging Markets Index 29.80% 15.70% 46.80% 30.60% 5.70% 6.30% -13.50% 14.70% 11.60% 12.20% 20.70%
XMWO LU MSCI World Index 8.10% 44.50% 42.00% 8.30% 13.60% 6.80% -20.90% 25.50% 7.40% 10.00% 18.20%
XUKX LU FTSE 100 Index 5.20% 30.50% 39.20% 5.50% 9.90% 5.60% -14.90% 16.20% 8.50% 10.70% 14.80%
XKSD LU MSCI Korea Index 1.40% -0.50% 35.80% 1.90% 0.80% 9.00% -20.30% 18.70% 11.00% 14.30% 29.20%
XMTD LU MSCI Taiw an Index 18.50% 24.30% 24.50% 19.40% 8.50% 9.20% -19.10% 12.80% 12.50% 13.60% 26.30%
DBXE LU EURO STOXX 50 Index 18.10% 59.60% 16.80% 18.90% 17.90% 3.40% -19.70% 18.00% 12.50% 13.50% 19.00%
XLA LU MSCI Emerging Markets Latin America Index 14.20% -21.10% -18.00% 16.30% -6.20% -0.10% -23.10% 16.10% 19.70% 17.40% 23.50%
XD3E LU EURO STOXX Select Dividend 30 Index 1.40% 15.20% -32.40% 2.30% 6.00% -2.60% -26.00% 35.20% 13.70% 14.80% 23.70%
XMBD LU MSCI Brazil Index -12.80% -45.30% -43.10% -9.40% -15.70% -1.70% -28.30% 21.10% 27.90% 24.20% 33.40%

Appendix B. ETFs correlation and Tracking Error for Ireland and Luxembourg

Track ing Error

Be nchm ark Tick e r Country Corre lation 1Y 3Y All tim e Y
DBXE LU 0.99 1.63% 1.50% 2.99%
EURO STOXX 50 Inde x
EUE IE 0.99 1.78% 1.59% 2.65%
XD3E LU 0.97 5.29% 5.61% 5.61%
EURO STOXX Se le ct Divide nd 30 Inde x
IDVY IE 0.99 3.74% 3.27% 2.97%
XUKX LU 0.99 1.23% 0.91% 2.49%
FTSE 100 Inde x
ISF IE 0.99 0.96% 1.06% 2.08%
XMCX LU 0.99 1.32% 0.92% 2.35%
FTSE 250 Inde x
MIDD IE 0.99 1.22% 1.07% 2.25%
XMBD LU 0.97 11.91% 7.39% 7.83%
MSCI Brazil Inde x
IBZL IE 0.98 5.25% 4.58% 6.60%
XEM LU 0.94 2.59% 2.62% 7.10%
MSCI Em e rging Mark e ts Inde x
IEMM IE 0.99 1.68% 2.49% 3.31%
XLA LU 0.98 3.93% 3.33% 4.71%
MSCI Em e rging Mark e ts Latin Am e rica Inde x
LTAM IE 0.97 4.70% 3.70% 6.19%
DBXA LU 0.99 0.87% 0.96% 1.98%
MSCI Europe Inde x
IMEU IE 0.99 2.29% 2.45% 2.57%
XJP LU 0.94 6.04% 4.57% 4.97%
MSCI Japan Inde x
IJPN IE 0.95 5.85% 4.37% 4.84%
XKSD LU 0.94 5.54% 7.09% 10.25%
MSCI Kore a Inde x
IKOR IE 0.98 2.74% 3.01% 6.50%
XMTD LU 0.97 3.65% 4.65% 6.51%
MSCI Taiw an Inde x
IDTW IE 0.98 2.68% 3.27% 4.96%
XMWO LU 0.98 3.10% 2.19% 3.56%
MSCI World Inde x
IDWR IE 0.99 1.99% 1.95% 3.00%
 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966 965

Appendix C. -HQVHQ¶V¶DOSKDIRUWKHVHOHFWHG(7)VIURPLuxembourg and Ireland

3Y All time
ETF Market Proxy ȕ E(RETF) ȝETF ĮJ ȕ E(RETF) ȝETF ĮJ
Domicile: Luxembourg
XLA MSCI EM Index 1.24 4.50% -6.18% -10.26% 1.13 5.04% -0.13% -4.94%
DBXE MSCI EMU Index 1.04 15.76% 17.89% 1.87% 0.99 1.98% 3.41% 1.40%
XD3E MSCI EMU Index 0.86 8.72% 5.96% -2.55% 1 2.38% -2.59% -4.87%
DBXA MSCI Europe Index 0.98 13.76% 17.41% 3.24% 1.02 3.80% 7.11% 3.20%
XUKX FTSE All Share Index 1.02 7.88% 9.89% 1.87% 0.95 4.23% 5.64% 1.35%
XJP MSCI Japan index 1.06 15.32% 16.07% 0.66% 0.96 6.11% 7.86% 1.65%
XMCX FTSE All Share Index 0.95 7.35% 18.03% 10.01% 1.08 4.67% 11.59% 6.64%
XKSD MSCI Korea Index 0.85 0.02% 0.85% 0.83% 0.92 5.03% 9.03% 3.83%
XMBD MSCI BRIC Index 1.25 -5.00% -15.70% -11.22% 1.11 -0.51% -1.72% -1.21%
XMTD MSCI Taiwan Index 1.12 6.27% 8.50% 2.11% 1.03 6.35% 9.20% 2.70%
XEM MSCI EM Index 1.03 3.82% 5.72% 1.84% 1 4.58% 6.26% 1.61%
XMWO MSCI World Index 0.93 10.74% 13.61% 2.61% 0.99 4.40% 6.84% 2.35%
Domicile: Ireland
LTAM MSCI EM Index 1.23 4.46% -7.98% -11.96% 1.15 5.10% -1.02% -5.85%
EUE MSCI EMU Index 1.04 15.83% 17.70% 1.64% 1.02 2.00% 2.68% 0.67%
IDVY MSCI EMU Index 0.91 9.21% 6.08% -2.89% 1.03 2.43% -2.34% -4.66%
IMEU MSCI Europe Index 0.95 8.57% 8.94% 0.34% 0.99 3.79% 3.91% 0.12%
ISF FTSE All Share Index 0.97 7.50% 9.61% 1.98% 0.98 4.32% 4.82% 0.48%
IJPN MSCI Japan index 1.04 15.01% 14.49% -0.46% 0.98 6.20% 6.35% 0.14%
MIDD FTSE All Share Index 0.93 7.20% 17.79% 9.94% 1.11 4.75% 10.87% 5.86%
IKOR MSCI Korea Index 1.02 -0.07% 0.01% 0.08% 1.08 5.75% 6.56% 0.78%
IBZL MSCI BRIC Index 1.33 -5.35% -16.95% -12.20% 1.14 -0.55% -3.40% -2.87%
IDTW MSCI Taiwan Index 1.05 5.90% 5.94% 0.04% 1.03 6.33% 6.42% 0.08%
IEMM MSCI EM Index 1.04 3.84% 3.70% -0.14% 1.03 4.68% 5.02% 0.33%
IDWR MSCI World Index 1.04 11.98% 11.69% -0.26% 1 4.44% 4.61% 0.17%

Appendix D. Modigliani performance measure M2 for 3 and 7 years for ETFs in Luxembourg and Ireland
2
Index Fund Domicile Higher M
Underlying Index Luxembourg Ireland
EURO STOXX 50 Index 16.86% DBXE 16.60% EUE Luxembourg
EURO STOXX Select Dividend 30 Index 6.07% XD3E 6.17% IDVY Ireland
FTSE 100 Index 9.58% XUKX 9.79% ISF Ireland
FTSE 250 Index 17.04% XMCX 17.14% MIDD Ireland
MSCI Brazil Index -7.21% XMBD -7.41% IBZL Luxembourg
MSCI Emerging Markets Index 5.44% XEM 3.51% IEMM Luxembourg
MSCI Emerging Markets Latin America Index -3.91% XLA -5.11% LTAM Luxembourg
MSCI Europe Index 17.68% IMEU 9.19% DBXA Luxembourg
MSCI Japan Index 14.36% XJP 13.22% IJPN Luxembourg
MSCI Korea Index 0.86% XKSD 0.03% IKOR Luxembourg
MSCI Taiwan Index 7.29% XMTD 5.53% IDTW Luxembourg
MSCI World Index 14.23% XMWO 11.11% IDWR Luxembourg
2
M calculated on 3Y data.
2
Index Fund Domicile Higher M
Underlying Index Luxembourg Ireland
EURO STOXX 50 Index 3.38% DBXE 2.62% EUE Luxembourg
EURO STOXX Select Dividend 30 Index -2.40% XD3E -2.12% IDVY Ireland
FTSE 100 Index 5.78% XUKX 4.86% ISF Luxembourg
FTSE 250 Index 9.93% XMCX 9.17% MIDD Luxembourg
MSCI Brazil Index -1.24% XMBD -2.58% IBZL Luxembourg
MSCI Emerging Markets Index 5.92% XEM 4.85% IEMM Luxembourg
MSCI Emerging Markets Latin America Index 0.08% XLA -0.62% LTAM Luxembourg
MSCI Europe Index 6.92% IMEU 3.91% DBXA Luxembourg
MSCI Japan Index 7.72% XJP 6.18% IJPN Luxembourg
MSCI Korea Index 9.25% XKSD 6.08% IKOR Luxembourg
MSCI Taiwan Index 8.69% XMTD 6.17% IDTW Luxembourg
MSCI World Index 6.80% XMWO 4.56% IDWR Luxembourg
M 2 calculated on 7Y data.
966 Simona Mihai Yiannaki / Procedia Economics and Finance 30 (2015) 955 – 966

References

Adjei, F. (2009). Diversification, Performance, and Performance Persistence in Exchange-Traded Funds. International Review of Applied Financial
Issues and Economics, 1(1), 4-19.
Agapova, A. (2011). Conventional Mutual Index Funds Versus Exchange Traded Funds. Journal of Financial Markets, 14(2), 323-343.
Aragon, G. O., & Ferson, W. E. (2006). Portfolio Performance Evaluation. Foundations and Trends in Finance, 2(2), 83-190.
Arugaslan, O., & Samant, A. (2014). Evaluating S&P 500 Sector ETFs Using Risk-Adjusted Performance Measures. Journal of Finance,
Accounting and Management, 5(1), 48-62.
Blitz, D., Huij, J., & Swinkels, L. (2012). The Performance of European Index Funds and Exchange-Traded Funds. European Financial
Management, 649-662.
Buetow, G. W., & Henderson, B. J. (2012). An Empirical Analysis of Exchange-Traded Funds. Journal of Portfolio Management, 38(4), 112-127.
Carne. (2012). UCITS Guide for Alternative Managers. Dublin: Carne.
&%2(9RODWLOLW\,QGH[Š9,;Š$FFHVVHG-XO\15 from https://www.cboe.com/micro/VIX/vixintro.aspx
Charupat, N., & Miu, P. (2013). Recent developments in exchange-traded fund literature. Managerial Finance, 39(5), 427-443.
Deutsche Bank. (2014). ETF Annual Review & Outlook . Deutsche Bank AG .
European Commission. (2015, 02 04). Taxation of the financial sector. Retrieved from European Commission:
http://ec.europa.eu/taxation_customs/taxation/other_taxes/financial_sector/index_en.htm#fate
Frino, A., & Gallagher, D. R. (2001). Tracking S&P 500 Index Funds. Journal of Portfolio Management, 28(1), 44-55.
Hwang, S., & Satchell , S. (2001). Tracking Error:Ex-Ante versus Ex-Post Measures. Journal of Asset Management, 2(3), 241-246.
Jensen, M. C. (1968). The Performance of Mutual Funds in the Period 1945-1964. Journal of Finance, 23(2), 389-416.
Lintner, J. (1965). The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets. The Review of
Economics and Statistics, 47(1), 13-37.
Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1), 77-91.
Mihai Yiannaki, S. (2013). Tales of Motives in Bank M&As in Emerging Markets. Thunderbird International Business Review, 55 (5). 579-592.
Modigliani, F., & Modigliani, L. (1997). Risk-Adjusted Performance: How to Measure It and Why. The Journal of Portfolio Management, 23(2),
45-54.
Petajisto, A. (2013). Inefficiencies in the Pricing of Exchange-Traded Funds. Social Science Research Network: Accessed 15 June 2015 from
http://dx.doi.org/10.2139/ssrn.1572907
PricewaterhouseCoopers.(2013). European ETF Listing & Distribution. pwc, Luxembourg.
Rompotis, G. G. (2011). Predictable Patterns in ETF's Return and Tracking Error. Studies in Economics and Finance, 28(1), 14-35.
Scozzari, A., Tardella, F., Paterlini, S., & Krink, T. (2013). Exact and Heuristic Approaches for The Index Tracking Problem with UCITS
Constraints. Annals of Operations Research, 205(1), 235-250.
Sharpe, W. F. (1966). Mutual Fund Performance. The Journal of Business, 39(1), 119-138.
Shiller R.J (2015). Irrational Exuberance. 3rd Ed., Princeton University Press, New York. U.S.
Smith S. (2015). Ossiam launches Shiller Barclays CAPE Sector Value Europe ETF, ETF Strategy,17 February 2015, Accessed 20 July 2015 from
http://www.etfstrategy.co.uk/ossiam-launches-shiller-barclays-cape-sector-value-europe-etf-41259/
Soe, A. M. (2014, June). Does Past Performance Matter? The Persistence Scorecard. Accessed January 15, 2015, from S&P Dow Jones Indexes:
http://www.spindexes.com/documents/spiva/persistence-scorecard-june-2014.pdf
Sortino, F. (2010). The Sortino Framework for Constructing Portfolios. Elsevier Inc.
Treynor, J. L. (1965). How to Rate Management of Investment Funds. Harvard Business Review, 43(1), 63-75.
Vardharaj, R., Fabozzi, F. J., and Jones, F. J. (2004). Determinants of Tracking Error for Equity Portfolios. Journal of Investing, 13(2), 37-47.

View publication stats