Measuring Tracking Efficiency in Chinese Equity ETFs

Similar documents
On The Right Track: Measuring Tracking Efficiency in ETFs

Morningstar s Guide to Investing in Chinese Equities via ETFs

Strategy Insights. Moving toward an all-market approach to investing in China

CSOP WTI Oil Annual Roll December Futures ER ETF.

RMB counter HKD counter. 100 Units- RMB counter 100 Units HKD counter 3.10% MSCI China A Index. Renminbi (RMB) 31 December

Mirae Asset Global Investments (Hong Kong) Limited. Annually at the Manager s discretion (May in each year) Financial year end of

Dividend Stocks The Best Way to Buy China

The Voices of Influence eighth annual guide to EXCHANGE EXCHANGE TRADEDFUNDS. & indexing innovations. Thought-Leading Sponsors

INDEX FUNDS AND EXCHANGE TRADED PRODUCTS COMPARED. Viewpoint IN THIS ISSUE. Examining different passive options for client portfolios

RYT Sector Weights. Price Chart

PRODUCT KEY FACTS Haitong CSI300 Index ETF a sub-fund of the Haitong ETF Series

Industrial and Commercial Bank of China Limited Dealing frequency: Daily on each business day *

ComStage 1. ComStage 1 DAX UCITS ETF. Issuer: Commerz Funds Solutions S.A. 29 June 2016

CONSIDERATIONS WHEN CONSTRUCTING A FOREIGN PORTFOLIO: AN ANALYSIS OF ADRs VS ORDINARIES

Trading of Listed Equity Products in Hong Kong, the UK and US Research Department, Supervision of Markets Division 1 July 2007

CIO Flash Chinese equities: what happens next? July 8, 2015

PRODUCT KEY FACTS BOCHK RMB Fixed Income Fund

Exchange Thoughts Brown Brothers Harriman s ETF Newsletter

INITIAL PUBLIC OFFERINGS

Buyers Guide to RMB Bonds. Main author: Bryan Collins

Fund Management Charges, Investment Costs and Performance

Yao Zheng University of New Orleans. Eric Osmer University of New Orleans

THE RELATIONSHIP BETWEEN MSCI EMERGING MARKETS INDEX, mini MSCI EMERGING MARKETS INDEX FUTURES AND THE ishares MSCI EMERGING MARKETS ETF

CIO Flash Chinese equities: what happens next? July 8, 2015

Under the surface. Focus on ETF Liquidity. For professional clients only

Update on HKEx Equity Derivatives Market. Derivatives Trading Global Markets Division 24 April 2015

Graduation Day: The Broadening Opportunity Set of MSCI Emerging Markets

J.P. Morgan Equity Risk Premium Multi-Factor (Long Only) Index Series

HANG SENG FTSE / XINHUA CHINA 25 INDEX ETF

Understanding Currency

Guide to Chinese Share Classes v1.1

China Universal Asset Management (Hong Kong) Company Limited 匯 添 富 資 產 管 理 ( 香 港 ) 有 限 公 司

Mirae Asset Global Investments (Hong Kong) Limited

Schroder International Opportunities Portfolio - Schroder Asian Income (the Fund )

China s Unwinding Stock Market Bubble

Understanding and accessing the Chinese equity market

Pros and Cons of Different Investment Options

Summary. Research Paper No. 22

LSEG Information Services Division. Investor and Analyst presentation. Mark Makepeace Group Director of Information Services, CEO of FTSE

Tracking Down the Right Index Fund The Morningstar Approach

Morningstar s Active/Passive Barometer A new yardstick for an old debate

Exchange Traded Funds: State of the Market, Regulation and Current Concerns

ETFs for private investors

2016 Summary Prospectus

Vanguard Investments Hong Kong Limited December 2015

What is the history and global performance of ETFS? What are ETFs? Assets Under Management (AUM) of ETFs: 2001 Q12013

PRODUCT HIGHLIGHTS SHEET

30% 5% of fixed income mutual funds paid capital gains in 2015

Completing Emerging Market Equity Allocations with Small Caps

Vanguard Emerging Markets Stock Index Fund

BMO Global Asset Management (Asia) Limited 11 February 2016

ETFs and Index Funds. Similarities and Differences. For professional clients only

Guotai Junan Assets (Asia) Limited (the Manager ) HSBC Trustee (Cayman) Limited

Module 1 Introduction to ETFs

Stock Exchange of Mauritius: Newsletter

4/26/2012. Navigating the ETF Landscape. The ETF revolution. ETF assets expected to approach $2 trillion by 2014 $2,500 1,200 AUM ($B) # of ETFs

BASKET A collection of securities. The underlying securities within an ETF are often collectively referred to as a basket

The FTSE China Onshore Bond Index Series

PRODUCT KEY FACTS Samsung TOPIX Daily (2x) Leveraged Product

SPDR S&P Software & Services ETF

PRODUCT KEY FACTS Samsung KOSPI 200 Daily (2x) Leveraged Product

TOOLS FOR EFFICIENTLY MANAGING BETA EXPOSURE: Index Funds, Futures, and Exchange-Traded Funds

ETFs as Investment Options in 401(k) Plans

Sizing Up Target Date Funds

FSDC Paper No. 16. Strengthening Hong Kong as a Capital Formation Centre for Exchange Traded Funds

9 Questions Every ETF Investor Should Ask Before Investing

INDUSTRIAL AND COMMERCIAL BANK OF CHINA ICBC: Your Global Portal to RMB Market. July 2012

ETF Total Cost Analysis in Action

9 Questions Every Australian Investor Should Ask Before Investing in an Exchange Traded Fund (ETF)

Exchange Traded Funds. An Introductory Guide. For professional clients only

Commerzbank Asset Management. Active. Passive. The perfect combination.

Impact of the Issuance of A-shares on the Turnover of H-shares Joseph Lee and Yan Yuhong 1 December 2003

InvestmentPerspectives October 2014

db x-trackers S&P 500 UCITS ETF (DR) Supplement to the Prospectus

BMO Global Asset Management (Asia) Limited 11 February 2016

Monthly Leveraged Mutual Funds UNDERSTANDING THE COMPOSITION, BENEFITS & RISKS

Xetra. The market. Xetra: Europe s largest trading platform for ETFs. ETF. One transaction is all you need.

Bosera ETFs. Bosera FTSE China A50 Index ETF

Latest developments in the international RMB market

Renminbi Depreciation and the Hong Kong Economy

The Year 2005 in Review. The Year 2005 in Review STOCK MARKET

Shanghai /Hong Kong Stock Connect

BMO Global Asset Management (Asia) Limited 11 February 2016

EXCHANGE TRADED FUNDS. Roberto Juanchito T. Dispo, President May 23, 2013

Evaluating Managers on an After-Tax Basis

BlackRock Diversified Income Portfolio. A portfolio from Fidelity Investments designed to seek income while managing risk

Treasury Bonds directly makes the Sub-Fund riskier than traditional exchange traded funds investing in A-Shares or in markets other than the PRC.

SPDR EURO STOXX 50 ETF

PART III TERMS APPLICABLE TO RESPECTIVE ACCOUNTS Schedule A Terms for Cash Account / Margin Account. 5. Pilot Programme for Trading US Securities

DCI Investment Trust. Da Cheng China RMB Fixed Income Fund. Addendum to the Explanatory Memorandum dated January 2012 ( Explanatory Memorandum )

Active indexing: Being passive-aggressive with ETFs

Exchange Traded Funds Tactical Asset Allocation Tools

Mirae Asset Global Investments (Hong Kong) Limited

Exchange Traded Funds

ETFs and Index Funds. Similarities and Differences. For professional clients only

Active vs. Passive Money Management

The following replaces similar text in the Investing With Vanguard section:

Risk Disclosures of Exchange Traded Fund (ETF)

RMB solutions for importers and exporters

Transcription:

Morningstar ETF Research On The Right Track: Measuring Tracking Efficiency in Chinese Equity ETFs Authors: Jackie Choy, CFA: ETF Strategist Ben Johnson, CFA: Director, Global Passive Fund Research Hortense Bioy, CFA: Director, European Passive Fund Research

Contents Executive Summary 3 The Subject of Our Study Chinese Equity ETFs 4 The Common Metrics 5 Are These ETFs on Track? 6 Caveats: A Few Words of Caution on Tracking 18 Conclusion 21 Appendix 1: Methodology 22 Appendix 2: Defining Common Metrics: Tracking Error 23 Appendix 3: Defining Common Metrics: Tracking Difference 27 Appendix 4: Chinese Equity ETFs 28

3 Executive Summary In February 2013, our European passive funds research team published a study titled On the Right Track: Measuring Tracking Efficiency in ETFs (the 2013 European Tracking Report ). The report examined the factors that influence the tracking performance of Europe-listed ETFs. In this study, we shift our focus to Chinese equity ETFs listed in various markets around the world as the prominence of Chinese equities has been increasing in the global investment arena. The universe we have examined in this study includes 33 exchange traded funds (ETFs) tracking five popular Chinese equity indices. In general, we have found that Chinese equity ETFs appear to be tracking their respective benchmarks reasonably well. As compared to the emerging markets equity ETFs we examined in our 2013 European Tracking Report, Chinese equity ETFs showed comparable levels of tracking efficiency. However, these funds have not tracked their benchmarks as closely as those ETFs tracking developed market equity benchmarks that we examined in our European study. These findings are not surprising as the Chinese equity market shares similar characteristics with its emerging-markets peers, characteristics which tend to lead to higher levels of tracking error and tracking difference as compared to ETFs tracking developed markets equity benchmarks. We also found that synthetic replication ETFs have tended to have greater levels of tracking difference relative to physical replication ETFs, despite the fact that they have tended to have lower levels of tracking error relative to their physical replication counterparts. In some cases, we believe the cost of derivatives in sythetic replication ETFs, which come in addition to the funds expense ratio, is the primary source of these relatively higher levels of tracking difference. Chinese equity ETFs have had lower levels of tracking error and tracking difference as compared to onshore Chinese equity ETFs. While onshore Chinese equity ETFs provide one of the few channels investors currently have to attain this unique exposure, investors are effectively paying more for this access. However, we believe this difference between offshore and onshore Chinese equity ETFs could be gradually narrowed as the Chinese onshore market opens further. sensitivity in a pair of hypothetical examples. In some cases, tracking difference can also be quite sensitive to changes in its inputs particularly the chosen start and end dates used in the calculation. It is vital that investors use accurate data in making the calculations. Inexplicably high levels of tracking error should be carefully examined, as there is a chance that they are a result of data errors, and might not be a true indication of poor tracking performance. While all the various tracking metrics we discuss in this paper are important to consider when evaluating an ETF, they are not the only metrics that investors should consider. Other key factors to assess are factors affecting trading costs such as commissions and bid-ask spreads, product and index construction, counterparty risk, and tax considerations, amongst others. Lastly, we see a number of important developments on the horizon that should ultimately serve to improve these funds ability to track their benchmarks over time. We also believe that there are some key steps that these fund s sponsors and regulators can take towards harmonising the calculation and dissemination of these metrics to allow investors to make more informed decisions: The ongoing liberalisation of the Chinese capital markets could lead to more competition and potentially drive down transaction costs and expense ratios charged by these ETFs. Harmonising the definitions of tracking error, tracking difference, expense ratio, etc, and the approach to calculating these figures would help investors to make apples-to-apples comparisons amongst different ETFs. We believe any move towards harmonising definitions and reporting will likely be driven by regulators. We believe the various ongoing fund passport initiatives could address these issues. There is a need for more transparent and accurate disclosures of NAV and index values by ETF providers. We also found that the tracking error calculation is extremely sensitive to minor changes in its inputs. We illustrate this

4 The Subject of Our Study Chinese Equity ETFs In this study, we shift our focus to Chinese equity ETFs listed in various markets around the world. Recall that, in general terms, Chinese equities are issued by companies doing business in the People s Republic of China (PRC). In more granular terms, Chinese equities can be classified as (1) onshore or (2) offshore: Onshore Chinese Equities: Companies incorporated in China and listed on the Shanghai or Shenzhen Stock Exchange. These are further classified into the A-Shares (quoted in Renminbi (Rmb) and only available to domestic Chinese investors, Qualified Foreign Institutional Investors (QFIIs) and RQFIIs ( R stands for Rmb)) and B-Shares (quoted in Hong Kong dollars or US dollars; but open to foreign investors; relatively small in size). Chinese Equities: Companies listed outside of China with significant business exposure in China (whether or not they are incorporated in China). These are further classified into H-Shares, Red Chips and P Chips (listed in Hong Kong); and also L-Shares (listed on the London Stock Exchange), N-Shares (listed on the NYSE or NASDAQ) and S-Shares (listed in Singapore). Our decision to narrow the scope of our study and focus on ETFs tracking Chinese equities was driven by a number of factors: The number of these ETFs has continued to grow since we published Morningstar s Guide to Investing in Chinese Equities via ETFs. Growth has been particularly pronounced in the RQFII space. As of March 2014, total AUM in RQFII ETFs stood at Rmb 35 bn (US$ 5.6 bn). Exhibit 1. Total Market Capitalisation of China and Hong Kong Markets China A-Shares B-Shares A-Shares (Shanghai & Shenzhen) + B-Shares (in RMB,billions) 23,511 152 23,663 (in USD, billions) 3,781 24 3,806 Hong Kong H-Shares Red Chips Mainland H-Shares Total Private + Red Chips Hong Enterprises + Mainland Private Enterprises (in HKD, billions) 4,603 4,420 4,009 13,032 23,065 (in USD, billions) 593 570 517 1,680 2,973 Source: Morningstar Direct, Morningstar Research (Data to end-mar 2014) Exhibit 2. RQFII ETF AUM, Quota Granted and Number of RQFII ETF Kong Chinese equities will likely continue to grow in prominence in the long run. In March 2014, MSCI launched a consultation on the proposed index inclusion roadmap for China A-Shares in the MSCI EM Index. More countries are being granted RQFII quota to invest directly in the onshore Chinese equity market. These include: Hong Kong (Rmb 270 bn), Taiwan (Rmb 100 bn, based on a favourable response from the Chinese government but not yet granted officially), UK (Rmb 80 bn), Singapore (Rmb 50 bn) and France (Rmb 80 bn). A number of A-Share ETFs were launched in the past 6 months outside of Hong Kong. In April 2014, the Shanghai-HK Stock Connect pilot programme was announced. The programme will allow domestic Chinese investors to invest in HK-listed stocks (a total quota of Rmb 250 bn has been granted) and Hong Kong investors to invest in Shanghai-listed stocks (a total quota of Rmb 300 bn). This marks another big stride towards opening up China s capital market. The Chinese equity market is becoming ever more important to the global investment landscape and the crop of ETFs offering exposure to Chinese equities continues to grow. In this context, we hope our various studies of this subject Chinese Equity ETFs will allow investors to gain a deeper understanding of these products as the industry continues to develop at a rapid pace. Source: HKEx, Morningstar Research

5 The Common Metrics There are two widely used metrics of index funds tracking efficiency: tracking error and tracking difference. Here, we provide a brief definition of each (a more detailed version can be found in the Appendix). Tracking Error Tracking error is a measure of the standard deviation of an ETF s excess returns. In this context, excess returns refer to the absolute difference between the ETF s performance and that of its benchmark. Lower tracking error is indicative of more consistency in the periodic deviations between the return of the fund and that of its benchmark. Tracking Difference Tracking difference is simply the annualised difference between a fund s actual return and its benchmark return over a specific period of time. A small tracking difference indicates that the ETF has done a good job matching its index over the period in question. Exhibit 3. Frequency Distribution of Daily Tracking Difference of a MSCI China Index ETF in a Year Tracking Error Measures the volatility of these daily differences Number of days 100 90 80 70 60 50 40 30 20 10-7.5 0 5 Daily tracking difference (bps) Tracking Difference Differences in return of the ETF against the benchmark index. Also the geometric sum of these daily differences. Source: Morningstar Direct, Morningstar Research (Data to end-jan 2014)

6 Are These ETFs on Track? In answering the big question How well did these ETFs track their benchmark indices? we will examine the two common metrics tracking error and tracking difference as well as a proprietary Morningstar data point Estimated Holding Cost. Tracking Error On average, ETFs tracking Chinese equity indices exhibited tracking error ranging from 64 basis points (for synthetic replication funds) to 102 basis points, or bps, (for physically replicated funds) over the period we studied. This is a narrower range than that seen amongst ETFs tracking the MSCI Emerging Markets Index. In our 2013 European Tracking Report, we found that these funds tracking error ranged from 17 bps (synthetic replication) to 177 bps (physical replication). It is important to note that we are sharing this data for purposes of generic comparison only and that we examined different measurement periods in each report. Synthetic Replication Offers Superior Tracking In general, ETFs utilising synthetic replication have lower tracking error. The lone exception is the XACT KINA ETF, which is listed in Sweden and utilises options and futures to replicate its benchmark, a process that results in high tracking error. On average, tracking error amongst synthetic ETFs was 15 68 bps lower than for physical ETFs. This is similar to what we found in Europe, where the difference was an average of 30 bps. In addition, this finding was, in general, universal among offshore and onshore Chinese equity ETFs. Specifically, the tracking error of offshore synthetic ETFs was anywhere from 15 33 bps lower than that measured amongst physical ETFs whereas the same figure among onshore synthetic ETFs was 42 68 bps lower than that for physical ETFs tracking onshore benchmarks. Chinese Equity ETFs Offer Superior Tracking Given the nature of onshore Chinese equity ETFs from the cross-market nature of the underlying securities to the more complex use of derivatives (China access products) by the synthetic ETFs it is not surprising that the offshore Chinese equity ETFs produced lower tracking error than the onshore Chinese equity ETFs. On average across the physical and synthetic ETFs, the difference was in excess of 110 bps. Exhibit 4. Average 1-Year Tracking Error for ETFs Tracking Chinese Equity Indices Onshore Index tracked by ETFs (excluding feeder funds) Physical Replication Synthetic Replication Daily Tracking Error % Weekly Tracking Error % Daily Tracking Error % Weekly Tracking Error % FTSE China 25 0.89 0.49 0.18 0.18 HSCEI 0.53 0.48 2.20 1.45 HSCEI (excl. XACT) 0.53 0.48 0.37 0.39 MSCI China 0.24 0.22 0.10 0.13 FTSE China A50 1.58 1.31 1.63 1.29 CSI 300 2.43 2.05 1.35 1.35 Average (excl. XACT) 1.02 0.80 0.64 0.62 Average Chinese Equity Indices (excl. XACT) 0.53 0.37 0.20 0.22 Average Onshore Chinese Equity Indices 2.09 1.75 1.42 1.34 Source: Morningstar Direct, Morningstar Research (Data to end-jan 2014)

7 Tracking Error By Index at the ETF Level Exhibit 5. Tracking Error for FTSE China 25 ETFs ETF Listing Countries Replication Method Daily Tracking Error % (since) Weekly Tracking Error % (since) Monthly Tracking Error % (since) Feb 2011 Feb 2013 Feb 2011 Feb 2013 Feb 2011 CIMB FTSE China 25 Malaysia Physical 2.64 2.69 1.30 1.34 0.67 db x-trackers FTSE China 25 UCITS ETF HK,SG, EU Synthetic 0.11 0.13 0.15 0.16 0.20 EasyETF FTSE China 25 EU Synthetic 0.42 0.24 0.28 0.19 0.23 Hang Seng FTSE China 25 Index ETF HK Physical 0.25 0.26 0.27 0.26 0.26 ishares China Large Cap UCITS ETF EU Physical 0.42 0.26 0.29 0.16 0.24 ishares China Large-Cap US, Chile, Mex, Aus Physical 0.30 0.35 0.21 0.20 0.21 ishares China Index CA Feeder Fund 23.42 17.34 12.61 9.64 6.18 Average FTSE China 25 3.94 3.04 2.16 1.71 1.14 Average FTSE China 25 (excluding feeder funds) 0.69 0.66 0.42 0.39 0.30 Average Physical Replicator 0.90 0.89 0.52 0.49 0.34 Average Synthetic Replicator 0.26 0.18 0.21 0.18 0.21 Source: Morningstar Direct, Morningstar Research (Annualised data to end-jan 2014) Exhibit 6. Tracking Error for HSCEI ETFs ETF Listing Countries Replication Method Daily Tracking Error % (since) Weekly Tracking Error % (since) Monthly Tracking Error % (since) Feb 2011 Feb 2013 Feb 2011 Feb 2013 Feb 2011 ComStage HSCEI UCITS ETF EU Synthetic 0.56 0.58 0.59 0.57 0.62 Hang Seng H-Share Index ETF HK, TW Physical 0.33 0.35 0.36 0.35 0.29 KODEX China H ETF KR Physical 0.80 0.72 0.59 0.62 0.45 Listed Index Fund China H-share (HSCE) JP Feeder Fund 36.97 32.57 16.32 17.67 9.78 Lyxor UCITS ETF China Enterprise (HSCEI) SG, EU Synthetic 0.18 0.16 0.22 0.21 0.29 XACT Kina ETF SE Synthetic N/A 5.87 N/A 3.57 N/A Average HSCEI 7.77 6.71 3.61 3.83 2.28 Average HSCEI (excl. XACT) 7.77 6.87 3.61 3.88 2.28 Average HSCEI (excluding feeder funds) 0.47 1.53 0.44 1.07 0.41 Average HSCEI (excluding feeder funds and XACT) 0.47 0.45 0.44 0.44 0.41 Average Physical Replicator 0.56 0.53 0.47 0.48 0.37 Average Synthetic Replicator 0.37 2.20 0.40 1.45 0.45 Average Synthetic Replicator (excl. XACT) 0.37 0.37 0.40 0.39 0.45 Source: Morningstar Direct, Morningstar Research (Annualised data to end-jan 2014)

8 Tracking Error By Index at the ETF Level (continued) Exhibit 7. Tracking Error for MSCI China ETFs ETF Listing Countries Replication Method Daily Tracking Error % (since) Weekly Tracking Error % (since) Monthly Tracking Error % (since) Feb 2011 Feb 2013 Feb 2011 Feb 2013 Feb 2011 db x-trackers MSCI China Index UCITS ETF HK, SG, EU Synthetic 0.10 0.10 0.11 0.13 0.15 Deka MSCI China UCITS ETF EU Physical 0.28 0.28 0.30 0.29 0.24 Horizons MSCI China ETF HK Physical Listed on 17 June 2013 HSBC MSCI CHINA UCITS ETF EU Physical 0.26 0.22 N/A 0.21 N/A HSBC MSCI China ETF HK Physical N/A 0.25 N/A 0.27 N/A ishares MSCI China US, Chile, Mex Physical N/A 0.26 N/A 0.19 N/A ishares MSCI China Index HK Physical 0.27 0.18 0.27 0.14 0.32 Source MSCI China ETF EU Synthetic 0.09 0.10 0.11 0.13 0.15 Average FTSE China 25 0.20 0.19 0.20 0.19 0.22 Average FTSE China 25 (excluding feeder funds) 0.27 0.24 0.28 0.22 0.28 Average Physical Replicator 0.09 0.10 0.11 0.13 0.15 Source: Morningstar Direct, Morningstar Research (Annualised data to end-jan 2014) Exhibit 8. Tracking Error for FTSE China A50 ETFs ETF Listing Countries Replication Method Daily Tracking Error % (since) Weekly Tracking Error % (since) Monthly Tracking Error % (since) Feb 2011 Feb 2013 Feb 2011 Feb 2013 Feb 2011 Onshore Bosera FTSE China A50 Index ETF HK Physical Listed on 9 December 2013 ComStage FTSE China A50 UCITS ETF EU Synthetic Listed on 8 October 2013 CSOP FTSE China A50 ETF HK, JP Physical N/A 1.79 N/A 1.74 N/A CSOP Source FTSE China A50 UCITS ETF EU Physical Listed on 8 January 2014 ishares FTSE A50 China Index ETF HK Synthetic 1.80 1.63 1.45 1.29 1.14 KODEX FTSE China A50 ETF KR Physical N/A 5.87 N/A 3.57 N/A Average FTSE China A50 1.80 1.60 1.45 1.30 1.14 Average Physical Replicator N/A 1.58 N/A 1.31 N/A Average Synthetic Replicator 1.80 1.63 1.45 1.29 1.14 Source: Morningstar Direct, Morningstar Research (Annualised data to end-jan 2014)

9 Tracking Error By Index at the ETF Level (continued) Exhibit 9. Tracking Error for CSI 300 ETFs ETF Listing Countries Replication Method Daily Tracking Error % (since) Weekly Tracking Error % (since) Monthly Tracking Error % (since) Feb 2011 Feb 2013 Feb 2011 Feb 2013 Feb 2011 Onshore ChinaAMC CSI 300 Index ETF HK, JP Physical N/A 2.27 N/A 2.22 N/A C-Shares CSI 300 Index ETF HK Physical Listed on 8 July 2013 db x-trackers CSI300 Index UCITS ETF HK, EU Synthetic 0.24 0.23 0.23 0.16 0.40 db X-trackers Harvest CSI 300 China A-Shares Fund US Physical Listed on 6 November 2013 db x-trackers Harvest CSI300 Index UCITS ETF EU Physical Listed on 7 January 2014 Fuh Hwa CSI300 A Shares ETF TW Physical N/A 2.37 N/A 1.71 N/A Haitong CSI300 Index ETF HK Physical Listed on 7 March 2014 ishares CSI 300 A-Share Index ETF HK Synthetic 6.04 1.57 5.27 1.44 4.54 KINDEX China A CSI300 ETF KR Physical N/A 2.66 N/A 2.23 N/A KraneShares Bosera MSCI China A ETF US Physical Listed on 5 April 2014 Listed Index Fund China A Share (Panda) CSI300 JP Feeder Fund 30.85 32.20 13.68 12.24 7.49 Lyxor UCITS ETF CSI 300 A-Share EU Synthetic Listed on 2 October 2013 Market Vectors ChinaAMC A-Share ETF US Physical 7.68 4.35 3.97 2.37 1.50 (formerly Market Vectors China ETF, prior to 7 January 2014) (formerly Synthetic) TIGER China A300 ETF KR Physical Listed on 17 February 2014 W.I.S.E. - CSI 300 China Tracker HK Synthetic 1.84 2.24 1.77 2.46 1.64 W.I.S.E. KTAM CSI 300 China Tracker TH Feeder Fund 12.34 9.29 10.16 7.35 8.71 W.I.S.E. Yuanta/P-shares CSI 300 ETF TW Feeder Fund 11.82 9.33 9.83 7.35 9.44 Average CSI 300 10.12 6.65 6.42 3.95 4.82 Average CSI 300 (ex-feeder funds) 3.95 2.24 2.81 1.80 2.02 Average Physical Replicator (excl. Market Vector) N/A 2.43 N/A 2.05 N/A Average Synthetic Replicator (excl. Market Vector) 2.71 1.35 2.42 1.35 2.19 Source: Morningstar Direct, Morningstar Research (Annualised data to end-jan 2014) Key Findings from Tracking Error at the ETF Level: Feeder funds exhibit very high daily tracking error These funds daily tracking error has ranged from 10% to 30% on an annualised basis over the trailing 1- and 3-year periods. This compares to a range of 0.10% to around 2% for their peers. Feeder funds generally invest in another ETF that tracks the index and hence the NAV of the feeder fund is based on the market price of the underlying ETF rather than the underlying securities of the index. The ETF that a feeder fund invests in may trade at a premium or discount to its NAV. This creates an additional source of tracking error for the feeder fund. The feeder fund s tracking error will likely be even larger if the trading hours for the exchange where the underlying ETF is listed do not overlap with the normal trading hours of the securities that comprise the fund s benchmark index. Daily/weekly/monthly data irrelevant In general, using weekly or monthly data instead of daily data results in similar annualised tracking error figures. The exception is once again the feeder funds. In their case, using weekly data tends to smooth-out the annualised daily tracking error and the monthly data further smooth this figure. This finding is somewhat similar to that which we shared in the 2013 European Tracking Report.

10 Tracking Error Varies Over Time In general, there were no obvious differences between the 1-year and 3-year annualised tracking errors amongst the ETFs we examined. This is consistent with our findings on the use of daily/weekly/monthly data. However, Exhibit 10 illustrates that tracking errors can vary considerably over time. This variation depends largely on the chosen beginning and ending points of the dataset. In our view, considerable changes in tracking error over time could be driven by a variety of factors, such as cash drag, NAV accounting adjustments, changes in replication methodology, inaccurate data, differences in data vendors, etc. For instance, the ishares CSI 300 A-Share Index ETF invested in other ishares sector ETFs to track its benchmark index prior to May 2012. It has since been utilising access products which has resulted in a significant reduction in tracking error. The fund s trailing 1-year tracking error has fallen sharply (1.57%) relative to the trailing 3-year figure (6.04%). Exhibit 10. Rolling One-Year Tracking Error for a Subset of MSCI China ETFs Source: Morningstar Direct, Morningstar Research

11 Tracking Difference Tracking difference the absolute level of annual out- or underperformance of each ETF relative to its benchmark varies considerably depending on the reference index in question. On average, ETFs tracking the Chinese equity indices exhibited tracking difference ranging from -95 bps (physical replication) to -200 bps (synthetic replication) over the measurement period under examination. This range is wider than that we discovered amongst ETFs tracking the MSCI Emerging Markets Index of -92 (synthetic replication) to -103 (physical replication) bps in the 2013 European Tracking Report. It is important to note that we are sharing this data for purposes of generic comparison only and that we examined different measurement periods in each report. Physical Replication Results in Lower Tracking Difference Amongst funds tracking all five indices within our study, physical replication ETFs exhibited lower tracking difference than synthetic replication ETFs. On average, tracking difference among physical ETFs was 105 bps lower than for synthetic ETFs. This difference was more pronounced amongst funds tracking onshore Chinese equity indices (-340 bps for synthetic replication funds; -104 bps for physical replication funds). We believe this is due to the embedded costs/spread associated with the use of derivatives to replicate the performance of onshore Chinese equity indices. As an example, the db x-trackers CSI300 Index UCITS ETF (listed in Hong Kong and various exchanges in Europe) discloses that there is an index replication cost of 2.65% in addition to the 0.50% TER that the ETF levies. Exhibit 11. Average 1-Year Tracking Difference for ETFs Tracking Chinese Equity Indices Onshore Index tracked by ETFs (excluding feeder funds) Tracking Difference % Physical Replication Synthetic Replication Average of Physical & Synthetic FTSE China 25-1.14-1.26-1.18 HSCEI -0.88-1.54-1.28 MSCI China -0.73-1.10-0.87 FTSE China A50-0.61-1.89-1.22 CSI 300-1.33-3.72-2.36 Average -0.95-2.00-1.40 Average Chinese Equity Indices -0.91-1.30-1.07 Average Onshore Chinese Equity Indices -1.04-3.40-2.09 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014)

12 Tracking Difference By Index at the ETF Level Exhibit 12. Tracking Difference for FTSE China 25 ETFs ETF Listing Countries Replication Method Expense Ratio % Tracking Difference % (since) Feb 2011 Feb 2013 CIMB FTSE China 25 Malaysia Physical 1.01-1.84-1.26 db x-trackers FTSE China 25 UCITS ETF HK,SG, EU Synthetic 0.60-0.99-1.05 EasyETF FTSE China 25 EU Synthetic 0.60-1.34-1.47 Hang Seng FTSE China 25 Index ETF HK Physical 0.89-1.26-1.35 ishares China Large Cap UCITS ETF EU Physical 0.74-0.89-0.96 ishares China Large-Cap US, Chile, Mex, Aus Physical 0.73-0.91-1.01 ishares China Index CA Feeder Fund 0.86-1.65-2.71 Average FTSE China 25-1.27-1.40 Average FTSE China 25 (excluding feeder funds) -1.21-1.18 Average Physical Replicator -1.22-1.14 Average Synthetic Replicator -1.17-1.26 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014) Exhibit 13. Tracking Difference for HSCEI ETFs ETF Listing Countries Replication Method Expense Ratio % Tracking Difference % (since) Feb 2011 Feb 2013 ComStage HSCEI UCITS ETF EU Synthetic 0.55-1.41-1.39 Hang Seng H-Share Index ETF HK, TW Physical 0.65-0.91-0.94 KODEX China H ETF KR Physical 0.37-0.85-0.83 Listed Index Fund China H-share (HSCE) JP Feeder Fund 0.55-1.11-1.12 Lyxor UCITS ETF China Enterprise (HSCEI) SG, EU Synthetic 0.65-1.25-1.43 XACT Kina ETF SE Synthetic 0.60 N/A -1.78 Average HSCEI -1.10-1.25 Average HSCEI (excl. XACT) -1.10-1.14 Average HSCEI (excluding feeder funds) -1.10-1.28 Average HSCEI (excluding feeder funds and XACT) -1.10-1.15 Average Physical Replicator -0.88-0.88 Average Synthetic Replicator -1.33-1.54 Average Synthetic Replicator (excl. XACT) -1.33-1.41 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014)

13 Tracking Difference By Index at the ETF Level (continued) Exhibit 14. Tracking Difference for MSCI China ETFs ETF Listing Countries Replication Method Expense Ratio % Tracking Difference % Feb 2011 Feb 2013 db x-trackers MSCI China Index UCITS ETF HK, SG, EU Synthetic 0.65-1.02-1.26 Deka MSCI China UCITS ETF EU Physical 0.65-0.95-1.11 Horizons MSCI China ETF HK Physical 0.25 Listed on 17 June 2013 HSBC MSCI CHINA UCITS ETF EU Physical 0.60-0.63-0.50 HSBC MSCI China ETF HK Physical 0.50 N/A -0.55 ishares MSCI China US, Chile, Mex Physical 0.73 N/A -0.67 ishares MSCI China Index HK Physical 0.59-0.84-0.82 Source MSCI China ETF EU Synthetic 0.65-1.20-1.17 Average MSCI China -0.93-0.87 Average Physical Replicator -0.81-0.73 Average Synthetic Replicator -1.11-1.10 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014) Exhibit 15. Tracking Difference for FTSE China A50 ETFs ETF Listing Countries Replication Method Expense Ratio % Tracking Difference % (since) Feb 2011 Feb 2013 Onshore Bosera FTSE China A50 Index ETF HK Physical 1.19 Listed on 9 December 2013 ComStage FTSE China A50 UCITS ETF EU Synthetic 0.40 Listed on 8 October 2013 CSOP FTSE China A50 ETF HK, JP Physical 1.15 N/A 0.13 CSOP Source FTSE China A50 UCITS ETF EU Physical 1.11 Listed on 8 January 2014 ishares FTSE A50 China Index ETF HK Synthetic 1.39-2.91-2.44 KODEX FTSE China A50 ETF KR Physical 0.99 N/A -1.34 Average FTSE China A50-2.91-1.22 Average Physical Replicator N/A -0.61 Average Synthetic Replicator -2.91-1.89 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014)

14 Tracking Difference By Index at the ETF Level (continued) Exhibit 16. Tracking Difference for CSI 300 ETFs ETF Listing Countries Replication Method Expense Ratio % Tracking Difference % (since) Feb 2011 Feb 2013 Onshore ChinaAMC CSI 300 Index ETF HK, JP Physical 0.85 N/A 0.87 C-Shares CSI 300 Index ETF HK Physical 0.79 Listed on 8 July 2013 db x-trackers CSI300 Index UCITS ETF HK, EU Synthetic 0.50-2.01-3.00 db X-trackers Harvest CSI 300 China A-Shares Fund US Physical 0.82 Listed on 6 November 2013 db x-trackers Harvest CSI300 Index UCITS ETF EU Physical 1.10 Listed on 7 January 2014 Fuh Hwa CSI300 A Shares ETF TW Physical 0.85 N/A -2.69 Haitong CSI300 Index ETF HK Physical 0.85 Listed on 7 March 2014 ishares CSI 300 A-Share Index ETF HK Synthetic 1.39-4.68-3.38 KINDEX China A CSI300 ETF KR Physical 0.70 N/A -2.17 KraneShares Bosera MSCI China A ETF US Physical 1.10 Listed on 5 April 2014 Listed Index Fund China A Share (Panda) CSI300 JP Feeder Fund 0.95-0.66-0.30 Lyxor UCITS ETF CSI 300 A-Share EU Synthetic 0.40 Listed on 2 October 2013 Market Vectors ChinaAMC A-Share ETF (formerly Market Vectors China ETF, prior to 7 January 2014) US Physical (formerly Synthetic) 0.72-1.55-1.38 TIGER China A300 ETF KR Physical 0.70 Listed on 17 February 2014 W.I.S.E. - CSI 300 China Tracker HK Synthetic 1.39-2.95-4.77 W.I.S.E. KTAM CSI 300 China Tracker TH Feeder Fund 1.50-6.73-8.16 W.I.S.E. Yuanta/P-shares CSI 300 ETF TW Feeder Fund 0.40-7.01-8.97 Average CSI 300-3.66-3.39 Average CSI 300 (ex-feeder funds) -2.80-2.36 Average Physical Replicator (excl. Market Vector) N/A -1.33 Average Synthetic Replicator (excl. Market Vector) -3.21-3.72 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014) Key Findings from Tracking Difference at the ETF Level: There is a Fairly Weak Relationship between Tracking Difference and Expenses Intuitively, an ETF s expense ratio should represent a good portion of its tracking difference given that expenses are the most explicit and predictable detractor from an ETF s performance. However, our samples showed that the relationship between tracking difference and expense ratios was fairly weak, contrary to our findings in the 2013 European Tracking Report. In our study, the two values only exhibited a correlation of 44% and a coefficient of determination, or R-squared, of 0.196, implying that just under 20% of a funds tracking difference could be explained by its expense ratio (contrary to a figure of 50% from our European study). Recall that other factors such as securities lending income, cash drag, taxes, accounting adjustments, rebalancing costs for physical ETFs and the cost of the derivatives underlying synthetic ETFs can all impact an ETF s tracking performance. In addition, our study has included ETFs which are

15 domiciled in various countries and listed on a number of different exchanges around the world. The definition and calculation of an ETF s expense ratio (ongoing charge, TER, etc.) may be different across geographies and go some way towards explaining the weaker statistical relationship. In particular, in many cases, the cost of swap and access products and some ancillary costs are not included in the stated expense ratio. In any case, our results underscore the fact that the stated expense ratio should not be the sole consideration when choosing an ETF. Tracking Difference Unstable Over Time Because tracking difference is a snapshot of difference between the return of the ETF and the benchmark over a specific time period, it is not surprising that this measure can be quite volatile as it is very sensitive to the selected beginning and end dates used in the calculation. Exhibit 18 is a prime example of this volatility, depicting the rolling 1-year tracking difference amongst a subset of ETFs tracking the MSCI China Index. Exhibit 17. Estimated Holding Cost and Tracking Difference of ETFs Tracking Difference % Physical Synthetic Tracking Difference % Onshore 6.0 6.0 4.0 4.0 2.0 2.0 0.0 0.0-2.0 0.5 1.0 1.5 Expense Ratio % -2.0 0.5 1.0 1.5 Expense Ratio % Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014)

16 Exhibit 18. Rolling 1-Year Tracking Difference for a Subset of MSCI China ETFs ETFs. In addition, physical replication ETFs exhibited lower Estimated Holding Costs with those tracking onshore Chinese equity indices in particular. Source: Morningstar Direct, Morningstar Research Estimated Holding Cost In the 2013 European Tracking Report, we introduced a proprietary Morningstar data point Estimated Holding Cost. We believe that estimated holding cost is a more reliable metric than tracking difference, as it takes a larger number of data points into account. Our methodology is based on twenty sets of trailing one-year returns derived from the prior twenty trading days. We then calculate twenty return ratios between the fund and its benchmark, and take the geometric average of these twenty return ratios. The result is a smoothed, annualised estimation of the divergence of the ETF s performance relative to its benchmark index over a year. Like tracking difference, Morningstar s Estimated Holding Cost accounts for all the various explicit (e.g. expense ratio) and implicit (e.g. turnover, cash drag, spread/cost of derivatives, etc.) costs of holding an ETF. Estimated holding cost excludes any costs associated with buying and selling ETFs. Our study shows that the Estimated Holding Costs for the ETFs we examined were fairly similar to their tracking differences. Similar to our findings on tracking differences, Estimated Holing Costs for onshore Chinese equity ETFs were higher as compared to the offshore Chinese equity

17 Exhibit 19. Estimated Holding Cost and Tracking Difference by Benchmark Index tracked by ETFs (excluding feeder funds) Estimated Holding Costs % Tracking Difference % Onshore FTSE China 25 1.37-1.18 HSCEI 1.50-1.28 MSCI China 0.95-0.87 FTSE China A50 1.88-1.22 CSI 300 3.06-2.36 Average 1.73-1.40 Average Chinese Equity Indices 1.23-1.07 Average Onshore Chinese Equity Indices 2.91-2.09 Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014) Exhibit 20. Estimated Holding Cost and Tracking Difference of ETFs Estimated Holding Costs % Physical Synthetic Estimated Holding Costs % Onshore 6.0 6.0 4.0 4.0 2.0 2.0 0.0 0.0-2.0-2.0 0.0 2.0 4.0 6.0-2.0-2.0 0.0 2.0 4.0 6.0 Tracking Difference % Tracking Difference % Source: Morningstar Direct, Morningstar Research (Data as of end-jan 2014)

18 Caveats: A Few Words of Caution on Tracking Exhibit 21. Impact of Data Accuracy Do we need accurate data? The answer is an unqualified yes, which would seem intuitive in conducting any form of analysis. However we have noticed that having accurate data is particularly important for purposes of analysing tracking error as this calculation is extremely sensitive to minor fluctuations in the values of its various inputs. Here, we illustrate the importance of accurate data with two hypothetical scenarios we implanted in the real life data set for the ishares MSCI China Index ETF which is listed in Hong Kong. Hypothetical Case 1: Stale NAV Data In this hypothetical example, we have built a scenario where the NAV of the ishares MSCI China Index ETF is stale by a single day. A stale NAV is not uncommon in the real world. It can arise from a number of sources, for example, exchange holidays might not be taken into account in the calculation. Errors might also arise at the level of the ETF and/or data provider. Tracking Difference % Daily Tracking Error % Scenario Since Feb 2011 Since Feb 2013 Since Feb 2011 Since Feb 2013 Real Life: ishares MSCI China Index ETF Hypothetic Case 1: One Day of Stale NAV -0.84-0.82 0.27 0.18-0.84-0.82 0.93 1.56 Meanwhile, the fund s tracking difference, being the product of a snapshot of a pair of year-on-year figures, was not affected, as the stale NAV did not fall on either of the dates at which the relevant data points were calculated. Should the stale NAV fall on the relevant data points for calculating tracking difference, tracking difference will be affected significantly. Hypothetical Case 2: Number of Decimal Places In this hypothetical example, we took the NAVs of the ishares MSCI China Index ETF during the measurement period and rounded the NAV to two decimal places from the four decimal places (in Hong Kong dollars) that were disclosed by the ETF provider. This hypothetical exercise had a significant effect on this fund s calculated tracking error. The fund s trailing 1-year tracking error increased from 18 bps to 36 bps; and its trailing 3-year annualised tracking error rose from 27 bps to 44 bps. The impact is less dramatic than that seen in our first hypothetical case, but the incremental difference of 17-18 bps could have caused an investor to choose another ETF in the event that they relied heavily on tracking error figures to inform their decision. On the other hand, tracking difference calculations were only slightly affected by the elimination of two decimal places from the raw NAV data. Hypothetical Case 2: 2 d.p. of Reported NAV -0.85-0.82 0.44 0.36 Source: Morningstar Direct, Morningstar Research In our hypothetical example, on the day (day T; we hypothetically chose it as three months into the third year of the measurement period) the stale NAV appeared in the data set, the benchmark index returned +1.1%. With a stale NAV, all else equal, on day T, it would appear that ETF lagged its benchmark index by 1.1%. On day T + 1, it the NAV would catch up and appear to generate an incremental 1.1 percentage points of performance. The end result of this hypothetical scenario is a fairly dramatic increase in the fund s trailing 1-year tracking error. This single stale NAV figure would cause the ETF s tracking error to increase from 18 bps to 156 bps. There would also be a less dramatic but also significant change for the fund s 3-year annualised tracking error which would rise from 27 bps to 93 bps.

19 Exhibit 22. 1-Year Rolling Tracking Error of the Hypothetical Scenarios are by no means limited to: Trading costs, including commissions, bid-ask spreads, and market impact ETFs market price relative to NAV over time (i.e. premiums and discounts) Counterparty risk Tax considerations Source: Morningstar Direct, Morningstar Research Some of these considerations have been the subject of previous Morningstar studies, while others will be covered in future research papers. Is it Desirable for an ETF to Outperform its Benchmark? As we mentioned in our conclusion of the 2013 European Tracking Report, in discussing tracking difference and Estimated Holding Cost, at some point it s valid to ask: is it desirable for an ETF to outperform its benchmark? Certainly, if you own the ETF, the extra basis points of return will augment your bottom line. And viewed through the traditional prism of investment management, a higher return is generally considered preferable to a lower one. But unlike actively-managed funds, ETFs have a clear objective to track not outperform a benchmark. Any deviation from that objective, on the upside or the downside, represents a subversion of their mission. To see why that distinction might matter, consider the market participant that sells an ETF short. That investor will be hurt, not helped, by the outperformance, and might justly criticise the fund for falling short of its stated objective. What Else Matters When Evaluating an ETF? Also from our conclusion of the 2013 European Tracking Report, it is important to mention that while all the tracking metrics we discussed in this paper, namely tracking error, tracking difference and Estimated Holding Cost are important factors to consider when evaluating an ETF, they are not the only metrics that investors should look at. Additional factors to take into consideration include, but

20 Tips for ETF Investors Use accurate data sources As illustrated in the hypothetical examples above, having accurate data sources is very important, both for the benchmark index and the ETF. Accurate data could be obtained from ETF providers (NAVs), index providers (index values) and trusted data providers (NAVs and index values). Providing accurate data NAVs and distributions should be disclosed accurately and consistently. If permitted, we would suggest ETF providers also provide index data as, in many cases; historical index data is not publicly available. In addition to using accurate data, it is also important to use the appropriate data. For example it is vital to measure the total return of an ETF (i.e. including the dividend distributions) against the appropriate version of its benchmark index (total return). Understand what the numbers are saying In analysing ETFs tracking performance, it is crucial that investors understand what the numbers are telling them. It is important to make apples-to-apples comparisons and assess the risks represented by the numbers. For example, some cases of inexplicably high tracking error could be the result of data errors rather than a true indication of poor tracking performance. This is just the start Tracking error, tracking difference, and estimated holding cost are just a few of the key metrics investors should employ when assessing the total cost of owning an ETF. It is also important to asses other metrics that serve to measure the frictional/transactional costs of ETF ownership. These include bid/ask spread, market impact, premium/discount, taxation, trading commissions, etc. Tips for ETF Providers Harmonise definitions ETF providers should align themselves in an effort to harmonise the definitions of the various metrics we ve discussed. These items include the specifics of what should be included in a fund s stated expense ratio, and could even go so far as regular disclosure of historical tracking error and tracking difference, and how they were calculated. We understand it might be difficult for ETF providers to form a consensus on definitions. As such, any move towards harmonising definitions and reporting will likely be driven by regulators.

21 Conclusion So, have Chinese Equity ETFs tracked their indices well? Chinese equities are included in MSCI s emerging markets classification while onshore Chinese equities have been included in the consultation for potential inclusion in this category. Chinese equities share many common characteristics with those already included in MSCI s emerging markets category, such as higher trading costs and lower levels of liquidity as compared developed equity markets. In addition, ETFs tracking Chinese equities tend to have similar fees as compared to emergingmarkets ETFs, which tend to be higher than those levied by ETFs tracking developed market equities. All of these factors contribute to higher tracking error and/or tracking difference amongst ETFs tracking Chinese or emergingmarkets equities as compared to ETFs tracking developed markets benchmarks. As a result, we judge that it is fairer to compare our findings on the tracking performance of Chinese equities ETFs against the tracking performance of emerging-markets ETFs. As compared to ETFs tracking the MSCI Emerging Markets Index, we have found that physically replicated Chinese equity ETFs have tracked their benchmarks about as well as physically replicated ETFs tracking the MSCI Emerging Markets Index, albeit with less volatility. Meanwhile, the synthetic replication ETFs tracking Chinese equities that we examined have demonstrated inferior tracking performance relative to synthetic replication ETFs benchmarked to the MSCI Emerging Markets Index that we studied in our European tracking study. Comparing to ETFs tracking various developed markets indices, Chinese equity ETFs produced less satisfactory results in terms of both tracking error and tracking difference. The chief drivers of this relatively poor performance include: these funds relatively higher expense ratios; their inability to benefit from or execute certain tax-optimisation techniques employed by other ETFs; less effective use of securities lending. Exhibit 23. Comparing to MSCI Emerging Market ETFs Scenario Annualised Tracking Error % Annualised Tracking Difference % Chinese Equity ETFs * 0.87-1.40 - Chinese Equity ETFs 0.69-1.07 - Onshore Chinese Equity ETFs 1.79-2.09 - Physical replicators 1.02-0.95 - Synthetic replicators 0.64-2.00 MSCI Emerging Markets ETFs ** 0.77-0.95 - Physical replicators 1.77-1.03 - Synthetic replicators 0.17-0.91 ETFs tracking various developed market indices ** 0.04 to 0.21 0.47 to -0.49 - Physical replicators 0.05 to 0.27 0.55 to -0.55 - Synthetic replicators 0.02 to 0.30 0.36 to -0.49 Notes: *As of end-jan 2014, ** Results from the 2013 European Tracking Report Source: Morningstar Direct, Morningstar Research While synthetic replication ETFs have generally produced less favourable results relative to their physical replication counterparts, some of them have had reasonable tracking performance. Hence, investors should be careful about making broad generalisations pertaining to synthetic funds tracking performance. Nevertheless, we are seeing a trend towards switching from synthetic replication to physical replication amongst ETF providers in Europe, with db X-trackers being an activist in this shift. ETFs tracking offshore Chinese equity indices have produced superior tracking performance relative to ETFs tracking onshore Chinese equity benchmarks. While these onshore Chinese equity ETFs provide one of the few channels through which investors can access this unique exposure, they are effectively paying a steeper price for this level of access. However, we believe this difference could be gradually narrowed as the Chinese onshore market opens further. Overall, Chinese equity ETFs appear to be tracking their benchmarks reasonably well. Ultimately, our hope is that as the Chinese onshore market opens further, and these ETFs continue to grow in both number and size, transaction costs and management fees will be reduced. This, in turn, should serve to drive down the total cost of owning Chinese equity ETFs.

22 Appendix 1: Methodology (A large portion of this section was first published in the On The Right Track: Measuring Tracking Efficiency in ETFs dated February 2013). Where: R nav is the single period total return of the fund s NAV For the purpose of this study we have tried to focus on ETFs covering a sample of the broadest and most widely used Chinese equity indices; hence our choice of the FTSE China 25, the Hang Seng China Enterprises Index (HSCEI), the MSCI China, the FTSE China A50 and the CSI 300. For each of these benchmarks, we have included all the ETFs around the world (outside of China) for which we have data and for which there is sufficient history. In total, the funds in our study constituted, at the time of writing, roughly 87% of the assets under management within ETFs (outside of China) tracking the Chinese equity indices. We have chosen a measurement period of three years. Wherever possible we have used the capitalising share class of a fund to compare it to total return benchmarks. In all other cases we have used an adjusted net asset value (NAV) that assumed immediate reinvestment of any distribution made by the ETF. R index is the single period total return of the index n is the number of observations per year For weekly results, we measured returns from Wednesday to Wednesday. For monthly tracking error, returns where calculated from the 13th of the month or preceding weekday. Tracking Difference To measure tracking difference, we have calculated the difference between the fund return and the benchmark return, and annualised the total period value: Tracking Difference = (1 + R nav R index )1/N 1 Where: R nav is the total return of the fund s NAV over the entire measurement period. As well, we ve excluded from our data set the values for any days when a ETF did not produce a net asset value but the index did produce a price, or vice versa. The holiday effect that would result from leaving this data in the calculation has a tendency to make tracking error look much higher than otherwise, when in fact it stems from the absence of a price on a particular day rather than reflecting the normal tracking error we are trying to measure. R nav is the total return of the index over the entire measurement period. N is the number of years. In all of our calculations, we have used the ETFs net asset value, rather than their closing price. We made this decision in order to be consistent for funds that trade on multiple exchanges, as a better reflection of the quality of ETF management rather than secondary market support. It is highly likely that our results would have looked markedly different had we used price data instead of NAV data. Tracking Error To measure tracking error, we have calculated the standard deviation of return differences between each fund and its benchmark, and annualised it: Tracking Error = std (R nav R index ) x n

23 Appendix 2: Defining Common Metrics: Tracking Error (A large portion of this section was first published in the On The Right Track: Measuring Tracking Efficiency in ETFs dated February 2013). Tracking Error Tracking error is often cited as one of the most important considerations when selecting an ETF. It measures the quality of index replication, i.e. how well a fund manager replicates the performance of a specific index. Investors typically expect their ETF to adhere tightly to an index. Different industry participants define tracking error in different ways. Some use the term to refer to the absolute difference in returns between an ETF and its benchmark over a period of time. In other words, they view it as a simple arithmetic exercise where the performance of the benchmark is subtracted from the fund s performance, with the difference representing tracking error. While this definition is easy to understand, it is not the most widely recognised way of calculating tracking error and is fraught with issues. For our purposes, we will refer to the result of this simple arithmetic as an ETF s tracking difference, which we will examine in more detail shortly. As it is more commonly defined, tracking error is a measure of the standard deviation of a fund s excess returns. In this context, excess returns refer to the absolute difference between the fund s performance and that of its benchmark. This is congruent with the definition that ESMA and IOSCO (International Organisation of Securities Commissions) communicated in their latest consultation papers on the subject: ESMA s consultation paper states that tracking error is the volatility of the difference of the returns of the fund and of the returns of the index. Lower tracking error is indicative of more consistency in the periodic deviations between the return of the fund and that of its benchmark. Indices representing certain segments of the market are inherently more difficult for managers of ETFs and other index funds to track than others. This is particularly the case in those instances where the benchmark in question has a very large number of constituents or when the index s components are illiquid or otherwise difficult to access. For example, as we saw in the 2013 European Tracking Report, ETFs tracking emerging market equities tend to exhibit higher tracking error than those based on developed market large cap indices like the DAX or the S&P 500. Sources of Tracking Error When it comes to index-tracking funds, tracking error is a risk, i.e. a risk that the fund s performance will diverge from that of its benchmark. When this occurs, it is crucial to understand why. Tracking error can be caused by many factors, and some are more likely than others to cause mistracking because of their unstable, non-recurring nature. Here is a list of the key factors influencing tracking error. Transaction and Rebalancing Costs Rebalancing costs are typically incurred by physical replication ETFs when an index s methodology requires a reweighting of its constituents or when market events force the rebalancing of a fund. Other transaction costs like stamp duties can increase rebalancing costs. Synthetic ETFs are not directly affected by transaction and rebalancing costs. The level of index turnover will be taken into consideration in the negotiation of the swap s price. Transaction and rebalancing costs could also manifest themselves as a premium between the ETF s market price and its net asset value in secondary market trading. Cash Drag Cash drag can result from periods when funds are forced to hold a portion of their portfolio in cash. This may occur during index rebalancing or be a result of the fund s dividend (or coupon) distribution policy. In the case that an index changes composition, for a physical replication ETF there may be a time lag between the liquidation of the index s old constituents and the addition of its new constituents. During this span, the fund will hold cash. Also, for those ETFs that regularly distribute income to shareholders, there can often be a lag between the time when the ETF receives dividends or coupon payments from its underlying holdings and the time that it ultimately distributes this income to its own investors. Between these dates, which can last from a few days to a few months, the pending distribution sits in an interest bearing