关于传统跟综误差与欧洲合成交易基金的关系分新

1. Introduction
Exchange traded funds (ETFs) were introduced in 1993 when the American Stock Exchange (AMEX) listed the Standard and Poor’s Depositary Receipts (SPDRs), which track the S&P 500. In Europe the first ETFs, tracking the Euro Stoxx 50 and the Stoxx Europe 50, were listed on the German market in 2000. Since the introduction of these products, the industry has experienced rapid growth, and according to a BlackRock Investment Institute report, at the end of June 2011, 1,185 and 1,039 ETFs were listed in Europe and in the USA, respectively. The estimated value of the assets under management for the European ETFs is $321.2 billion, while in the US it is $973.5 billion.
1。介绍
在1993年被引入美国证券交易所（AMEX）上市的标准普尔存托凭证（SPDRs的），跟踪标准普尔500指数的交易所买卖基金（ETF）。在欧洲第一的ETF，跟踪Euro Stoxx 50指数和欧洲斯托克50，于2000年在德国市场上上市。推出这些产品以来，该行业已经历了快速的增长，并根据向贝莱德投资研究所的报告，在2011年6月底，1,185和1,039只ETF上市，在欧洲和美国。欧洲交易所买卖基金所管理的资产估计价值为321.2十亿美元，而在美国，它是973.5十亿美元。

This study investigates the tracking error of traditional and synthetic ETFs2 traded in Europe. The objective of ETFs is to track the returns of the benchmark index as closely as possible. Traditional ETFs attempt to fulfill this objective by holding the benchmark underlying securities, while synthetic ETFs use derivatives contracts, primarily total return swaps. However, neither type of ETF can guarantee that their performance perfectly matches the returns of the benchmark index.In Europe, ETFs are UCITS funds.3 One interesting feature of the European market that does not exist in the US because of regulatory constraints,4 is the synthetic replication method that was introduced in the French market in 2001. Synthetic ETFs hold a basket of securities that usually do not match the index’s underlying securities and than swap their return with that of the benchmark index. The swap counterparty is usually the parent company of the ETF provider. http://www.ukthesis.org/dissertation_writing/Ecommerce/
本研究探讨在欧洲交易的传统和合成ETFs2的跟踪误差。交易所买卖基金的目的是尽可能地跟踪基准指数的回报。传统ETF试图达到这个目的，通过持有标的证券基准，
而合成交易所买卖基金使用衍生工具合约，主要是总回报掉期。然而，无论是类型的ETF可以保证他们的表现完美的的基准index.In欧洲的回报相匹配，交易所买卖基金的UCITS funds.3欧洲市场的一个有趣的功能，在美国不存在由于监管方面的限制，4是合成复制的方法，于2001年在法国市场推出。合成ETF持有一篮子证券，通常不匹配指数的标的证券和不是交换与基准指数收益率。掉期对手通常是ETF供应商的母公司。


Synthetic ETFs appear to have some advantages over traditional funds, but they can also incorporate additional risks. The synthetic replication model makes it possible to create ETFs that track indices that otherwise would be very difficult to reach because of restrictions on foreign investments. Moreover, synthetic ETF providers claim that the synthetic replication method is more efficient and produces a lower tracking error when compared with traditional ETFs. However, the main concern with this replication strategy is counterparty risk. According to UCITS regulations, counterparty risk cannot exceed 10% of the ETF’s net asset value (NAV). However, if the swap counterparty defaults on its obligations, the ETFs might face a loss failing to track the return of the benchmark index. Ramaswamy (2011) notes that the synthetic replication strategy transforms the tracking error risk into counterparty risk and highlights the potential systemic risks that this replication strategy can create. Large liquidation of synthetic ETFs in periods of higher counterparty risk could force to sell the collateral, often illiquid assets. This in turn can hinder the correct functioning of the markets. He concludes that the market risk of these products can be underestimated. Furthermore, the synthetic replication method represents a serious threat to the traditional flagship qualities of ETFs, i.e., simplicity and transparency.
  合成ETF似乎比传统的资金有一定的优势，但他们也可以加入额外的风险。合成复制模型使得它可以创建的ETF，跟踪指数，否则会很难达到，因为对外国投资的限制。此外，合成ETF供应商声称，合成复制的方法更有效，并产生一个较低的跟踪误差时，与传统ETF相比。但是这种复制策略，主要关注的是交易对手风险。根据UCITS规例，
交易对手风险不能超过ETF的资产净值（NAV）的10％。然而，如果掉期对手方未能履行其责任，交易所买卖基金可能面临亏损，未能跟踪基准指数的回报。拉马斯瓦米（2011）指出，合成复制策略，将追踪误差风险，交易对手风险，并强调潜在的系统性风险，这种复制策略可以创建。大清算合成ETF的交易对手风险较高的时期，可以强制出售抵押品，往往是流动性不足的资产。反过来，这可能会妨碍市场的正常运作。他得出结论认为，这些产品的市场风险可能被低估了。此外，合成复制的方法，代表传统的旗舰本色交易所买卖基金，即，简单性和透明度的严重威胁。
  
Previous empirical studies identify the main factors that give rise to tracking errors, including transaction costs, index-composition changes, corporate activity, fund cash flows, index volatility, the reinvestment of dividends, and index replication strategies (Chiang 1998; Elton,
Gruber, Comer, and Li 2002; Frino and Gallagher 2002). The volatility of the exchange rate is a further source of tracking error (Shin and Soydemir 2010).

This article finds that both traditional and synthetic European ETFs are affected by a significant tracking error. This analysis provides evidence that ETFs that follow a synthetic replication strategy rather than holding the underlying benchmark securities, enjoy a lower tracking
error and a higher tax efficiency. Furthermore, they are particularly efficient when tracking emerging-market benchmarks. However, synthetic ETFs underperform both the benchmarks and the traditional counterparts

This paper extends the previous literature by examining the tracking ability of traditional and synthetic European ETFs offered by the leading providers in Europe. The synthetic replication method, although widely used by European ETF providers, has never been deeply analyzed by previous studies. However, synthetic ETFs became a serious concern for regulators6 also because of the potential implications for the stability of the financial system. This research is motivated by the need to provide investors, who seek a transparent and cost-efficient passive investment strategy, with more insight regarding the tracking ability of traditional and synthetic ETFs.

The remainder of the paper is organized as follows. The next section reviews the related literature. Section 3 presents the data and the empirical design. Section 4 reports the results and the final section concludes.

2. Literature Review
Previous academic research, based primarily on US and Australian markets, documents significant tracking errors generated by index funds and ETFs (Elton, Gruber, Comer, and Li 2002; Frino and Gallagher 2001 and 2002). Elton, Gruber, Comer, and Li (2002) and Harper, Madura, and Schnusenberg (2006) compare the return of the ETFs and the corresponding index return. Roll(1992), Pope and Yadav (1994) and Larsen and Resnick (1998) identify three metrics to measure the tracking error as the dispersion of the fund’s NAV return relative to the benchmark return. Tracking error can also be evaluated using market prices instead of the NAV (Harper, Madura, and Schnusenberg 2006). Any market-price deviations from NAVs should disappear quickly because of the in-kind and in-cash creation/redemption processes (Elton, Gruber, Comer, and Li 2002; Engle and Sarkar 2006), but the market price tracking error can substantially deviate from the NAV tracking error. DeFusco, Ivanov, and Karels (2011) show that the pricing deviations of the Spiders, Diamonds and Cubes7 are different from zero. The authors claim that the pricing deviation can be considered an additional cost of administering an ETF. Furthermore, seasonal patterns in tracking errors have been detected (Frino and Gallagher 2001 and 2002; Frino, Gallagher, Neubert, and Oetomo 2004; Rompotis 2010).

Chu (2011) investigates the tracking errors of ETFs traded in Hong Kong and finds that they are higher compared with those in the US and Australia. He assumes that one possible explanation could be the use of synthetic investment tools instead of holding the underlying stocks. Other recent research by Blitz, Huij and Swinkels (2010) examines the tracking error of
European index funds and ETFs as measured by their underperformance against the gross total return indices. They find that European funds underperform their benchmarks and that dividend withholding taxes and fund expenses have similar explanatory power.

3. Data and Method
In this section, the data and the method used in this research are described.
3.1. Data
The analyzed sample comprises traditional and synthetic ETFs listed in Europe. According to the BlackRock Investment Institute report, at the end of June 2011, 1,185 ETFs were listed in Europe. For the purposes of this study, the ETFs that track the major European and global stock market indices are selected. To ensure a significant data history, only the ETFs that were created before September 2007 are considered to allow for the analysis of a common period of four years
that starts with September 2007 and ends in August 2011. This four-year period has been selected as a trade-off because it offers a meaningful data history while also including a significant number of both traditional and synthetic ETFs from the leading providers of ETFs in Europe.

The final sample consists of 48 ETFs that track 20 different benchmark indices: 21 are traditional ETFs and 27 use a synthetic replication method. Providers comprise Blackrock (iShares),
Sociétè Générale (Lyxor), Deutsche Bank (db x-trackers), State Street Global Advisors, BBVA, UBS, Amundi, EasyEtf, Powershares, and Credit Suisse. The replication method, as well as total
expense ratio (TER) and other characteristics are acquired from the fund prospectuses. The funds are domiciled in Ireland, France, Spain, and Luxembourg. This sample is one of the most diverse for ETFs in the academic literature in terms of the benchmark indices, the providers and the
replication strategies included. Table 1 reports the profiles of the 48 ETFs, including the name, replication method, net or gross benchmark index, primary listing, inception date, and total expense ratio (TER).


For the other three regressions that test the significance of the determinants of the tracking error measured by TEAAD,TESDRD and TESER, the coefficients on TER are statistically indistinguishable from zero. Thus, while fund expenses generate the tracking error measured by
TERD, i.e., reduce the performance of the ETF, they do not generate variability of the return differences of the ETF. The coefficient on SYNT is negative and significant. This also confirms the descriptive statistics. Synthetic ETFs are more efficient in tracking benchmark indices because they generate a lower tracking errors compared to traditional ETFs. The coefficients on OPT and EMERG are positive and significant, that shows ETFs that track the underlying benchmark by holding a subset of the constituent securities and ETFs that track emerging-market indices, generate a higher tracking error. The coefficient on the interaction term SYNT × EMERG is negative and statistically significant as expected. It shows that synthetic ETFs are more efficient when tracking emerging markets benchmark indices compared to the ETFs that follow a traditional replication method. Finally, the coefficients on the dummy variables LU, FR, and ES are significant. Therefore the domicile of the fund is significant in explaining the tracking error.