The Research on Tracking Error of Shen 100 ETF
|Keywords||ETF Tracking error Stationary Error correction model|
ETF is a kind of index fund that have been swiftly developing from the 1990s.It can be requested or redeemed in the primary market, also can be bought or sold in the secondary market. An ETF targets copying the trend of the targeted index, and it’s key to exist is following the targeted index closely. Therefore, testing how closely that an ETF follows the targeted index is the most important norm that test whether it was operated successfully, and the tracking error is the norm that has been used widely.In China, the most researches about ETF are in limitations that they regress the ETFs’price and the targeted index by OLS simply, and they didn’t analysis the time-series’stationary, so the spurious regression problem may come into being. In this article, I take Shen 100 ETF as the research object, which is the outstanding achievement champion fund in the first half of 2009. I analyzed Shen 100 ETF’s data in latest 3 years that began the year it listed, and calculate its tracking error of net value by Tracking Difference and MSD separately. The tracking error is very small, which conforms to the standard in Shen 100 ETF’s collection directions. Then, the time-series of net value and index are tested by stationary, and both are I(1) series are obtained by ECM. The relationship is close. However, when we replace the net value of Shen 100 ETF with its price, we found that the error correction term of price is much bigger than the error correction term of the net value. This conclusion can explain that the fluctuation of the price is fiercer than the net value in short time. Therefore, the effect of copying the index by Shen 100 ETF in the primary market is greater than in the secondary market.In the end of this article, the reason why the tracking error is emerged is analyzed. We found that the tracking error is affected outstanding by the management cost and the share bonus of the component, and the proportions of the influence by the two factors are close. However, the influence of the net request/redeem to its tracking error is not very clearly.