eThesis - Kauppakorkeakoulun sähköiset gradut

In this thesis I examine the daily price deviations of exchange-traded funds (ETF) in the U.S. market using the most recent available data. The ETF market has grown explosively in 2010s, which makes it an interesting subject to research. This study brings an updated situation from Petäjistö (2013) study that examined the price deviations in the U.S. market from 2007 to 2010. I bring additional contribution to the price deviations of newly listed ETFs in their early trading days and the examination of the relationship between fund flows and ETF premiums.

I use the daily prices and net asset values of 1,480 U.S.-domiciled ETFs for determining the price deviations. Data sample period is from 1 January 2014 to 31 July 2015 and the data are retrieved from Thomson Reuters Datastream. I divide the ETFs into subcategories based on Morningstar categorization to see the results for different types of ETFs. Besides the long-term averages, I look at the premiums of newly listed ETFs in their early trading days compared to the long-term averages. In addition, the price dispersions i.e. the daily standard deviations across the sample are examined in relation to the overall market conditions measured by the VIX volatility index and TED spread.

I find the average daily deviations have decreased across the whole sample and subsamples compared to Petäjistö (2013). This could partly be caused by the lack of high-volatility or uncertain market conditions such as 2008-2009 during my sample period but it could also reflect a more efficient arbitrage mechanism due to market growth. I also find that new ETFs generally start trading at a higher premium which normally converges towards the long-term average after a few months of trading. However, the difference from the early days is economically not very significant, mainly 10-20 basis points.

The daily price dispersions in ETF premiums appear to be more related to the TED spread than the VIX index. This could also be explained by the lack of a high-volatility period in my sample. Regression results show that one percentage point in TED spread causes around 10 basis points increase in the price dispersion, while one point increase in the VIX index affects the dispersions by over 2 basis points.

ETF premiums do not seem to be related with the fund flows. According to the (highly significant) regression results, doubling of an ETFs assets would mean around 8 basis points increase in the following day's price premium but around 8 basis points decrease in the current day's premium. The results are mixed and slightly unexpected, but the economical magnitude suggests that there is practically no impact on the ETF premiums.