High-frequency analysis of exchange traded funds' pricing deviation

Purpose – The purpose of this study is to extend the work of DeFusco, Ivanov and Karels by examining pricing deviation of DIA, SPY and QQQQ on intradaily basis. Design/methodology/approach – The DIA is designed to be one hundredth of the DJIA, the SPY is designed to be one tenth of the S&P 500 and QQQQ is designed to be one fortieth of the NASDAQ 100. This feature of ETFs requires the estimation of the difference between the proportional level of the index and the price of the ETF, which is the ETF pricing deviation. Findings – The paper finds that the DIA, SPY and QQQQ pricing deviations are 0.0429, -0.0743 and 0.4298, respectively. The findings indicate that the prices of DIA and QQQQ are on average lower than the underlying indexes. SPY is the exception having a price which is higher than the theoretical price of the S&P 500 index. The author finds that this is due to the increased demand for the SPY. Additionally, the paper provides an explanation for the large change (increase) in the pricing deviation of QQQQ after December 1, 2004 which DeFusco, Ivanov and Karels could not explain. On December 1, 2004 QQQQ trading was consolidated on NASDAQ. The paper finds negative growth in the volume of QQQQ after December 1, 2004 indicating decrease in popularity of this ETF. The decrease in popularity of QQQQ might explain the increase in its pricing deviation. Research limitations/implications – The paper uses high frequency data in the analysis of pricing deviation which might be artificially deflating standard errors and thus inflating the t-test significance values. Originality/value – The paper contributes to the ongoing search in the finance literature of precision ETF performance metrics.