Purpose The purpose of this paper is to present the results of a study on investor behavior in exchange-traded fund (ETF) markets. The standard feedback trading model of Sentana and Wadhwani (1992) is used in a sample of 18 ETFs contracts in Brazil, China, South Africa, Korea, Mexico and India, as well as three ETFs contracts in the US market. Design/methodology/approach The sample includes data on daily closing prices and net asset values (NAVs) for three ETFs from each of the emerging markets of Brazil, China, Mexico, Korea and India, as well as on three ETFs from the US market. The authors used the earliest start date available in the Thomson Reuters database pertaining to all of the ETFs, and all series ended on May 5, 2017, and applied the well-established Santana and Wadhwani (1992) seminal model to evaluate evidence of feedback trading in the sample. Findings The empirical analysis suggests that there is evidence of feedback trading in emerging markets such as Brazil, Korea, Mexico and India, while there is no such evidence for the US market. The results are consistent with the view that developed markets investors are prone to pursue fundamental-driven investment strategies, while emerging markets investors appear to have informational guided behavior. Research limitations/implications Emerging markets still make up a very small part of the global ETF market, led by the USA. Nevertheless, it is extremely important that studies of this nature be gradually expanded as these markets grow, in order to verify how emerging markets compare to their developed counterparts in terms of the efficiency of information sharing and rationalization of its operations. Practical implications Emerging markets policy makers could benefit from these findings by stimulating new mechanisms that could minimize informational asymmetry and the persistence of so-called noise traders, a phenomenon observed recently in studies regarding ETF markets (Brown, Davies and Ringgenberg, 2018). Originality/value The behavior of investors was investigated by analyzing a sample of 18 ETFs from the emerging markets of Brazil, China, South Africa, Korea, India and Mexico, as well as three ETFs from the US market. Despite of being investigated separately both emerging (Charteris et al., 2014) and developed markets (Chau et al., 2011), the innovation consists in comparing those markets in a single study, pursuing to explain potential reasons for the differences observed between developed and emerging markets.

This study examines the tracking efficiency of a sample of exchange-traded funds (ETFs) from seven different emerging and developed markets, in bullish and bearish market conditions, using the data from daily closing prices. It seeks to address two major questions. First, do ETFs from both developed and emerging markets have the same behaviour regarding tracking their underlying indexes? Second, is ETFs tracking ability affected by events that could change market conditions from a bullish to a bearish pattern, and vice-versa? Our findings suggest that tracking error appears to be higher in emerging markets when compared to developed markets. Furthermore, tracking error was found to be relatively higher in bearish conditions for developed markets, while this was quite the opposite in emerging markets.