| dc.description.abstract | Since the listing of the Satrix 40 in November 2000, Exchange Traded Fund (ETFs) have grown to become an investment vehicle of choice amongst retail and institutional investors of the Johannesburg Securities Exchange (JSE). Albeit gaining such an enormous traction, investors' remains curious about ETFs ability to successfully replicate the movements of their target benchmark indices and also their capability to yield arbitrage profit opportunity through mispricing. In addition to that, investors are also interested to know whether ETFs as an index tracking investment vehicle are resilient in variously cycles of the economy. Motivated by this gap in the body of knowledge, this research undertakes to evaluate the tracking ability and pricing efficiency of 19 ETFs listed on the JSE over various cycles of the economy. According to Faulkner, Loewald and Makrelov (2013) South African economy experienced the effect of the 2008 global financial crisis between 1 September 2008 and 30 June 2009. For that reason, the examination period of this research is segmented into four main categories namely: full examination period which spans from the launch date of each of the ETF under review until 30 September 2015, pre-crisis period that is between the launch date and 29August 2008, crisis-period dated 1 September 2008 and 30 June 2009 and the post-crisis or the recovery phase being 1 July 2009 through 30 September 2015. The tracking ability results across all the sub-periods suggested that, on average, ETFs yields daily returns which closely resemble that of their target benchmark indices but with relatively high level of volatility. With regard to the tracking error as another tracking ability measurement, it was discovered that the ETFs under review were inadequately replicating the movements of their target benchmark indices irrespective of the economic cycle. In tandem with the evidence documented by Mateus and Rahmani (2014) from the London Stock Exchange (LSE), tracking errors were substantially high during the 2008 global financial crisis as opposed to the prior and the post crisis period. Across all the examination periods, sizeable amount of tracking error was found to be associated to the ETFs which mimics the international broad-market access underlying indices. Amongst other things, the diversity of these indices as well as the trading hours overlap between the JSE and their host market were found to be the key attributing factors. On the contrary, ETFs which replicates most liquid target benchmark indices such as the FTSE/JSE Top 40 index appeared to have lower tracking error on relative basis. In this regard, the liquidity of the FTSE/JSE Top 40 index proved to be the main attribute. Apart from the diversity or the liquidity of indices, the length of the examination period also had a significant influence towards the magnitude of tracking errors. In this instance, shorter examination period were found to be characterised by noise or volatility in the market which makes it difficult for the ETFs providers to promptly rebalance their portfolios and align them to their target benchmark indices. Over and above these factors, this research discovered that tracking errors across all the sub-periods were largely driven by management fees and daily volatility of the ETFs market prices, more especially during the crisis period. On the one hand, trading volume and the effect of dividends distribution had a negative influence towards the magnitude of tracking errors. On the question of how efficient these 19 ETFs are, the empirical findings revealed that significant deviation between the ETFs closing price and the Net Asset Value (NAV) does exist either being a discount or premium. In line with the prior work on the JSE by Charteris (2013), ETFs which mimics local based indices were found to be trading mostly on a discount to the NAV whilst the opposite was true in the case of the international broad-market access ETFs. At the same token, international broad-market access ETFs portrayed sizeable amount of premiums across all the cycles of the economy. In line with the analysis of tracking errors, such enormous premiums were mainly driven by lack of synchronicity in the trading hours between the JSE and host market wherein these ETFs target benchmark indices are listed. Empirical literature suggests that ETFs that exhibit discount and premium which fails to persist for more than one trading day are deemed to be efficiently priced since there is limited opportunity to arbitrage. On that note, this research found that mispricing of ETFs which mimics most liquid indices such as the domestic broad-market access and sectorial indices disappears within a period of one trading day. For that reason, majority of these ETFs were considered to be efficiently priced against their NAV. Contrarily, discounts and premiums exhibited by ETFs which mostly replicate style based and the international broad-market access indices appeared to be persistent even to the fifth trading day. From the attribution point of view, the complexity of these ETFs underlying indices as well as the trading hours overlap between the JSE and the host market of these indices were found to be the main drivers of such level of mispricing. In addition to that, attribution analysis through linear regression proved that transaction cost (bid-ask spread), daily volatility of the ETFs market prices as well as the impact of trading volume had a positive influence towards the existence of discounts and premiums observed across all sub-periods. | en_US |
