Thesis (MBA)--Stellenbosch University, 2000. / ENGLISH ABSTRACT: Similar trading patterns, or so-called herding by institutional investors has interested market participants and academics for some time. Recent international research has established empirical evidence of this phenomenon. The context of this study is to observe evidence of herding by institutional investors in the South African Unit Trust Industry during the period December 1991 to September 1999. Furthermore, it investigates possible relationships between institutional herding and other topics of interest. These include risk profile of funds, market volatility, house view of management companies, size of herd, returns on hypothetical portfolios and on individual shares. Empirical results indicate evidence of herding over the period, albeit at a relatively small level. The average count herding measure of 2,4% is similar to levels recorded in the American Mutual Fund and Pension Fund Industries. A currency imbalance ratio is also used to measure herding, and indicates average levels of herding of 8,3%. This measure is deemed to reflect greater accuracy. as it measures actual currency movements (size of positions) as opposed to number of funds active in a share. On average, herding is present as often on the buy side as on the sell side of the market. Intuition would suggest that as the size of a herd increases (number of funds taking similar positions), the degree of herding increases. Evidence from this study corroborates with an American study, indicating to the contrary. There is a statistically significant negative relationship between the size of a herd and degree of herding. An observation of herding measures relative to the risk profile of funds indicates the presence of a statistically significant positive relationship. Highest levels of herding are recorded in aggressive growth funds and lowest levels in income/growth funds. This supports the rationale that aggressive growth funds, by nature of their investment objective, follow high growth firms. Analysts possess less accurate information regarding future earnings; hence the greater herding levels, for whatever reason. House views, imposed by individual management companies, may also lead to higher herding levels. There is tentative evidence of the presence of this practice, particularly in two of the seven companies observed. The relationship between herding and equity market volatility indicates an interesting phenomenon. There is the presence of a strong positive relationship between quarterly volatility estimates and levels of herding. This relationship changes materially for volatility levels in excess of 9%, indicating the clear presence of a structural breakpoint. For quarterly volatility estimates greater than 9%, the relationship weakens substantially and the slope of the relationship flattens. A quarterly time series of portfolio returns is calculated relative to levels of herding to observe trading strategies practiced by fund managers. The findings indicate weak evidence of funds following positive feedback trading strategies. Furthermore, funds trade in past winners more often than in past losers, indicating the absence of window dressing strategies. There is strong evidence of funds following profit taking strategies at quarterly intervals. The absence of return reversals indicates the absence of over reaction at quarterly intervals. / AFRIKAANSE OPSOMMING: Soortgelyke handelspatrone, of sogenaamde samedromming van institusionele beleggers, is al geruime tyd vir markdeelnemers en akademici van belang. Onlangse internasionale navorsing het empiriese bewys van hierdie fenomeen opgelewer. Hierdie studie is daarop gefokus om bewyse van samedromming deur institisionele beleggers waar te neem, binne konteks van die Suid Afrikaanse Effektetrustbedryf gedurende die periode Desember 1991 tot September 1999. Daar word verder ondersoek ingestel na 'n moontlike verhouding tussen institusionele samedromming en ander onderwerpe van belang. Ingesluit hierby is die risikoprofiel van fondse, markonbestendigheid, 'house view' van bestuursmaatkappye, grootte van die samedromming, opbrengste van hipotetiese portefeuljes en individuele aandele. Empiriese resultate dui aan dat bewyse van samedromming gedurende die relevante periode wel bestaan, alhoewel dit op 'n relatiewe klein skaal plaasvind. Die gemiddelde telling samedrommingsmaatstaf van 2,4% is soortgelyk aan die vlakke waargeneem in die Amerikaanse Mutualfonds en Pensioenfondsbedrywe. 'n Geldeenheid onewewigtigheidsverhouding word ook gebruik as maatstaf van samedromming en dui 'n gemiddelde samedrommingsvlak van 8,3% aan. Die laasgenoemde maatstaf word as meer akkuraat beskou aangesien dit werklike geldeenheidbewegings (grootte van die posisies) reflekteer, in teenstelling met die aantal fondse wat aktief betrokke is by 'n spesifieke aandele. Samedromming is oor die algemeen ewe veel teenwoordig tydens die koop en verkoop van aandele in die mark. Intuïsie dui aan dat soos die grootte van die samedromming toeneem (aantal fondse wat dieselfde posisie inneem), die intensietyd van samedromming ook toeneem. Bewyse van hierdie studie bevestig die bevindinge van 'n Amerikaanse studie wat aandui dat die teenoorgestelde waar is. Statisties gesproke is daar 'n wesenlike negatiewe verhouding tussen die grootte en intensietyd van samedromming. 'n Waarneming van samedrommingsmaatstawwe, relatief tot die risikoprofiel van fondse, dui die teenwoordigheid van 'n statistiese wesenlike positiewe verhouding aan. Die hoogste vlakke van samedromming word waargeneem by aggresiewe groeifondse en die laagste vlakke by inkomste-/groeifondse. Hierdie bevinding staaf die gedagte dat aggresiewe groeifondse, as gevolg van die aard van hul beleggingsdoelwit, hoë groei maatskappye volg. Ontleders beskik oor minder akkurate inligting ten opsigte van toekomstige opbrengs en gevolglik is daar groter samedrommingsvlakke vir watter rede ookal. 'House views', soos voorgeskryf deur individuele bestuursmaatskappye, mag ook tot hoër vlakke van samedromming lei. Voorlopige bewyse ten opsigte van die teenwoordigheid van hierdie praktyk bestaan, veral in twee van die sewe maatskappye waargeneem. Die verhouding tussen samedromming en aandelemarkonbestendigheid reflekteer 'n interessante fenomeen. 'n Sterk positiewe verhouding is teenwoordig tussen kwartaalikse onbestendigheidsskattings en vlakke van samedromming. Hierdie verhouding verander wesenlik vir onbestendigheidsvlakke groter as 9%, wat die teenwoordigheid van 'n strukturele breukpunt duidelik aantoon. 'n Kwartaallikse tydreeks van portefeulje-opbrengste word bereken relatief tot die vlak van samedromming om handelstrategië, soos deur fondsbestuurders toegepas, waar te neem. Bevindinge dui aan dat daar gebrekkige bewyse is van fondse wat positiewe terugvoerstrategië volg. Daar is ook gevind dat fondse meer gereeld handel in gewese wenners as in gewese verloorders wat 'n afwesigheid van uitstallingsstrategië aandui. Daar is besliste bewyse van fondse wat winsbejagstrategië volg met kwartaalikse tussenposes. Die afwesigheid van omgekeerde opbrengste dui die afwesigheid van oorreaksie aan met kwartaalikse tussenposes.

This thesis adopts an empirical approach to examine various market microstructure issues, using data from the National Stock Exchange of India (NSE). Whilst the respective empirical analyses may be considered as self-contained investigations, they are primarily linked through the common objective of understanding the mechanics of the pricing process as it occurs on actual markets, using the NSE as exemplar. The first major focus of the dissertation is non-synchronous trading: empirical evidence of nonsynchronicity is obtained by testing for predictability as between indices of different levels of liquidity. A simple test of the analysis of trading-break returns is proposed to infer whether predictability may be mainly attributable to non-synchronous trading or whether it constitutes a delayed adjustment of traders' expectations. The second question tackled in the thesis is whether volatility on the NSE may be considered as justified or excessive. Rathert han adopting the established methodology of comparing stock price changes to information about expected dividends, the research question is split up into two subsidiary ones. The first question is whether volatility is related to information flows, whilst the second related questionc oncernst he relationship betweenv olatility and returns. Three sources of excessive volatility are pin-pointed. Monday effects are found in index data but not in the underlying stocks-indicating index fluctuations which are not information-related. A second indicator of excessive price movements is the pronounced volatility which coincides with the fiscal year end of quoted companies but which is not accompanied by a similar increase in long-term returns. A third indication of unjustified price fluctuations is that volatility seems unrelated to returns when considering a long-term time series. The third topic of the thesis relates to the efficacy of opening and closing call auctions. This issue may be considered as the crux of the dissertation and it is tackled by analysing the effects of the suspension of a call auction system on NSE. Changes in volatility, efficiency and liquidity following the suspension are analysed, and an event study is presented. The relationship between call auctions and long-term volatility is also investigated. The findings suggest that the expected benefits of call auctions may not always materialise, possibly due to an inappropriately structured auction, or because a liquidity threshold for stocks must be surpassed for the expected benefits to accrue.

參考Fang 2003年研究方法架構，我們檢驗了結合策略(結合技術分析法則和時間序列模型)應用在六個亞洲股票市場。由於技術分析法則和時間序列模型皆可利用過去歷史資訊來預測報酬，所以結合策略的實證結果優於技術分析法則和時間序列模型。此篇中超額報酬的計算是與買進持有相比較下未考慮交易成本的超額報酬。實證結果顯示，結合策略在完整樣本中可以成功的預測資產報酬，在六個國家的平均上，結合策略的超額報酬為0.19%優於技術交易法則下的0.13%和時間序列模型下的0.17%。並且，發現在新興國家如台灣、泰國、馬來西亞和南韓的預測能力比在已開發國家市場如香港和日本還要來的好。預測能力可被低階的自我相關係數解釋。除此之外，發現我們的預測能力受到非同步交易的影響。非同步交易所造成的衡量誤差使得超額報酬下降，但是我們的預測能力還是存在的。 / Following Fang and Xu (2003), we examine trading strategies combining technical trading rules and times series forecasts on six Asian stock markets. Since both technical trading rules and time series models can exploit predictable components as function of past prices or returns, the combined strategies outperform both technical trading rules and time series forecasts. The excess returns before transaction costs for each rule and country are compared to a passive buy-and-hold strategy. The combined strategies are quite successful in predicting asset returns in full samples. On average the buy-sell returns for combined strategies are 0.19% much higher than 0.13% for technical trading rules and 0.17% for time series models. Besides, we also find that all three rules have more explanatory power in emerging markets such as Taiwan, Thailand, Malaysia and Korea than more developed markets such as Japan and Hong Kong. The predictability can be explained by significant low-order autocorrelations in returns. Moreover, excess returns (pre-trading costs) for both time series models and combined strategies can be partially attributed to the measurement errors arising from non-synchronous trading. The non-synchronous trading bias reduces but does not eliminate the predictive power of combined strategies.