Browsing by Author "Van Rensburg, Paul"

Now showing 1 - 20 of 43
  • No Thumbnail Available
    Item
    Open Access
    A comparative analysis of machine learning models for forecasting JSE Stock Returns
    (2025) Muir, Cameron James; Van Rensburg, Paul
    This study examines the application of machine learning models to predict the cross-section of Johannesburg Stock Exchange (JSE)- listed share returns. Four models are developed and compared using monthly data from 2005 to 2021: neural networks, random forest, long short- term memory (LSTM) networks, and conventional linear regression. The explanatory variables comprise nine firm-specific financial metrics, motivated by prior research. The sample is divided into a training period (2005–2016) and a testing period (2016–2021), further split into 1-year, 3-year, and 5-year testing intervals. The results show that the LSTM model performsbest across most evaluation metrics and investment scenarios, with the random forest model close behind, offering slightly better risk-adjusted returns.
  • Thumbnail Image
    Item
    Open Access
    Analysis of the cross-section of equity returns on the JSE Securities Exchange based on linear and nonlinear modeling techniques
    (2010) Hodnett, Kathleen E; Van Rensburg, Paul
    This research investigates the relationship between firm-specific style attributes and the cross-section of equity returns on the JSE Securities Exchange (JSE) over the period from 1 January 1997 to 31 December 2007. Both linear and nonlinear expected returns forecasting models are constructed based on the cross-section of equity returns. A blended approach combining a linear modeling technique with a nonlinear artificial neural network technique is developed to identify future potential top performing shares on the JSE. 1. Both linear and nonlinear models identify book-value-to-price and cash flow-to-price as significant styles attributes that distinguish near-term future share returns on the JSE. 2. This thesis found updating the identity of attributes is equally important as updating the factor payoffs of attributes in applying the stepwise regression approach. 3. Nonlinearity on the JSE equity returns is found to complement the forecasting power of linear factor models. 4. In terms of artificial neural network modeling, the extended Kalman filter learning rule introduced in the thesis is found to outperform the traditional back-propagation approach. 5. This thesis found that updating the identity of attributes via a genetic algorithm in the nonlinear forecasting models is superior to the static nonlinear forecasting models. 6. Both linear and nonlinear models are found to be more adequate in identifying future outperformers than identifying future underperformers on the JSE. The results of the research provide for potential alpha generating stock selection techniques for active portfolio managers in the South African equity market using the blended linear-nonlinear approach.
  • Thumbnail Image
    Item
    Open Access
    Applications of global equity style indices in active and passive portfolio management
    (2010) Hsieh, Heng-Hsing; Van Rensburg, Paul
    The success of the Fama and French 3-factor model in explaining empirical anomalies of the Capital Asset Pricing Model (CAPM) suggests that style investing which places portfolios out-of-sync with the broad market has the potential to generate significant alpha. Since momentum abnormal return is the only anomaly that is not explained by the 3-factor model, it could well be the third style-based factor in addition to the size and the value factors to complete the model. With the goal of searching for practical mean-variance efficient allocation mechanisms in the global capital market, this study develops and examines the long-only, long-short leverage and market neutral strategies from the global size, value and momentum proxies along with the Morgan Stanley Capital International World Index over the examination period, 1 January 1991 to 31 December 2008.
  • Thumbnail Image
    Item
    Open Access
    An assessment of the style and performance of South African institutional fund managers
    (2013) Moore, David; Van Rensburg, Paul
    This paper aims to expand on the growing area of fund style classification and benchmarking research in developed markets by extending such analyses to the South African context. ... A differentiating feature of this study is both the style indices used and the sample of fund manager return data in the South African context. The style indices used were sourced from A-DEX, which unlike those used in Scher and Muller (2005) comprise a greater sample of JSE listed companies and are fully tradable. Furthermore, the data sample compiled by RisCura Solutions (Pty) Ltd and contains returns from a total of sixty South African institutional fund managers. ... The current study analyses one of the largest samples of institutional manager return data in the South African context.
  • Thumbnail Image
    Item
    Open Access
    Asset allocation and Regulation 28
    (2016) Mjebeza, Athenkosi; Van Rensburg, Paul
    This paper aims to determine the impact Regulation 28 has on optimal asset allocation. The revised Regulation 28 of the pensions fund act came into effect as of 1 July 2011 which imposed certain restrictions or constraints on pension funds under direct control of trustees. This study evaluates some of the constraints imposed on the Regulation 28 through the use of Markowitz (1952) efficient frontier framework and a non-parametric model. With offshore allocation increased to 25% and an additional 5% to African (ex SA) markets the study also explores the diversification prospects to international, emerging and African (exSA) markets. The findings suggest that international markets bring about increase benefits to South African markets; however, when the Regulation 28 constraint is imposed the benefits slightly diminishes. Further analysis show that emerging and African markets bring little to no benefits to optimal South African pension fund allocation. Locally, the study looked at the gold index and the findings suggest that the gold asset class increases the welfare of an investor and it's a safe haven asset class.
  • Thumbnail Image
    Item
    Open Access
    Asset allocation in the South African environment
    (2014) Mahoney, Kevin; Van Rensburg, Paul
    The aim of this paper is to find solutions to the asset allocation problem in the South African environment. These solutions look at a variety of different investor's preferences. These include an investor's age, risk aversion and required levels of returns. To do this, an analysis was done of prior research, so the most up to date mean-variance asset allocation model could be developed. Returns from 10 different indices, over different asset classes were gathered. The indices of importance were found to be: All Bond Index (ALBI), Inflation Linked All Maturities Index (ILB), Salient's Momentum Active Index Fund (MOME), Salient's Value Active Index Fund (VAL), South African Short Term Fixed Interest Index (STEFI) and South African Property Index (SAPY).
  • Thumbnail Image
    Item
    Open Access
    The behaviour of style anomalies in worldwide sector indices : a univariate and multivariate analysis
    (2007) Acres, Daniel Nigel Gerard; Van Rensburg, Paul
    The aim of this thesis is to explain the cross-section of International Classification Benchmark (ICB) level 4 (sector) index returns. A worldwide study of 48 developed and emerging countries is conducted, considering up to 38 sector indices per country. In cluster and factor analyses of the sector returns all the developed markets are found to cluster together, as are the emerging markets, suggesting diversificationary benefits from investing across the two. The one-month-ahead return forecasting power of 35 sector-specific attributes is investigated over an in-sample period from 31 January 1995 to 31 December 2001 and an out-sample period from 31 January 2002 to 31 December 2005. The data is adjusted for look-ahead bias, outliers, influential observations and non-uniformity across markets. Monthly sector returns are cross-sectionally regressed on the attributes in a similar fashion to Fama and MacBeth (1973). Sector returns are considered both before and after risk adjustment with the Capital Asset Pricing Model (CAPM), the Arbitrage Pricing Theory (APT) model and Solnik's (2000) version of the International CAPM (ICAPM). The ICAPM is found to be the best performing model but, in general, the evidence does not support covariance-based models of asset pricing. Nine attributes are found to be significant and robust over the two sample periods namely cash earnings per share to price (CP), dividend yield (DY), cash earnings to book value (CB), 6 and 12-month growth in cash earnings, to price (C-6P & C-12P), 12 and 24-month growth in dividends, to price (D-12P & D-24P), the payout ratio (PO) and 12-month prior return (MOM-12). All the significant attributes from the univariate regression tests are found to payoff consistently in the positive direction when tested with the nonparametric Sign Test. Nine of the significant attributes namely book value per share to price (BP), dividend yield (DY), earnings yield (EY), 6-month growth in cash earnings, to price (C-6P), cash earnings to book value (CB), 24-month growth in dividends, to price (D-24P), 24-month growth in earnings, to price (E-24P), 12-month and 18-month prior return (MOM-12 & MOM-18) are also found to have significantly low frequencies of changes in payoff direction when assessed with the nonparametric Runs Test. Seven style timing models are developed, all of which produce significantly accurate payoff direction forecasts for most of the significant attributes. The timing models are however generally inaccurate in forecasting the magnitude of the payoffs. Very little seasonality is observed in the payoffs to the significant attributes. Two sets of seven 'stepwise optimal' and 'control' multivariate models are constructed from the significant univariate in-sample attributes in order to forecast the payoffs to the factors in a controlled multifactor setting. The stepwise optimal models are derived from a stepwise procedure, whilst the 'control' models comprise all the attributes which are found to be significant in one or more of the 'optimal' models. The forecasting power of the all the models is found to be below an exploitable level; of the 'control' models the single exponential smoothing model is the most accurate outsample performer. Weighted Least Squares (WLS) models are used to allow for the possibility of heteroskedasticity, which may exist in the cross-section of worldwide sector returns. The WLS models are ineffective in improving forecasting power when the inverse of the 12-month rolling standard deviation of the residuals is used as the weight series.
  • Thumbnail Image
    Item
    Open Access
    The behaviour of style anomalies on the Australian Stock Exchange : a univariate and multivariate analysis
    (2005) Janari, Emile; Van Rensburg, Paul
    Recent attempts to empirically verify the Sharpe (1964), Lintner (1965), Moss in (1966), and Black (1972) Capital Asset Pricing Model (CAPM) have identified numerous inconsistencies with the model's predictions. A number of variables have displayed evidence of the ability to explain the cross-sectional variation in share returns beyond that explained by data. These anomalous effect have become known as "style effects " or "style characteristics". This thesis sets out to examine the existence and behaviour of these style-characteristics over the period June 1994 to May 2004. A data set of 207 firm-specific attributes is created for all Australian Stock Exchange (ASX) All Ordinaries stocks listed on 1 September 2004. The data are adjusted for both thin trading and look-ahead bias. The study largely follows the tests of van Rensburg and Robertson (2003) who adopt the characteristic-based approach of Fama and Macbeth (1973). Attributes are tested for the ability to explain the cross-sectional variation in ASX share returns beyond that explained by the CAPM and a principal-components-derived APT model. Similar significant characteristics are found when unadjusted and both risk-adjusted returns sets are examined. The set of significant characteristics d e rived from the unadjusted returns test is then simplified using correlation analysis and an agglomerative hierarchical clustering algorithm, resulting in a list of 27 variables that are not highly correlated with each other. These characteristics are divided into nine interpretation groups or combinations thereof, namely: (1) Liquidity; (2) Momentum; (3) Performance; (4) Size; (5) Value; (6) Change in Liquidity; (7) Change in Performance; (8) Change in Size; and (9) Change in Value. While the existence of the anomalies found in prior Australian literature (size, price-per-share, M/B, cashflow-to-price, and short- to medium-term momentum) is confirmed, the PIE effect is not found to be significant in this study. As these previously documented anomalies only cover five of the final 27 characteristics, this paper identifies 2 2 new Australian anomalies. Six style-timing models are evaluated for the ability to forecast the monthly payoffs to the 27 characteristics. A twelve-lag autoregressive model convincingly displays the best performance against moving average and historic mean models. Parametric and nonparametric tests find inconclusive evidence of seasonality in the monthly payoffs to the attributes. The 27 significant style characteristics are then used to construct a multifactor style-characteristics model which comprises a set of factors that are significant when simultaneously cross-sectionally regressed on share returns. The employed construction method yields a five-factor style model for the ASX and comprises: (1) prior twelve-month momentum; (2) book-to-market value; (3) two-year percentage change in dividends paid; (4) cashflow-to-price; and (5) two-year percentage change in market-to-book value. Finally, a step wise procedure is performed using six style-timing models. Five dynamic multifactor expected return models are created and contrast with a static multifactor expected return model similar to that used in van Rensburg and Robertson (2003). The derived expected return models have between three and thirteen factors. While all six models display good forecasting ability, the dynamic (trailing moving average) models all perform better than the static (historic mean) model. This is convincing evidence that the asset pricing relationship follows a dynamic model.
  • Thumbnail Image
    Item
    Open Access
    Comparative performances of capital protection strategies in the South African market
    (2015) Du Plessis, Richard Michael; Van Rensburg, Paul
    The performance of cash protection strategies implemented in the South African market are investigated in order to establish if investors are able to add value through the use of dynamic portfolio insurance methods. The analysis is performed, using monthly data, from January 1961 to August 2014 using six alternative methodologies including both a Fixed Rate and Rolling Average Stop-Loss approach, a Lock-In approach, a Constant Mix strategy, a Constant Proportion Portfolio Insurance ("CPPI") approach and an alternative CPPI approach using a Ratchet mechanism. The results indicate that the use of such cash protection strategies can markedly improve portfolio performance from a risk return perspective compared to a pure diversified investment strategy. Notably, the use of older, simpler trading strategies such as the Stop-Loss and Lock-In approaches at optimum threshold levels can still offer investors higher risk to reward benefits with less commitment required. These strategies, though, lack the flexibility observed with the more recently developed dynamic trading strategies in terms of providing for varying risk appetites.
  • Thumbnail Image
    Item
    Open Access
    Cross-sectional and time-series momentum on the JSE
    (2016) Lockhart-Ross, Simon; Van Rensburg, Paul
    This research report documents multiple accounts of past return-based momentum strategies employed on South African-listed equities over the period 2002.02-2015.05. Two cross-sectional momentum approaches-strategies that go long (short) in assets with relative formation period out performance (underperformance) of peer stocks to make the winner (loser) portfolio-and four time-series approaches-strategies that go long (short) in assets with formation period outperformance (underperformance) of a hurdle rate to make the winner (loser) portfolio-are employed in this report. This report finds that both the top decile winner portfolio and top half winner portfolio long-only cross-sectional momentum strategies outperform the benchmark. The 12-month formation period top decile winner achieves the highest long-only excess return of 30.21% per annum, whilst all the loser cross-sectional portfolios constitute a return-reducing funding portfolio when conducting a n investment-neutral winner minus loser approach. Short-term zero investment exposure cross-sectional momentum strategies earn strong negative returns, thus presenting contrarian investment opportunities The two exposure-neutral winner minus loser time-series strategies exhibit similar results to the corresponding cross-sectional strategies, however the variable exposure strategies earn positive returns for every formation period-the 12-month formation period strategy being the best earner (25.92% p.a.). These variable exposure strategies earn time-varying returns from the market due to their non-zero net long market exposure as well as some residual return. This premium is left uncaptured by all investment-neutral app roaches and is a strong cause of the lack of skewness of the variable exposure strategies' returns. All of the examined exposure-neutral strategies exhibit significant leftward skewness due to two incidences of extreme and sustained drawdowns. Both incidences occur as a result of the momentum strategy holding market beta exposure of the opposite sign to the market's drastic turn ; the first: positive exposure and market downturn, the second : negative exposure and positive upturn. These drawdowns are reduced when employing strategies of a more intermediate-term formation period such as the 12-month formation strategy. This report's findings confirm the existence of cross-sectional and time-series momentum in South African-listed equities, as well as the case of equity momentum crashing. Further, it provides evidence for both explained and unexplained variations between the two types of momentum trading, with possibilities for further profitability when combining the two.
  • Thumbnail Image
    Item
    Open Access
    An evaluation of analysts' expectational data regarding firms listed on the JSE securities exchange
    (2004) Prayag, C; Van Rensburg, Paul
    This study empirically investigates the usefulness of South African stockbroker analysts' two primary forms of expectational output: earnings forecasts and investment (buy, hold, and sell) recommendations. Instead of focusing on individual stockbroker analysts' forecasts and recommendations, the consensus estimates are examined as they are accessible to a large community of investors.
  • Thumbnail Image
    Item
    Open Access
    The existence and behaviour of style anomalies in the global equity market : a univariate and multivariate analysis
    (2014) Baars, Monique; Van Rensburg, Paul
    Style anomalies comprise patterns and relationships found in the cross-section of stock returns data, which contradict the existing asset-pricing models. They have proven to be reasonably effective at explaining the return-genera ting process of ordinary shares, and have bro ad uses within modern finance. Empirically, style anomalies are found to have statistically significant rewards in individual markets and s mall market groupings, and are found to be significant at a sector level on a global scale, but have not been tested at a firm level on a global scale. The aim of this study is to explain the cross-section of returns of the 1468 largest global firms by market capitalisation. The worldwide study considers stocks from 53 different countries and 112 industries, and investigates the end of month return forecasting power of 44 different firm-specific attributes over the period August 2003 to August 2013. A univariate analysis is performed through a cross-sectional regression of the forward stock re turns on the firm-specific attributes in a similar method to Fama and MacBeth (1973). A ‘Full Data’ regression is also conducted, and results are presented both before and after a beta-adjustment for market risk. Following this, a multivariate analysis is conducted and a forward stepwise procedure is used to construct a multi-factor model. According to the results of this study, style anomalies exist and have a statistically significant reward at a firm level on a global scale. In a univariate setting there are 25 firm-specific style factors that have a significant return payoff at a 5% level of significance. The specific style groups containing significant firm-specific attributes are the Value, Growth, Momentum, Size and Liquidity, Leverage, and Emerging Market groupings. Ten attributes within these style groupings are found to be robust as they are highly significant both before and after beta-adjustment, and within both a univariate and multivariate setting, namely: EBITDA to Share Price (EBP), Emerging Market (EM), CAPEX to Sales (CXS), Sales to Total Assets (STA), Payout Ratio (PR), 24-month growth in Turnover by Volume (TVO24), Sales to Share Price (SP), 6-month growth in Earnings (E6), 1-month prior return (MOM1), and 3-month prior return (MOM3). This confirms that style effects exist both independently, in a univariate setting, and in a multi-factor model. The results of this study show that the Value and Emerging Market styles have the highest cumulative payoffs over the 10-year period, and the evidence of strong correlation between attributes within specific styles gives further validation to the traditional style groupings. The behaviour of, and relationships between the firm-specific style factors give great insight into the payoffs to investing in different style factors over time, and are key to the construction of a multi-factor model. The fifteen firm-specific style factors that are significant in a multivariate setting form the core of a multi-factor style model, which can potentially be used to explain a degree of unexplained returns, predict returns, give insight into global market behaviour, and price global assets for use within a global portfolio. These firm-specific attributes include: EBITDA to Share Price (EBP), Emerging Market (EM), CAPEX to Sales (CXS), Sales to Total Assets (STA), Payout Ratio (PR), 24-month growth in Turnover by Volume (TVO24), Sales to Share Price (SP), 6-month growth in Earnings (E6), 1-month prior return (MOM1), 3-month prior return (MOM3), the natural log of Enterprise Value (LNEV), Interest Cover before Tax (ITBT), 6-month prior return (MOM6), Price-to-Book value (PTB), and Cash Flow-to-Price (CFP).
  • Thumbnail Image
    Item
    Open Access
    Factor-based replication of hedge funds using a state space model
    (2016) Noakes, Michael A; Van Rensburg, Paul
    It has been suggested that the Kalman filter technique may be used to improve the quality of hedge fund replication, compared to existing replication techniques. This study uses the Kalman filter technique, along with three variations of the rolling-window regression technique, to create clones which attempt to replicate the returns of various categories of hedge fund indices. These clones are created over several scenarios and are used to compare the ability of the Kalman filter and rolling-window regression techniques. The clones are constructed using South African specific asset class and investment style factors. This study finds that the Kalman filter does not provide the expected improvement in replication ability over the rolling-window regression, for the hedge fund indices analysed. The competing techniques appear to each be better suited to replicating different hedge fund index strategies and may, therefore, be used in combination. While some of the hedge fund clones offer desirable risk characteristics, they offer lower mean returns and underperform their indices in most periods. As such, the hedge fund clones constructed in this study require further refinement and are not yet equipped for use in practice.
  • Thumbnail Image
    Item
    Open Access
    Firm-specific attributes and the cross-section of equity returns on the Tokyo Stock Exchange
    (2006) Velaers, Juliette F; Van Rensburg, Paul
    This thesis follows the methodologies of Fama and Macbeth (1973) and Robertson (2003) and empirically investigates the cross-sectional relationship between firm-specific attributes and stock returns on the Tokyo Stock Exchange (TSE). A dataset of 226 firm-specific attributes are constructed and tested. The data are adjusted for thin-trading and outliers are free from look-ahead bias. Two separate time periods are investigated in order to reduce the likelihood of data snooping. The in-sample regressions are run over the 1 January 1993 to 31 December 2000 time period and out-sample regressions cover 1 January 2001 to 31 December 2004.
  • Thumbnail Image
    Item
    Open Access
    Firm-specific attributes and the cross-section of JSE securities exchange returns
    (2002) Robertson, Michael N; Van Rensburg, Paul
    The empirical counterpart of a theory of asset prices is a model of the cross-section of security returns. Empirical tests of the Capital Asset Pricing Model (CAPM) of Sharpe (1964), Lintner (1965), Mossin (1966) and Black (1972) using cross-sectional methodologies have identified numerous cases where variables apart from beta explain expected returns. Initially termed 'anomalies', these empirical violations of the theory are frequently associated with firm-specific attribute data. Traditional views of market efficiency however rule out the ability to predict risk-adjusted asset returns and the source and exploitability of these anomalies remains controversial. This thesis empirically investigates the cross-sectional relationship between twenty-four firm-specific attributes and stock returns on the JSE Securities Exchange (JSE). The results are evaluated to determine whether they conform to the predictions of the CAPM and a two factor Arbitrage Pricing Theory (APT) model suggested by prior research. In monthly univariate tests similar to Fama and MacBeth (1973), price-to-NAV, dividend yield, price-to-earnings, cashflow-to-price, price-to-profit and size are found to have statistically significant predictive power in the cross-section of returns using monthly data from July 1990 to June 2000. After conducting risk-adjustments using both the CAPM and a two factor APT model these effects are found to persist. Multivariate testing suggests a two factor model with the size and price-to-earnings attributes as explanatory variables. When the cross-sectional tests are compared across the broad industry sectors of the JSE, value effects are observed as being stronger in the financial-industrial sector than in the resource sector, the small size effect is roughly the same across sectors while the premium earned by higher debt levels is negative across all sectors except the financial sector. The results are not reconcilable with the predictions of the CAPM. Simulated portfolios constructed using a methodology similar to Fama and French (1992) reveal that beta, if anything, is inversely related to average returns on the JSE. Contrary to the international evidence, greater average returns accruing to small and low price-to- earnings firms are not commensurate with higher betas. Controlling for the small size and low price-to-earnings effects does not improve the explanatory power of beta. An application of the Daniel and Titman (1997) methodology suggests that average returns on the JSE are not compensation for factor risk, but instead are associated with the exposure to the attributes themselves. The findings of this study present a serious empirical challenge for covariance-based asset pricing models on the JSE.
  • Thumbnail Image
    Item
    Open Access
    A framework for evaluating the benchmark risk of South African equity portfolios
    (2005) Kruger, Ryan; Van Rensburg, Paul
    The aim of this study is to identify and quantify those primary aspects of risk which impact on the construction of benchmark ind ices as well as active portfolios in the South African market. The appropriateness of tile application of the new FTSE classification structure with regard to the particular structure of the local exchange on 30 June 2002 has been placed in question. An initial cluster analysis of the index returns underlying the new classification demonstrated that there were significant behavioural anomalies amongst the new index structure with many Financial-Industrial indices now grouped closely with Resources stocks. A principal factors analysis of the market sectors indicated that the strong Financial-Industrials and Resources dichotomy was present within the market but also demonstrated that a number of Financial-Industrial indices, most notably Basic Industries and Cyclical Consumer Goods, demonstrated either loadings on both factors or loaded solely on the Resources factor rather than their own Financial-Industrials factor. An investigation on a share level found that in most cases one or two large cap shares were responsible for the behaviour of their sectors as a whole and that each of the shares in question was either dual-listed or had significant exposure to foreign markets.
  • Thumbnail Image
    Item
    Open Access
    Hedge funds and higher moment portfolio selection
    (2005) Bergh, G; Van Rensburg, Paul
    This study confirms the findings of Davies, Kat and Lu (2003) and Feldman, Chen and Goda (2002) that Global Macro and Equity Market-Neutral strategies are crucial constituents in a fund of hedge funds portfolio. When comparing optimised multi-asset class portfolios including an allocation to hedge funds, the results show that meanvariance optimisation overallocates to the hedge fund class on the basis of its high reward to volatility ratio.
  • No Thumbnail Available
    Item
    Open Access
    Hedge funds and higher moment portfolio selection
    (2005) Bergh, Gregory; Van Rensburg, Paul
    Recent studies by Amin and Kat (2001) and Lo (2001) show that, notwithstanding the central limit theorem, the returns of several hedge fund indices exhibit distributional characteristics inconsistent with normality. In this context, this study empirically compares the Markowitz (1952) mean-variance optimisation technique with a higher moment methodology recently proposed by Davies, Kat and Lu (2005). It extends the methodology to optimise portfolios without a unity-variance constraint. In addition, this study augments the application of Davies et al (2005) beyond that of fund of hedge fund portfolio construction to also incorporate the traditional asset classes of equities, bonds and cash. The descriptive statistics show that hedge fund strategies of Fixed Income Arbitrage and EventDriven while displaying low volatility, also exhibit latent higher moment risk in their negative skewness and high kurtosis. These two higher moments collectively suggest an increase in the probability of extreme adverse returns to the investor that is not revealed in traditional mean-variance analysis. Confirming the findings of Amin and Kat (2001) and Lo (2002), Jarque-Bera tests find that only two out of the fourteen hedge fund indices used in this study are normal at the 5% level. Applying Markowitz (1952) mean-variance portfolio selection to an array of published hedge fund indices produces portfolios with higher ex-post returns but nai've exposure to undesirable higher moment risks. When the higher moments of hedge fund index return distribution are accounted for in the portfolio optimisation algorithm, the resultant portfolios have improved diversification and higher moment statistics. This study confirms the findings of Davies, Kat and Lu (2003) and Feldman, Chen and Goda (2002) that Global Macro and Equity Market-Neutral strategies are crucial constituents in a fund of hedge funds portfolio. When comparing optimised multi-asset class portfolios including an allocation to hedge funds, the results show that mean-variance optimisation overallocates to the hedge fund class on the basis of its high reward to volatility ratio. The higher moment optimised portfolios all outperform the mean-variance comparatives when evaluated on an Omega function basis. More generally, the results suggest that when assembling portfolios that include hedge funds, higher-order optimisation makes a meaningful difference to portfolio composition.
  • Thumbnail Image
    Item
    Open Access
    The hedonic valuation of South African wine brands
    (2010) Priilaid, David A; Van Rensburg, Paul; Robb, Frank
    This study aims to value South African wine brands. Deploying blind and sighted versions of hedonic quality it defines (1) ‘functional’ wine brands as those with consistently higher levels of intrinsic quality as proxied by their blind tasting scores, and (2) placebo-type ‘symbolic’ wine brands as those with statistically significant positive predictive differences between their blind and sighted scores. Through a series of econometric analyses applied to 8225 wines sampled over the eight year period from 2000 through 2007, a higher proportion of functional-to-symbolic brands is notified. Bi-polar clustering is observed in both brand-classes, with positive and negative brand-effects yielding positive and negative ‘non’ brands, respectively. Such clustering extends to those brands presenting simultaneously with both functional and symbolic brand-effects. Here brands decompose into zones of either Symbolic Values (with positive placebos and weak intrinsics) or Functional Values (with negative placebos and strong intrinsics). When these zones are graphed relative to their intrinsic blind to sighted-minus-blind scores, no brands appear to occupy the middleground. Each zone is located approximately one standard deviation left and right of the grand-sample mean intrinsic score. This functional-to-symbolic typology confirms and extends the literature on brands in general (Bhat and Reddy, 1998), and wine brands in particular (Mowle and Merrilees, 2005). Two wine brand valuation techniques are subsequently presented and comparatively assessed. Each is based on the combined use of non-ordinal wine valuation models and discounted cash-flow methodologies. The first price-premium approach defines brand equity value (per bottle) as the difference between a wine’s price and a valuation of its intrinsic worth as measured by blind ratings. The second quality-premium approach defines brand equity value as the difference between a wine’s intrinsic value and (instead of price) the value of its perceived quality when sampled sighted.
  • Thumbnail Image
    Item
    Open Access
    The impact of firm-specific factors on the cross- sectional variation in Johannesburg security exchange listed equity returns
    (2014) Van Heerden, Jakobus Daniël; Van Rensburg, Paul
    The aim of this study is to examine the impact of technical and fundamental (referred to as firm-specific) factors on the cross-sectional variation in equity returns on the Johannesburg Securities Exchange (JSE). Three approaches to address this objective were identified through an extensive literature study covering more than half a century’s research, namely a cross-sectional regression approach, a factor portfolio approach and an extreme performer approach. All three approaches are applied in this study, allowing for comparison and robustness- tests to be performed on the JSE for the first time. In addition to factors identified through the literature review, factors that make economic sense from a South African point of view have also been included in the dataset, resulting in a total of fifty firm-specific factors to be examined. A fresh data set was created by collecting monthly data through numerous data sources on all shares listed on the JSE for the period January 1994 through May 2011, for these factors. The seventeen and a half year period is the longest period used to date (to the author’s knowledge) for the kind of research conducted in this thesis. Furthermore, the data has been prepared to correct for potential statistical biases that may affect the results, including data snooping, infrequent trading, survivorship bias, look-ahead bias and outliers. This lengthy period further allows for the formation of two independent subsamples, each covering a full investment cycle, enabling in- and out of- sample empirical research and testing to be conducted on the JSE for the first time.