In this thesis, we study random bridges of a certain class of Lévy processes and their applications to credit risky asset pricing. In the first part, we construct the compound random bridges(CLRBs) and analyze some tools and properties that make them suitable models for information processes. We focus on the Markov property, dynamic consistency, measure changes and increment distributions. Thereafter, we consider applications in credit risky asset pricing. We generalize the information based credit risky asset pricing framework to incorporate prematurity default possibilities. Lastly we derive closed-form expressions for default trends and intensities for credit risky bonds with CLRB as the background partial information process. We obtain analytical expressions for specific CLRBs. The second part looks at application of stochastic filtering in the current information based asset pricing framework. First, we formulate credit risky asset pricing in the information-based framework as a filtering problem under incomplete information. We derive the Kalman-Bucy filter in one dimension for bridges of Lévy processes with a given finite variance.

The purpose of this study is to investigate the determinants of technical efficiency, using data envelopment analysis and the Tobit regression model, of the six largest listed South African banks for the years 2008-2018. An input-oriented intermediary constant-return-to-scale approach was followed to determine technical efficiency scores. After technical efficiency scores were obtained, a binary data set was created by assigning a score of 1 to all observations that were regarded as technical efficient, whereas all observations that were regarded as technically inefficient were assigned a score of 0. Thereafter, a Tobit regression analysis was performed to test the following hypotheses: skimping hypothesis, diversification hypothesis, bad management hypothesis and the funding hypothesis. The results of the regression analysis show that the skimping, diversification, and bad management hypotheses were not relevant for the six largest South African banks over the period under review. Regression results pointed towards the funding hypothesis being applicable to the six largest listed banks over the review period. It can therefore be suggested that the banks under review were generally well managed with a keen focus on expense control and thorough underwriting. To ensure the efficiency of large listed banks, it is proposed that regulators continue to monitor large banks as evidence of the study suggests that as deposit bases grow, a deterioration in technical efficiency is experienced. Generally, the results of the study indicate that the six large listed banks are overall relatively efficient over the review period.

Includes bibliographical references / This study examines a multifactor model of the Johannesburg Stock Exchange (JSE) framed within the Arbitrage Pricing Theory (APT). The APT has been set up such that it can be able to separate the beta for the precious metal factor within the model. The process goes via the investigation of macrovariables (with precious metals used as one of the macrovariables) and their effect on market (JSE) returns. A complete analysis and modeling of this relationship is likely to yield unparalleled rewards and cost-effective risk management, monitoring and mitigation. Using monthly data for the period 31/07/2002 to 30/04/2013 the dissertation focuse d on using a market (JSE) representative index as a basis for creating a wholly functioning APT model. This included creating a more liquid representative of the JSE All Share Index (A LSI) by using the top 100 stocks by market capitalization. Principal Components Analysis (PCA) was applied to the variables to ascertain a proper model for the JSE return structure. However, in the end an appropriate econometric structure in the form of Autoregressive Conditional Heteroscedastic (ARCH) and Generalized Autoregressive Conditional Heteroscedastic (GARCH) models was used and applied to test and create the APT model to address the objective. The other purpose of this dissertation was to separate beta attributable to the precious metal macrovariable within the model. This is based on the establishment of the JSE in the late 1880s being primarily due to the discovery of precious metals in the former Transvaal (North West) and Pretoria, Witwatersrand and Vereeniging (PWV) region now Gauteng. This is to ascertain whether these metals still have as much influence on the JSE as they did for over half a century. The results show that macroeconomic variables do influence the return generating process of the JSE, explaining almost 80% of variation in returns. The results show that the ALSI is characterized by a seven factor APT with, industrial production, money supply, SA consumer price index, ZARUSD exchange rate, crude oil, MSCI ACWI and precious metals statistically significant.

The maintenance of asset quality, efficiency and profitability is a vital requirement for the survival and development of banks. Loans are the main asset class from which banks generate their major portion of income and also signify the greatest risk to banks. There has been significant indication that the financial crises in the USA, Sub-Saharan Africa and East Asia were signalled by high levels of non-performing loans (NPLs). Due to the detrimental effect that these loans have on a bank's revenue and the economic welfare of a country, it is essential to examine and investigate the determinants of NPLs in the banking industry of any country. This study examines Botswana, a developing country in Southern-Africa and is stimulated by the assumption that both the industry level variables and macroeconomic variables have an effect on NPLs. Secondary data of the banking sector was obtained from Botswana's central bank, the Bank of Botswana. Correlation and regression analysis were carried out over a period of ten years (2005-2014), using quarterly data. It was found that the following industry level variables (i.e. credit growth, industry size and profitability) and macroeconomic variables (i.e. real gross domestic product (GDP) growth, inflation, real interest rates and the unemployment rate) have a statistically significant impact on the NPL rate. On the other hand, capitalization and diversification had a statistically insignificant relationship with NPLs. The banking industry in Botswana should carefully monitor the household loan portfolio as well as their credit advancement policies with regard to the aforementioned variables to help lower their NPL ratios. This study is the first of its kind in the Botswana banking industry and therefore will provide scholars with the opportunity to enrich their knowledge and serve as a reference for other researchers in the related area while also providing a foundation for further studies.

The ability of active managers to produce consistent benchmark-beating returns is a topic that has been widely debated with increasing interest over the past decade. The majority of previous studies in which persistence of performance is tested consider a fund's ability to maintain its relative ranking over various time periods amongst its peer group. This study adds to the literature by considering the persistence of alpha, where alpha is defined as the out- or under-performance of a market-related benchmark. Persistence of alpha for South African general equity unit trusts is tested over six-month, one-, two- and three-year formation and holding periods using a similar methodology to that of Collinet & Firer (2003). Alpha is found to persist most prominently in tests of one-year periods, with other period lengths yielding less significant results. Additionally, using the methodology of Malkiel (1995), certain funds which have demonstrated statistically significant persistent alpha over various periods are identified.

The financial crises of 2008 increased the focus around financial distress and even more so on predicting financially distressed companies prior to the fact. This research paper investigates using recursive partitioning to predict financially distressed companies on the Johannesburg Stock Exchange, taking different business cycle periods into account over the time period 1997-2014. The updated as well as longer time period over which the analysis is conducted distinguishes this research paper from prior research. This paper employs both the CART and CHAID algorithm and obtains financially distressed prediction models which have a higher correct classification rate than chance alone and prior literature in South Africa. This paper also makes use of a matched data sample approach and the manner in which missing data is addressed makes a valuable contribution to financial distress prediction research. Furthermore, support is found for prior literature in that financial variables are statistically significant in predicting financial distress.

This study investigates the effect that announcements of rights issues have on abnormal share price returns on the JSE over the period January 2009 to December 2014. This study will focus specifically on the equity element of the capital structure and the issuing of new equity in the form of rights issues. There have been a few studies done in this regard in the South African context but the prior papers have been conducted over significantly different time periods and data samples and there is therefore merit in combining all four approaches into a single study focused on one consistent data sample. Secondly, this study also investigates the impact the motivation for the rights issue as provided by the issuer, has on the share price returns of the issuers. Thirdly, this study investigates the effect of the "financial health" of the rights issuer, as measured in terms of the Altman Z-score, has on the abnormal returns of the share prices of the issuers. The final area of investigation is to test the ability of rights issuers on the JSE to time the market when performing rights issues.

This paper aims to investigate the change in corporate debt levels in South Africa from 1994 to 2016 as well as analyse certain factors that play a role in the decision making of corporates when it comes to the all important decision of capital structure. The study uses data from large capitalisation, retail and food producing firms listed on the Johannesburg Stock Exchange. Four different leverage measures are used to determine the change in capital structure over the period under review as well as six of the most common determinants of capital structure used in literature. The analysis shows that South African corporates have drastically increased their appetite for debt funding compared to equity funding over the last two decades. Large capitalisation stocks reflected the largest increase in the use of debt, whilst food producers showed the smallest yet still significant increase in debt. Analysis has also shown that firms have changed their maturity profile of their debt significantly since the 2008 financial crises. Results from the analysis on determinants varied with some determinants showing statistical significance.

Using a Principal Component Analysis (PCA) approach, we investigate the sovereign yield spread term structure of the BRICS economies against the U.S. We show that the term structure for these markets are primarily driven by three latent factors which can be classified as the spread level, slope and curvature factors. We further postulate that a country’s yield curve contains valuable information about its future economic state and as such the PCA derived spread factors, which are based on the differences between sovereign yield curves, encapsulates material macro-economic information between the countries. In light of this, we show that augmenting the traditional Uncovered Interest Rate Parity model (UIRP) with these factors improves the models predictive accuracy of exchange rate movements.

In this study it was examined whether Pairs trading is a potentially profitable trading strategy on the Johannesburg Stock Exchange. Pairs trading is a quantitative based trading strategy, in which shares are paired up based on a historic price relationship and traded accordingly, in a contrarian manner, when they diverge from said historical relationship. The essence of Pairs trading is to take advantage of perceived market inefficiencies, which is a direct contradiction of the Efficient Markets Hypothesis (even in its weak form). This study tested Pairs trading on both an unrestricted (any two shares can be paired), as well as a sector-restricted (only pairs within the RESI and the FINDI sectors could be paired), sample of shares (the JSE Top80 - based on market capitalization). Furthermore, a number of different signals (which are based on standard deviations) to open and close pairs were tested, on both the unrestricted and sector-restricted samples. The aim of using different samples of shares, as well as different trading signals, was to determine whether or not different strategies could serve to bolster the performance of a Pairs trading strategy.