京都大学 / 0048 / 新制・課程博士 / 博士(経済学) / 甲第21523号 / 経博第591号 / 新制||経||288(附属図書館) / 京都大学大学院経済学研究科経済学専攻 / (主査)教授 梶井 厚志, 教授 原 千秋, 教授 若井 克俊 / 学位規則第4条第1項該当 / Doctor of Economics / Kyoto University / DFAM

Should cryptocurrencies populate modern portfolios, and to what degree? How can von Neumann-Morgenstern utility theory determine which portfolio is best? For this thesis, we take six cryptocurrencies and six stocks to create optimal portfolios from each and a combination of both. Then we compare the expected utility from each portfolio, with and without short selling, to a benchmark. As concluded from our data, cryptocurrencies should be a part of a modern portfolio to increase the Sharpe ratio and expected utility even if they do not take a majority proportion.

Increased volatility of agricultural commodity prices as well as market linkages between the agricultural and energy markets expose producers to different types of systematic price risk. Producers that operate on margins involving both input and output price uncertainty are perhaps the most adversely affected by these volatility changes. The beef cattle feeding industry is one such example. This research focuses on how expected margins in the beef cattle backgrounding and finishing stages are affected by output and input price uncertainty.

The average length of life from birth until death in a human population is a single statistic that is often used to characterise the prevailing health status of the population. It is one of many statistics calculated from an analysis that, for each age, combines the number of deaths with the size of the population in which these deaths occur. This analysis is generally known as life table analysis. Life tables have only occasionally been produced specifically for South Australia, although the necessary data has been routinely collected since 1842. In this thesis, the mortality pattern of South Australia over the period of 150 years of European settlement is quantified by using life table analyses and estimates of average length of life. In Chapter 1, a mathematical derivation is given for the lifetime statistical distribution function that is the basis of life table analysis, and from which the average length of life or current expected life is calculated. This derivation uses mathematical notation that clearly shows the deficiency of current expected life as a measure of the life expectancy of an existing population. Four statistical estimation procedures are defined, and the computationally intensive method of bootstrapping is discussed as an estimation procedure for the standard error of each of the estimates of expected life. A generalisation of this method is given to examine the robustness of the estimate of current expected life. In Chapter 2, gender and age-specific mortality and population data are presented for twenty five three-year periods; each period encompassing one of the colonial (1841-1901) or post-Federation (1911-96) censuses that have been taken in South Australia. For both genders within a census period, four types of estimate of current expected life, each with a bootstrap standard error, are calculated and compared, and a robustness assessment is made. In Chapter 3, an alternate measure of life expectancy known as generation expected life is considered. Generation expected life is derived by extracting, from official records arranged in temporal order, the mortality pattern of a notional group of individuals who were born in the same calendar year. Several estimates of generation expected life are calculated using South Australian data, and each estimate is compared to the corresponding estimate of current expected life. Additional estimates of generation expected life calculated using data obtained from the Roll of Honour at the Australian War Memorial quantify the reduction in male generation expected life for 1881-1900 as a consequence of military service during World War I, 1914-18, and the Influenza Pandemic, 1919. / Thesis (M.Sc.) -- University of Adelaide, School of Applied Mathematics, 2003.

För att värna om finansiell stabilitet utförs kontinuerligt bedömningar utav de risker och hot som återfinns mot det finansiella systemet, man granskar också det finansiella systemets motståndskraft emot dessa. För att förhindra att nya bankkriser uppstår införs med jämna mellanrum nya regleringar. Däribland skall en ny redovisningsmodell implementeras för redovisning utav kreditförluster – Expected loss model. Denna redovisningsmodell ger upphov till ett ökat inslag av bedömningar vid redovisning utav kreditförluster. Det blir därmed intressant att försöka förklara banktjänstemäns och revisorers uppfattningar angående subjektiva bedömningar och dess påverkan på en rättvisande bild vid redovisning utav kreditförluster. Genom att tillämpa en abduktiv ansats i studien har en växling mellan empirisk och teoretisk reflektion möjliggjorts vid skapande utav hypotesen. Med hjälp utav ett webbaserat frågformulär har man undersökt banktjänstemän och revisorers uppfattningar angående i vilken utsträckning subjektiva bedömningar leder till en rättvisande bild vid redovisning utav kreditförluster. Frågeformuläret tilldelades banktjänstemän som dagligen arbetar med förlustavsättningar och samtliga auktoriserade revisorer vilka är medlemmar i FAR. Empirin har sedan bearbetats statistiskt för att kunna testa den hypotes som tagits fram med bakgrund i studiens teoretiska referensram. Genom denna studie har man statistiskt kunnat säkerställa att det återfinns en skillnad i uppfattningar mellan banktjänstemän och revisorer angående hur subjektiva bedömningar speglar en rättvisande bild utav bankens kreditkvalitet. Man kan således i enlighet med studiens hypotes påvisa att banktjänstemän i större utsträckning anser att subjektiva bedömningar leder till en mer rättvisande bild utav bankens kreditkvalitet. Då bedömningar med stor sannolikhet förändras i samklang med rådande marknadsklimat och aktuella redovisningsstandarder anser man att studien fyller en kunskapslucka inom området. / To protect the financial stability, assessments of the risk and threats to the financial system are carried out continuously. Also, the financial system’s resilience against these risks and threats is reviewed. To prevent creation of new bank crisis new regulation are implemented continuously. Among these, a new accounting model is implemented for the accounting of credit losses - Expected loss model. This accounting model gives rise to an increase in the use of assessments in the accounting of credit losses. This makes it interesting to try explaining bank office workers and auditors’ perceptions regarding subjective judgments and their impact on a true and fair view in accounting of credit losses. The application of an abductive view in the study made it possible to switch between empirical and theoretical reflections to be used in the creation of the hypothesis. An investigation of bank office workers and auditors’ perceptions regarding the extent to which subjective judgments impacts a true and fair view in accounting of credit losses was carried out through a questionnaire. The questionnaire was handed to bankers who in their daily work are in contact with loss provision and to all certified public accountants that are members of FAR. The result was then statistically processed in order to test the hypotheses which have been created from the study’s theoretical framework. This study has statistically made it possible to show that there is a difference in perceptions between bank office workers and auditors’ regarding how subjective judgments reflect a true and fair value of the bank’s credit quality. Therefore it is possible in accordance with the study’s hypothesis to prove that bank office workers to a greater extent believe that subjective judgments lead to a more true and fair value in accounting for credit losses. Since assessments with high probability change in consistence with current market conditions and accounting standards, it is believed that the study fills a knowledge gap in the field.

The average length of life from birth until death in a human population is a single statistic that is often used to characterise the prevailing health status of the population. It is one of many statistics calculated from an analysis that, for each age, combines the number of deaths with the size of the population in which these deaths occur. This analysis is generally known as life table analysis. Life tables have only occasionally been produced specifically for South Australia, although the necessary data has been routinely collected since 1842. In this thesis, the mortality pattern of South Australia over the period of 150 years of European settlement is quantified by using life table analyses and estimates of average length of life. In Chapter 1, a mathematical derivation is given for the lifetime statistical distribution function that is the basis of life table analysis, and from which the average length of life or current expected life is calculated. This derivation uses mathematical notation that clearly shows the deficiency of current expected life as a measure of the life expectancy of an existing population. Four statistical estimation procedures are defined, and the computationally intensive method of bootstrapping is discussed as an estimation procedure for the standard error of each of the estimates of expected life. A generalisation of this method is given to examine the robustness of the estimate of current expected life. In Chapter 2, gender and age-specific mortality and population data are presented for twenty five three-year periods; each period encompassing one of the colonial (1841-1901) or post-Federation (1911-96) censuses that have been taken in South Australia. For both genders within a census period, four types of estimate of current expected life, each with a bootstrap standard error, are calculated and compared, and a robustness assessment is made. In Chapter 3, an alternate measure of life expectancy known as generation expected life is considered. Generation expected life is derived by extracting, from official records arranged in temporal order, the mortality pattern of a notional group of individuals who were born in the same calendar year. Several estimates of generation expected life are calculated using South Australian data, and each estimate is compared to the corresponding estimate of current expected life. Additional estimates of generation expected life calculated using data obtained from the Roll of Honour at the Australian War Memorial quantify the reduction in male generation expected life for 1881-1900 as a consequence of military service during World War I, 1914-18, and the Influenza Pandemic, 1919. / Thesis (M.Sc.) -- University of Adelaide, School of Applied Mathematics, 2003.

This thesis aims at implementing and evaluating the performance of multivariate Expected Shortfall models on high frequency foreign exchange data. The implementation is conducted with a unique portfolio consisting of five foreign exchange rates; EUR/SEK, EUR/NOK, EUR/USD, USD/SEK and USD/NOK. High frequency is in this context defined as observations with time intervals from second by second up to minute by minute. The thesis consists of three main parts. In the first part, the exchange rates are modelled individually with time series models for returns and realized volatility. In the second part, the dependence between the exchange rates is modelled with copulas. In the third part, Expected Shortfall is calculated, the risk contribution of each exchange rate is derived and the models are backtested. The results of the thesis indicate that three of the five final models can be rejected at a 5% significance level if the risk is measured by Expected Shortfall (ES0:05). The two models that cannot be rejected are based on the Clayton and Student’s t copulas, the only two copulas with heavy left tails. The rejected models are based on the Gaussian, Gumbel-Hougaard and Frank copulas. The fact that some of the copula models are rejected emphasizes the importance of choosing an appropriate dependence structure. The risk contribution calculations show that the risk contributions are highest from EUR/NOK and USD/NOK, and that EUR/USD has the lowest risk contribution and even decreases the portfolio risk in some cases. Regarding the underlying models, it is concluded that for the data used in this thesis, the final combined time series and copula models perform quite well, given that the purpose is to measure the risk. However, the most important parts to capture seem to be the fluctuations in the volatilities as well as the tail dependencies between the exchange rates. Thus, the predictions of the return mean values play a less significant role, even though they still improve the results and are necessary in order to proceed with other parts of the modelling. As future research, we first and foremost recommend including the liquidity aspect in the models. / Syftet med denna masteruppsats är att implementera och utvärdera multidimensionella Expected Shortfall-modeller på högfrekvent växelkursdata. Implementeringen och utvärderingen utförs med en unik portfölj bestående av fem växelkurser; EUR/SEK, EUR/NOK, EUR/USD, USD/SEK och USD/NOK. Högfrekventa observationer är i denna uppsats definierade som sekundvisa upp till minutvisa observationer. Uppsatsen består av tre huvuddelar. I den första delen modelleras växelkurserna individuellt med tidsseriemodeller för växelkursförändringarna i form av avkastning och realiserad volatilitet. I del två modelleras beroendestrukturerna mellan de olika växelkurserna med hjälp av copulas. I den tredje och sista delen beräknas Expected Shortfall och riskbidragen från de enskilda växelkurserna, varefter modellerna utfallstestas. De slutgiltiga resultaten indikerar att tre av de fem föreslagna modellerna kan förkastas vid en signifikansnivå på 5% om risken mäts med Expected Shortfall (ES0:05). De två modeller som inte kan förkastas är baserade på Clayton och Student’s t copulas, vilka särskiljer sig från övriga copulas genom att de har tjocka vänstersvansar. De modeller som förkastas är baserade på Gaussian, Gumbel-Hougaard och Frank copulas. Det faktum att några copula-modeller förkastas betonar vikten av att välja en lämplig beroendestruktur. Riskbidragsberäkningarna visar att EUR/NOK och USD/NOK bidrar mest till den totala risken i portföljen och att EUR/USD har det lägsta riskbidraget, där EUR/USD till och med minskar risken i vissa fall. Vad gäller underliggande modeller så visas det att för den tillgängliga datan i den här uppsatsen så fungerar tidsseriemodeller i kombination med copulas bra, givet att syftet är att mäta risk. Dock tyder resultaten på att volatilitetsfluktuationer samt svansberoenden mellan växelkurserna är de mest väsentliga delarna att fånga. Väntevärdesprognoserna för avkastningarna har mindre inverkan på de slutgiltiga beräkningarna, även om de fortfarande förbättrar resultaten och i sig är nödvändiga för fortsatt modellering. För framtida studier rekommenderar vi först och främst att inkludera likviditetsaspekter i modellerna.

The efficient market hypothesis and behavioural finance have been the cause of much debate for decades, with one theory advocating market efficiency and the other opposing it. The efficient market hypothesis (EMH) assumes that investors always act rationally and stock prices adjust rapidly to new information and should reflect all available information. In contrast, behavioural finance suggests that markets are not rational and investors make irrational decisions, which may lead them to over- or under-price stocks. Researchers for years have been empirically testing these assumptions in stock markets. However, there has been no consensus on which asset-pricing models perform better in capturing the effect of market anomalies and what impact these market anomalies have on the expected returns of different stock market’s sectors. The aim of the study was to test the effect of selected market anomalies on expected return in different sectors of the Johannesburg Stock Exchange (JSE). More specifically, the study aimed to compare the performance of different asset-pricing models and their ability to account for market anomalies in different sectors of the JSE. Additionally, this study tested the applicability of the recent Fama and French five (FF5-factor) model, in estimating the expected return on the JSE. The study used a quantitative approach with secondary data over a period of 12 years starting from January 2002 to December 2014. The sample used in the study consists of monthly data obtained from McGregor BFA and the South African Reserve Bank. The study examined for the effects of size, value, January and momentum variables across six sectors of the JSE. This was accomplished by the use of various asset-pricing models such as the Capital asset pricing model (CAPM), the Fama and French three-factor model (FF3-factor), the Carhart four-factor model (C4F) and the recent five-factor model of Fama and French (FF5-factor). The study showed that whenever the asset-pricing models were not restricted, they tend to capture the market anomalies in four out of the six sectors examined. However, no market anomalies were found present in two of the six sectors analysed. In contrast, when the asset-pricing models are restricted, the asset-pricing models only seem to capture the effects of market anomalies in one of the six examined sectors. The findings in this study suggest that market anomalies are sensitive to model specifications, as restricting the models tends to capture the different market anomalies across the sectors of the JSE. The study also found that market anomalies differ across sectors and that some sectors are more efficient than others. The study also reveals that the FF5-factor model is able to account for expected returns on the JSE. In addition, the FF5-factor model tends to perform better when the model is restricted. It is also evident from the findings presented in this study, that the value anomaly loses its predictive power when profitability and investment variables are included in the model. Overall, the study illustrated that market anomalies have an effect on returns of the JSE, that the model specifications play an important role in an asset-pricing model and that the FF5-factor model is applicable on the JSE, however, it is not certain whether four or five factors apply to the South African market.

The efficient market hypothesis and behavioural finance have been the cause of much debate for decades, with one theory advocating market efficiency and the other opposing it. The efficient market hypothesis (EMH) assumes that investors always act rationally and stock prices adjust rapidly to new information and should reflect all available information. In contrast, behavioural finance suggests that markets are not rational and investors make irrational decisions, which may lead them to over- or under-price stocks. Researchers for years have been empirically testing these assumptions in stock markets. However, there has been no consensus on which asset-pricing models perform better in capturing the effect of market anomalies and what impact these market anomalies have on the expected returns of different stock market’s sectors. The aim of the study was to test the effect of selected market anomalies on expected return in different sectors of the Johannesburg Stock Exchange (JSE). More specifically, the study aimed to compare the performance of different asset-pricing models and their ability to account for market anomalies in different sectors of the JSE. Additionally, this study tested the applicability of the recent Fama and French five (FF5-factor) model, in estimating the expected return on the JSE. The study used a quantitative approach with secondary data over a period of 12 years starting from January 2002 to December 2014. The sample used in the study consists of monthly data obtained from McGregor BFA and the South African Reserve Bank. The study examined for the effects of size, value, January and momentum variables across six sectors of the JSE. This was accomplished by the use of various asset-pricing models such as the Capital asset pricing model (CAPM), the Fama and French three-factor model (FF3-factor), the Carhart four-factor model (C4F) and the recent five-factor model of Fama and French (FF5-factor). The study showed that whenever the asset-pricing models were not restricted, they tend to capture the market anomalies in four out of the six sectors examined. However, no market anomalies were found present in two of the six sectors analysed. In contrast, when the asset-pricing models are restricted, the asset-pricing models only seem to capture the effects of market anomalies in one of the six examined sectors. The findings in this study suggest that market anomalies are sensitive to model specifications, as restricting the models tends to capture the different market anomalies across the sectors of the JSE. The study also found that market anomalies differ across sectors and that some sectors are more efficient than others. The study also reveals that the FF5-factor model is able to account for expected returns on the JSE. In addition, the FF5-factor model tends to perform better when the model is restricted. It is also evident from the findings presented in this study, that the value anomaly loses its predictive power when profitability and investment variables are included in the model. Overall, the study illustrated that market anomalies have an effect on returns of the JSE, that the model specifications play an important role in an asset-pricing model and that the FF5-factor model is applicable on the JSE, however, it is not certain whether four or five factors apply to the South African market.

We examine the task of feature selection, which is a method of forming simplified descriptions of complex data for use in probabilistic classifiers. Feature selection typically requires a numerical measure or metric of the desirability of a given set of features. The thesis considers a number of existing metrics, with particular attention to those based on entropy and other quantities derived from information theory. A useful new perspective on feature selection is provided by the concepts of partitioning and encoding of data by a feature set. The ideas of partitioning and encoding, together with the theoretical shortcomings of existing metrics, motivate a new class of feature selection metrics based on conditional entropy. The simplest of the new metrics is referred to as expected partition entropy or EPE. Performances of the new and existing metrics are compared by experiments with a simplified form of part-of-speech tagging and with classification of Reuters news stories by topic. In order to conduct the experiments, a new class of accelerated feature selection search algorithms is introduced; a member of this class is found to provide significantly increased speed with minimal loss in performance, as measured by feature selection metrics and accuracy on test data. The comparative performance of existing metrics is also analysed, giving rise to a new general conjecture regarding the wrapper class of metrics. Each wrapper is inherently tied to a specific type of classifier. The experimental results support the idea that a wrapper selects feature sets which perform well in conjunction with its own particular classifier, but this good performance cannot be expected to carry over to other types of model. The new metrics introduced in this thesis prove to have substantial advantages over a representative selection of other feature selection mechanisms: Mutual information, frequency-based cutoff, the Koller-Sahami information loss measure, and two different types of wrapper method. Feature selection using the new metrics easily outperforms other filter-based methods such as mutual information; additionally, our approach attains comparable performance to a wrapper method, but at a fraction of the computational expense. Finally, members of the new class of metrics succeed in a case where the Koller-Sahami metric fails to provide a meaningful criterion for feature selection.