La thèse comporte deux parties. Dans la première partie (Chapitres 1 et 2), un examen des déterminants des interventions officielles sur le marché des changes est proposée. Dans la second partie (Chapitres 3 et 4), c'est la problématique des interventions dites « secrètes » qui est étudiée. <p><p>Chapitre 1: « Talks, financial operations or both »<p><p>Ce chapitre propose une nouvelle approche aux fonctions de réaction permettant d’examiner, dans un même modèle, les déterminants des différents types d’interventions (les interventions effectives et les interventions orales). Le modèle permet de mieux comprendre les choix stratégiques des autorités (opérations financières ou simple politique de communication) et d’en évaluer le degré de substituabilité ou de complémentarité.<p><p>Chapitre 2 :« The institutional organization underlying interventions »<p><p>La structure institutionnelle sous-jacente au processus d’intervention (interactions entre le Ministère des finances et la banque centrale) est explicitement incorporée dans le modèle proposé dans ce chapitre. Cette approche permet d’évaluer, dans quelle mesure, le Ministère des finances (l’autorité responsable de la politique de change), en intervenant sur le marché, internalise les objectifs de la banque centrale(l’agent du Ministère pour l’implémentation des ordres d’intervention).<p><p>Chapitre 3 :« The secrecy puzzle »<p><p>Ce chapitre propose une évaluation empirique des différents arguments théoriques expliquant le recours aux interventions secrètes. Le travail repose sur l’examen économétrique d’une fonction de stratégie, dans laquelle, des déterminants relatifs à la décision d’intervenir secrètement d’une part et, d’autre part, des déterminants relatifs à la détection des interventions par le marché sont incorporés.<p><p>Chapitre 4 :« A unified approach to interventions »<p><p>Un modèle unique, permettant d’expliquer les trois étapes du processus d’intervention, est proposé dans ce chapitre. Ces trois étapes sont relatives (i) au choix d’intervenir, (ii) au choix d’intervenir de façon secrète et (iii) à la perception des interventions par le marché. Grâce à l’inclusion de déterminants spécifiques pour ces différentes étapes, cette approche multidimensionnelle permet d’appréhender leurs interrelations et, donc, de mieux comprendre les différents arbitrages réalisés par les autorités lorsqu’elles décident d’intervenir. / Doctorat en Sciences économiques et de gestion / info:eu-repo/semantics/nonPublished

Cette thèse de doctorat composée de trois chapitres contribue au développement de la problématique sur la sélection de modèle de volatilité de type GARCH. Le premier chapitre propose une étude de simulation sur la sélection de modèles dans le cadre spécifique des modèles à changement de régimes. On propose des expériences de simulation permettant de mettre en évidence l'inefficacité des critères de sélection usuels dans des cas particuliers, ce qui peut conduire à des erreurs de spécification lors du choix de modèle. Le deuxième chapitre propose un test du multiplicateur de Lagrange de mauvaise spécification dans les modèles GARCH univariés. L'hypothèse nulle admet que le processus générateur des données est un modèle GARCH linéaire tandis que sous l'hypothèse alternative il correspond à une forme fonctionnelle inconnue qui est linéarisée à l’aide d’un développement de Taylor. On illustre le test dans une application empirique sur les taux de change. Le dernier chapitre étudie l'impact du prix du pétrole sur les spreads de Credit Default Swaps souverains de deux pays exportateurs de pétrole: le Vénézuela et la Russie. Utilisant des données récentes, nous trouvons que les rendements du prix du pétrole impactent les spread de CDS souverains du Vénézuela directement alors que cela passe par le canal du taux de change pour la Russie. Ce chapitre emploie des méthodes statistiques avancées, notamment l'utilisation de modèles à changement de régimes Markoviens. Finalement, l'appendice propose le manuel de la toolbox MSGtool (Matlab) qui propose une collection de fonctions pour l'étude des modèles à changement de régimes Markoviens. La toolbox est très user-friendly. / This Ph.D. thesis composed by three chapters contributes to the development of model selection in GARCH-type models.The first chapter investigates whether the most common selection criteria lead to choose the right specification in a regime switching framework. We propose simulation experiments which reveal the inefficiency of some selection criteria in particular cases which lead to misspecification. Depending on the Data Generating Process used in the experiments, great care is needed when choosing a criterion.In the second chapter, a misspecication test for GARCH-type models is presented. We propose a Lagrange Multiplier type test based on a Taylor expansion to distinguish between (G)ARCH models and unknown nonlinear GARCH-type models. This test can be seen as a general misspecication test. We investigate the size and the power of this test through Monte Carlo experiments. We show the usefulness of our test with an illustrative empirical example based on daily exchange rate returns.In the third chapter, we study the impact of oil price returns on sovereign Credit Default Swaps (CDS) spreads for two major oil producers, Russia and Venezuela. Using daily spreads from 2008 to 2015, we find that crude oil price returns are a critical determinant of Venezuela CDS spreads changes, but does not explain significantly Russian CDS spreads. Indeed, oil prices seem to impact Russian CDS spreads through the exchange rates canal. Finally, we propose as an appendix the manual of the MSGtool, a MATLAB toolbox, which provides a collection of functions for the simulation and estimation of a large variety of Markov Switching GARCH (MSG) models.

本研究要探討的是在新聞、論壇和社群媒體討論的相關訊息是否真的會影響匯率的運動的假設。對於這樣的研究目標，我們建立了一個實驗，首先以文字探勘技術應用在新聞、論壇與社群媒體來產生與匯率相關的數值表示。接著，機器學習技術應用於學習得到的數值表示和匯率波動之間的關係。最後，我們證明透過檢驗所獲得的關係的有效性的假設。在此研究中，我們提出一種兩階段的神經網路來學習與預測每日美金兌台幣匯率的走勢。不同於其他專注於新聞或者社群媒體的研究，我們將他們進行整合，並將論壇的討論納為輸入資料。不同的資料組合產生出多種觀點，而三個資料來源的不同組合可能會以不同的方式影響預測準確率。透過該方法，初步實驗的結果顯示此方法優於隨機漫步模型。 / This study wants to explore the hypothesis that the relevant information in the news, the posts in forums and discussions on the social media can really affect the daily movement of exchange rates. For such study objective, we set up an experiment, where the text mining technique is first applied to the news, the forum and the social media to generate numerical representations regarding the textual information relevant with the exchange rate. Then the machine learning technique is applied to learn the relationship between the derived numerical representations and the movement of exchange rates. At the end, we justify the hypothesis through examining the effectiveness of the obtained relationship. In this paper, we propose a hybrid neural networks to learn and forecast the daily movements of USD/TWD exchange rates. Different from other studies, which focus on news or social media, we integrate them and add the discussion of forum as input data. Different data combinations yield many views while different combination of three data sources might affect the forecasting accuracy rate in different ways. As a result of this method, the experiment result was better than random walk model.

This study examines the nature of the relationship which exists between mining sector production and development of the financial systems in South Africa. This is particularly important in that the mining sector is considered to be one of the major contributors to the country’s overall economic growth. South Africa is also considered to have a very well developed financial system, to the point where the dominance of one over the other is difficult to identify. Therefore offering insight into the nature of this relationship will assist policy makers in identifying the most effective policies in order to ensure that the developments within the financial systems impact appropriately on the mining sector, and ultimately on the economy. In addition to using the conventional proxies of financial system development, this study utilises the principal component analysis (PCA) to construct an index for the entire financial system. The multivariate cointegration approach as proposed by Johansen (1988) and Johansen and Juselius (1990) was then used to estimate the relationship between the development of the financial systems and the mining sector production for the period 1988-2008. The study reveals mixed results for different measures of financial system development. Those involving the banking system show that a negative relationship exists between total mining production and total credit extended to the private sector, while liquid liabilities has a positive relationship. Similarly, with the stock market system, mixed results are also obtained which reveal a negative relationship between total mining production and stock market capitalisation, while a positive relationship is found with secondary market turnover. Of all the financial system variables, only that of stock market capitalisation was found to be significant. The result with the financial development index reveals that a significant negative relationship exists between financial system development and total mining sector production. Results on the other variables controlled in the estimation show that positive and significant relationships exist between total mining production and both nominal exchange rate and political stability respectively. Increased mining production therefore takes place in periods of appreciating exchange rates, and similarly in the post-apartheid era. On the other hand, negative relationships were found for both trade openness and inflation control variables. The impulse response and variance decomposition analyses showed that total mining production explains the largest amount of shocks within itself. Overall, the study reveals that the mining sector might not have benefited much from the development in the South African financial system.

Monetary policy in South African is the primary means by which the authorities can influence activity in the overall economy. The South African Reserve Bank accommodates banks through repo transactions for which they charge the repo rate. The most important market in the transmission of the repo rate to the rest of the economy is the interbank market. As such, a detailed discussion of this market is given. In September 200 I the monetary authorities made certain adjustments to the repo system of accommodation, which included changing the repo rate from a floating rate to a fixed rate that would be administratively determined by the MPC. This was done to address certain weaknesses in the floating rate system. This thesis examines and compares the period before and after the adjustments to the repo system, with the aim of determining whether or not the monetary authorities achieved the goals intended from making this change. The repo rate, prime interbank rate, 3-month NCO rate and the prime lending rate are analysed using the Engle-Granger two variable approach and an ECM model to test for causality. It was found that the monetary authorities did not achieve their intended goals as the relationship between the repo rate and the interbank rate was more significant in the first period. Furthermore, the direction of causality the authorities hoped to achieve by implementing the changes were in fact already in place. As such the adjustments to the system changed the transmission mechanism from the one desired by the authorities to one that was not intended. The conclusions reached by this study show that, in terms of the objectives of the monetary authorities, the previous repo system functioned better. / KMBT_363 / Adobe Acrobat 9.54 Paper Capture Plug-in

This study assessed the effects of currency volatility on the Johannesburg Stock Exchange. An evaluation of literature on exchange rate volatility and stock markets was conducted resulting into specification of an empirical model.The Generalised Autoregressive Conditional Heteroskedascity (1.1) (GARCH) model was used in establishing the relationship between exchange rate volatility and stock market performance. The study employed monthly South African data for the period 2000 – 2010. The data frequency selected ensured an adequate number of observations. A very weak relationship between currency volatility and the stock market was confirmed. The research finding is supported by previous studies. Prime overdraft rate and total mining production were found to have a negative impact on Market capitalisation. Surprisingly, US interest rates were found to have a positive impact on Market capitalisation. This study recommended that, since the South African stock market is not really exposed to the negative effects of currency volatility, government can use exchange rate as a policy tool to attract foreign portfolio investment. The weak relationship between currency volatility and the stock market suggests that the JSE can be marketed as a safe market for foreign investors. However, investors, bankers and portfolio managers still need to be vigilant in regard to the spillovers from the foreign exchange rate into the stock market. Although there is a weak relationship between rand volatility and the stock market in South Africa, this does not necessarily mean that investors and portfolio managers need not monitor the developments between these two variables.

This work examined the optimality of the inclusion of the exchange rate in the reaction function of the Central Bank in an inflation targeting framework. The study attempts to answer the question whether the exchange rate should have an independent role in an open economy Taylor-type rule. To this end, a Taylor-type rule is incorporating the exchange rate is estimated by the cointegration and vector error correction modeling (VECM) using quarterly data for the period of 1995 to 2009. The empirical studies point out the importance of the exchange rates in explaining and forecasting the behaviour of the South African Reserve Bank monetary policy control variable.

The thesis focuses on modelling and forecasting the exchange rate time series volatility. The basic approach used for the conditional variance modelling are class (G)ARCH models and their variations. Modelling of the conditional mean is based on the use of AR autoregressive models. Due to the breach of one of the basic assumption of the models (normality assumption), an important part of the work is a detailed analysis of unconditional distribution of returns enabling the selection of a suitable distributional assumption of error terms of (G)ARCH models. The use of leptokurtic distribution assumption leads to a major improvement of volatility forecasting compared to normal distribution. In regard to this fact, the often applied GED and the Student's t distributions represent the key-stones of this work. In addition, the less known distributions are applied in the work, e.g. the Johnson's SU and the normal Inverse Gaussian Distribution. To model volatility, a great number of linear and non-linear models have been tested. Linear models are represented by ARCH, GARCH, GARCH in mean, integrated GARCH, fractionally integrated GARCH and HYGARCH. In the event of the presence of the leverage effect, non-linear EGARCH, GJR-GARCH, APARCH and FIEGARCH models are applied. Using suitable models according to the selected criteria, volatility forecasts are made with different long-term and short-term forecasting horizons. Outcomes of traditional approaches using parametric models (G)ARCH are compared with semi-parametric neural networks based concepts that are widely applicable in clustering and also in time series prediction problems. In conclusion, a description is given of the coincident and different properties of the analyzed exchange rate time series. The author further summarized the models that provide the best forecasts of volatility behaviour of the selected time series, including recommendations for their modelling. Such models can be further used to measure market risk rate by the Value at Risk method or in future price estimating where future volatility is inevitable prerequisite for the interval forecasts.

Ce travail de thèse poursuit une perspective double dans l'usage conjoint des méthodes de Monte Carlo séquentielles (MMS) et de l'algorithme Espérance-Maximisation (EM) dans le cadre des modèles de Markov cachés présentant une structure de dépendance markovienne d'ordre supérieur à 1 au niveau de la composante inobservée. Tout d'abord, nous commençons par un exposé succinct de l'assise théorique des deux concepts statistiques à Travers les chapitres 1 et 2 qui leurs sont consacrés. Dans un second temps, nous nous intéressons à la mise en pratique simultanée des deux concepts au chapitre 3 et ce dans le cadre usuel ou la structure de dépendance est d'ordre 1, l'apport des méthodes MMS dans ce travail réside dans leur capacité à approximer efficacement des fonctionnelles conditionnelles bornées, notamment des quantités de filtrage et de lissage dans un cadre non linéaire et non gaussien. Quant à l'algorithme EM, il est motivé par la présence à la fois de variables observables, et inobservables (ou partiellement observées) dans les modèles de Markov Cachés et singulièrement les modèles de volatilité stochastique étudié. Après avoir présenté aussi bien l'algorithme EM que les méthodes MCS ainsi que quelques une de leurs propriétés dans les chapitres 1 et 2 respectivement, nous illustrons ces deux outils statistiques au travers de la calibration d'un modèle de volatilité stochastique. Cette application est effectuée pour des taux change ainsi que pour quelques indices boursiers au chapitre 3. Nous concluons ce chapitre sur un léger écart du modèle de volatilité stochastique canonique utilisé ainsi que des simulations de Monte Carlo portant sur le modèle résultant. Enfin, nous nous efforçons dans les chapitres 4 et 5 à fournir les assises théoriques et pratiques de l'extension des méthodes Monte Carlo séquentielles notamment le filtrage et le lissage particulaire lorsque la structure markovienne est plus prononcée. En guise d’illustration, nous donnons l'exemple d'un modèle de volatilité stochastique dégénéré dont une approximation présente une telle propriété de dépendance. / This thesis pursues a double perspective in the joint use of sequential Monte Carlo methods (SMC) and the Expectation-Maximization algorithm (EM) under hidden Mar­kov models having a Markov dependence structure of order grater than one in the unobserved component signal. Firstly, we begin with a brief description of the theo­retical basis of both statistical concepts through Chapters 1 and 2 that are devoted. In a second hand, we focus on the simultaneous implementation of both concepts in Chapter 3 in the usual setting where the dependence structure is of order 1. The contribution of SMC methods in this work lies in their ability to effectively approximate any bounded conditional functional in particular, those of filtering and smoothing quantities in a non-linear and non-Gaussian settings. The EM algorithm is itself motivated by the presence of both observable and unobservable ( or partially observed) variables in Hidden Markov Models and particularly the stochastic volatility models in study. Having presented the EM algorithm as well as the SMC methods and some of their properties in Chapters 1 and 2 respectively, we illustrate these two statistical tools through the calibration of a stochastic volatility model. This application is clone for exchange rates and for some stock indexes in Chapter 3. We conclude this chapter on a slight departure from canonical stochastic volatility model as well Monte Carlo simulations on the resulting model. Finally, we strive in Chapters 4 and 5 to provide the theoretical and practical foundation of sequential Monte Carlo methods extension including particle filtering and smoothing when the Markov structure is more pronounced. As an illustration, we give the example of a degenerate stochastic volatility model whose approximation has such a dependence property.

Thesis (M. Commerce (Economics)) -- University of Limpopo, 2016 / The purpose of this paper is to ascertain the impact of the real effective exchange rate on South Africa‟s trade balance and whether the J-curve phenomenon and the Marshal-Lerner condition are satisfied in the economy. Using data spanning the period 1980Q1 – 2014Q4, the Autoregressive Distributed Lag (ARDL) bounds test as well as the Johansen cointegration test were employed to test for the long run cointegrating relationship between the variables. The ARDL approach was employed to estimate both the long run and short run models as well as to ascertain whether the Marshal – Learner condition as well as the J-curve phenomenon are satisfied in the RSA economy. The results from the cointegration tests show that there is a stable long run equilibrium relationship between the trade balance, real effective exchange rate, domestic GDP, money supply, terms of trade and foreign reserves. The results from the Autoregressive Distributed Lag long run model show that a depreciation of the ZAR improves the trade balance, thus confirming the MarshalLerner condition. The results further reveal that domestic GDP and money supply both have a significant negative impact on the trade balance in the long run with the terms of trade reported positive as well. Foreign reserves were not found to significantly affect the trade balance in the long run. In the short run, the ARDL error correction model shows that a ZAR depreciation leads to a deterioration of the trade balance, thus confirming the J-curve effect for the RSA economy. The terms of trade effect was reported positive in the short run, thus confirming the Harberger-LaursenMetzler effect (HLME) in the process. Money supply, domestic GDP and foreign reserves are also found to have a significant negative impact on the trade balance in the short run. Finally, the error correction model reveals that about 26% of the disequilibrium in the trade balance model is corrected in each quarter.