
Nové metody ve schvalování úvěrů
Rychnovský, Michal ; Arlt, Josef (advisor) ; Pecáková, Iva (referee) ; Veselý, Petr (referee)
This thesis contributes to the field of applied statistics and financial modeling by analyzing mathematical models used in retail credit underwriting processes. Specifically, it has three goals. First, the thesis aims to challenge the performance criteria used by established statistical approaches and propose focusing on predictive power instead. Secondly, it compares the analytical leverage of the established and other suggested methods according to the newly proposed criteria. Third, the thesis seeks to develop and specify a new comprehensive profitabilitybased underwriting model and critically reflect on its strengths and weaknesses. In the first chapter I look into the area of probability of default modeling and argue for comparing the predictive power of the models in time rather than focusing on the random testing sample only, as typically suggested in the scholarly literature. For this purpose I use the concept of survival analysis and the Cox model in particular, and apply it to a real Czech banking data sample alongside the commonly used logistic regression model to compare the results using the Gini coefficient and lift characteristics. The Cox model performs comparably on the randomly chosen validation sample and clearly outperforms the logistic regression approach in the predictive power. In the second chapter, in the area of loss given default modeling I introduce two Coxbased models, and compare their predictive power with the standard approaches using the linear and logistic regression on a real data sample. Based on the modified coefficient of determination, the Cox model shows better predictions. Third chapter focuses on estimating the expected profit as an alternative to the risk estimation itself and building on the probability of default and loss given default models, I construct a comprehensive profitability model for fixterm retail loans underwriting. The model also incorporates various related riskadjusted revenues and costs, allowing more precise results. Moreover, I propose four measures of profitability, including the riskadjusted expected internal rate of return and return on equity and simulate the impact of the model on each of the measures. Finally, I discuss some weaknesses of these approaches and solve the problem of finding default or fraud concentrations in the portfolio. For this purpose, I introduce a new statistical measure based on a predefined expert critical default rate and compare the GUHA method with the classification tree method on a real data sample. While drawing on the comparison of different methods, this work contributes to the debates about survival analysis models used in financial modeling and profitability models used in credit underwriting.


Selected problems of financial time series modelling
Hendrych, Radek ; Cipra, Tomáš (advisor) ; Arlt, Josef (referee) ; Prášková, Zuzana (referee)
Title: Selected problems of financial time series modelling Author: Radek Hendrych Department: Department of Probability and Mathematical Statistics (DPMS) Supervisor: Prof. RNDr. Tomáš Cipra, DrSc., DPMS Abstract: The present dissertation thesis deals with selected problems of financial time series analysis. In particular, it focuses on two fundamental aspects of condi tional heteroscedasticity modelling. The first part of the thesis introduces and discusses selfweighted recursive estimation algorithms for several classic univariate conditional heteroscedasticity models, namely for the ARCH, GARCH, RiskMetrics EWMA, and GJRGARCH processes. Their numerical capabilities are demonstrated by Monte Carlo experiments and real data examples. The second part of the thesis proposes a novel approach to conditional covariance (correlation) modelling. The suggested modelling technique has been inspired by the essential idea of the multivariate orthogonal GARCH method. It is based on a suitable type of linear timevarying orthogonal transformation, which enables to employ the constant conditional correlation scheme. The correspond ing model is implemented by using a nonlinear discretetime state space representation. The proposed approach is compared with other commonly applied models. It demon strates its...


Methods for periodic and irregular time series
Hanzák, Tomáš ; Cipra, Tomáš (advisor) ; Arlt, Josef (referee) ; Prášková, Zuzana (referee)
Title: Methods for periodic and irregular time series Author: Mgr. Tomáš Hanzák Department: Department of Probability and Mathematical Statistics Supervisor: Prof. RNDr. Tomáš Cipra, DrSc. Abstract: The thesis primarily deals with modifications of exponential smoothing type methods for univariate time series with periodicity and/or certain types of irregularities. A modified Holt method for irregular times series robust to the problem of "timeclose" observations is suggested. The general concept of seasonality modeling is introduced into HoltWinters method including a linear interpolation of seasonal indices and usage of trigonometric functions as special cases (the both methods are applicable for irregular observations). The DLS estimation of linear trend with seasonal dummies is investigated and compared with the additive HoltWinters method. An autocorrelated term is introduced as an additional component in the time series decomposition. The suggested methods are compared with the classical ones using real data examples and/or simulation studies. Keywords: Discounted Least Squares, Exponential smoothing, HoltWinters method, Irregular observations, Time series periodicity


Use of Interest Rate Models for Interest Rate Risk Management in the Czech Financial Market Environment
Cíchová Králová, Dana ; Arlt, Josef (advisor) ; Cipra, Tomáš (referee) ; Witzany, Jiří (referee)
The main goal of this thesis is to suggest an appropriate approach to interest rate risk modeling in the Czech financial market environment in various situations. Three distinct periods are analyzed. These periods, which are the period before the global financial crisis, period during the financial crisis and in the aftermath of the global financial crisis and calming subsequent debt crisis in the eurozone, are characterized by different evaluation of liquidity and credit risk, different relationship between financial variables and market participants and different degree of market regulations. Within this goal, an application of the BGM model in the Czech financial market environment is crucial. Use of the BGM model for the purpose of predicting a dynamics of a yield curve is not very common. This is firstly due to the fact that primary use of this model is a valuation of interest rate derivatives while ensuring the absence of arbitrage and secondly its application is relatively difficult. Nevertheless, I apply the BGM model to obtain predictions of the probability distributions of interest rates in the Czech and eurozone market environment, because its complexity, direct modeling of a yield curve based on market rates and especially a possibility of parameter estimation based on current swaptions volatilities quotations may lead to a significant improvement of predictions. This improvement was also confirmed in this thesis. Use of swaptions volatilities market quotations is especially useful in the period of unprecedented mone tary easing and increased number of central banks and other regulators interventions into financial markets that occur after the financial crisis, because it reflects current market expectations which also include future interventions. As a consequence of underdevelopment of the Czech financial market there are no market quotations of Czech koruna denominated swaptions volatilities. I suggest their approximations based on quotations of euro denominated swaptions volatilities and also using volatilities of koruna and euro forward rates. Use of this approach ensures that predictions of the Czech yield curve dynamics contain current market expectations. To my knowledge, any other author has not presented similar application of the BGM model in the Czech financial market environment. In this thesis I further predict a Czech and Euro area money market yield curve dynamics using the CIR and the GP models as representatives of various types of interest rates models to compare these predictions with BGM predictions. I suggest a comprehensive system of three criteria, based on comparison of predicti ons with reality, to describe a predictive power of selected models and an appropria teness of their use in the Czech market environment during different situations in the market. This analysis shows that predictions of the Czech money market yield curve dynamics based on the BGM model demonstrate high predictive power and the best 8 quality in comparison with other models. GP model also produces relatively good qua lity predictions. Conversely, predictions based on the CIR model as a representative of short rate model family completely failed when describing reality. In a situation when the economy allows negative rates and there is simultaneously a significant likelihood of their implementation, I recommend to obtain predictions of Czech money market yield curve dynamics using GP model which allows existence of negative interest rates. This analysis also contains a statistical test for validating the predictive power of each model and information on other tests. Berkowitz test rejects a hypothesis of accurate predictions for each model. However, this fact is common in real data testing even when using relatively good model. This fact is especially caused by difficult fulfilment of test conditions in real world. To my knowledge, such an analysis of the predictive power of selected interest rate models moreover in the Czech financial market environment has not been published yet. The last goal of this thesis is to suggest an appropriate approach to obtaining pre dictions of Czech government bonds risk premium dynamics. I define this risk premium as a difference between government bond yields and fixed rate of CZK IRS with the same length. I apply the GP model to describe the dynamics of this indicator of the Czech Republic credit risk. In order to obtain a time series of the risk premium which are necessary for estimation of GP model parameters I firstly estimate yield curves of Czech government bonds using Svensson model for each trading day since 2005. Resulting si mulations of risk premium show that the GP model predicts the real development of risk premiums of all maturities relatively well. Hence, the proposed approach is suitable for modeling of Czech Republic credit risk based on the use of information extracted from financial markets. I have not registered proposed approach to risk premium modeling moreover in the Czech financial market environment in other publications.


Nonnegative linear operators and their use in econometric and statistic models
Horský, Richard ; Arlt, Josef (advisor) ; Vrabec, Michal (referee) ; Klazar, Martin (referee)
Nonnegative operators, in special case nonnegative matrices, are an interesting topics for many scientists and scientific teams from the beginning of the 20th century. It is not suprising because there are a lot of applications in different areas of science like economy, statistics, linear programming, computer science and others. We can give as the particular example the theory of the Markov chains in which we deal with nonnegative matrices, so called transition matrices. They are of the special form and we called them stochastic matrices. Another example is given by the nonnegative operator on spaces of infinite dimension which is employed in the theory of stochastic processes. It is the backward shift operator called the lag operator as well. The nonnegativity in these examples is considered as the piecewise nonnegativity. Another type of nonnegativity is that in the sense of inner products. In the case of matrices we talk about positivedefinite or positivesemidefinite matrices. A typical example is the covariance matrix of a random vector or symmetrization of any linear operator, for instance the symmetrization of the difference operator. The terms inverse problem or illposed problem have been gaining popularity in modern science since the middle of the last century. The subjects of the first publications in this area were related to quantum scattering theory, geophysics, astronomy and others. Thanks to powerful computers the chances for applications of the theory of inverse and illposed problems has extended in almost all fields of science which use mathematical methods. Illposed problems bear the feature of instability and there is the need of regularization if we want to get some reasonable solution. A typical example of the regularization is the differencing of stochastic process with the purpose to obtain a stationary process. Another concept of regularization used for solving e.g. integral equations with compact operators consists in application of regularization method as truncated singular value decomposition, Tichonov regularization method or Landweber iteration method. Mathematical tools employed in this work are those of the functional analysis. It is the area of mathematics in which distinct mathematical structures meet each other. They are structures built within different mathematical disciplines as mathematical analysis, topology, theory of sets, algebra (mainly linear algebra) and theory of measure (probability). The functional analysis framework enables us to obtain right formulations of definitions and problems providing the general view on the notions and problems of the theory of stochastic processes.


Construction of Linear Stochastic Models of SARIMA Class Time Lines – Automatized Method
Trcka, Peter ; Arlt, Josef (advisor) ; Hindls, Richard (referee)
This work concerns the creation of automatized procedure of ARIMA and SARIMA class model choice according to BoxJenkins methodology and in this connection, also deals with force testing of unit roots and analysis of applying of informatics criteria when choosing a model. The goal of this work is to create an application in the environment R that can automatically choose a model of time array generating process. The procedure is verified by a simulation study. In this work an effect of values of generating ARMA (1,1) model processes parameters is examined, for his choice and power of KPSS test, augmented DickeyFuller and PhillipsPeron test of unit roots.

 
 

Analysis of delays in the econometric model
Arlt, Josef ; Radkovský, Štěpán
The study deals with the detection time delay chrakter econometric model captures the relationship between the time series. In three parts includes the mean delay, delay variance, transforming time series analysis of the relationship and the time delay between the time series of interest rates.
Fulltext: PDF

 