Centre for Modelling and Analysis of Big Data in Finance and Economics

https://university.innopolis.ru/en/about/structure/director

Research Novelty and Contributions

The existing approaches to the analysis in finance and economics usually consider the problems of modeling the effects of crises and those of interdependence of financial and economic markets separately from each other. In particular, powerful machinery for modeling and analysis of the effects of crises on financial and economic variables and markets is provided by the theory of heavy-tailed distributions. Together with the development of its applications in finance and economics, the research and literature in this fields and their applications in practice have focused on approaches to modeling of market interdependence that are based on the theory of copulas, that is, functions that characterize all the dependence properties of financial and economic variables considered. A problem of significant theoretical and practical interest is the development of new methods of modeling and the analysis of financial and economic markets that combine the above approaches. One of the main goals of the present project is the analysis of the above research problems using computer and statistical modeling of wide classes of copula structures of interdependence of financial and economic variables and indicators with heavy-tailed distributions, including copula models that generate tail dependence and, thus, financial contagion effects. One of important problems in the modern literature in finance, economics, risk management and insurance consists in extensions of the results for independent risks and random variables affected by crises to the case of dependence. One of the main directions of the research on the project is the analysis of robustness of a number of important models in these fields under crises, data heterogeneity, heavy-tailed distributions and interdependence. The analysis of robustness properties under heavy-tailed distributions and dependence in the project will include the study of optimality of diversification in the value at risk (VaR) framework, insurance and re-insurance models for catastrophe risks firm growth theory, optimal bundling models and widely used statistical and econometric inference methods. The wide spectrum of dependence structures considered in the project will be analyzed using different classes of widely used copulas, including copula structures that generate tail dependence and financial contagion effects. The research on the project will focus on computer and mathematical modeling and statistical analysis of the effects of crises and financial contagion and interdependence properties on different developed and emerging markets, including Russia, its regions and post-Soviet economies. It will provide robust estimates of heavy-tailedness parameters and their dynamics for key economic and financial indicators in the markets considered. The projects' results will also include applications of the obtained statistical results in the analysis of macroeconomic effects of crises and large fluctuations on economic and financial markets. Similar estimates will also be provided for copula models describing interdependence properties of key financial and economic variables and financial contagion phenomena in different markets. Dependence, heterogeneity and presence of outliers in data on many key financial and economic variables and indicators significantly complicate their statistical analysis. Unfortunately, many approaches to the analysis of dependent and heterogeneous data have poor statistical properties in finite samples that are typically observed in practice. Therefore, it is of significant interest to consider and focus on development and applications of new robust inference methods with better statistical properties. Further, important practical problems consist in incorporation of new and improved approaches to robust inference into widely used statistical software, development of new statistical software packages and toolboxes on their basis and the use of modern computer technologies in the analysis of properties of robust inference and their applications in the study of financial and economic phenomena and markets. The project will develop a wide range of robust inference approaches for dependent, heterogeneous and heavy-tailed data. In particular, it will focus on further analysis and applications of the t-statistic based approaches in Ibragimov & Müller (2010) to robust large sample inference. The research in this direction will also focus on the development of general related robust inference procedures that rely on conservativeness of test statistics employed in the analysis. In particular, the project will develop robust tests for equality of two or more parameters of interest using new probabilistic results on conservativeness of the two-sample t-statistic for testing equality of two means and their extensions. The latter robust tests will allow one to conduct robust large sample inference on equality of two parameters of interest in the following simple way: Partition the data into some number of groups, estimate the considered parameters for each group, and then conduct the standard two-sample t-test with the resulting group estimators. The research on the project will further provide a wide range of applications of the robust inference approaches, including those in the robust analysis of effects of crises and other structural breaks and changes, robust analysis of treatment effects and evaluation of effectiveness of economic policies, tests for equality of parameters of interest in two or more heavy-tailed samples and the analysis of volatility and covariance stationarity. The study will also focus on extensions of the robust inference approaches to other research problems of key interest in statistics and econometrics. The project will further provide robust estimates and the analysis of the dynamics of economic and financial markets over time and their differences across economies. The research in this direction, in particular, will focus on the analysis of changes of key economic and financial indicators, including high-frequency financial returns and foreign exchange rates. Among other contributions, the study will considerably complement the results available in the literature, especially in terms of the focus on applications of robust inference methods in emerging and transition markets with pronounced heterogeneity, dependence and large fluctuations in economic and financial variables, including those in Russia and other post-Soviet countries. The research on the project will include construction of robust confidence intervals for parameters of interest as well as the robust analysis of changes in the parameters due to crises and other structural breaks and changes. The project will further provide comparisons of conclusions of the research with those obtained using the traditional statistical analysis methods. The study is expected to be complete in the sense of providing new methods for robust statistical inference, empirical and economic motivation for their use as well as important empirical applications for real-world data on financial and economic markets affected by crises, their propagation and financial contagion. An important factor of the novelty and contribution of the research on the project consists in the wide use of modern computer technologies and high-performance (parallel and distributed) computing systems. The research on the project in this direction will include adaptation of modern approaches to computer and mathematical modeling of and robust statistical inference on financial and economic markets for effective analysis on high-performance computing systems and development of the relevant software. The developed new robust statistical and econometric methods will be incorporated into widely used software packages for computer, mathematical and statistical modeling and analysis; the work on the project will further provide new software on the base of the new robust approaches. The wide of computer technologies and high-performance computing systems and their software in the project will include broad applications of new approaches to robust statistical analysis and mathematical modeling of financial and economic markets as well as the numerical analysis of properties of the newly developed robust inference methods and their comparisons with the existing robust statistical and econometric procedures.