CEBA Talks

Authors: Dean Fantazzini

Abstract: This paper investigates the relationship between the bitcoin price and the hashrate by disentangling the effects of the energy efficiency of the bitcoin mining equipment, bitcoin halving, and of structural breaks on the price dynamics. For this purpose, we propose a methodology based on exponential smoothing to model the dynamics of the Bitcoin network energy efficiency. We consider either directly the hashrate or the bitcoin cost-of-production model (CPM) as a proxy for the hashrate, to take any nonlinearity into account. In the first examined subsample (01/08/2016–04/12/2017), the hashrate and the CPMs were never significant, while a significant cointegration relationship was found in the second subsample (11/12/2017–24/02/2020). The empirical evidence shows that it is better to consider the hashrate directly rather than its proxy represented by the CPM when modeling its relationship with the bitcoin price. Moreover, the causality is always unidirectional going from the bitcoin price to the hashrate (or its proxies), with lags ranging from one week up to six weeks later. These findings are consistent with a large literature in energy economics, which showed that oil and gas returns affect the purchase of the drilling rigs with a delay of up to three months, whereas the impact of changes in the rig count on oil and gas returns is limited or not significant.

Authors: Igor Kheifets and Peter C.B. Phillips

Abstract: Multicointegration is traditionally defined as a particular long run relationship among variables in a parametric vector autoregressive model that introduces links between these variables and partial sums of the equilibrium errors. This paper departs from the parametric model, using a semiparametric formulation that reveals the explicit role that singularity of the long run conditional covariance matrix plays in determining multicointegration. The semiparametric framework has the advantage that short run dynamics do not need to be modeled and estimation by standard techniques such as fully modified least squares (FM-OLS) on the original I(1) system is straightforward. The paper derives FM-OLS limit theory in the multicointegrated setting, showing how faster rates of convergence are achieved in the direction of singularity and that the limit distribution depends on the distribution of the conditional one-sided long run covariance estimator used in FM-OLS estimation. Wald tests of restrictions on the regression coefficients have nonstandard limit theory which depends on nuisance parameters in general. The usual tests are shown to be conservative when the restrictions are isolated to the directions of singularity and, under certain conditions, are invariant to singularity otherwise. Simulations show that approximations derived in the paper work well in finite samples. We illustrate our findings by analyzing fiscal sustainability of the US government over the post-war period.

Link to paper: https://cowles.yale.edu/publications/cfdp/cfdp-2210

Authors: Andrei Kostyrka and Dmitry Malakhov.

Abstract: We propose a univariate conditional density model where asset returns are decomposed into a sum of copula-connected unobserved positive and negative shocks, both continuous and discrete, thus yielding up to 4 distinct shocks. The ‘Bad environments, good environments’ model is a special case of our model with zero-mean uncorrelated shocks, dynamic shape parameters, and without jumps. We compare our models with different marginal distributions and copulæ to 40 well-established GARCH variants (4 distributions, 10 volatility dynamics) by backtesting them on a sample of S&P500 daily data. Our models with dynamic scale parameters and without jumps perform better both in sample and out of sample compared with standard models. However, all dynamic-shape models have on average poor out-of-sample performance. Using the best-performing model, we reveal some hidden characteristics of returns behaviour. We show that the independence assumption for signed shocks does not hold: models with correlated shocks perform better, covariance is an important component of total variance, and it is time-dependent with a leverage-like effect. Conditional skewness behaviour reveals naïve investors' expectations. The U.S. market on average has a propensity for bull trends and a lower possibility of a bear trend during crisis times. The relation between returns and volatility is either very non-linear or insignificant. In this draft we also show preliminary results for models with jumps which indicate that introduction of discrete jumps does not improve the model performance; however, negative jumps have greater sizes, and occur more frequently.

Authors: Stanislav Anatolyev

Abstract: We propose, derive and analyze AICm. Also, we do simulations with AICm.

Authors: Sofya Budanova

Abstract: Economists often model unobserved heterogeneity using finite mixtures. In practice, the number of mixture components is rarely known. Model parameters lack point-identification if the estimation includes too many components, thus invalidating the classic properties of maximum likelihood estimation. I propose a penalized likelihood method to estimate finite mixtures with an unknown number of components. The resulting Order-Selection-Consistent Estimator (OSCE) consistently estimates the true number of components and achieves oracle efficiency. This paper extends penalized estimation to models without point-identification and to mixtures with growing number of components. I apply the OSCE to estimate players’ rationality level in a coordination game.

Link to paper: https://drive.google.com/file/d/1kA5X3EkJgJ3lq9AcW05ynTz5QYbLNSAh/view?usp=sharing

Authors: Robert James, Henry Leung and Artem Prokhorov

Abstract: We design an adaptive framework for detection of predatory trading behavior. Its key component is an extension of a pattern recognition tool, originating from the fields of engineering and signal processing and adapted to modern electronic systems of securities trading. The new methodology combines flexibility of dynamic time warping, rigor of extreme value theory and richness of order book data of an exchange to accurately identify predatory trading without access to many confirmed illegal transactions for training. The method is shown to achieve significant improvements over alternative approaches in the identification of illegal insider trading cases included in a high-frequency dataset provided by an large investment bank.

Link to paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3722391.

Authors: Alexei Kolokolov and Roberto Renò

Abstract: Even moderate amounts of zero returns in financial data, associated with stale prices, are heavily detrimental for reliable jump inference. Price staleness is easily mistaken for jumps by statistics based on multipower variation. We harness staleness-robust estimators to re-appraise the statistical features of jumps in financial markets. We find that jumps are much less frequent and much less contributing to price variation than what found by the empirical literature so far. In particular, the empirical finding that volatility is driven by a pure jump process is actually shown to be an artifact due to staleness.

Authors: Daniil Lomonosov, Andrey Polbin, Nikita Fokin

Abstract: The paper considers a simple Bayesian vector autoregressive model for the Russian economy based on data for real GDP, GDP deflator and oil price as an exogenous variable that acts as a proxy variable for the terms of trade. Along with the impact of oil price shocks, the model estimates the impact of supply and demand shocks, the identification of which is based on the approach of sign restrictions. According to the results obtained, at the end of 2014 and in 2015, demand shocks had a positive impact on GDP growth, which can be interpreted as a positive effect of the ruble devaluation at the end of 2014. In the next years, demand shocks led mainly to a slowdown in economic growth. The paper also attempts to identify monetary policy shocks and assesses their impact on GDP, household consumption and investment. According to the results, the effect of monetary shocks in 2015—2019 on all endogenous variables was negative. However, an increase in the interest rate at the end of 2014 is identified mostly as an endogenous reaction to other shocks, and the effect of the monetary shock on GDP in 2015 is nearly zero. In 2017, monetary shocks slowed down GDP by 0.92 percentage points.

Link to paper: https://www.vopreco.ru/jour/article/view/2938

Authors: Martin Bladt and Alexander J. McNeil

Abstract: An approach to the modelling of volatile time series, such as financial returns, using a class of uniformity-preserving transforms for uniform random variables is proposed. V-transforms describe the relationship between quantiles of the marginal distribution and quantiles of the distribution of a predictable volatility proxy variable constructed as a function of the data. They permit a copula-based approach to volatile time series in which arbitrary marginal distributions may be combined with copula processes describing the serial dependence structure of the data. The idea is illustrated using stationary d-vine copula processes which model higher-order Markov dependence using a decomposition of the joint density into pair copulas. Estimation of the models can be carried out by maximum likelihood or by a sequential method-of-moments procedure using rank correlations. In combination with suitably-chosen parametric marginal distributions, it is shown that the resulting models can rival and often outperform well-known models in the extended GARCH family.

Link to paper: http://arxiv.org/abs/2006.11088

Authors: Aleksandr Alekseev, Mikhail Sokolov

Abstract: This paper contributes to the theory of average rate of change (ARC) measurement. The contribution is twofold. First, it relates ARC measurement to intertemporal choice. We show that an ARC of a variable can be identified with a discount rate which makes an economic agent indifferent between the initial and final temporal states of the variable. Furthermore, there is a one-to-one correspondence between ARC measures and one-parameter families of time preferences indexed by a discount rate. Second, we employ an axiomatic approach to generalize the conventional ARC measures (such as the difference quotient and the continuously compounded growth rate) in several directions: to variables with arbitrary connected domains, to not necessarily time-shift invariant dependence on dates, to sets of time points other than an interval, to a benchmark-based evaluation. The generalized ARC measures turn out to correspond to the existing time preference models such as the discounted utility and the relative discounting model of Ok and Masatlioglu [Ok, E.A., Masatlioglu, Y., 2007. A theory of (relative) discounting. J. Econ. Theory 137(1), 214–245].

Link to paper: https://eusp.org/sites/default/files/econpapers/Ec-2020_01.pdf

Authors: Ramis Khabibullin and Sergei Seleznev

Abstract: We illustrate the ability of the stochastic gradient variational Bayes algorithm, which is a very popular machine learning tool, to work with macrodata and macromodels. Choosing two approximations (mean-field and normalizing flows), we test properties of algorithms for a set of models and show that these models can be estimated fast despite the presence of estimated hyperparameters. Finally, we discuss the difficulties and possible directions of further research.

Link to paper: https://www.cbr.ru/Content/Document/File/112571/wp-61_e.pdf

Authors: Nagler, T., Krüger, D., Min, A.

Abstract: Multivariate time series exhibit two types of dependence: across variables and across time points. Vine copulas are graphical models for the dependence and can conveniently capture both types of dependence in the same model. We derive the maximal class of graph structures that guarantees stationarity under a condition called translation invariance. Translation invariance is not only a necessary condition for stationarity, but also the only condition we can reasonably check in practice. In this sense, the new model class characterizes all practically relevant vine structures for modeling stationary time series. We propose computationally efficient methods for estimation, simulation, prediction, and uncertainty quantification and show their validity by asymptotic results and simulations.

Link to paper: https://arxiv.org/abs/2008.05990

Authors: Rustam Ibragimov, Artem Prokhorov and Johan Walden

Abstract: We develop a general framework for modelling the optimal bundling strategies of a multiproduct monopolist providing goods that have extreme and dependent valuations by consumers. The strategies crucially depend on the degree of heavy-tailedness of consumers' valuations, their dependence structure, and the degrees of complementarity and substitutability among the goods provided. For substitutes with suciently high degree of substitutability, the seller's optimal strategy is to provide goods separately. Provision of the goods as a single bundle is optimal for the seller in the case of complements and substitutes with relatively low degree of substitutability.

Authors: Kien Tran

Abstract: This paper proposes a semiparametric spatial autoregressive stochastic frontier model, where the functional form of the frontier is modeled nonparametrically. A three-step estimation procedure is considered where in the first two steps, a constrained semiparametric profile GMM is used to obtain the estimates of the spatial parameter and the unknown smooth function of the frontier; whilst in the final step, MLE is used to obtain the remaining parameters of the model. We derive the limiting distributions of the proposed estimators for both parametric and nonparametric components in the model. Monte Carlo simulations reveal that our proposed estimators perform well in finite samples.

Authors: Galina Besstremyannaya & Sergei Golovan

Abstract: This note discusses two errors in the approach proposed in Canay (2011) for constructing a computationally simple two-step estimator in a quantile regression model with quantile-independent fixed effects. Firstly, we show that Canay’s assumption about n/Ts → 0 for some s > 1 is not strong enough and can entail severe bias or even the non-existence of the limiting distribution for the estimator of the vector of coefficients. The condition n/T → 0 appears to be closer to the required set of restrictions. These problems are likely to cause incorrect inference in applied papers with large n/T, but the impact is less in applications with small n/T. In an attempt to improve Canay’s estimator, we propose a simple correction that may reduce the bias. The second error concerns the incorrect asymptotic standard error of the estimator of the constant term. We show that, contrary to Canay’s assumption, the within estimator has an influence function that is not i.i.d. and this affects inference. Moreover, the constant term is unlikely to be estimable at rate $\sqrt{nT}$, so a different estimator may not be available. However, the issue concerning the constant term does not have an effect on slope coefficients. Finally, we give recommendations to practitioners and conduct a meta-review of applied papers that use Canay’s estimator.

Link to paper: https://www.nes.ru/files/Preprints-resh/WP249.pdf

Authors: Artem Prokhorov and Jian Zhai

Abstract: Production systems that account for technical and allocative inefficiencies offer a natural way to model dependence using vine copulas. We construct such vine copulas using a recently proposed family of bivariate copulas that permit dependence between the magnitude (but not the sign) of the allocative inefficiency and the magnitude of the technical inefficiency. We show how to estimate such models and argue that they better reflect dependencies that arise in practice. Such models also allow for significant improvements in precision of inefficiency estimation.

Link to paper: https://sites.google.com/site/artembprokhorovv0/papers/vineSFA.pdf

Authors: Alexander Semenov, Gaurav Pandey, Maciej Rysz, Guanglin Xu

Abstract: Contextual bandit techniques have recently been used for generating personalized user recommendations in situations where collaborative filtering based algorithms may be inefficient. They are often used in cases when input data are dynamically changing as new users and content items constantly change. One such setting involves recommending news articles to users on the basis of context, i.e., user and article features. Contextual bandit methods sequentially select articles for recommendation to a user and continuously modify their strategies so as to present users with articles that maximize clicks. However, exclusively focusing on maximizing the number of clicks can lead to over-exposure of certain articles, while under-representing others. In an era of ever growing demand for digital news delivery, this, in turn, invokes the important notion of presenting news content to users in a ``socially responsible'' way. To this effect, we introduce a technique based on the contextual bandit framework that, in addition to maximization of the click rate, also considers historical frequency of an article as the ``cost'' associated with recommending it. It is demonstrated that this approach results in a more balanced distribution and a diverse set of recommended articles. Experiments utilizing a benchmark news dataset demonstrate the trade-off between clicks and diversity of recommended articles.

Link to paper: TBA

Authors: Alexandr Popkov

Abstract: This report largely describes a master thesis performed at the department of business Informatics, economic faculty, St. Petersburg State University on the theme of computer vision in the creation of services to research clients and growth of service quality. One of the modern sources of information about customers of a certain service is images-photos or videos that record events in the store or on its territory. The report presents options for practical solutions based on computer vision for registering consumer facts with Gazprom Neft data and possible improvements to the recommendations service with Airbnb open data.

Authors: Siyun He and Rustam Ibragimov

Abstract: The paper provides a comparative empirical study of predictability of cryptocurrency returns and prices using econometrically justified robust inference methods. We present a robust econometric analysis of predictive regressions incorporating factors that were suggested by Liu and Tsyvinski (2018) as useful predictors for cryptocurrency returns, including cryptocurrency momentum, stock market factors, an analogue of price-to-acceptance ratio, and Google trends measure of investors' attention. Due to inherent heterogeneity and dependence properties of returns and other time series in financial and crypto markets, we provide the analysis of the predictive regressions using heteroskedasticity and autocorrelation consistent (HAC) standard errors. We further present the analysis of the predictive regressions using recently developed t-statistic robust inference approaches (Ibragimov and Müller, 2010, 2016). We provide comparisons of robust predictive regression estimates between different cryptocurrencies and their corresponding risk and factor exposures. In general, the number of significant factors decreases as we use more robust t-tests, and the t-statistic robust inference approaches appear to perform better than the t-tests based on HAC standard errors in terms of pointing out interpretable economic conclusions.

Abstract: Целью проекта является возможность предсказать цену картин, формируемую на аукционах, с помощью предварительно обученных систем распознавания объектов на изображении. Мы используем модели, натренированные на датасете ImageNet, которые умеют с некоторой точностью относить предмет на изображении к одному из 1000 классов объектов, представленных в ImageNet. Предсказание цены строится с помощью линейной регрессии на основе извлеченных таким образом визуальных признаков.

Authors: Mikhail Mamonov and Anna Pestova

Abstract: Following the recent revival of the endogenous business cycle theories which highlight the predictability of changes of the business cycle phases, we provide a first comprehensive test of the associated ``too good is bad" hypothesis. The literature so far has convincingly shown that a rapid growth of credit predicts financial instability and is followed by a deeper recession. We employ the concept of ``exuberance indicators" and extend the list of these indicators beyond the credit by considering several other domestic overheating and external imbalance indicators. In a panel of 25 countries, we show that our exuberance indicators --- besides credit market exuberance --- convey important information about the probability of future recession, controlling for the classical recession predictors (term spread, stock market return, short-term interest rates) and sentiment indicators. In a causal analysis, we show that domestic credit supply and global financial shocks push exuberance indicators to hazardous values associated with a higher recession risk. Our counterfactual analysis indicates that, if we shut down global financial shocks during 2002--2006, the probability of the 2007--2009 recession would decline on average by 11 percentage points.

Link to paper:

Abstract: Cобираюсь рассказывать про свой опыт работы и разработки в различных компаниях. Первая компания про которую я хотел бы рассказать, это Unilever. Здесь я занимался оптимизацией производственной линии, а конкретно регистрацией неисправностей на производстве. Так же я собираюсь рассказать про работу на gazporm export. Здесь задача состояла в прогнозировании объёмов закачки/ выкачки сжиженного природного газа из газонефтехранилищ. При решении задачи использовались линейные модели прогнозирования, а так же экзогенные переменные различных типов и областей. В работе рассматривается построение, применение и оценка качества улучшения прогноза посредством включения в него экзогенных переменных.

Authors: Mikhail Mamonov and Anna Pestova

Abstract: In this paper, we study the effects of financial sanctions on the Russian economy at the bank level. Financial sanctions against largest Russian banks were imposed at different points in time from 2014 to 2019, thus leaving a room for yet-not-treated banks to adapt their international and domestic assets and liabilities. We use detailed monthly balance sheets, apply matching and difference-in-differences approaches and show that indeed such an informational effect of sanctions existed and pushed the banks to reduce their foreign assets and increase, rather than decrease, their foreign liabilities. Nonetheless, after the sanctions had been imposed, the treated banks further reduced their foreign assets and turned to decreasing their foreign liabilities. We show that, as a result of negative informational effects of the first portion of sanctions, yet-not-treated banks had faced panic runs of retail and corporate depositors, amounted to -5 and -10 percentage points for sectoral- and entity-sanction banks' assets. The government then stepped in and supported the banks. However, in the short run the banks were forced to reduce loans to non-financial firms by -1.8 and -7 percentage points of respectively sectoral- and entity-sanction banks' assets. With our SVAR-analysis, we show these figures imply that Russian GDP could had lost up to 2.3 percentage points, as average for 2014-2015. This number implies the sanctions had moderate though significant effect on the Russian economy.

Abstract: Nowadays, people's interest in contemporary art is growing all over the world, and Russia is not the exception. People turn to contemporary art for vivid emotions, but in order to avoid superficial attitude, cultural spaces use special participatory practices to create a better experience of interacting with art. Despite the general increase in the attendance of cultural spaces and the increased interest in contemporary art, cultural leisure is not a regular activity for the majority of Russians. The visitor's short-term relationship to the cultural space may end as soon as the emotion economy trend passes. That can lead to the risk for cultural spaces being left without benefits from their now increased popularity. In my work, using the example of a young contemporary art gallery - MYTH, I tried to find out how relevant practices of engagement are for their visitors and how this can help the gallery.

Authors: Dilyara Salakhova and Madina Karamysheva

Abstract: Due to the lack and poor access to the data on real exposures between banks, several methods have been proposed to reconstruct a network using market data. However, the question is what does the market-based network price? Does it represent well exposure-based networks? Which market-based approach is better? In this paper, we replicate several well-known market-based networks and build true exposure networks. We provide network characteristics comparison across different types of networks. We also conduct a regression analysis for finding the relative importance of different exposures. We find that networks evolve over time, while the global network structure remains stable. Regression analysis shows that the market identifies two banks as connected when they have similar business models defined by overlapping portfolios.

Authors: Svetlana Litvinova and Mervyn Silvapulle

Abstract: In risk management areas such as reinsurance, the need often arises to construct a confidence interval for a quantile in the tail of the distribution. While different methods are available for this purpose, doubts have been raised about the validity of full-sample bootstrap. In this paper, we first obtain some general results on the validity of full sample bootstrap for the tail quantile process. This opens the possibility of developing bootstrap methods based on tail statistics. Second, we develop a bootstrap method for constructing confidence intervals for high-quantiles of heavy-tailed distributions and show that it is consistent. In our simulation study, the bootstrap method for constructing confidence intervals for high quantiles performed overall better than the data tilting method; the data tilting method appears to be currently the preferred choice. The applicability of the bootstrap method is illustrated using the Danish fire insurance data.

Link to paper: https://www.monash.edu/business/ebs/our-research/publications/ebs/wp12-2018.pdf

Authors: Anton Skrobotov, Rasmus Pedersen and Rustam Ibragimov

Abstract: Many key variables in finance, economics and risk management, including financial returns and foreign exchange rates, exhibit nonlinear dependence, heterogeneity and heavy-tailedness of some usually largely unknown type. The presence of non-linear dependence (usually modelled using GARCH-type dynamics) and heavy-tailedness may make problematic the analysis of (non-)efficiency, volatility clustering and predictive regressions in economic and financial markets using traditional approaches that appeal to asymptotic normality of sample autocorrelation functions (ACFs) of returns and their squares. The paper presents several new approaches to deal with the above problems. We provide the results that motivate the use of measures of market (non-)efficiency, volatility clustering and nonlinear dependence based on (small) powers of absolute returns and their signed versions. The paper provides asymptotic theory for sample analogues of the above measures in the case of general time series, including GARCH-type processes. It further develops new approaches to robust inference on them in the case of general GARCH-type processes exhibiting heavy-tailedness properties typical for real-world financial markets. The approaches are based on robust inference methods exploiting conservativeness properties of t-statisticsIbragimov and Muller (2010,2016) and several new results on their applicability in the settings considered. In the approaches, estimates of parameters of interest (e.g., measures of nonlinear dependence given by sample autocorrelations of powers of the returns' absolute values) are computed for groups of data and the inference is based on t-statistics in resulting group estimates. This results in valid robust inference under a wide range of heterogeneity and dependence assumptions satisfied in financial and economic markets. Numerical results and empirical applications confirm advantages of the new approaches over existing ones and their wide applicability in the study of market (non-)efficiency, volatility clustering, nonlinear dependence, and other areas.

Authors: Masayuki Hirukawa, Di Liu and Artem Prokhorov

Abstract: Economists often use matched samples, especially dealing with earning data where some observations are missing in one sample and need to be imputed from another sample. Hirukawa and Prokhorov (2018) show that the ordinary least squares estimator using matched samples is inconsistent and propose two consistent estimators. We describe a new Stata command, msreg, which implements these two consistent estimators based on two samples. The estimators attain the parametric convergence rate if the number of continuous matching variables is no greater than four.

Link to paper

Abstract: According to the recent data, there are more than 1200 state art museums in Russia. Their activity, in contrast to the private ones, is governed by special conditions of museum objects and collections purchasing, and is restricted by the need to meet the approved government plans of spreading of cultural values and increasing of museum attendance. For these reasons, nowadays, the state museums need to apply modern marketing and management techniques in order to successfully compete with alternative recreational venues as well as private museums and galleries. The primary goal of this research is to give recommendations for the Russian state museums using the results obtained through an empirical study, which was aimed at revealing of the stylistic preferences of the target audience. The empirical part of this study is based on the findings of the Laboratory of Empirical Visual Aesthetics of the Vienna University using the variance analysis (ANOVA) and the structural equation modelling (SEM) to determine the factors influencing people’s appreciation of three art styles (modern, abstract, and classical) within two groups of respondents – experts and non-experts. Despite a relatively small sample size, the results demonstrated a more positive perception of the classical and the abstract art among the Russian respondents. Moreover, the relationship between “interest in” and “knowledge of” the art was proved to be statistically significant.

Authors: Evgenii Gilenko and Aleksandra Chernova

Abstract: Understanding of types and determinants of saving behavior of people is important for securing financial stability of the person, individually, and the country, at large. The commonly accepted viewpoint is that a higher level of financial literacy (as brought by the relevant economic education) via, in particular, better saving, leads to increasing financial well-being of people. But, as we discuss in this paper, this relation may have a more complicated nature: in some cases, financial literacy may have an adverse effect on people’s financial well-being. Moreover, to secure the positive effect of financial literacy on financial well-being, specifically, via more active saving, the appropriate programs should be introduced at the early stages of education of a person (e.g., in school). We use a representative sample of Russian high school students to test the above-mentioned ideas, accounting for the endogenous nature of financial literacy and, thus, employing a copula-based bivariate logistic regression to detect the actual magnitude of influence of financial literacy on the willingness to make savings. As a result, we demonstrate that for the studied cohort of adolescents, this magnitude of influence is much higher when we appropriately control for the endogeneity effect. Necessary policy recommendations are provided.

Authors: Stanislav Anatolyev and Vladimir Pyrlik

Abstract: Copulas are a convenient framework to synthesize joint distributions, particularly in higher dimensions. Currently, copula-based high dimensional settings are used for as many as a few hundred variables and require large data samples for estimation to be precise. In this paper, we employ shrinkage techniques for large covariance matrices in the problem of estimation of elliptical copulas whose dimensionality goes well beyond that typical in the literature. Specifically, we use the covariance matrix shrinkage of Ledoit and Wolf to estimate large matrix parameters of Gaussian and t copulas for up to thousands of variables, using up to 20 times lower sample sizes. The simulation study shows that the shrinkage estimation significantly outperforms traditional estimators, both in low and especially high dimensions. We also apply this approach to the problem of allocation of large portfolios.

Authors: Mardi Dungey, Raisul Islam, Vladimir Volkov

Abstract: We develop a means of visualizing the vulnerability of complex systems of financial interactions around the globe using Neural Network clustering techniques. We show how time-varying spillover indices can be translated into two dimensional crisis maps. The crisis maps have the advantage of showing the changing paths of vulnerability, including the direction and extent of the effect between source and affected markets. Using equity market data for 31 global markets over 1998–2017 we provide these crisis maps. These tools help portfolio managers and policy makers to distinguish which of the available tools for crisis management will be most appropriate for the form of vulnerability in play.

Abstract: In this paper we study statistical properties of the weighted network formed by 40.000 entrepreneurship communities from social media site VK.com. The goal of the research is to verify if the behavior of digital consumers has changed in the times of crisis and pandemic. We analyze two snapshots, the first one was collected in Autumn 2019, and the second one on April 2020, amidst COVID-19 pandemic. We compare such characteristics of the network as degree distribution, distribution of edge weights, and multivariate distribution of weights and degrees, and analyze how these characteristics changed during the pandemic.

Abstract: Several predictive regressions with variables such as dividend to price ratio used as predictors are by now well-established for financial returns in developed markets. Due to the autocorrelation and heterogeneity of the data, the standard approaches to the analysis of statistical significance of predictive regressors and their coefficients based on independent and identically distributed standard errors are not directly applicable in the case of stock returns. At the same time, while there are quite many researches devoted to developed markets in the field of return predictability, there are quite a few for emerging ones based only on conventional methods. This research is focusing on application of several methods to the Russian market data, providing inference about the predictability of stock returns. The methods include widely used heteroskedasticity and autocorrelation consistent (HAC) standard errors, t-statistics robust inference method and recently developed test relying on nonparametric correction of volatility.

Authors: Rustam Ibragimov, Jihyun Kim, Anton Skrobotov

Abstract: We propose two robust methods for testing hypotheses on unknown parameters of predictive regression models under heterogeneous and persistent volatility as well as endogenous, persistent and/or fat-tailed regressors and errors. The proposed robust testing approaches are applicable both in the case of discrete and continuous time models. Both of the methods use the Cauchy estimator to effectively handle the problems of endogeneity, persistence and/or fat-tailedness in regressors and errors. The difference between our two methods is how the heterogeneous volatility is controlled. The first method relies on robust t-statistic inference using group estimators of a regression parameter of interest proposed in Ibragimov & Muller (2010). It is simple to implement, but requires the exogenous volatility assumption. To relax the exogenous volatility assumption, we propose another method which relies on the nonparametric correction of volatility. The proposed methods perform well compared with widely used alternative inference procedures in terms of their finite sample properties.

[link to paper]