Pavel Krupskii

Factor copula models for multivariate data

General conditional independence models for d observed variables, in terms of p latent variables, are presented in terms of bivariate copulas that link observed data to latent variables. The representation is called a factor copula model and the classical multivariate normal model with a correlation matrix having a factor structure is a special case. Dependence and tail properties of the model are obtained. The factor copula model can handle multivariate data with tail dependence and tail asymmetry, properties that the multivariate normal copula does not possess. It is a good choice for modeling high-dimensional data as a parametric form can be specified to have O(d) dependence parameters instead of O(d^2) parameters. Data examples show that, based on the Akaike information criterion, the factor copula model provides a good fit to financial return data, in comparison with related truncated vine copula models.

Structured factor copula models

In factor copula models for multivariate data, dependence is explained via one or several common factors. These models are flexible in handling tail dependence and asymmetry with parsimonious dependence structures. We propose two structured factor copula models for the case where variables can be split into non-overlapping groups such that there is homogeneous dependence within each group. A typical example of such variables occurs for stock returns from different sectors. The structured models inherit most of dependence properties derived for common factor copula models. With appropriate numerical methods, efficient estimation of dependence parameters is possible for data sets with over 100 variables. We apply the structured factor copula models to analyze a financial data set, and compare with other copula models for tail inference. Using model-based interval estimates, we find that some commonly used risk measures may not be well discriminated by copula models, but tail-weighted dependence measures can discriminate copula models with different dependence and tail properties.

Tail-weighted measures of dependence

Multivariate copula models are commonly used in place of Gaussian dependence models when plots of the data suggest tail dependence and tail asymmetry. In these cases, it is useful to have simple statistics to summarize the strength of dependence in different joint tails. Measures of monotone association such as Kendalls tau and Spearmans rho are insufficient to distinguish commonly used parametric bivariate families with different tail properties. We propose lower and upper tail-weighted bivariate measures of dependence as additional scalar measures to distinguish bivariate copulas with roughly the same overall monotone dependence. These measures allow the efficient estimation of strength of dependence in the joint tails and can be used as a guide for selection of bivariate linking copulas in vine and factor models as well as for assessing the adequacy of fit of multivariate copula models. We apply the tail-weighted measures of dependence to a financial data set and show that the measures better discriminate models with different tail properties compared to commonly used risk measures – the portfolio value-at-risk and conditional tail expectation.

Factor Copula Models for Replicated Spatial Data

ABSTRACT We propose a new copula model that can be used with replicated spatial data. Unlike the multivariate normal copula, the proposed copula is based on the assumption that a common factor exists and affects the joint dependence of all measurements of the process. Moreover, the proposed copula can model tail dependence and tail asymmetry. The model is parameterized in terms of a covariance function that may be chosen from the many models proposed in the literature, such as the Matérn model. For some choice of common factors, the joint copula density is given in closed form and therefore likelihood estimation is very fast. In the general case, one-dimensional numerical integration is needed to calculate the likelihood, but estimation is still reasonably fast even with large datasets. We use simulation studies to show the wide range of dependence structures that can be generated by the proposed model with different choices of common factors. We apply the proposed model to spatial temperature data and compare its performance with some popular geostatistics models. Supplementary materials for this article are available online.

A copula model for non-Gaussian multivariate spatial data

We propose a new copula model for replicated multivariate spatial data. Unlike classical models that assume multivariate normality of the data, the proposed copula is based on the assumption that some factors exist that affect the joint spatial dependence of all measurements of each variable as well as the joint dependence among these variables. The model is parameterized in terms of a cross-covariance function that may be chosen from the many models proposed in the literature. In addition, there are additive factors in the model that allow tail dependence and reflection asymmetry of each variable measured at different locations, and of different variables to be modeled. The proposed approach can therefore be seen as an extension of the linear model of coregionalization widely used for modeling multivariate spatial data. The likelihood of the model can be obtained in a simple form and, therefore, the likelihood estimation is quite fast. The model is not restricted to the set of data locations, and using the estimated copula, spatial data can be interpolated at locations where values of variables are unknown. We apply the proposed model to temperature and pressure data, and we compare its performance with that of a popular model from multivariate geostatistics.

Tail-weighted dependence measures with limit being the tail dependence coefficient

ABSTRACT For bivariate continuous data, measures of monotonic dependence are based on the rank transformations of the two variables. For bivariate extreme value copulas, there is a family of estimators , for , of the extremal coefficient, based on a transform of the absolute difference of the α power of the ranks. In the case of general bivariate copulas, we obtain the probability limit of as the sample size goes to infinity and show that (i) for is a measure of central dependence with properties similar to Kendalls tau and Spearmans rank correlation, (ii) is a tail-weighted dependence measure for large α, and (iii) the limit as is the upper tail dependence coefficient. We obtain asymptotic properties for the rank-based measure and estimate tail dependence coefficients through extrapolation on . A data example illustrates the use of the new dependence measures for tail inference.

Extreme-value limit of the convolution of exponential and multivariate normal distributions: Link to the Hüsler–Reiß distribution

The multivariate Husler–Reis copula is obtained as a direct extreme-value limit from the convolution of a multivariate normal random vector and an exponential random variable multiplied by a vector of constants. It is shown how the set of Husler–Reis parameters can be mapped to the parameters of this convolution model. Assuming there are no singular components in the Husler–Reis copula, the convolution model leads to exact and approximate simulation methods. An application of simulation is to check if the Husler–Reis copula with different parsimonious dependence structures provides adequate fit to some data consisting of multivariate extremes.