Estimating the probability of simultaneous rainfall extremes within a region: a spatial approach

In this paper we investigate the impact of model mis-specification, in terms of the dependence structure in the extremes of a spatial process, on the estimation of key quantities that are of interest to hydrologists and engineers. For example, it is often the case that severe flooding occurs as a result of the observation of rainfall extremes at several locations in a region simultaneously. Thus, practitioners might be interested in estimates of the joint exceedance probability of some high levels across these locations. It is likely that there will be spatial dependence present between the extremes, and this should be properly accounted for when estimating such probabilities. We compare the use of standard models from the geostatistics literature with max-stables models from extreme value theory. We find that, in some situations, using an incorrect spatial model for our extremes results in a significant under-estimation of these probabilities which – in flood defence terms – could lead to substantial under-protection.     

On the estimation and application of max-stable processes

The theory of max-stable processes generalizes traditional univariate and multivariate extreme value theory by allowing for processes indexed by a time or space variable. We consider a particular class of max-stable processes, known as M4 processes, that are particularly well adapted to modeling the extreme behavior of multiple time series. We develop procedures for determining the order of an M4 process and for estimating the parameters. To illustrate the methods, some examples are given for modeling jumps in returns in multivariate financial time series. We introduce a new measure to quantify and predict the extreme co-movements in price returns.     

On the performance of the Bayesian composite likelihood estimation of max-stable processes

The max-stable process is a natural approach for modelling extrenal dependence in spatial data. However, the estimation is difficult due to the intractability of the full likelihoods. One approach that can be used to estimate the posterior distribution of the parameters of the max-stable process is to employ composite likelihoods in the Markov chain Monte Carlo (MCMC) samplers, possibly with adjustment of the credible intervals. In this paper, we investigate the performance of the composite likelihood-based MCMC samplers under various settings of the Gaussian extreme value process and the Brown–Resnick process. Based on our findings, some suggestions are made to facilitate the application of this estimator in real data.     

Extremes of Markov chains with tail switching potential

Summary.  A recent advance in the utility of extreme value techniques has been the characteri- zation of the extremal behaviour of Markov chains. This has enabled the application of extreme value models to series whose temporal dependence is Markovian, subject to a limitation that prevents switching between extremely high and extremely low levels. For many applications this is sufficient, but for others, most notably in the field of finance, it is common to find series in which successive values switch between high and low levels. We term such series Markov chains with tail switching potential, and the scope of this paper is to generalize the previous theory to enable the characterization of the extremal properties of series displaying this type of behaviour. In addition to theoretical developments, a modelling procedure is proposed. A simulation study is made to assess the utility of the model in inferring the extremal dependence structure of autoregressive conditional heteroscedastic processes, which fall within the tail switching Markov family, and generalized autoregressive conditional heteroscedastic processes which do not, being non-Markov in general. Finally, the procedure is applied to model extremal aspects of a financial index extracted from the New York Stock Exchange compendium.     

Investigating precipitation extremes in South Carolina with focus on the state's October 2015 precipitation event

The October 2015 precipitation event in the Southeastern United States brought large amounts of rainfall to South Carolina, with particularly heavy amounts in Charleston and Columbia. The subsequent flooding resulted in numerous casualties and hundreds of millions of dollars in property damage. Precipitation levels were so severe that media outlets and government agencies labeled this storm as a 1 in 1000-year event in parts of the state. Two points of discussion emerged as a result of this event. The first was related to understanding the degree to which this event was anomalous; the second was related to understanding whether precipitation extremes in South Carolina have changed over recent time. In this work, 50 years of daily precipitation data at 28 locations are used to fit a spatiotemporal hierarchical model, with the ultimate goal of addressing these two points of discussion. Bayesian inference is used to estimate return levels and to perform a severity-area-frequency analysis, and it is determined that precipitation levels related to this event were atypical throughout much of the state, but were particularly unusual in the Columbia area. This analysis also finds marginal evidence in favor of the claim that precipitation extremes in the Carolinas have become more intense over the last 50 years.     

多元极值的参数建模方法及其金融应用：最新进展述评

 由于现实中的极值事件往往倾向于同时或相继发生,因此多元极值研究正成为极值统计学的理论前沿和研究热点。本文对该领域中参数建模方法的最新进展做了系统性述评,包括经典多元极值理论、Ledford-Tawn-Ramos方法和Heffernan和Tawn条件法等,并指出了这些建模方法的优缺点以及未来可能的理论突破点。本文还全面分析了近年来多元极值分析方法在金融领域的国内外应用现状,并探讨其未来的应用前景,可能是在金融传染、组合问题和系统性风险管理等方面。     

A Bayesian approach to zero-inflated data in extremes

Abstract

The generalized extreme value (GEV) distribution is known as the limiting result for the modeling of maxima blocks of size n, which is used in the modeling of extreme events. However, it is possible for the data to present an excessive number of zeros when dealing with extreme data, making it difficult to analyze and estimate these events by using the usual GEV distribution. The Zero-Inflated Distribution (ZID) is widely known in literature for modeling data with inflated zeros, where the inflator parameter w is inserted. The present work aims to create a new approach to analyze zero-inflated extreme values, that will be applied in data of monthly maximum precipitation, that can occur during months where there was no precipitation, being these computed as zero. An inference was made on the Bayesian paradigm, and the parameter estimation was made by numerical approximations of the posterior distribution using Markov Chain Monte Carlo (MCMC) methods. Time series of some cities in the northeastern region of Brazil were analyzed, some of them with predominance of non-rainy months. The results of these applications showed the need to use this approach to obtain more accurate and with better adjustment measures results when compared to the standard distribution of extreme value analysis.     

Diagnostics for dependence within time series extremes

Summary. The analysis of extreme values within a stationary time series entails various assumptions concerning its long- and short-range dependence. We present a range of new diagnostic tools for assessing whether these assumptions are appropriate and for identifying structure within extreme events. These tools are based on tail characteristics of joint survivor functions but can be implemented by using existing estimation methods for extremes of univariate independent and identically distributed variables. Our diagnostic aids are illustrated through theoretical examples, simulation studies and by application to rainfall and exchange rate data. On the basis of these diagnostics we can explain characteristics that are found in the observed extreme events of these series and also gain insight into the properties of events that are more extreme than those observed.     

A semi-parametric model for multivariate extreme values

Threshold methods for multivariate extreme values are based on the use of asymptotically justified approximations of both the marginal distributions and the dependence structure in the joint tail. Models derived from these approximations are fitted to a region of the observed joint tail which is determined by suitably chosen high thresholds. A drawback of the existing methods is the necessity for the same thresholds to be taken for the convergence of both marginal and dependence aspects, which can result in inefficient estimation. In this paper an extension of the existing models, which removes this constraint, is proposed. The resulting model is semi-parametric and requires computationally intensive techniques for likelihood evaluation. The methods are illustrated using a coastal engineering application.     

A model for estimating point sources in spatial data

We introduce a point source model which may be useful for estimating point sources in spatial data. It may also be useful for modelling general spatial data, and providing a simple explanatory model for some data, whilst in other cases it may give a parsimonious representation. The model assumes that there are point sources (or sinks), usually at unknown positions, and that the mean value at a site depends on the distance from these sources. We discuss the general form of the model, and some methods for estimating the sources and the regression parameters. We demonstrate the methodology using a simulation study, and apply the model to two real data sets. Some possibilities for further research are outlined.     

Modelling individual fertility levels in Malawian women: a spatial semiparametric regression model

We applied semiparametric spatial Poisson models to analyse fertility decisions at individual level in Malawi. We used the 2000 Malawi Demographic and Health Survey (MDHS) to investigate determinants of fertility, in the model that allowed for nonlinear, fixed and spatial risk factors. Inference was based on the Bayesian approach. The unstructured spatial effects were modelled using the exchangeable prior, while for the structured spatial effects we used the intrinsic conditional autoregressive models. Nonlinear effects were modelled using P-splines. Results showed non-linear declining trends of fertility with year of marriage and increasing trends with age at marriage. We also observed significant unstructured effects, however, no significant spatial autocorrelated effects were displayed. Overall, total spatial effects were significantly different at district level.     

A conditional approach for multivariate extreme values (with discussion)

Summary.  Multivariate extreme value theory and methods concern the characterization, estimation and extrapolation of the joint tail of the distribution of a d -dimensional random variable. Existing approaches are based on limiting arguments in which all components of the variable become large at the same rate. This limit approach is inappropriate when the extreme values of all the variables are unlikely to occur together or when interest is in regions of the support of the joint distribution where only a subset of components is extreme. In practice this restricts existing methods to applications where d is typically 2 or 3. Under an assumption about the asymptotic form of the joint distribution of a d -dimensional random variable conditional on its having an extreme component, we develop an entirely new semiparametric approach which overcomes these existing restrictions and can be applied to problems of any dimension. We demonstrate the performance of our approach and its advantages over existing methods by using theoretical examples and simulation studies. The approach is used to analyse air pollution data and reveals complex extremal dependence behaviour that is consistent with scientific understanding of the process. We find that the dependence structure exhibits marked seasonality, with ex- tremal dependence between some pollutants being significantly greater than the dependence at non-extreme levels.     

A spatial structural equation model for health outcomes

Population level risk factors in spatial epidemiology (e.g. socioeconomic deprivation) are often not directly available but indirectly measured through census or other sources. This paper considers multiple health outcomes (e.g. mortality, hospital admissions) in relation to unmeasured latent constructs of population morbidity, established as relevant to explaining spatial contrasts in such health outcomes. The constructs are derived using a factor analytic approach in which observed area social indicators are measures of a smaller set of latent constructs. The constructs are allowed to be spatially correlated as well as correlated with one another. The possibility of nonlinear construct effects is considered using a spline regression. A case study considers suicide mortality and self-harm contrasts in 32 London boroughs, in relation to two latent constructs: area deprivation and social fragmentation.     

Parallel algorithms for Markov chain Monte Carlo methods in latent spatial Gaussian models

Markov chain Monte Carlo (MCMC) implementations of Bayesian inference for latent spatial Gaussian models are very computationally intensive, and restrictions on storage and computation time are limiting their application to large problems. Here we propose various parallel MCMC algorithms for such models. The algorithms' performance is discussed with respect to a simulation study, which demonstrates the increase in speed with which the algorithms explore the posterior distribution as a function of the number of processors. We also discuss how feasible problem size is increased by use of these algorithms.     

Statistics for Extreme Sea Currents

Estimates of various characteristics of extreme sea currents, such as speeds and their directions, are required when designing offshore structures. This paper extends standard statistical methods for extreme values to handle the directionality, temporal dependence and tidal non-stationarity that are present in sea current extremes. The methods are applied to a short period of data from the Inner Dowsing Light Tower in the North Sea. Substantial benefits, over existing methods, are obtained from our analysis of the sea current by decomposing it into tide and surge currents. In particular, we find that at the Inner Dowsing the strong directionality in extreme sea current speeds is completely explained by the tidal current and directionality in the non-extreme surge currents. This finding aids model fitting and extrapolation.     

Sparse moving maxima models for tail dependence in multivariate financial time series

The multivariate maxima of moving maxima (M4) model has the potential to model both the cross-sectional and temporal tail-dependence for a rich class of multivariate time series. The main difficulty of applying M4 model to real data is due to the estimation of a large number of parameters in the model and the intractability of its joint likelihood. In this paper, we consider a sparse M4 random coefficient model (SM4R), which has a parsimonious number of parameters and it can potentially capture the major stylized facts exhibited by devolatized asset returns found in empirical studies. We study the probabilistic properties of the newly proposed model. Statistical inference can be made based on the Generalized Method of Moments (GMM) approach. We also demonstrate through real data analysis that the SM4R model can be effectively used to improve the estimates of the Value-at-Risk (VaR) for portfolios consisting of multivariate financial returns while ignoring either temporal or cross-sectional tail dependence could potentially result in serious underestimate of market risk.     

Multivariate autoregressive extreme value process and its application for modeling the time series properties of the extreme daily asset prices

Abstract

In this article we suggest a new multivariate autoregressive process for modeling time-dependent extreme value distributed observations. The idea behind the approach is to transform the original observations to latent variables that are univariate normally distributed. Then the vector autoregressive DCC model is fitted to the multivariate latent process. The distributional properties of the suggested model are extensively studied. The process parameters are estimated by applying a two-stage estimation procedure. We derive a prediction interval for future values of the suggested process. The results are applied in an empirically study by modeling the behavior of extreme daily stock prices.     

Inference of a hidden spatial tessellation from multivariate data: application to the delineation of homogeneous regions in an agricultural field

Summary.  In a precision farming context, differentiated management decisions regarding fertilization, application of lime and other cultivation activities may require the subdivision of the field into homogeneous regions with respect to the soil variables of main agronomic significance. The paper develops an approach that is aimed at delineating homogeneous regions on the basis of measurements of a categorical and quantitative nature, namely soil type and resistivity measurements at different soil layers. We propose a Bayesian multivariate spatial model and embed it in a Markov chain Monte Carlo inference scheme. Implementation is discussed using real data from a 15-ha field. Although applied to soil data, this model could be relevant in areas of spatial modelling as diverse as epidemiology, ecology or meteorology.