We consider daily rainfall observations at 32 stations in the province of North Holland (the Netherlands) during 30 years. Let T be total this area on one day. An important question is: what is amount that exceeded once 100 years? This clearly a problem belonging to extreme value theory. Also, it genuinely spatial problem. Recently, theory extremes continuous stochastic processes has been developed. Using ideas and much computer power (simulations), we have able come up with reasonable answer above.

We consider three measures on the systemic importance of a financial institution within interconnected system. Based measures, we study relation between size and its importance. From both theoretical model empirical analysis, find that in analyzing risk posed by one to system, should not be considered as proxy In other words, "too big fail" argument is always valid, alternative considered. provide estimation methodology under multivariate Extreme Value Theory (EVT) framework.

Abstract We test for the presence of a systematic tail risk premium in cross section expected returns by applying measure sensitivity assets to extreme market downturns, beta. Empirically, historical betas help predict future performance stocks downturns. During crash, with historically high suffer losses that are approximately 2 3 times larger than their low-tail-beta counterparts. However, we find no evidence associated betas. The theoretically additive and empirically persistent can...

Summary Denote the loss return on equity of a financial institution as X and that entire market Y. For given very small value p > 0, marginal expected shortfall (MES) is defined E{X|Y>QY(1−p)}, where Q Y(1 − p) (1 p)th quantile distribution The MES an important factor when measuring systemic risk institutions. wide non-parametric class bivariate distributions, we construct estimator establish asymptotic normality ↓ sample size n → ∞. Since are in particular interested case =...

Summary We extend classical extreme value theory to non-identically distributed observations. When the tails of distribution are proportional much statistics remains valid. The proportionality function for can be estimated non-parametrically along with (common) index. For a positive index, joint asymptotic normality both estimators is shown; they asymptotically independent. also establish forecasted high quantile and develop tests validity model. show through simulations good performance...

Flexible large-scale WLANs are now widely deployed in crowded and highly mobile places such as campus, airport, shopping mall company etc.But network management is hard for due to uneven interference throughput among links.So the traffic difficult predict accurately.In paper, through analysis of two real WLANs, Granger Causality found both scenarios.In combination with information entropy, it shows that prediction target AP considering can be more predictable than utilizing alone, or...

Summary In this paper, we decompose banks' systemic risk into two dimensions: the of a bank (“bank tail risk”) and link to system in financial distress (“systemic linkage”). Based on extreme value theory, estimate measure that can be decomposed subcomponents reflecting these dimensions. Empirically, assess relationships business models dimensions risk. The observed differences partly explain why micro‐ macroprudential perspectives sometimes have different implications for banking regulation.

We handle two major issues in applying extreme value analysis to financial time series, bias and serial dependence, jointly. This is achieved by studying correction methods when observations exhibit weak the sense that they come from $\beta$ -mixing series. For estimating index, we propose an asymptotically unbiased estimator prove its asymptotic normality under condition. The procedure dependence structure have a joint impact on variance of estimator. Then construct high quantiles. apply...

Classical extreme value statistics consists of two fundamental approaches: the block maxima (BM) method and peak-over-threshold (POT) approach. It seems to be general consensus among researchers in field that POT approach makes use observations more efficiently than BM method. We shed light on this discussion from three different perspectives. First, based recent theoretical results for method, we provide a comparison i.i.d. scenarios. argue data generating process may favour either one or...

This paper analyzes the robustness of standard techniques for risk analysis, with a special emphasis on Basel III measures. We focus difference between value-at-risk and expected shortfall, their small sample properties, scope manipulating measures how estimation can be improved. Overall, find that forecasts are extremely uncertain at low sizes, more accurate than while is easily manipulated without violating regulations. Finally implications practitioners regulators discussed along best...

We consider extreme value analysis for independent but nonidentically distributed observations. In particular, the observations do not share same index. Assuming continuously changing indices, we provide a nonparametric estimate functional Besides estimating index locally, also global estimator trend and its joint asymptotic theory. The theory can be used testing prespecified parametric in indices. it applied to test whether remains at constant level across all

A two-dimensional random vector in the domain of attraction an extreme value distribution G is said to be asymptotically independent (i.e. tail) if product its marginal functions. Ledford and Tawn (1996) discussed a form residual dependence this case. In paper we give characterization phenomenon (see also Ramos (2009)), offer extensions higher-dimensional spaces stochastic processes. Systemic risk banking system treated similar framework.

The efficiency of simulation algorithms for max-stable processes relies on the choice spectral representation: different choices result in sequences finite approximations to process. We propose a constructive approach yielding normalized representation that solves an optimization problem related simulating processes. algorithm based can be regarded as max-importance sampling. Compared other hitherto, our has at least two advantages. First, it allows exact comprising class Second, stopping...

Problem definition: This study examines the effects of mobile money on value chain, that is, network operators (MNOs), banks, and end users. Mobile is an innovative technology bundled with related services designed to provide cost-efficient financial inclusion for underserved populations in developing world. Academic/practical relevance: By studying interaction between chain structures service bundle compositions, we expand our understanding revenue allocation mechanisms economies. provides...

Abstract This paper considers the problem of estimating a linear model between two heavy-tailed variables if explanatory variable has an extremely low (or high) value. We propose estimator for coefficient by exploiting tail dependence and prove its asymptotic properties. Simulations show that our estimation method yields lower mean-squared error than regressions conditional on observations. In empirical application, we illustrate better performance approach relative to regression in...