The con­fer­ence is go­ing to take place at Eötvös Loránd Uni­ver­sity, Bu­dapest, Hun­gary Pázmány Péter Sétány 1/​A, Gömb Aula.

If you need free park­ing place for the time of the con­fer­ence, please send your li­cense plate num­ber to riskconf@math.elte.hu.

Paul Em­brechts (ETH Zürich)

Basel III, Sol­vency II and Be­yond: a Crit­i­cal Ap­praisal

The reg­u­la­tory land­scape is un­der­go­ing con­sid­er­able changes world­wide. The 2007-2009 Fi­nan­cial Cri­sis brought into ques­tion sev­eral as­pects of reg­u­la­tion for bank­ing, the so-called Basel and Sol­vency guide­lines. At the same time, de­mo­graphic and eco­nomic de­vel­op­ments (e.g. longevity, low in­ter­est rates) are caus­ing ma­jor prob­lems for the in­sur­ance in­dus­try, and this mainly, but not ex­clu­sively, for life in­sur­ance. Added to these we do wit­ness im­por­tant changes to so­ci­ety at large, also dri­ven by in­for­ma­tion tech­nol­ogy. Be­sides the ob­vi­ous so­cial and po­lit­i­cal changes ex­pe­ri­enced world­wide, we should add more tech­no­log­i­cally dri­ven ones like net­work vul­ner­a­bil­ity and sys­temic risk, new prod­ucts, large data (data sci­ence, ma­chine learn­ing), block-chain tech­nol­ogy, cy­ber se­cu­rity. These de­vel­op­ments will no doubt have a con­sid­er­able im­pact on the fi­nan­cial and in­sur­ance in­dus­try both at the busi­ness as well as at the reg­u­la­tory level. In this talk I will dis­cuss some of the un­der­ly­ing is­sues from a more per­sonal per­spec­tive as a re­searcher in Quan­ti­ta­tive Risk Man­age­ment.

Slides

Gá­bor Vígh - Dóra Bagyin­szki - Nor­bert Hári (Mor­gan Stan­ley)

The im­pact of sto­chas­tic LI­BOR-OIS ba­sis on coun­ter­party risk

Be­fore the on­set of the credit crunch in 2007, the dif­fer­ence be­tween Lon­don In­ter­bank Of­fered Rate (LI­BOR) and the Overnight In­dexed Swap Rate (OIS) was neg­li­gi­ble. In the cri­sis the spreads be­tween the two rates sud­denly started to widen and since then it has been evolv­ing ran­domly. Prices of in­stru­ments linked LI­BOR rates started to re­flect sto­chas­tic spreads, a new risk fac­tor emerged. The dis­con­nect be­tween the two rates re­quired the in­dus­try to re­vise the mod­el­ing as­sump­tions, pric­ing for­mu­las and hedg­ing strate­gies.

This re­search in­ves­ti­gates the im­pact of the sto­chas­tic LI­BOR-OIS spreads on fu­ture coun­ter­party ex­po­sure dis­tri­b­u­tions by com­par­ing the re­sults ob­tained from a sto­chas­tic spread model to the in­dus­try wide de­ter­min­is­tic spread as­sump­tion. We con­sid­ered the sto­chas­tic ba­sis model pro­posed by Mer­cu­rio and Li (2016) with two ba­sis dy­nam­ics: the ex­tended Va­sicek and ex­tended CIR mod­els. Both mod­els are cal­i­brated to two dis­tinct his­tor­i­cal ba­sis time se­ries: i.) to the fi­nan­cial cri­sis pe­riod and ii.) to a more re­cent and sta­ble pe­riod. The analy­sis fo­cuses on a sin­gle tenor, on a sin­gle cur­rency and a vanilla Zero Coupon Swap in­stru­ment.

Petr Jakubík (EIOPA and Charles Uni­ver­sity Prague)

Stress tests and their con­tri­bu­tion to fi­nan­cial sta­bil­ity

The cur­rent mod­er­ate eco­nomic growth com­ple­mented with low yields and in­creased pol­icy re­lated eco­nomic un­cer­tainty poses many chal­lenges for the Eu­ro­pean fi­nan­cial sec­tor. Hence, it is im­por­tant to as­sess prop­erly and timely all po­ten­tial risks for fi­nan­cial sta­bil­ity. Stress test ex­er­cises as one of the most com­plex as­sess­ment tools could help to iden­tify the key risks and vul­ner­a­bil­i­ties and as­sess their po­ten­tial im­pacts on fi­nan­cial sys­tems. It is im­por­tant to con­duct such ex­er­cises not only at na­tional level, but also at the Eu­ro­pean level. Ad­di­tion­ally, the in­creased in­ter­con­nect­ed­ness be­tween bank­ing, in­sur­ance and other fi­nan­cial sec­tors call for de­vel­op­ing new method­olo­gies that would al­low to con­duct cross sec­toral stress test ex­er­cises.

Slides

Ti­bor Szen­drei - Katalin Varga (Cen­tral Bank of Hun­gary)

A fac­tor based in­dex of sys­temic stress in the fi­nan­cial sys­tem

Track­ing and mon­i­tor­ing stress within the fi­nan­cial sys­tem is a key com­po­nent for fi­nan­cial sta­bil­ity and macro­pru­den­tial pol­icy pur­poses. Fi­nan­cial stress mea­sures are im­por­tant as in­di­ca­tors mea­sur­ing ma­te­ri­alised risks, en­abling pol­icy mak­ers to take cor­rec­tive mea­sures in time. This pre­sen­ta­tion in­tro­duces a new mea­sure of con­tem­po­ra­ne­ous stress within the Hun­gar­ian fi­nan­cial sys­tem named Fac­tor based In­dex of Sys­temic Stress (FISS). Its sta­tis­ti­cal de­sign is a dy­namic Bayesian fac­tor method. The main method­olog­i­cal in­no­va­tion of the FISS is the abil­ity to fully cap­ture in­for­ma­tion con­tained in per­sis­tent, high-fre­quency data with the us­age of com­mon sto­chas­tic trends as fac­tors. The FISS is planned to be a key el­e­ment of the Hun­gar­ian macro­pru­den­tial toolkit. Apart from its pol­icy use the FISS can also be utilised as a thresh­old vari­able for VAR mod­els, aim­ing to quan­tify the ef­fects of macro­pru­den­tial in­stru­ments.

Di­ane Pier­ret (HEC Uni­ver­sity of Lau­sanne)

Stressed Banks (joint work with Roberto Steri)

We in­ves­ti­gate the risk tak­ing in­cen­tives of “stressed banks” — the banks that are sub­ject to an­nual reg­u­la­tory stress tests in the U.S. since 2011. Stressed banks are sub­ject to more strin­gent cap­i­tal re­quire­ments com­pared to other banks de­pend­ing on the as­sessed risk­i­ness of their port­fo­lios to a reg­u­la­tory stress sce­nario. On one hand, strin­gent cap­i­tal re­quire­ments pro­vide stressed banks with mo­tives to in­vest in riskier as­sets with higher ex­pected re­turns to off­set their in­creased cost of fund­ing orig­i­nat­ing from costly eq­uity in­jec­tions. On the other hand, stressed banks are sub­ject to more in­va­sive mon­i­tor­ing by the reg­u­la­tor, who im­poses bank-spe­cific cap­i­tal re­quire­ments on the ba­sis of the as­sessed risk of their in­di­vid­ual port­fo­lios. Mon­i­tor­ing gives in­cen­tives to stressed banks to in­vest in low-risk as­sets that are less sen­si­tive to the reg­u­la­tory stress sce­nario in or­der to re­duce their cap­i­tal re­quire­ment. Our re­sults high­light the im­por­tance of reg­u­la­tory mon­i­tor­ing of banks’ port­fo­lios in par­al­lel to set­ting more strin­gent cap­i­tal re­quire­ments.

Slides

Thomas Ver­braken (MSCI)

Stress test­ing in prac­tice: A real-life case study

Dri­ven by reg­u­la­tory re­quire­ments, fi­nan­cial in­sti­tu­tions ex­pe­ri­ence the need to im­prove and ex­pand their stress test­ing prac­tices. The topic of this pre­sen­ta­tion is a real-life case study of a fi­nan­cial in­sti­tu­tion set­ting up a new frame­work for stress test­ing their client port­fo­lios. We will dis­cuss the pro­posed method­ol­ogy to ar­rive at a rel­e­vant set of stress tests suited to their use case. This, among other things, en­tails the de­f­i­n­i­tion of his­tor­i­cal stress test sce­nar­ios as well as the qual­ity as­sess­ment of these sce­nar­ios.

Zoltán Fóris (Mor­gan Stan­ley)

Mod­el­ing chal­lenges in stress test­ing

Com­pre­hen­sive Cap­i­tal Analy­sis and Re­view (CCAR), the an­nual stress test­ing ex­er­cise of the Fed­eral Re­serve Board (FRB) to be per­formed by all ma­jor US banks, is not only about cal­cu­lat­ing stress loss and cap­i­tal re­quire­ment. The FRB re­quires banks to demon­strate that mod­els used for stress loss cal­cu­la­tions are ap­pro­pri­ate. This may im­pose sig­nif­i­cant chal­lenge to mod­el­ers. We will see ex­am­ples of FRB re­quire­ments for the as­sess­ment of model as­sump­tions and lim­i­ta­tions, and bench­mark­ing, and dis­cuss some po­ten­tial ap­proaches to sat­isfy these re­quire­ments.

Balázs Székely (MSCI)

Min­i­mally bi­ased back­test for Ex­pected Short­fall

Re­cent re­sults (Acerbi, Szekely 2017) have shown that a back­test of Ex­pected Short­fall (ES) is nec­es­sar­ily ap­prox­i­mated, in the sense that it’s un­avoid­ably sen­si­tive to pos­si­ble er­rors in the pre­dic­tion of Value at Risk (VaR). We in­tro­duce the back­test for ES which min­i­mizes such sen­si­tiv­ity. The bias is small and pru­den­tial: any im­per­fect VaR pre­dic­tion re­sults in a more puni­tive test against ES and the ef­fect is gen­er­ally neg­li­gi­ble for small VaR dis­crep­an­cies. For this rea­son the back­test qual­i­fies as an ap­pro­pri­ate end-to-end model val­i­da­tion tool for ES based mod­els, no­tably Basel IV in­ter­nal mod­els for which the ques­tion is still largely open. Fur­ther­more, the ES back­test, as op­posed to the com­mon VaR back­test, es­ti­mates not only model ac­cep­tance prob­a­bil­ity, but also the pre­dic­tion dis­crep­ancy mag­ni­tude. This means that the back­test au­to­mat­i­cally mea­sures port­fo­lio-spe­cific cap­i­tal mul­ti­pli­ers. For the same rea­son, a new no­tion of “re­al­ized ES” emerges, anal­o­gous to the clas­si­cal re­al­ized vari­ance, which does not ex­ist for VaR.

• Al­bum 1
• Al­bum 2

Pro­gram com­mit­tee:

• L. Márkus (chair),
• N.M. Arató,
• J. Gáll,
• Gy. Michalet­zky,
• G. Mol­nár-Sáska,
• V. Prokaj,
• M. Rá­sonyi

Lo­cal or­gan­is­ers:

• A. Zem­pléni (chair),
• Á. Back­hausz,
• V. Csiszár

email: riskconf@math.elte.hu

The main spon­sor is This work­shop is also sup­ported by COST Ac­tion TD1409, Math­e­mat­ics for In­dus­try Net­work (MI-NET)