Enhancing Value at Risk Models Using Extreme Value Theory in Extreme Market Conditions : Empirical Evidence from Finland during COVID-19 crisis
Ikäheimonen, Miro (2025-05-28)
Enhancing Value at Risk Models Using Extreme Value Theory in Extreme Market Conditions : Empirical Evidence from Finland during COVID-19 crisis
Ikäheimonen, Miro
(28.05.2025)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
avoin
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2025060661957
https://urn.fi/URN:NBN:fi-fe2025060661957
Tiivistelmä
Periods of extreme market volatility, such as the COVID-19 crisis, challenge the reliability of traditional risk management tools in the financial sector. Sudden and severe downturns reveal the limitations of widely used models, raising questions about how well financial institutions can anticipate and prepare for such events.
One commonly used tool in risk management has been Value at Risk (VaR), which es- timates the potential loss in value of a portfolio over a given period. However, standard VaR models, particularly those based on the assumption of normally distributed returns, often underestimate the likelihood and impact of rare, extreme events. Although recent regulations, such as Basel III, have shifted focus toward more conservative measures like Expected Shortfall, VaR remains a widely recognized and applied metric. Improving its reliability during turbulent times is therefore both a practical and academic concern.
This thesis explores how Extreme Value Theory (EVT) can enhance the performance of VaR during times of financial stress, using the COVID-19 market crash in Finland as a case study. EVT is specifically designed to model rare and extreme outcomes, offering tools that may better capture the risks traditional models overlook.
The aim of the thesis is to examine whether EVT can improve the accuracy of VaR in extreme market conditions. The analysis compares different EVT approaches, such as the Peaks-Over-Threshold (POT) and Block Maxima methods, and evaluates how their results differ from conventional VaR calculations. The backtesting results indicate that EVT-based VaR, particularly the POT method, outperforms traditional VaR approaches during the COVID-19 crisis in Finland. Among the traditional models, the Historical Sim- ulation VaR demonstrated the best performance. The findings contribute to the broader discussion on risk modeling by providing insight into the strengths and weaknesses of both traditional and EVT-based methods in periods of market distress. Yksi yleisimmin käytetyistä riskienhallinnan malleita on Value at Risk (VaR), joka arvioi sijoitussalkun mahdollisen arvonmenetyksen tietyllä todennäköisyystasolla ja aikavälillä. Perinteiset VaR-mallit, erityisesti normaalijakaumaan perustuvat menetelmät, ali- arvioivat usein äärimmäisten tapahtumien todennäköisyyksiä ja vaikutuksia. Vaikka uudempi sääntely, kuten Basel III -kehikko, on siirtänyt painopistettä kohti varovaisempia menetelmiä, kuten odotettavissa olevaa tappiota (Expected Shortfall), VaR säilyttää asemansa keskeisenä ja laajasti sovellettuna riskimittarina. Sen luotettavuuden parantaminen kriisiaikoina onkin sekä käytännöllisesti että akateemisesti merkittävä tavoite. Tämä tutkielma tarkastelee, kuinka ääriarvoteorian (Extreme Value Theoryn, EVT) soveltaminen voi parantaa VaR-mallien tarkkuutta poikkeuksellisissa markkinaolosuhteissa. Ääriarvoteoria tarjoaa työkaluja harvinaisten ja äärimmäisten arvojen mallintamiseen, mikä voi auttaa tunnistamaan riskejä, jotka jäävät perinteisten mallien ulottumattomiin. Tutkielman tavoitteena on selvittää, voidaanko ääriarvoteorian avulla parantaa VaR- mallien suorituskykyä äärimmäisissä markkinatilanteissa. Analyysissa vertaillaan ääri- arvoteorian eri lähestymistapoja, kuten Peaks-Over-Threshold (POT) ja Block Max- ima -menetelmiä, sekä arvioidaan niiden antamia tuloksia suhteessa perinteisiin VaR- laskelmiin. Testitulokset osoittavat, että ääriarvoteoria-pohjaiset VaR-mallit, erityisesti POT-menetelmä, suoriutuivat paremmin kuin perinteiset VaR-mallit COVID-19-kriisin aikana Suomessa. Perinteisistä malleista historialliseen dataan perustuva VaR tuotti parhaan tuloksen backtestauksessa. Tutkimus tarjoaa näkökulmaa riskimallinnuksen kehittämiseen suomalaisilla rahoitusmarkkinoilla COVID-19-kriisin aikana.
One commonly used tool in risk management has been Value at Risk (VaR), which es- timates the potential loss in value of a portfolio over a given period. However, standard VaR models, particularly those based on the assumption of normally distributed returns, often underestimate the likelihood and impact of rare, extreme events. Although recent regulations, such as Basel III, have shifted focus toward more conservative measures like Expected Shortfall, VaR remains a widely recognized and applied metric. Improving its reliability during turbulent times is therefore both a practical and academic concern.
This thesis explores how Extreme Value Theory (EVT) can enhance the performance of VaR during times of financial stress, using the COVID-19 market crash in Finland as a case study. EVT is specifically designed to model rare and extreme outcomes, offering tools that may better capture the risks traditional models overlook.
The aim of the thesis is to examine whether EVT can improve the accuracy of VaR in extreme market conditions. The analysis compares different EVT approaches, such as the Peaks-Over-Threshold (POT) and Block Maxima methods, and evaluates how their results differ from conventional VaR calculations. The backtesting results indicate that EVT-based VaR, particularly the POT method, outperforms traditional VaR approaches during the COVID-19 crisis in Finland. Among the traditional models, the Historical Sim- ulation VaR demonstrated the best performance. The findings contribute to the broader discussion on risk modeling by providing insight into the strengths and weaknesses of both traditional and EVT-based methods in periods of market distress.