Erwan Koch
EPFL SB MATH MATH-GE
MA A1 354 (Bâtiment MA)
Station 8
CH-1015 Lausanne
Web site: Web site: https://math.epfl.ch/
Web site: Web site: https://sma.epfl.ch/
Awards
SCOR Actuarial Awards France 2014 for the PhD thesis "Tools and Models for the Study of some Spatial and Network Risks: Application to Climate Extremes and Contagion in Finance".Research interests
- Weather extremes- Spatial, weather and climatic risks
- Extreme-value statistics
- Spatio-temporal statistics
- Data science and machine learning
- Actuarial science
- Green finance
Invited talks at conferences or workshops
CMStatistics 2017, London, Dec. 2017.A central limit theorem for functions of stationary max-stable random fields on R^d.
EVA 2017, Delft (Netherlands), June 2017.
Space-time max-stable models with spectral separability.
Workshop on Stochastic Weather Generators, Vannes (France), May 2016.
A frailty-contagion model for multi-site hourly precipitation driven by atmospheric covariates.
Workshop Mathematical Foundations of Heavy Tailed Analysis, University of Copenhagen, June 2015.
Spatial risk measures and applications to max-stable processes.
First Meeting AMERISKA Network, Paris, May 2015.
A frailty-contagion model for multi-site hourly precipitation driven by atmospheric covariates.
Uncertainty, Risk and Robustness Workshop at Columbia University, New York, Apr. 2015.
Spatial risk measures and applications to max-stable processes.
PEPER Workshop organized by Prof. Dr. P. Naveau, Aussois (France), Dec. 2013.
Spatial risk measures based on max-stable processes.
CFE 2013, London, Dec. 2013.
Spatial risk measures and applications to max-stable processes.
ISI WSC 2013, Hong Kong, Aug. 2013.
A multi-site hourly precipitation generator based on a frailty-contagion approach.
Economics and Management Doctoral School Workshop, Lyon, June 2012.
Estimation of max-stable processes by simulated maximum likelihood.
Contributed talks at conferences or workshops
Validation of Ensemble Forecasting Workshop, Aussois (France), Mar. 2019.Trends in the extremes of environments associated with severe US thunderstorms.
EVA 2015, Ann Arbor (USA), June 2015.
Spatial risk measures and applications to max-stable processes.
EVA 2013, Shanghai, July 2013.
Estimation of max-stable processes by simulated maximum likelihood.
12th IMSC, Jeju (Korea), June 2013.
A multi-site hourly precipitation weather generator based on a frailty-contagion approach.
30th French Finance Association Conference, Lyon, May 2013.
Endogenous Financial Network Formation for Solvency and Liquidity Shocks.
Symposium on Recent Advances in Extreme-Value Theory honouring Prof. Ross Leadbetter, Lisbon, March 2013.
Estimation of max-stable processes by simulated maximum likelihood.
PARTY (Perspectives on Actuarial Risks in Talks of Young Researchers) 2013, Ascona (Switzerland), Jan. 2013.
Estimation of max-stable processes and applications to insurance.
ERCIM 2012, Oviedo (Spain), Dec. 2012.
Estimation of max-stable processes by simulated maximum likelihood.
Entretiens Jacques Cartier 2012 (Applied Mathematics to Risk Management), Lyon, Nov. 2012.
Estimation of max-stable processes and applications to insurance.
Talks at seminars
Center for Research in Economics and Statistics (CREST) seminar in financial econometrics, Paris, May 2019.Infinitesimal perturbation analysis for risk measures based on the Smith max-stable random field.
Chalmers Statistics and Biomathematics Seminar, Gothenburg, May 2018.
Spatial risk measures induced by powers of max-stable random fields.
Columbia University Applied Mathematics Colloquium, New York, Apr. 2016.
Space-time max-stable models with spectral separability.
CREST seminar in financial econometrics, Paris, Jan. 2016.
Space-time max-stable models with spectral separability.
LMU (Ludwig Maximilian University of Munich) and TUM (Technical University of Munich) seminar in finance and insurance mathematics, Munich, Dec. 2015.
Spatial risk measures and applications to max-stable processes.
EPFL statistics seminar, Lausanne, Sept. 2015.
Space-time max-stable models with spectral separability.
Seminar at CNRM (National Center for Meteorological Research) Météo France, Toulouse, Feb. 2015.
Spatial risk measures for economic impacts due to extreme climatic events.
Seminar at Laboratoire de Mathématiques de Besançon, Dec. 2014.
Spatial risk measures and applications to max-stable processes.
Seminar of the working group on the theory of extreme values at LSTA (Theoretical and Applied Statistics Laboratory), Paris, Nov. 2014.
A multi-site hourly precipitation generator based on a frailty-contagion approach.
CREST seminar in financial econometrics, Paris, Oct. 2013.
Spatial risk measures based on max-stable processes.
IRMA (Research Institute in Advanced Mathematics) statistics seminar, Strasbourg, Nov. 2012.
Estimation of max-stable processes by simulated maximum likelihood.
CREST seminar in financial econometrics, Paris, Oct. 2012.
Estimation of max-stable processes and applications to insurance.
ISFA seminar, Lyon, Sept. 2012.
Estimation of max-stable processes and applications to insurance.
Seminar at LSCE (Laboratory of Climatic and Environmental Sciences), Gif-sur-Yvette (France), July 2012.
Financial models with frailty and contagion.
Teaching
One-semester (2 hours/week) course on “Probability and Statistics” for students in EnvironmentalSciences and Engineering (bachelor level), EPFL, Feb.–June 2022.
One-semester (2 hours/week) course on “Multivariate Statistics” for students in Mathematics, Applied
Mathematics, Physics and other programs (master level), EPFL, Feb.–June 2022.
One-semester (2 hours/week) course on “Probability and Statistics” for EPFL students in Chemistry and
Chemical Engineering and UNIL students in Forensic Science (bachelor level), EPFL, Sep.–Dec 2021.
One-semester (2 hours/week) course on “Statistical Theory” for students in Mathematics, Applied
mathematics, Data Science and other programs (master level), EPFL, Sep.–Dec 2021.
One-semester (2 hours/week) course on “Probability and Statistics” for students in Environmental
Sciences and Engineering (bachelor level), EPFL, Feb.–June 2021.
One-semester (2 hours/week) course on “Multivariate Statistics” for students in Mathematics, Applied
Mathematics, Physics and other programs (master level), EPFL, Feb.–June 2021.
One-semester (2 hours/week) course on “Probability and Statistics” for EPFL students in Chemistry and
Chemical Engineering and UNIL students in Forensic Science (bachelor level), EPFL, Sep.–Dec 2020.
One-semester (2 hours/week) course on “Statistical Theory” for students in Mathematics, Applied
mathematics, Data Science and other programs (master level), EPFL, Sep.–Dec 2020.
One-semester (2 hours/week) course on “Probability and Statistics” for students in Environmental
Sciences and Engineering (bachelor level), EPFL, Feb.–June 2020.
One-semester (2 hours/week) course on “Multivariate Statistics” for students in Mathematics, Applied
Mathematics, Physics and other programs (master level), EPFL, Feb.–June 2020.
One-semester (2 hours/week) course on “Probability and Statistics” for EPFL students in Chemistry and
Chemical Engineering and UNIL students in Forensic Science (bachelor level), EPFL, Sep.–Dec 2019.
One-semester (2 hours/week) course on “Statistical Theory” for students in Mathematics, Applied
Mathematics, Data Science and other programs (master level), EPFL, Sep.–Dec 2019.
One-semester (2 hours/week) course "Probability and Statistics" (bachelor level), EPFL, Feb.–June 2019.
Main teaching assistant for the course "Risk, rare events and extremes" (master level) taught by Prof. Dr. A. C. Davison, EPFL, Sep.-Dec 2018.
Main teaching assistant for the course "Probability and Statistics" (bachelor level) taught by Dr. E. Thibaud, EPFL, Feb.-June 2018.
Teaching assistant for the course "Geometry I" (bachelor level) taught by Prof. Dr. P. Michel, EPFL, Sept.-Dec. 2017.
Respectively 10, 4 and 2 hours lecturing (replacement of Prof. Dr. P. Embrechts) for the courses "Probability and Statistics" (bachelor level), "Quantitative Risk Management" (master level) and "An Introduction to the Modelling of Extremes" (master level), ETH Zurich, 2014–2017.
Teaching assistant for the courses "An Introduction to the Modelling of Extremes" (master level) and "Quantitative Risk Management" (master level) taught by Prof. Dr. P. Embrechts, ETH Zurich, 2014–2017.
One-semester (2 hours/week) on "Multivariate extreme value theory and max-stable processes" (master and PhD level), ETH Zurich, Feb.-June 2015.
Practical work for the course "Financial Econometrics" taught by Prof. Dr. J.-M. Zakoian at ENSAE (French National School in Economics and Statistics), 2012-2013 and 2013-2014.
Mentoring
Master thesis of V. Gonzalo, with Prof. R. Zagst and M. Wahl, on the topic "HARA utility maximizationin a Markov-switching financial market", EPFL, 2022.
Semester project (bachelor level) of R. Khani on the topic "Univariate extreme-value theory and applications to snow depth and temperature", EPFL, 2022.
Master thesis of F. Dutto, on the topic "Tail index regression with unobserved heterogeneity for panel data", EPFL, 2021--2022.
Semester project of B. Rashiti, on the topic "Statistical modeling of spatio-temporal rainfall data in Lausanne", EPFL, 2021--2022.
Semester project of S. Legrosdidier (master level), on the topic "Modeling of wind damage using machine learning", EPFL, 2021.
Master thesis of R. Cotsakis, on the topic “Inference for space-time max-stable processes with the Markov property in time”, EPFL, 2020–2021.
Semester project (master level) of R. Cotsakis, on the topic “Literature review of space-time max-stable
processes”, EPFL, 2020.
PhD thesis of J. Koh, with Prof. A. C. Davison (informal), on the topic “Spatiotemporal modelling of
extreme wildfires and severe thunderstorm environments”, EPFL, 2017–2021.
Master thesis of J. Koh, with Prof. Dr. P. Embrechts, on the topic “An extreme-value theory based model for proxies of hail”, ETH Zurich, 2016-2017.
Master thesis of Z. Chen, with Prof. Dr. P. Embrechts and Dr. R. Muraviev, on the topic "An extreme-value theory based hurricane model with applications to Cat bonds", ETH Zurich, 2016.
Master thesis of A. Gabrielli, with Prof. Dr. P. Embrechts, on the topic "Temporal dependence in max-stable processes", ETH Zurich, 2014-2015.
Master thesis of T. Descombes, P. Mabille and O. Wang, with Dr. J. C. Héam, on the topic "Financial networks", ENSAE, 2012-2013.
Refereeing
Annals of Applied Statistics.ASTIN Bulletin.
Dependence Modeling.
Econometrics and Statistics.
Environmental Modeling & Assessment.
European Actuarial Journal.
Extremes.
Journal of Agricultural, Biological, and Environmental Statistics.
Journal of Climate.
Journal of the American Statistical Association.
Journal of the Royal Statistical Society.
Journal of Time Series Analysis.
Scandinavian Journal of Statistics.
Stochastics.
Discussions of papers presented in conferences
30th French Finance Association Conference, May 2013.Liquidity Supply across Multiple Platforms by L. Lescourret and S. Moinas.
6th Financial Risks International Forum, March 2013.
- Acceptability indexes via g-expectations: an application to liquidity risk by E. Rosazza Gianin and C. Sgarra.
- Valuing financial assets with liquidity discount: an implication to Basel III, by R. Chen et al.
Organization of workshops and conferences
Columbia University-ETH Zurich workshop; Extreme Environmental Risks: Statistical Modeling and Insurability, with Prof. Dr. P. Embrechts and G. Shoham.Professional course
Post-doctoral Fellow/Scientist/Lecturer under the supervision of Prof. Dr. P. Embrechts
ETH Zurich (Department of Mathematics, RiskLab)
2014-2017
Post-doctoral Fellow under the supervision of Prof. Dr. J. H. Blanchet and Prof. Dr. R. A. Davis
Columbia University (Department of Statistics)
2016 (one month and a half)
Post-doctoral Fellow/Scientist under the supervision of Prof. Dr. A. C. Davison
EPFL (Chair of Statistics STAT
2017-2019
Education
Publications
Selected publications
| Koch, E. and Robert C. (2022) European Journal of Operational Research |
Stochastic derivative estimation for max-stable random fields |
| Koch, E., Koh, J., Davison, A. C., Lepore, C., and Tippett, M. K. (2020) Journal of Climate 34(4):1259–1272 |
Trends in the extremes of environments associated with severe US thunderstorms |
| Davison, A. C., Koch, E. and Koh, J. (2019) Statistical Science 34(4):584-590 |
Comment: Models Are Approximations! |
| Koch, E. (2019) Risks 7(2):52 |
Spatial risk measures and rate of spatial diversification |
| Koch, E. and Robert C. Y. (2019) Statistics and Probability Letters, 145:43-49 |
Geometric ergodicity for some space-time max-stable Markov chains |
| Koch, E., Dombry, C. and Robert, C. Y. (2019) Stochastic Processes and their Applications, 129(9):3406-3430 |
A central limit theorem for functions of stationary max-stable random fields on R^d |
| Koch, E. (2017) Extremes, 20(3):635–670 |
Spatial risk measures and applications to max-stable processes |
| Embrechts, P., Koch, E. and Robert, C. Y. (2016) Advances in Applied Probability, 48(A):77-97 |
Space-time max-stable models with spectral separability |
Other publications
In peer-reviewed journals
Koch, E. and Naveau, P. (2015). A frailty-contagion model for multi-site hourly precipitation driven by atmospheric covariates. Advances in Water Resources, 78:145–154.
Submitted to or in revision in peer-reviewed journ
Koch, E. (2022). Correlation of powers of Hüsler–Reiss vectors and Brown–Resnick fields, and
application to insured wind losses. arXiv preprint arXiv:2203.00455.
Koh, J., Koch, E. and Davison A. C. (2022). Space-time extremes of severe US thunderstorm environments. arXiv preprint arXiv:2201.05102.
Koch, E. (2019). Extremal dependence and spatial risk measures for insured losses due to extreme winds. arXiv preprint
arXiv:1804.05694.
Other publications ((*) when reviewed)
Koch, E. (2022). Code and data for the journal article “Stochastic derivative estimation for max-stable
random fields”. Zenodo. https://doi.org/10.5281/zenodo.5913865.
(*) Koch, E. (2016). Spatial risk measures and applications to max-stable processes. SCOR Papers, number 36.
Héam, J. C. and Koch, E. (2015). Diversification and endogenous financial networks. arXiv preprint arXiv:1408.4618.
(*) Koch, E. (2015). Outils et modèles pour l'étude de quelques risques spatiaux et en réseaux: application aux extrêmes climatiques et à la contagion en finance. L'actuariel, number 15, p. 34-36.
(*) Koch, E. (2014). Tools and Models for the Study of some Spatial and Network Risks: Application to Climate Extremes and Contagion in Finance. PhD thesis, Claude Bernard Lyon 1 University. Published by SCOR.
(*) Héam, J. C. and Koch, E. (2014). La recherche d'une diversification explique-t-elle l'interconnexion bancaire ? Les Cahiers Louis Bachelier, number 13, p. 10-11.
(*) Koch, E., Ribereau, P. and Robert, C. Y. (2013). Une tempête dans l'économie. Mathématiques de la Planète Terre 2013.
Koch, E. (2011). Etude de faisabilité d'une assurance rendement basée sur indice climatique. Actuarial thesis.
Research
Interests
- Weather extremes- Spatial, weather and climatic risks
- Extreme-value statistics
- Spatio-temporal statistics
- Data science and machine learning
- Actuarial science
- Green finance
Teaching & PhD
Teaching
Mathematics
Courses
Probability and statistics
The course introduces the basic notions of probability and statistical inference, with an emphasis on the main concepts and the most commonly used methods.
Statistical theory
The course aims at developing certain key aspects of the theory of statistics, providing a common general framework for statistical methodology. While the main emphasis will be on the mathematical aspects of statistics, an effort will be made to balance rigor and intuition.
Multivariate statistics
Multivariate statistics refers to data in vector form. The main objective is to uncover the associations between the components of the vectors. The course introduces the major types of statistical methodology for this purpose as well as some applications to quantitative risk management.