CFM Talks To: Professor Johannes Muhle-Karbe
We had the pleasure of welcoming Professor Johannes Muhle-Karbe, the new director of the CFM-Imperial Institute of Quantitative Finance to our Paris office to talk about the future ambitions of the Institute, his research and his view on the state of quantitative finance. Johannes joins Imperial College London from Carnegie Mellon University where he taught courses on Stochastic Control and its Application in Finance, Market Microstructure and Advanced Derivative Models amongst others. During his time in Pittsburgh, he was also a member of the steering committee for the Center for Computational Finance. Johannes took up his current role as Director of the Institute in January after the departure of Professor Rama Cont.
CFM: Thank you for coming to speak to us about the Institute and the work you will be doing there. Can you briefly tell us a bit about your background, and what led you to the Institute?
JMK: My journey started during my undergraduate degree in mathematics at TU Munich, where I became fascinated with probability and stochastic processes. The professor who taught these courses later became my PhD advisor and also introduced me to Mathematical Finance. And I got hooked. I started doing stochastic optimisation and research on transaction costs during my PhD and completed my degree in 2009, also at TU Munich. After a one year post-doc in Vienna, I moved to Switzerland to take up an assistant professorship at ETH Zurich. I stayed for five years and then moved to the US, where I first joined the University of Michigan and later Carnegie Mellon.
Wanting to move back to Europe, Imperial offered me an incredible opportunity. It has a fantastic math department with a Fields medallist in stochastic analysis, for example. Along with the connection to CFM through the institute and the finance industry in London, it was a perfect fit.
CFM: The Institute is part of the Mathematics department of Imperial?
JMK: Yes. The mathematics department at Imperial has a number of different sections: pure and applied mathematics, statistics, and mathematical finance. The Institute is a partnership between Imperial College’s Mathematical Finance Group and CFM. The mathematical finance section is one of the largest research groups in this field worldwide, and also has close connections to industry through the Master in Quantitative Finance we offer. As a hub for quantitative research in London, it was a natural partner for CFM.
CFM: What are the key objectives and mandates of the Institute?
JMK: The Institute is an initiative to promote research in quantitative finance and modelling. Particular emphasis lies on interdisciplinary projects that have mathematical aspects, but also connections to statistics, econophysics, finance, and real-life problems from industry.
CFM: How does the institute specifically promote these research initiatives? What ambitions do you have for the institute?
JMK: One of the mainstays of the Institute has been the biannual ‘Market Microstructure Workshop’, which alternates between Paris and London. The next edition will take place in London on December 12-13 later this year. It is perhaps the only, truly interdisciplinary conference where academics with backgrounds in math, physics, finance, and other disciplines get together with practitioners from the finance and asset management industry.
Another key outreach project is the monthly ‘CFM-Imperial Seminar’, where high profile speakers are invited to give lectures on a wide range of contemporary topics. The talks are deliberately aimed at a diverse audience, and are scheduled in the early evening to also allow practitioners to attend. The goal is to have big picture, broad-audience talks to ignite discussion about the future challenges in Quantitative Finance. The first two seminars for the fall have been scheduled beginning in October. Details can be found on the Institute’s website.
CFM: So it seems fair to say that the Institute places a high premium on pushing interdisciplinary studies and research?
JMK: Yes. But it always is a challenge to engage with people who don’t have similar training. Whilst my background is in mathematics, I gradually became more interested in economics and finance. To get up to speed with research in these communities, I had to spend a lot of time talking to researchers with this kind of training and reading their papers. Interdisciplinary scientific work is challenging, because it is like learning a new language: the problems are different, the style is different, and the traditions are different. But, if you can manage to establish a working relationship with a diverse set of researchers, it can be extremely fruitful.
CFM: Do you have any ideas of how to leverage the Institute to address the difficulties of promoting interdisciplinary collaboration?
JMK: One of the things I’m currently pushing for is to establish joint activities with the Imperial College Business School. Imperial has a fantastic finance group, with research projects similar to the ones pursued within the institute. Market microstructure is one prominent example. Historically, these two groups have operated largely on their own, but the institute is a perfect vehicle to foster collaboration between them. In May 2020, for instance, we are jointly organising a conference on ‘Frictions in Finance’, where junior researchers in finance and mathematics will have the chance to learn about topical research questions and methods from some of the leading figures in both fields.
CFM: Any particular research area that is ripe for such a tie-up?
JMK: There are researchers in both groups, the business school and mathematical finance that work on analysing and understanding the flaws and frictions observed in financial markets, such as asymmetric information, transaction costs etc. The style of research is a bit different, and the strengths each brings to the table are different. However, once the two sides get to know each other, we will hopefully be able to realise a lot of synergies. I had a very positive experience in Pittsburgh, where the Masters programme (Carnegie Mellon’s MSc in Computational Finance) has had great success as a collaborative effort between four departments: mathematics, statistics, computer science and the business school. The idea right from the beginning was that learning about finance requires a broad curriculum, often campaigning competing points of view. Of course, another benefit is that the master programme forms a connection between the different departments. In London, the CFM-Imperial Institute should ideally become such a point where faculty can naturally rally around.
CFM: Business schools’ teaching of 50 years ago is vastly different than what the representative student will encounter today, with more focus on quantitative techniques, interdisciplinary teaching and the like. What do you think was the genesis of this new paradigm?
JMK: Yes, I agree, and business schools have accordingly become more technical. When modern business schools and MBA programmes first took off, their curricula were not very technical. MIT, Stanford, and also Carnegie Mellon then played big roles in hiring mathematically well-trained faculty and students, pushing academic teaching of finance into a much more quantitative direction. Now again, with the rise of Artificial Intelligence, Machine Learning, and Big Data, graduates are under increasing pressure to acquire these skills. I think the skills from traditional MBA training are still crucial, but demand from industry is making proficiency in advanced programming and data manipulation indispensable as well. At Imperial, we just restructured our master programme in Quantitative Finance to make sure that our students remain up to date both with respect to programming and statistics. It is clear that such cross-disciplinary training is vital for a successful career in the finance industry.
CFM: And talking about these new buzzwords of finance, Machine Learning and Big Data. Barely a day goes by without an article, or new research being published on these (or related fields) and how it is set to dramatically alter the world of finance. Is there a similarly robust focus within academia, as there seems to be in the industry?
JMK: I would argue there is generally more inertia in academia, but there is now a large push towards machine learning and AI as well. Many academics, myself included, have started to use and experiment with some of the new numerical methods. However, in my own research, I feel the jury is still out on which research questions you may want to answer with these new tools. It is clear to me that we academics need to learn about these new developments to teach them to our students. Calibration, optimisation problems along with data analysis etc. - there is no doubt in my mind that these will continue to grow in importance in the industry. Regarding academic research, the path ahead is less clear. What is the general principle I could uncover, what is the mechanism I could understand with these methods? What problem should we try to solve with these new tools? In this context, machine learning is like a big, brand new hammer. But what is the right nail we should try to hit with it?
There is exciting ongoing work in computer science, complexity theory, and mathematics on when and why these new tools work. But more from the finance side of things, what are the economic questions we can answer with all this computational power? I think this is a key question and the answer is not clear to me yet.
CFM: With this advent of Big Data came an industry awash with ‘alternative’ data providers, offering new data on consumer spending patterns, satellite imagery showing levels of oil and gas storage etc. These are just a few among a near countless assortment, all holding the promise of profiting from advanced knowledge when using these tools. Do you believe that this data may hold some new information?
JMK: This is something that I believe holds promise. I think if you want to make short term predictions using this data, by employing non-linear learning algorithms, one could very well be able to make more sophisticated forecasts. I think this is a well-defined problem, and the industry relevance is clear. However, as a researcher, you would also like to uncover the mechanism by which, for example, consumer behaviour fluctuates. Or, as a policy maker, you would like to understand how consumer spending may be stimulated. Bringing tools from data science to bear on questions of this kind is more difficult, but an exciting challenge with potentially high payoff.
CFM: Crowding, especially in alternative risk premia, is a topic that makes many investors nervous. With many practitioners starting to use similar tools and techniques, do you think this will ultimately lead to a crowded space?
JMK: I have done quite a bit of work on game-theoretic models where a number of strategic agents trade in the same market, accessing the same trading opportunities and the same pool of liquidity. This naturally leads to crowding. However, to bring such models closer to practice, a key feature will be to model endogenous information acquisition, such as investment in new data sources or trading algorithms.
CFM: Do you think the finance industry has reached a point where asset managers who fail to adopt this slew of new tools are bound to become irrelevant in coming years?
JMK: I think technical know-how is now clearly more important than ever. The competition from the tech industry, in terms of talent attraction and breakthrough research, is contributing to this trend. One should however not discount the ‘old school’ skillset which will remain important. So, while the momentum of change is undisputed, I don’t expect things will change as quickly as many claim, also because many of the key decision makers and processes are unlikely to change overnight. Nor will non-technical skills suddenly become unimportant.
CFM: You mentioned some inertia with regards to academia. Do you observe this from policy makers too?
JMK: Monetary policymaking, for instance, is still very much dominated by economists, with mathematicians, physicists, computer scientists and others very much underrepresented. Accordingly, remaining up to date regarding new tools and theoretical discoveries will be a key challenge for policy makers and regulators, too.
Thus, another goal of the Market Microstructure Conference in December will be to explore how artificial intelligence is shaping the industry, how its consequences should be measured and interpreted, and how it should be regulated. We hope to bring people from technology companies, regulators, and from the finance industry together to discuss this.