My main areas of research are currently fixed income on the one hand, and stress testing on the other. To be more precise, in fixed income I am working on risk premia (and risk factors) for Treasuries and risky bonds. This has both a time-series dimension (when is it a good time to invest?) and a cross-sectional angle (if we ‘had to’ invest today, which assets would give us a most attractive compensation for the risk we bear?).
Of course, the cross-sectional studies are another incarnation of ‘smart-beta’ strategies, this time applied to fixed income. In both cases (time-series and cross-sectional) we strive for a financial understanding of the factors, and we are very mindful of the twin dangers of data mining and over-engineered proxies. To be more specific, I am working (with my PhD students) on liquidity and liquidity risk premia and on a variation on the theme of Downside CAPM. As for stress testing, I am working on the application of Bayesian-net technologies to this domain. I think it is fair to say that the ‘conceptual case for stress testing’ has been won – we all agree that, for macro- and micro-prudential purposes stress testing can be very useful. What stands in the way of more widespread application, I think, is a number of ‘engineering’ problems. Clearly, speaking of engineering solutions only makes sense in the context of a specific technology, and the technology that I prefer is the Bayesian-net one. However, the solutions I am developing can be applied to most stress testing programmes. This should be of interest to portfolio managers, regulators, and risk managers for buy- and sell-side firms.
In the course I have tried to share with my students the excitement in term structure modelling of the last decade or so. After the work by Cochrane and Piazzesi (2004) on the one hand, and on fixed-income smart-beta on the other, a field that until recently had been rather staid and ‘boring’ has become red hot with ideas and innovations. I also shared with these PhDs students the latest findings, and explained why they matter: for the conduct of monetary policy, for instance, for investing, and to understand price formation.
Teaching the EDEHC PhD students has been a real pleasure because they come with a wealth of professional experience that makes them a particularly deep and perceptive cohort of students. They immediately understand why a result is important, and they can see the connection between theory, measurement and financial reality straight away. Working with them is truly a give-and-take process.
I would recommend that they highly prize (and not hide) their professional experience. It is a great asset. It is easy to brush up on rusty maths, but experience cannot be improvised.
My strongest recommendation is to make sure that the results they present really improve our understanding of a topic of finance. It does not matter if the contribution is small, but it should be ‘honest’. By this I mean that analytical pyrotechnics to show the cleverness of the author or over-stylised cases of the “let’s assume a horse is a sphere’’ type may be fun games to play, but leave no lasting mark. A simple insight, a clear result, a useful technical suggestion all make a positive impact, and improve our understanding (and our practice) of finance. Also, the style should be ‘professional’, but not stuffy. When the reasoning becomes complex, don’t be shy to use examples (Many times a referee has thanked me for that). And, please use simple words whenever you can. The goal is to be understood, not to baffle.
Not surprisingly, I talked about smart-beta in fixed income, and about stress testing with Bayesian nets.