I was not formally taught by David, nor wrote any articles with him. But, in his gentle way, like so many, he changed me as a person and as a scholar—all for the much better.
When he was the head of Nuffield College, Oxford, with his colleague Sir David F. Hendry, he asked Lord David Sainsbury (through his charity, the Gatsby Charitable Foundation) to fund a research fellowship to work independently in econometrics. I knew neither of the Davids personally, but was in awe of both—I still am. I had been an assistant professor at the London School of Economics and Political Science (LSE) since 1988, doing my PhD from 1987 to 1990, and felt I needed some time doing full-time research. I left the LSE in 1991 for Oxford, on leave to take up the research fellowship, and never returned.
I would talk with David over lunch or afternoon tea most days we were not traveling. David worked very long hours in the office and at home, but was very social. He always gave the impression he had time for each person, despite being the warden of a college, editor of Biometrika, and typically teaching courses in the Oxford Statistics Department and writing 10 papers a year! He would come to econometrics seminars and I learned from listening to his questions and asking him what he thought of the papers, and more broadly and usefully, his taste for problems. Many of his research breakthroughs came from his background in stochastic processes, which connected to some of my own research interests. Often conversations ended up with a discussion of wool fibers from his time at the Wool Industries Research Association.
During my postdoc David was working with the late great Danish mathematical statistician Ole Eiler Barndorff-Nielsen on the second volume of their books on asymptotics and statistical inference. I asked David if I could read the draft book. He was kind enough to let me and I worked through it, carefully (I thought I found an error, went to see David, he said “that’s interesting, I think I proved that three different ways, but let us have a look…” I leave it to the reader to guess who was right). This worked well for the first four chapters, then there was a change and it was mighty hard and very rewarding work. I later learned David mostly wrote the first four chapters and Ole mainly worked on the last pieces.
In 1993 David was kind enough to support my appointment to a permanent position at Nuffield College, Oxford, as an Official Fellow in Economics. This gave me space to mostly focus on research for the next 20 years. David organized a workshop at Nuffield College with Ole and they invited me to give a series of lectures on stochastic volatility.
During the workshop Ole got very interested in the way I thought about stochastic volatility. I later understood it connected to his earlier deep probabilistic work on mixtures and infinite divisibility. After the workshop Ole and I started working together on problems in financial econometrics. Ole was so inspiring and extraordinarily intellectually brave, diving into original ways of thinking about problems, knowing his clarity of thought and intellectual power would deliver something interesting. He taught me a totally different way of being a scholar. I got to work with him because of David. In all, Ole and I wrote 30 papers together and I think it includes some of my best work so far.
David’s leadership strategy seems simple to me. Provide as much intellectual space and protection as possible for young scholars, add encouragement, urge people to finish things, inspire by personal example (work hard, be modest, focus on the core ideas, formalizing early can be counterproductive, do work that can broadly impact applied research) and provide a stream of interesting visitors. I vividly recall his protection of the young and new. I had a friend who was a postdoc working with a very grand professor who thought she should work on his problems 24/7. My friend viewed her duties to him as her first 40 hours of the week, after that she could work on her own material. Much steam was generated. David, as warden, sided with my friend. The grand professor left the college. In my view, David was right. He was nearly always right. I tried, the best I could, to use these lessons when I ran, no doubt imperfectly, the Oxford-Man Institute and later when I was chair of Harvard’s Department of Statistics.
Recently I have been teaching an Introduction to Statistics Inference to hundreds of first- and second-year Harvard undergraduates. David’s ideas appear through logistic regression, noninterference in causal studies, and the behavior of maximum likelihood estimation (MLE) for quasi-likelihoods.
Neil Shephard has no financial or non-financial disclosures to share for this article.
©2023 Neil Shephard. This article is licensed under a Creative Commons Attribution (CC BY 4.0) International license, except where otherwise indicated with respect to particular material included in the article.