I dabbed a little into land value surfaces estimations using ML/neural nets. The implementation was surprisingly straightforward and results don’t look completely insane. For instance, the surfaces are not as smooth as kriging (good!) but I have no idea how reliable the ML estimates really are. More work is needed, for sure, but it is exciting to already see the spatial and the temporal dynamics.
We add to the emerging literature on machine learning empirical asset pricing by analyzing a comprehensive set of return prediction factors on REITs. We show that machine learning models are superior to traditional ordinary least square models and we find that REIT investors experience significant economic gains when using machine learning forecasts. Comparing to the stock market, we show that REITs are more predictable than stocks, and that the higher predictability is stable across time and across industry types.
It was such a joy to welcome back our MPhil alumni and former colleagues for the 2024 MPhil reunion! I truly enjoyed the evening, hearing all about your impressive careers—but, more importantly, seeing old friendships glowing strong and new bonds forming.Sometimes, the beauty of working in education gets lost in the daily grind. Being there with you wonderful people, I felt grateful, sentimental, and curious about your next steps. Thanks to all for the reunion, and I hope to see you again soon!
As I sat in the dentist’s chair, waiting for the local anesthetic to kick in, I finally asked the dentist something I had wondered about a few times before: “How do you keep up with new techniques and stay on top of your field?” I inquired, half-expecting a mention of professional courses or medical journals. His response was immediate and enthusiastic: “YouTube!” And that was reassuring.
The Cambridge Endowment for Research in Finance (CERF) has published a blog post on Kahshin Leow’s and my research on predictable REIT returns: https://www.cerf.cam.ac.uk/blog/blog-2024
