Simon Donker van Heel is a final-year PhD candidate in time series econometrics at the Econometric Institute and the Tinbergen Institute. He develops methods to track quantities that change over time and are not observed directly, such as underlying inflation, the volatility of a share price and the trend in temperature.
Job market paper
Filtering without recursion and some of its uses in financial economics (with Neil Shephard)
The best-known time series filters, such as the exponentially weighted moving average, GARCH and the Kalman filter, are fast to compute, but a single outlier can distort their estimates long after it occurred. Robust filters exist. Many of them, however, have to go through all past observations again at every update, which is not computationally feasible for a long time series. This paper replaces that full pass by a random sample from the past, in which recent observations are more likely to be drawn than older ones. Applied to the prices of the 1.5 million trades of one stock on a single day, the resulting robust filter removes so much noise that volatility can be estimated from ten times as many price changes as a linear filter allows.
Link to paper (pdf)
CV
Download CV (pdf)
References
- Prof. Dick van Dijk, Erasmus University Rotterdam
- Dr. Rutger-Jan Lange, Erasmus University Rotterdam
- Prof. Neil Shephard, Harvard University
- Prof. Siem Jan Koopman, Vrije Universiteit Amsterdam
Contact
Email: donkervanheel@ese.eur.nl
Personal website: Simon Donker van Heel