Torben G. Andersen

Overview

Torben G. Andersen is a financial econometrician at the Kellogg School of Management, Northwestern University. He is best known for co-developing the realized variance/covariance framework using high-frequency intraday data, and for establishing that returns standardized by realized volatility are approximately normally distributed.

Key Contributions

• Answering the Skeptics (with Bollerslev 1998): Resolved the paradox that GARCH models with significant parameters yield low $R^2$ in forecast evaluations. Showed analytically that low $R^2$ is a mathematical implication of correctly specified GARCH — not misspecification — because squared daily returns are a very noisy proxy for latent variance. Introduced realized volatility (cumulative intraday squared returns) as the proper evaluation criterion: at 5-minute frequency, $R^2$ rises from $\approx 0.05$ to $\approx 0.48$, close to the theoretical maximum derived from the Nelson (1990) diffusion limit. See GARCH and BEKK-GARCH.
• Realized volatility (with Bollerslev, Diebold, Labys 2001–2003): Showed that realized variance from 30-minute intraday returns is a near-unbiased measure of integrated volatility; established log-normality of realized volatility and near-normality of RV-standardized returns.
• Andersen-Bollerslev-Christoffersen-Diebold (2004): Survey of practical volatility and correlation modeling for risk management; critique of HS-VaR and RiskMetrics; promotion of DCC, realized variance, and filtered historical simulation.

Related

• Tim Bollerslev
• Peter F. Christoffersen
• Francis X. Diebold
• Paul Labys
• Realized Volatility
• Quadratic Variation
• Andersen-Bollerslev-Diebold-Labys (2003)
• GARCH and BEKK-GARCH
• Value at Risk