Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation
Read the paperopens doi.org in a new tab
What they found
Engle introduced the ARCH model, the first formal way to describe a process whose variance changes over time in a predictable way: today's variance depends on the size of recent shocks. The application was U.K. inflation, but the framework was quickly adopted for financial returns, where the observation that large moves cluster together had been noted since Mandelbrot. Engle received the 2003 Nobel Prize for this work, which underlies nearly all modern volatility modeling and risk management.
What you can use
- Volatility is forecastable even when returns are not: big moves are followed by big moves.
- Any risk model that assumes constant volatility will understate risk after a shock and overstate it in calm periods.
- Position sizing should adapt to current conditional volatility, which is exactly what ARCH-type models estimate.
Caveats
Econometric theory paper; the original application is macroeconomic. Practitioners use the GARCH extension in practice.
Tags: volatility, arch, econometrics, foundations
Summaries are our own reading of the paper, not the authors' words. Educational only, not advice. Discuss it in Book Club.