Forecasting Volatilities in Equity, Bond, and Money Markets: A Market-Based Approach

Abstract

This study examines the forecasting power of the most popular volatility forecasting models in the S&P 500 index market, Eurodollar futures market, and 30-year US T-Bond futures market. A new way to evaluate volatility forecasting models by applying the out-of-sample testing techniques in the context of option pricing is proposed. The approach develops Karolyi's (1993) option pricing error approach empirically. Spurious regressions biases and biases of measurement of volatility forecasts are controlled for. The evidence in this paper supports use of implied volatility as a proxy for market volatility, as it works best in forecasting future volatility. It is also concluded that volatilities in the three markets follow a Stochastic Volatility process, as an AR(1) best fits the implied volatility series in each of the markets. These empirical results are consistent with the predictions of Rational Expectations theory. Directions for further investigation are noted.

Keywords: Market-based, Volatility Models, Forecasting, Option Pricing

JEL Classification: G12, G13, C22