In this dense and intriguing paper, the authors construct a novel investor risk aversion, or volatility risk premium index (VRP). Essentially, their index is a metric for implied and realized volatilities after adjusting for shocks to economic state variables. The measure allows the authors to link financial markets and the overall economy.

The math behind the construction of the VRP is highly technical, but in brief, the authors start by working under the assumption that they can estimate the VRP for a security from observed high-frequency intraday stock price returns and the prices of listed options. Based on a small-scale Monte Carlo experiment, they determined that a five-minute sample period for security prices in each month should be adequate for their purposes. They then applied their analysis to the S&P 500 and found that the estimate of a constant volatility risk from the basic dataset was statistically significant. They then created an enhanced estimator of a time-varying volatility risk premium by extending the dataset to include a number of macro-finance variables, including realized volatility, Moody’s AAA bond spread, housing starts, P/E ratio, industrial production, producer price index (PPI), and payroll employment.

The authors also used the index to predict stock market returns and find it has twice the predictive power of the best macro-finance variable (P/E ratio) and far outperforms industrial production, non-farm payrolls, and dividend yield (though some macro-finance variables are already incorporated in the volatility premium).

This paper was a finalist for the 2011 Whitebox Prize for the best financial research of 2011. To see the top 10, click here.