GEOVOL: A NEW STATISTICAL MODEL FOR GEOPOLITICAL RISK

New York University, New York NY

Investigators

Abstract

This proposal develops a new statistical measure of geopolitical risk called GEOVOL, which will have broad implications for understanding, measuring, and hedging geopolitical risk. GEOVOL will involve multiple factors and data sets, a deeper set of measures of geopolitical risks can be presented. By examining shocks that move all markets, this measure focuses on material news. Society has become used to news stories that appear to have massive implications, but do not move the markets. The spread of the coronavirus provided an extraordinary shock to virtually all financial markets in 2020. Similar to Brexit election and 9-11 Bombings, these can be interpreted as truly geopolitical events which were essentially unpredictable. This research will help to understand the distribution and impact of such events. In addition to developing the statistical model, this research will also collect and store data on geopolitical risk. By posting daily estimates of GEOVOL on the V-LAB web site, investors world-wide will be able to see the extent of this geopolitical risk. The web site will also incorporate several popular measures that have been developed by other research teams. This presentation will help firms and individuals everywhere to make better decisions. The GEOVOL measure is based on the insight that geopolitical risk is an innovation to the idiosyncratic volatility of financial assets in all countries, sectors, factors and asset classes. The proposal utilizes a statistical model, developed by Susana Martins and Robert Engle, extending a conventional framework: a set of risk factors generate contemporaneous correlation among financial assets. Volatilities of these assets net of factors are predictable and the innovation to these volatilities are the squared standardized residuals. These standardized residuals will be cross-sectionally uncorrelated but their squares can be and turn out to be correlated. A day with a geopolitical shock will be a day when most of the squared standardized residuals are above average. The common factor GEOVOL and factor loading can be estimated consistently as the number of assets and time series become large. The proposal requests support to 1) complete a paper documenting the analysis of the global equity market data, 2) take this model and put it into V-LAB where estimates are updated daily and published on the internet, 3) extend the model to adjust for different time zones, 4) apply the model to commodities, currencies and cross asset class data, 5) examine the portfolio implications by back-testing the risk reductions that come from the theory and 6) extend the theory to include multi- factor models. This model is a major advance in understanding multivariate volatility models and how idiosyncratic volatilities of different assets and countries become correlated. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.