Distinguishing upside potential from downside risk

Abstract

Purpose

The purpose of this paper is to develop new tools to interpret changes in risk neutral probability distributions (RNPDs). It distinguishes between changes attributable to upside potential and those attributable to downside risk, and shows that the distinction is supported empirically.

Design/methodology/approach

This paper estimates pricing kernels and RNPDs from option price data, then studies the expected excess returns on a fixed‐strategy reference portfolio composed of the claims defined by the RNPDs. The portfolio is disaggregated so that realized returns can be expressed as a value‐weighted average of returns to upside (investment) and downside (insurance) sub‐portfolios, respectively. An upside sub‐portfolio can be interpreted as defining payoffs to a call option, a downside sub‐portfolio as payoffs to a short put position.

Findings

Empirical results indicate that the realized excess returns on the reference portfolios are significantly and negatively related to both S&P index growth and volatility (measured by the Chicago Board Options Exchanges (CBOEs) volatility index (VIX)) in the original data, but neither variable is significant in regressions on data first differences. However, in regressions on both the original data and first differences, realized excess returns on the investment sub‐portfolios are significantly and negatively related to both S&P index growth and volatility, whereas the realized excess returns to insurance sub‐portfolios are significantly and positively related only to the VIX. In regressions on both original data and its first differences the ratio of realized insurance excess return to total return is positively and significantly related only to the VIX. Constant terms are significant in about half of all the regressions, suggesting the presence of additional explanatory factors not captured in currently available data.

Originality/value

The paper shows that upside and downside sub‐portfolios have different return distributions in different market regimes, and that while returns to upside claims depend significantly on both S&P index growth and volatility, returns to downside claims depend significantly on just S&P index volatility. Thus realized excess returns to sub‐portfolios convey more nearly precise information about changes in market attitudes than do realized excess returns to entire portfolios. Although concepts of aggregating and disaggregating information have been investigated in the context of annual earnings announcements in other research, they have not previously been applied to realized portfolio returns in the manner used here. If the paper's findings are sustained in further empirical analyses, they can potentially provide information regarding both the Grossman‐Zhou and Holmstrom‐Tirole theories of claim pricing. Overall, because they distinguish between upside potential and downside risk, these methods contribute to more discriminating ways of understanding reference portfolio returns. In contrast, the CAPM measures of return variance do not distinguish between the risks of returns fluctuating on the upside from the risk of returns fluctuating on the downside.

Citation

Neave, E.H., Ross, M.N. and Yang, J. (2009), "Distinguishing upside potential from downside risk", Management Research News, Vol. 32 No. 1, pp. 26-36. https://doi.org/10.1108/01409170910922005