Investment Analysis of Autocallable Contingent Income Securities

Abstract

Autocallable contingent income securities (autocalls) have payouts contingent on the performance of an underlying asset and give investors an opportunity to earn high yields in a low-interest environment. The authors collected data on US-issued autocalls and modeled a typical autocall under various assumptions, finding that they are issued on underlying assets that display high volatility, high prices, and negative skewness. Incorporating stochastic volatility into the model explains some of the overpricing routinely reported in prior studies.

Autocallable contingent income securities, or autocalls, are a relatively new type of structured finance security whose payout is contingent on the performance of an underlying asset and that gives investors an opportunity to earn high yields in a low-interest environment. These complex products, which have many payoff features to consider, are often targeted primarily at retail investors, which, as a result, has provided challenges for both investors and regulators. We collected data on autocalls issued in the United States and described their contractual properties and the properties of their underlying assets at issuance. We found that the autocalls are generally issued on underlying assets that display high volatility, high prices, and negative skewness. Following our analysis of autocall characteristics at issuance, we modeled a typical autocall under different assumptions about the price of the underlying asset, analyzed the rationale behind the characteristics of the underlying asset at issuance, and discussed the valuation of autocalls using various models. We conclude that the traditional use of the geometric Brownian motion model is inappropriate because of several factors, including (1) our empirical findings regarding underlying assets’ price characteristics at issuance that suggest underwriters do not choose underlying assets at random, (2) the large body of evidence of stochastic volatility showing differences in short- and long-term volatility, and (3) the vast evidence suggesting reversals in stock prices. Although the literature has consistently found that structured products are overpriced, we found that incorporating stochastic volatility into the pricing model eliminates some of the overpricing routinely reported in prior studies.

We thank Kyle Darres for his excellent research assistance, Rachael Christensen and Scott McEwan of Morgan Stanley for providing data, and Harvey Boshart, CFA, of JPMorgan Chase for his generous assistance and review of the project. We are particularly grateful to Steve Ross and the Squire Ridge Company, LLC, for their financial assistance. In addition, Rui Albuquerque acknowledges the Portuguese Science Foundation for financial support under the project PTDC/EGE-GES/120282/2010, and research conducted by Raquel Gaspar was partially supported by the Portuguese Science Foundation under the SANAF project UTA_CMU/MAT/0006/2009.