Asian Review of Financial Research Vol.29 No.1 pp.77-112
The Investment Benefits of Structured Products : Auto-Callable Equity Linked Securities
Key Words : Auto-Callable Equity Linked Security,Portfolio Selection,Expected Utility Theory,Cumulative Prospect Theory,Safety First Theory
Structured derivatives markets, including equity linked securities (ELS), derivatives linked securities, structured notes and credit linked notes have grown dramatically since the mid-2000s in Korea, but little attention has been paid to how much these securities contribute to the improvement of investors' performance from a portfolio perspective. This study fills the gap by examining the optimal portfolio choice of investors who are allowed to invest not only in stocks and bonds, but also in ELS (especially auto-callable ELS). As ELS are generally regarded as alternative assets to enhance portfolio performance, their economic benefits should be considered from a portfolio perspective rather than on a stand-alone basis. In addition, even though the majority of ELS in Korea include an auto-callable feature, there is little study for this specific type of product. This study is the first step in understanding the investment benefits of ELS. In our analysis, we first estimate the return distributions of the KOSPI200 index, a risk-free asset, and a typical auto-callable security linked to the KOSPI200 index. Taking the complexity of the payoff structure into consideration, it is very difficult to theoretically determine the return distributions of the auto-callable ELS. Unlike stocks and bonds, even from an empirical point of view, any standard statistical method to derive the empirical distributions from the observed historical returns cannot be applied to the auto-callable ELS as we do not have a large enough sample (i.e., independent or non-overlapping return data) to obtain a reliable estimate. For example, even if we assume that all of the auto-callable ELS issued during the past 10 years were exercised early on the first possible exercise date, the maximum number of independent returns we can observe is only about 20, and thus we cannot estimate the return distributions in a valid manner. To reconcile this problem, we estimate the GJR-GARCH (1,1) model from the observed returns of the KOSPI200 index from 2003 through 2015, and then derive the empirical return distributions of the ELS via Monte Carlo simulations using the estimated GARCH model. Second, we use three portfolio selection models to derive investors' optimal portfolio choice given access to the ELS market: (1) the conventional expected utility theory, (2) the prospect theory of Kahneman and Tversky (1979), and (3) the safety first theory of Telser (1956), which is the cornerstone of the behavioral portfolio theory with mental accounts proposed by Das, Markowitz, Scheid, and Statman (2010). Our main empirical findings are as follows. First, auto-callable ELS are shown to be unnecessary for the construction of the optimal portfolio for all investors trying to maximize their expected utility, regardless of their degree of risk aversion. Second, the auto-callable ELS do not improve performance for the majority of loss-averse investors. That is, the auto-callable ELS are regarded as a redundant asset according to both expected utility theory and prospect theory. Third, we find that auto-callable ELS are valuable assets that play a key role in improving the portfolio performance of the majority of investors who make investment decisions based on the safety first theory. This suggests that auto-callable ELS can be a very effective investment tool for investors who try to maximize the expected returns of a portfolio with a restricted probability of failing to reach a pre-specified threshold return. The difference in our empirical results depending on the portfolio selection models arises fundamentally from the structural characteristics of the auto-callable ELS. Their risk and return profile indicates that losses occur infrequently, but when they do, the expected losses can be considerably large. Similarly, although gains occur frequently, they tend to be very marginal. These characteristics of auto-callable ELS are very similar to those of selling deep out-of-the-money (OTM) put options. In this sense, for investors assumed by safety first theory, who measure risk by the frequency rather than the amount of expected losses, auto-callable ELS can be effective in enhancing the investment opportunity set. However, for investors with expected utility theory or prospect theory preferences, the relative advantages of auto-callable ELS over common stocks and bonds are weakened as the portfolio risk is generally recognized and measured by the expected losses rather than the frequency of losses. Finally, our robustness test results indicate that the findings remain valid when we consider other types of auto-callable ELS, issuing costs, and the effect of the global financial crisis.