Fortune may favor the bold, but in investing, those who gauge their risks smartly will be better off. Experts at Wharton and elsewhere have developed a novel model to identify latent (or hidden) risk factors that are common across investments in the three asset classes of stocks, bonds, and options. That is an advancement over conventional asset pricing research, much of which has focused on risk factors specific to particular asset classes.
The research findings are contained in a paper titled, “Common Risk Factors in the Returns on Stocks, Bonds (and Options), Redux,” authored by Wharton doctoral student Zhongtian Chen, Wharton finance professor Nikolai Roussanov, a professor at the Hong Kong University of Science and Technology Xiaoliang Wang, and Wharton doctoral student Dongchen Zou. It won the 2024 Best Paper prize from Wharton’s Jacobs Levy Equity Management Center for Quantitative Financial Research. The paper’s significance can be understood in the context of factor investing, which has been an important component of quantitative investing for decades.
The findings could help investors refine the ways in which they manage the interplay of risks, volatility, and returns on their investments. The risk factors that the paper tracked are called “latent factors” because they are unobservable. “Latent factors are hidden inside the returns; you don’t see them until you apply a statistical method to extract them,” Roussanov explained. The authors used a bottom-up approach to identify latent factors, where they analyzed monthly returns for the three asset classes from June 2004 to December 2021. That approach was in contrast to the traditional, top-down approach of applying risk factors — such as financial market uncertainty — across asset classes, or using factors that are built from portfolios that are specific to each asset class.
Those traditional factors essentially reflect characteristics of an investment, and can be specific to the type of security — be it stocks, bonds, or options — or the firm issuing the security. The study covered 35 characteristics for stocks, 26 characteristics for corporate bonds, and 19 characteristics for stock options.
The authors’ approach instead relies on common movement among characteristic portfolios across the three asset classes, identifying the components that explain a large amount of variation in all of the asset classes simultaneously.
The study found about a half-dozen firm-level characteristics that are useful predictors in all three asset classes. But those factors get granular with specific characteristics for each type of security, Roussanov noted. For instance, a company may issue short-maturity bonds and long-maturity bonds, each of which comes with different risks. Similarly, a company’s stock options may have different strike prices with varied risk profiles.
“Latent factors are hidden inside the returns; you don’t see them until you apply a statistical method to extract them.”– Nikolai Roussanov
Understanding the Main Findings
The main finding of the research was strong commonality among risk factors across the three asset classes. In particular, the largest factor was very closely related to market-wide financial uncertainty, which is one of the biggest drivers of returns. Investors typically manage risks by diversifying their investments, but there are limits to that exercise, especially when some risk factors find strong representation in all the three asset classes the paper tracked — stocks, bonds, and stock options. “Exploiting these factors jointly across asset classes is a little bit more profitable than just looking at one asset class,” Roussanov said.
The factors tracked by the study do not explain all of the average returns across assets. That is a pointer to some unrealized “alpha,” which is a measure of how much the returns on an investment outperform or underperform a passive benchmark that relies on a set of pre-specified factors beyond a simple index such as the S&P 500. Factor ETFs that track factors related to “value,” “quality,” or “momentum” have become popular recently. The authors’ findings, however, suggest that “momentum,” or the tendency of out- or under-performing securities to out- or under-perform, has little relation to pervasive common factors, and appears to be more like an “alpha,” or a true “anomaly” that is unrelated to systematic risk.
The authors argue that evaluating factors can still be useful for generating average returns, Roussanov said. “They are also useful for constructing optimal portfolios that take advantage of the predictability of returns with various stock or bond characteristics. [That is because] imposing the factor structure reduces the number of correlations that need to be estimated, which is one of the most challenging elements of portfolio construction in practice,” he added.
The latent factor model in the paper demonstrates how investors can better calibrate the diversification of their investments across the three asset classes. While alphas are calculated relative to benchmark indices, the Sharpe ratio captures the risk-adjusted returns of securities.
Roussanov offered a glimpse of the size of the untapped opportunity. Over the roughly 18-year study period, the equity market as a whole generated a Sharpe ratio of 0.5. The factors that the research tracked revealed Sharpe ratios of between 2 and 3, or between four and six times the Sharpe ratio that the equity market delivered. Most of that Sharpe ratio of between 2 and 3 is seen in equity and options, but very little in corporate bonds, he added.
“Much of the alpha that we document in our study period has been arbitraged away over time by quantitative investors, hedge funds, and the like.” – Nikolai Roussanov
So, is there big money lying on the table, waiting for savvy investors? Not really. Quantitative investing has witnessed huge growth in recent times, and it has enabled investors to exploit some of the predictability in returns that factor evaluation makes possible, Roussanov said. “Much of the alpha that we document in our study period has been arbitraged away over time by quantitative investors, hedge funds, and the like,” he added. “In particular, there is very little alpha at the moment in corporate bonds; it’s close to a relatively efficient market. In equity, there is still some alpha, and in options, there is quite a bit.”
Takeaways for Investors
Sophisticated investors like institutions are well positioned to apply learnings from research such as this paper, but there are some pickings for retail investors, too. “For retail investors, the takeaway is that having a well-diversified portfolio is key,” Roussanov said. “I would not worry about corporate bonds and options markets, which are pretty hard for retail investors to access, because much of the benefit in terms of average returns that can be generated is already available in equities.”
Investors may also see a window of opportunity in one surprising aspect of the findings. “There is a fair amount of mispricing that is related to the security level, but it’s very different across the asset classes,” Roussanov said. “There’s more in some asset classes like options, but a lot less in corporate bonds.”
Compared with prior work on factor investing, the key differentiator in the research by Roussanov and his colleagues is its methodology and the large amounts of computing power it needed to process a large data set. “Modern econometric tools allow researchers to extract factors in a very robust way so that it’s not subject to a lot of statistical biases,” he said. “It allows us to analyze the contributions of individual factors and produces better performance in explaining returns than existing factors. They are also precise in extracting common-factor information from asset returns [in] that they allow us to reveal that there is still a substantial unexplained component; once the factors are hedged, you see that there is this large alpha that remains.”
