Raman Uppal: "Unsystematic risk should not be dismissed when designing an optimal portfolio: it could drive up to 80% of the performance"
A risk that is unique to a company is called “unsystematic risk”. For several decades, financial portfolio choice models have ignored this risk. Raman Uppal, EDHEC professor, along with his co-authors, has been building an alternative approach where unsystematic risk is considered (1). Here is how and why in a few questions.
What is an optimal portfolio and why is it important?
An optimal stock portfolio is one that has the best possible (expected) return for a given level of risk. This optimal portfolio is the foundation of financial economics. Why? Because it determines the relationship between the risk and return of individual stocks, which is how we determine the price of a stock.
What are the existing theories of asset pricing and portfolio choice? How do they perform in practice?
The first theory to explore the relationship between risk and returns is the Capital Asset Pricing Model (Sharpe, 1964), for which Sharpe received a Nobel prize. According to this theory, the only systematic risk factor that matters is the return on the market portfolio, and unsystematic risk earns no returns. Thus, investors should diversify unsystematic risk (because it earns no returns) and differences in returns on assets should depend only on their sensitivities to market returns.
But the Capital Asset Pricing Model performs poorly in explaining differences in the returns of different stocks. Thus, researchers have considered other risk factors; for example, Fama and French have developed models with three factors, five factors, and six factors. By now, other researchers have examined more than four hundred risk factors. Despite this, existing models cannot explain differences in returns across assets.
Why do you think these models perform poorly in explaining differences in returns across assets?
In our view, all these models have focused their attention on identifying the correct sources of systematic risk that matter for pricing assets. But, just like Sharpe, they have continued to assume that there is zero compensation for bearing unsystematic risk. So, the key feature of our work (1) that distinguishes it from existing work is that we allow for compensation for unsystematic risk. This is a significant departure from the existing way of thinking about the relationship between risk and return.
Why do you believe unsystematic risk should be included in asset-pricing models?
Unsystematic risk, which can be defined as a risk that is unique to a company, is typically dismissed when pricing assets because if one assumes, as Sharpe did, that there is no return for bearing idiosyncratic risk, then indeed it is optimal to diversify away this risk by holding a portfolio with a large number of assets.
But, in the data, investors do not hold a large number of assets in their portfolios. In fact, the median number of stocks in the portfolio of retail/household investors is only three. And even many institutional investors hold fewer than one hundred stocks. Moreover, there are substantial costs and frictions that make it difficult to invest in a large number of stocks, which means that retail and institutional investors do not fully diversify away from unsystematic risk, and hence, demand compensation for it.
So, how do you include unsystematic risk in designing an optimal portfolio?
In contrast to existing models, our hypothesis is that bearing unsystematic (i.e., asset-specific) risk is rewarded. We then validate this assumption with our empirical analysis. If unsystematic risk is compensated with a non-zero return, we show that the optimal portfolio consists of two components. The first component is a portfolio that depends only on systematic risk factors and the return for bearing this risk; this is exactly the portfolio considered in existing models of asset pricing and portfolio choice. The second component is new. It is a portfolio that depends only on unsystematic risk and the return for bearing this risk.
When combined, these two portfolios effectively generate the efficient frontier, which is the portfolio that offers the optimal trade-off between risk and expected return. Our work thus provides a new method for measuring accurately the return for bearing idiosyncratic risk and for exploiting this optimally.
What is the relative importance of the portfolio components based on systematic and unsystematic risk?
Theoretically, we show that as the number of stocks increases, the weights of the unsystematic-risk portfolio increase in importance relative to the weights of the systematic-risk portfolio. Empirically, we find that, even in a portfolio with only 30 assets, the weights of the unsystematic-risk portfolio contribute to about 80% of the overall portfolio weight in each stock on average, while the weights in the systematic-risk portfolio contribute only about 20%.
So, even in a diversified portfolio, unsystematic risk is actually rewarded?
Precisely. In fact, in terms of performance, the unsystematic-risk portfolio dominates the systematic-risk portfolio. That is, most of the performance comes from the unsystematic-risk portfolio, with the systematic-risk portfolio contributing only a small fraction (less than 20%). We confirm this finding in four different datasets.
What does this mean for portfolio choice models?
Thanks to this study, we can conclude that what has generally been viewed as pricing errors is in reality compensation for unsystematic risk and an integral part of the asset pricing model. Our theoretical and empirical work demonstrates that unsystematic risks in individual stocks are actually compensated, so it is important to include in one's portfolio the component that exploits unsystematic risk rather than dismiss it.
(1) What is Missing in Asset-Pricing Factor Models? Massimo Dello Preite, Raman Uppal, Paolo Zaffaroni et Irina Zviadadzea (May 2023). Available at SSRN: https://ssrn.com/abstract=4135146