Welcome to module two, which has a focus on factor investing. Let's start to think about why factor investing is important and why it matters. Well, securities earn their risk premium through the exposure to a small number of underlying risk factors. When you think about it this way, what matters when you invest in stocks, for example, is not so much the stocks that you're holding in your portfolio. This is actually the implicit underlying factor exposure that you're holding when you're holding this particular portfolio of stocks. One way to think about it is that factors are to asset exactly what nutrients are to food. At the end of the day, if you're looking, for example, at this very good-looking cake, what matters more is not necessarily the exact ingredients that we see or the food labels that we can think of when looking at the cake. What matters more is the underlying nutrients, right? That will prove to be important when we think about the benefits of factor investing, to keep in mind that what eventually matters is having the right factor exposure. So eating right is actually absorbing the right nutrients, and investing right means getting the right factor exposure. Now one important distinction exist between macroeconomic factors and microeconomic factors. Macro-factors are the one that can explain time series, and differences, and cross-sectional differences in risk and returns across asset classes. So we want to think in terms of macroeconomic indicators, like for example GDP growth, real rates, or inflation, for example. We do understand that these macroeconomic factors have an impact on all asset classes. Now, conversely, when zooming in within a given asset class, well, let's say let's focus on equities, for example, then within this equity asset class, we have a number of meaningful macro-factors, which are actually attributes, which allow us to understand the differences in risk and return. In the commonly used attributes value and size, from the seminal research from Fama and French, in the '90s, and then we can think about momentum, past winners verses past losers. We can think about Low vol as well, versus high vol, and other such meaningful attributes. We know now that these attributes actually explain, to a fair degree, the differences in risk and return within equity markets. Now, what are the benefits of factor investing? Well, one key benefit of factor investing is to try and obtain a better diversification. It is often the case that the seemingly well-diversified portfolio at the asset level eventually ends up loading on a very limited set of factors. So in other words, even though you think your well-diversified when you're looking at your portfolio, because the factor on loading is actually converge towards this unique factor in master asset classes that you're holding, you actually end up holding an extremely concentrated portfolio. So thinking about investing through the factor lens will help you better understand the kind of portfolio exposure or the kind of diversification benefits you can enjoy. Now, it's not only useful for risk analysis, it's also extremely useful for performance analysis. So if we think in particular about the current situation with extremely low interest rates and a pretty low bond risk premium, and unfortunately, we also have reasons to believe that the equity risk premium is also not extremely high. Well then, in the ability for, let's say, an unfunded pension plan, to try and identify rewarded risk factors across and within asset classes on top of the traditional bond and equity risk premium. Well, that will prove helpful. Now, last but not least, there are a number of asset classes which are actually end up loading on a high number of risk factors that are of a different types. So there are a lot of hybrid securities. If you think, for example, about a high-yield bond, well, it tends to behave in a bond-like way during normal periods. But then, impaired of crashes, the underlying equity exposure is going to start increasing. So it's only if we have a proper understanding of this multi-factor perspective that we can better understand what these hybrid securities are. By the way, all the alternative asset categories that pension funds are increasingly holding now, are actually only explainable if we look at those multiple factors. John, do you want to maybe talk about these alternative asset classes? Sure. Over the past 70 years, there's been a massive change in the asset allocation for institutional investors. Back in the early days in the 1960's, 1970's, most institutional investors had stocks and bonds, and maybe some real estate. Those three asset categories were the major components of investment approaches by institutional investors. Over the past period, however, if you look at university endowments, you see a dramatic shift towards factor, towards alternative investments. In particular, you see on this chart that the top 50 universities have over 51 percent of their money, of their capital invested in alternatives. This is a dramatic shift, as I said, by university endowments. In many ways, these universities have been the leading lights in terms of asset allocation going forward. If you see them investing in a particular way, you see many other institutional investors coming along to that same idea. Particularly you see here, the largest endowments have 57 percent of their money in alternatives. Princeton itself, Princeton University endowment, has almost 70 percent in alternatives. So this gives us motivation to think about what are the underlying risk factors underneath those asset categories? Now the approach we use here is going to be a combination of traditional statistics and machine learning. In particular, we're going to use regression techniques, which are commonly used for identifying factors, in fact, the most commonly used approach, and we're going to add on the idea of regularization terms or penalty terms on top of it. Also the idea of taking out subsets of the data and trying to analyze the out-of-sample analysis that we can. In particular, we're going to use these machine learning ideas in conjunction with traditional regression. This is going to be a hybrid approach, taking traditional statistics, and applying machine learning to enhance its capabilities. We will see other applications of this technology as we move along through the other applications and modules in this online course.
