Seeking Alpha
Macro, economy, long only
Profile| Send Message|
( followers)  

Risk never gives investors a break. It's constantly baiting us, dispensing false signals and generally throwing landmines on the path that appears as a smooth trek to easy gains.

Consider that the past 20 years have been particularly fruitful in financial economics for identifying sources of partial predictability in certain data series. A few examples include dividend yield and other fundamental valuation measures that have shown robust results for estimating the equity premiums for medium- to long-term horizons. The shape of the yield curve and the spread between interest rates on corporate and Treasury bonds have also shown encouraging results as predictors.

In fact, there are a number of factors that are worth monitoring on a regular basis for estimating the price of risk in the major asset classes. No surprise, then, that watching dividend yield, the term structure of interest rates and a host of other metrics, and inferring risk premia and asset allocation from the analysis informs much of our work in The Beta Investment Report. The danger is thinking that the investment challenge has been solved.

It hasn't, nor can it be, lest the risk premia that investors chase is arbitraged into oblivion. But fear not: nothing approaching that future is possible in this universe. For every clue identified in the financial literature as productive for estimating risk and return, another intrepid researcher presents the case for staying cautious if not skeptical on reading too much into the previous studies. That's the nature of risk, and it forever stalks the best laid plans of mice, men and optimistic investors everywhere.

An recent example comes by way of a study that's forthcoming in the Journal of Quantitative Analysis (a working copy is available from the Federal Reserve). After studying 20,000 data points from 40 international markets, the paper advises that the partial predictability of the dividend yield that's been identified in the U.S. equity market fades when applied in foreign bourses. The "empirical results" are a bit more encouraging for using interest rates and yield curves as tools for peering into the future, although the limited power of these factors seems to work best in developed markets.

The study's message raises fresh questions about the reliability of dividend yield, interest rates and yield curves for estimating return and risk. Previous research, starting with a number of studies in the late-1980s, suggested that studying such factors was useful. Should we now jettison 20 years of near consensus in financial economics and look elsewhere for discounting the future?

Actually, that's the wrong way to look at designing and developing investment strategies. There's always someone throwing cold water on your assumptions, or offering a new idea that promises new and improved results.

A better approach is to recognize that there are no silver bullets, even though we're all susceptible to thinking otherwise. To the extent that we can develop context and perspective about forecasting risk premia, the insight is likely to come from multiple sources. What's more, there's no assurance that the list won't evolve through time.

Any one factor will wax and wane as a window into the future. What looks more like an enduring principle is the idea that we should estimate return premia indirectly, by way of monitoring and modeling risk factors. That's in contrast to attempting the Houdini-like trick of predicting returns directly.

That said, we need to recognize that the relevant risk factors for estimating return premia aren't a stable lot. Fortunately, the degree of relevance for individual factors doesn’t change overnight, but it does change. Our only hope for keeping the instability on the margins is by watching a mix of factors, each chosen because there are compelling reasons for doing so.

The reasoning is that while some variables will rise and fall through time as useful measures, absolute failure across the board is highly unlikely. Assuming, of course, you build models with a robust mix of risk variables that exhibit minimal cross dependence. Ideally, we're looking for risk factors whose relevance can't be arbitraged away easily, if at all. A quick example: the so-called January effect is relatively ephemeral, since ambitious traders can simply buy sooner. On the other hand, risk premia related to the business cycle is far more durable.

So, yes, we're constantly wondering if this or that variable has lost some or all of its power as a predictor. Even under the best of circumstances, a lone factor offers limited visibility into the future. None of this should come as a surprise. If you go fishing for risk premia, there's always a chance that you'll come up empty handed. That's why they call it risk premia. Yet risk premia aren't a figment of our imagination: they do exist and they can be captured. It's not easy, nor is it for the feint of heart. And what appears to be a skillful harvesting of premia in the short run tends to be an illusion.

In other words, prudence, modesty and a healthy respect for the unknown are critical for generating success in the money game. And that starts by recognizing that risk premia are constantly being repriced and that the rules that govern success and failure in reverse engineering the associated prices are subject to change without advance notice. Other than that, investing's a breeze.

Source: Risk: Problem, Solution - It's Always There