5 Décember 2017
Risk Premia Investing means fundamentally considering in the investment process the fact that risk premia are not stable but vary greatly over time.
It is well known in the academic literature that all the volatility of assets’ prices cannot be fully explained by changes in expected pay-offs and short-term interest rates (current and expected). A significant part of markets’ volatility is triggered by changes in risk premia, markets moving from “cheap” to “expensive” and vice-versa.
For a discussion of the reasons why risk premia change, see this background paper. In our Yield Curve Model, we find that broadly half of the volatility of 10-year government bonds can be explained by changes in risk premia (the rest coming from expected changes in monetary policies).
Obviously, it is always difficult to estimate very precisely the risk premia one can expect by investing in various markets. Experts can disagree on the details and there is always a healthy debate going on. Yet, there is often a qualitative agreement to qualify various assets classes as reasonably valued, cheap or expensive (or even sometimes in bubble territory).
The volatility of risk premia has three main consequences for investors.
The first two are well recognized. But, in our view, the last one does not receive the attention it deserves.
1/ Optimal strategic asset allocations for long-term investors vary over time.
Obviously, it does not make sense to invest the same way in cheap or expensive markets. Yet, very often investors have rigid strategic asset allocations or benchmarks and believe that changes in risk premia could be accommodated by some short-term limited tactical twists to their rigid strategic allocation. This approach generally does not recognize how powerful the impact of changes in risk premia should be: large cycles in asset prices justify more than a few percentage points added or subtracted in the weight given to various asset classes in the investment process.
2/ Moreover, this time-varying optimal asset allocation is very dependent on the investment horizon.
There are two main reasons why a long-term investor should not have the same portfolio as a short-term investor.
They do not face the same risks.
For a short-term investor, lower asset prices are painful whatever the fundamental reasons behind the loss. There is no difference between falls triggered by lower future expected pay-offs, a more hawkish expected monetary policy or higher risk premia. Yet, for a long-term investor, the origin of potential short-term losses is very important. Lower expected pay-offs mean that, barring a new positive surprise, the loss is durable and will not be recouped over time. But if pay-offs are stable and that the discount factor (thanks to a more hawkish monetary policy and/or higher risk premia) has simply increased, the short-term loss is likely to be compensated overt time. Thus, the long-term investors will overall face less risks and will be able, for a given risk aversion, to invest more in risky assets. Moreover, the optimal combination of these risky assets will also depend on the horizon. The more an asset class is sensitive to risk premia shocks relative to pay-offs shocks, the more the long-term investor will overweight this asset class in its portfolio compared to what a similar short-term investor would do.
They should not use the same forecasts and risk premia.
Short-term investors are only interested in the short-term expected performances of various asset classes. Thus, they need what we call “spot” risk premia: the short-term expected performances of various assets relative to a risk-free interest rate. Long-term investors are obviously also interested in these short-term forecasts, as they can take some tactical positions: even if an asset class is very cheap, they will be cautious if they believe that prices are likely to fall even lower in the short term. Yet, this is not the end of the story for long-term investors. If an asset is cheap, they fear to miss out future high expected returns. Thus, one should recognize that underweighting a cheap asset is risky for long-term investors. As a result, long-term investors will use both short-term financial forecasts and long-term estimates of risk premia. Indeed, as all long-term fund managers know, the tricky question is to optimally weight this two information when they send conflicting signals (i.e. in an apparently rising market which is already expansive – the US stock market at the beginning of 2018? – or in a falling market which is already quite cheap – stock markets at the beginning of 2009?).
3/ Thanks to varying risk premia, markets may face “valuation crisis” and tactical asset allocations should take this risk into account.
A key belief of Risk Premium Invest is that markets are not very efficient to ride smoothly changes in risk premia (see Why Risk Premia Vary and Our Yield Curve Model). A good test of this view may come soon with the end of Quantitative Easing and the realignment of risk premia this could trigger! We also wait with some worries how the foreign exchange market will react in the coming few years to the worsening US fiscal position. When risk premia change, markets tend to underreact in a first phase. Then, they often find very difficult to assess how equilibrium valuation have changed (see Our guide to understand and trade the US Treasuries yield curve). In this second phase, markets may become very unstable. We believe that short-term forecasts should be very attentive to the risk premia dynamic and try to assess where markets stand in this process of integrating underlying changes in risk premia. This approach of financial forecasts is specific to Risk Premium Invest.
How to take into account these three key consequences of the variability of risk premia? We believe that as a result the asset allocations should be based on a three-stage process.
1/ Building an explicit and consistent central scenario for assets’ returns at various horizons.
What we call an “explicit and consistent” scenario is a scenario in which the key drivers of asset prices are clearly analyzed at various horizons and fully coherent with the scenario. In other words, expected future prices should be based on assumptions relative to 1/ the future relevant pay-offs (depending on the asset class under consideration: the dividends for equities, the risks of default for bonds, and the long-term equilibrium exchange rates for currencies), 2/ the expected path of monetary policies (a key driver of discount factors for all asset classes) and, last but not least, 3/ a detailed forward-looking analysis of risk premia, especially in times of turmoil.
2/ Building explicit and consistent set of alternative scenarios to measure risks at various horizons.
Here again the consistency depends on the use of mainstream valuation models and the dynamic analysis of how various shocks impact asset prices at various horizons. A negative shock on pay-offs will have a durable impact while shocks on monetary policies and risk premia will be corrected over time (an extreme case being risk-free government bonds: held to maturity the long-term return is 100% sure while short-term investors may face a lot of risks as the bond market reacts to changes in monetary policies or the duration risk premia).
Building this dynamically coherent description of risk is not easy, but Risk Premium Invest has developed a generator of scenarios which has the required properties: shocks are related to pays-offs, monetary policies and risk premia, and they are dynamically coherent. Obviously, the main difficulty is to calibrate this generator of scenario: how the volatility of markets will change in the future? How much of the current and future volatility will come from changes in expected pay-offs versus changes in risk premia or destabilizing news from monetary policy?
To help answering this difficult question, Risk Premium Invest monitors closely the dynamic of financial markets, and based on various models and an analysis of new information, try to identify the source of the observed volatility.
3/ Using a powerful optimizer to extract optimal portfolios depending on investors’ risk aversions and investment horizons.
Once this set of central and alternative scenarios are available, the question is to find the optimal dynamic allocations considering the risk aversions and horizons of the various investors. As already explained, for long term investors, the optimal portfolio will depend both on the short-term financial forecasts and on the long-term risk premia.
A key aspect of a dynamic allocations is that they are expected to change significantly over time. If risk premia on some assets increase without a commensurate increase in risk, the investor may increase its allocation to these risky assets to benefit from better expected return. Thus, not only long-term investors are not that penalized by a fall in prices triggered by an increase in risk premia, they can even benefit from such a situation as new opportunities emerge. Volatility is the friend of the long-term investors as long as volatility comes from risk premia and not drastic changes in expected pay-offs. A powerful optimizer must to take into account how this option to benefit from higher risk premia in the future impacts as soon as today the optimal portfolio. This is the case of the optimizer used by Risk Premium Invest based on a stochastic programming approach.