"We have a natural right to make use of our pens as of our tongue, at our peril, risk and hazard." - Voltaire
Courtesy of our friends at Rcube Global Macro, please find enclosed their latest publication where Paul Buigues looks at Risk Parity strategies:
(for PDF please use the following link: http://www.rcube.com/docs/Rcube_Is_Risk_Parity_a_Scam.pdf)
Risk parity strategies experienced large drawdowns between early May and late June due to a combination of rising government yields and falling equities.
This rather significant correction raised quite a few eyebrows, particularly because risk parity strategies are often marketed as being able to withstand a wide range of economic environments (and, unlike 2008, today’s environment is rather benign).
Although it would be preposterous to disparage a strategy based on two months of negative returns, this drawback gave us the impetus to express our thoughts on risk parity as an investment strategy, as it emerged from relative obscurity just a few years ago, only recently becoming fairly popular among investors.
Like other passive asset allocation strategies,1 the basic premise of risk parity is that asset returns are unpredictable, at least in the short] and medium]term. Consequently, investors should only attempt to capture risk premia, without wasting their time and energy trying to forecast the behavior of specific asset classes.
According to the Modern Portfolio Theory (which is, itself, based on a dozen theoretical assumptions), the only rational choice for an investor is consequently to own the gmarket portfolioh which contains every asset available in the market, weighed according to its relative size. Because this is difficult to implement in practice, investors often settle for a (generally more granular) version of the 60/40 allocation between equity and fixed income.
Risk parity is a different viewpoint on how not to exert judgment on any asset class. According to risk
parity proponents, investors should try to own all major investable asset classes on an equal risk basis.
Supposedly, this results in portfolios that have better risk/reward characteristics than traditional asset allocations. Moreover, as mentioned above, some argue that risk parity portfolios can generate quasi-absolute performances, even in the face of stormy markets.
Before going any further, it is worth stating that implementing a portfolio that contains all assets on
an equal-risk basis is even more challenging to implement than implementing the "market portfolio".
This explains the existence of many different variants of risk parity.2
Recap: Portfolios that express a neutral view on future asset class returns
After selecting a specific variant of risk parity, many implementation choices need to be made:
‐ What universe of assets should be used, and how should they be regrouped them in asset classes?
‐ Should asset class correlations be taken into account? And if so, how?
- How should we define risk? In our understanding, most risk parity implementations use volatility,
which obviously exists in many different varieties (historical, implied, predicted, GARCH, etc.) and
calculation horizons.
- What leverage should be applied to the portfolio for it to reach an acceptable rate of return? (Risk
parity generally involves leverage.)
- What frequency should be used for portfolio rebalancing and volatility calibration?
----------------------------------------------------------------------------------------------------------------------------------
1 Although risk parity strategies have to be managed actively (if only to equalize risk levels on a regular basis), we consider them to be passive, in the sense that they are not based on trying to forecast future asset returns.
2 Here are just a few implementations of risk parity: the “All weather” portfolio, classical risk parity, cluster risk parity, risk factor parity, and equal risk contribution.
To our understanding, the “All weather” strategy is not risk parity in the strict sense. From the way it has been described in various papers, it basically consists in choosing a set of asset classes, and in leveraging each of them to obtain a common expected return (generally the expected return of equities). In that sense, this strategy should be called return parity, rather than risk parity. Unless we expect all asset classes to have the same Sharpe ratio, these two approaches are not equivalent.
Due to this large number of degrees of freedom and parameters, this paper will present risk parity from a generic viewpoint. It will contain case studies and thought experiments rather than backtests (as we will see, backtests are generally biased towards risk parity strategies).
Although the term “risk parity” was only introduced in 2005, we can trace the origins of the concept
to a strategy that Ray Dalio (3) started using in 1996 to manage his family trust. Despite Bridgewater’s success in generating sizeable alpha for their clients, Dalio wanted to create an investment process that would not depend on his own ability to manage funds or to select managers (as he wouldn’t be able to do so after his death).
The strategy (named the “All Weather portfolio”) also had to deliver returns, regardless of economic
conditions. Dalio therefore concluded that the portfolio should maintain 25% of the portfolio’s risk in
each of the four following quadrants:
This is clearly an excessively simplified portrayal of a strategy that now has $70Bn under management and that has generated an annualized performance of around 8.5% with a volatility of around 10% since 1996, inspiring many fund managers and institutional investors to run the same type of strategies in-house.
However, despite its commercial and financial success, many observers consider risk parity to be an
investment scam. Finding a strategy that might dominate the classical 60/40 portfolio is one thing. Pretending that this strategy is able to produce stable returns (without attempting to predict those returns) sounds a lot like a "get rich steadily and without effort" scheme.
Even though wefre not into passive asset allocation strategies (otherwise, we would look for another
line of work), we will try to contribute to the debate. We will organize our thoughts by looking at
three intertwined dimensions of risk parity: diversification, returns, and risk. In each section, we will
express our opinion on the conceptual merits of risk parity, as well as its prospects in the current
environment.
----------------------------------------------------------------------------------------------------------------------------------
(3) Ray Dalio is Bridgewater’s founder and one of risk parity’s pioneers. Despite the criticism against risk parity expressed in this paper, Dalio is at the very top of our pantheon of financial thinkers.
1. Diversification
From a passive asset allocation standpoint, it is hard to argue against diversification, which
constitutes the core of risk parityfs philosophy.
The idea of spreading risk among different asset classes obviously precedes risk parity by a few millennia, as we can find references to it in the Talmud ("One's assets should be divided into thirds: 1/3rd in land, 1/3rd in business, 1/3rd in gold") or in the Ecclesiast ("Divide your investments among many places, for you do not know what risks might lie ahead").
In the early 1980s, Harry Browne introduced the gpermanent portfolioh, an investment strategy whose aim was to withstand all sorts of economic environments, and which was originally composed of an equally weighted portfolio of four asset classes: 25% US stocks, 25% long-term bonds, 25% cash, and 25% precious metals.
However, it is worth noting that these simple equal]weight approaches only aim at minimizing the risk of ruin from a personal wealth standpoint, which is not the modern view of how portfolios should be managed (i.e., maximizing investment returns for a given level of risk).
In terms of diversification, the major innovation of risk parity over these early approaches resides in
equally weighting risks, instead of allocations.
In that respect, risk parity proponents are indisputably right when they state that traditional 60/40 asset allocations are not truly diversified, as they have had a correlation of 0.90 with equities over the last 40 years.
That being said, we believe that placing diversification above everything else can lead to unpleasant consequences. The main advantage of a passive gmarket portfolioh approach is that, by definition, it does not disturb the market's equilibrium, as every asset class is weighted according to its relative importance in the market. On the opposite side, once it becomes popular, any other passive investment process that significantly deviates from market weights can wreak havoc in market valuations, precisely because passive investment processes entail not caring about valuations.
For example, letfs take a small exotic asset class (public Timber REITS, for instance), which would display nice diversification properties in the eyes of many different diversification]minded managers. Although each individual manager might decide not to own more than 1% of the total float, their combined buying power could very well provoke a bubble in the asset class.
A real-life example of the damage that can be caused by a blind quest for diversification can be found
in the way in which CDOs used to be managed before the credit crisis. In order to increase their contractual Moody's "diversity score", CDO managers were forced to diversify their exposures in terms of industries. As a consequence, some industries that had little outstanding debt became heavily sought after and, therefore, completely mispriced. In the end, a supposedly superior diversification did not help CDO managers, as correlations converged towards 1.00 during the 2008 credit crunch.
To a certain extent, the appeal of diversification might also explain investorsf willingness to buy TIPS
at negative yields (down to around -1% for the 10 years recently). While being a relatively small part of government debt (around 10%), TIPS' characteristics make them very attractive in the eyes of
investors who value diversification far above everything else, including valuation (in this particular
case, however, the jury is still out in determining whether we’re all “turning Japanese”).
One last word about diversification: as we will see in our next section, we’re not convinced that financial markets offer a sufficient number of uncorrelated risk premia in order to be able to reach a “true” diversification.
2. Returns
2.1. Risk premia
Like any other passive asset allocation strategies, risk parity relies on the assumption that some asset
classes should structurally outperform the risk‐free rate. Although there are theoretical justifications and ample empirical evidence for some of these risk premia, their number and their magnitude is ‐ and will always be ‐ subject to intense debate.
To us, the most convincing and economically meaningful risk premium resides in equities. Because of
the high covariance of corporate asset values with the state of the economy, equities have to compensate investors for the risk they take (no one wants to lose his job and experience portfolio losses at the same time). We can obviously only make rough estimates of the forward equity risk premium (letfs settle for 5% on a global basis), but we do have little doubt about its existence.
Even if they might offer some diversification benefits from a marked]to]market perspective, we believe that many asset classes (e.g., high yield bonds, REITS, or private equity) have a risk premium that originates from the same covariance with the state of the economy. Whether they should be considered as completely separate assets classes is, therefore, debatable. In fact, this question is specifically addressed by newer risk parity implementations, such as cluster risk parity and equal risk contribution.
For some asset classes, the very existence of a positive risk premium can be questioned. In the case
of commodities, for instance, the classical justification for a risk premium (i.e., Keynesf "normal backwardation") is nowadays dubious, as an increasing number of investors have been willing to take hedgersf opposite side. Roll yields, which had been the sole source of excess returns for commodities, have been centered on zero for the last 10 years.
For other asset classes, risk premium prospects currently look rather grim, the most obvious example
being Treasuries. If we look at 10]year Treasuries, their historical long-term return over short-term
rates has been around 1.6% since 1920. Since the early 1980s however, 10-year Treasuries have produced far higher excess returns (around 5%), as 10]year yields went from 15.8% to the current 2.5%.
Although there are only a few things about which we can be certain in finance, we can safely proclaim the mathematical impossibility of getting 5% excess returns by rolling 10-year treasuries over the next 10 years.
Therefore, because risk parity strategies always overweigh fixed income assets due to their low volatility, we can ascertain that this source of outperformance against conventional 60/40 allocations has dried up, even without invoking a gbig rotationh that would bring 10-year yields back to a theoretical long]term equilibrium value.
There are obviously many other sources of risk premia. However, most of them (liquidity‐based ones,
for instance) are the “bread and butter” of specialized hedge funds. Therefore, they are outside of the scope of risk parity, which is not a bad thing, as many of these arcane risk premia tend to display a very negative skewness.
Our main point is that, even if we consider a large universe of asset classes, it’s not as if there were dozens of investable and economically meaningful risk premia waiting to be harvested by passive investors. In the end, when we take into account the fact that many risk premia actually originate from the same basic sources, we might end up with just a few investable risk premia. Additionally, as more people reach for diversification, those few risk premia tend to become more correlated over time.
2.2. Leverage
One important point regarding returns resides in the fact that risk parity strategies generally involve
leverage—that is, unless the investor is satisfied with long‐term returns of 2 to 2.5% over the risk-free rate.
Risk parity practitioners generally characterize leverage as a mere “implementation tool”, and they
believe that their superior diversification outweighs the disadvantages of running a levered strategy.
Although a reasonable use of leverage might not be fatal to a portfolio, it can irremediably hurt its
returns. Indeed, as we will see in our section about risk, leverage introduces a path dependency issue.
We can very well imagine a “black swan” situation, in which a supposedly safe asset class experiences a price trajectory that forces a deleveraging of the portfolio and, therefore, wipes out a large chunk of it.
3. Risk
We believe that the subject of risk is the one wherein risk parity is the most open to criticism.
Indeed, to reach the gparityh in risk parity, one has to reduce the risk of an asset to a single number
one way or another (generally a specific variant of the assetfs volatility). Although it is not a very original point of view, we believe that the risk of an asset cannot be quantified in this simplistic way.
Despite the fact that there is a certain level of stickiness in an assetfs risk (or volatility), every now
and then, assets - even supposedly gsafeh ones - have the nasty habit of breaking the parameters of
the equations that are supposed to describe their behavior (especially if these equations do not take
into account skewness).
To illustrate this point with a little story, letfs imagine a situation that could very well have happened
during the last decade:
In the aftermath of the 2000s tech crash, John becomes yet another young unemployed electrical
engineer (as Taleb, the inventor of the Black Swan theory, likes to characterize most quants). He decides to start a new career by getting a masterfs degree in finance. Armed with his solid math skills, John quickly digests modern portfolio theory, basic statistics, and all varieties of volatility calculations. He finds a job at an institutional investor and quickly moves up the corporate ladder.
In 2006, John convinces his board to apply a risk parity strategy to manage the firm's portfolio. Because he has a fresh and open mind about finance, he decides to spice up the asset mix by adding an exposure to mortgage]backed securities in the form of newly-minted ABX indices.
Who could blame him, based on the information available in 2006?
- The total size of the US mortgage debt is huge ($13 trillion in 2006), comparable to US equities, and larger than government debt.
- ABX indices are highly diversified, as each index is based on 20 distinct RMBS transactions. Each RMBS containing a minimum of $500 million worth of homes, an ABX investor is exposed to more than 50,000 homeowners throughout the US. What can possibly go wrong with such a diversified pool of debtors?- ABX products are rated by respectable institutions, such as Standard & Poorfs (1860) and Moody's (1909), and they offer a wide variety of risk levels (from AAA to BBB).- The volatility of the underlying financial products that compose the index is minuscule (they always trade around par).
Even if John had opted to buy the safest AAA ABX tranches (with, consequently, a high allocation due to their glowh risk), he would have experienced heavy losses during the 2007-2008 crisis. Additionally, he would have been forced to drastically reduce his allocation to the asset class as the gtrueh risk (or volatility) of ABXs revealed itself, preventing it from benefiting from any subsequent recovery.
Consequently, given that he was running a leveraged portfolio, John would have been forced to crystallize his losses.
This story might sound far]fetched, but we could have invented a similar story about Georgios implementing a risk parity strategy for a Greek institutional investor by leveraging on domestic government debt.
Some might argue that both of these examples involve blatantly asymmetric assets, which could have
easily been filtered out (especially in retrospect) by an experienced risk parity practitioner.
However, we can also imagine a forward‐looking scenario that would involve one of the most respectable assets on earth ‐ US Treasuries ‐ as the main culprit of a risk parity carnage:
Let’s imagine that, a few years down the road, Bernanke’s successor has to manage another “great
recession”. This time, the Fed decides to go beyond QE by pegging long‐term rates at a very low level (let’s say 0.5% for the 10year).4
As the Treasury remains stuck at 0.5%, there is no more volatility on Treasuries.
According to the risk parity playbook, an investor should therefore increase his exposure to Treasuries alongside the Fed. In exchange for a minuscule return, the investor would, thus, face a substantial jump risk if the Fed had to apply a hurried “exit strategy” due to a surge in inflation…
From a broader perspective, we consider risk parity to be the antithesis of Minsky’s “financial instability hypothesis”. According to this view, investors increase their leverage when they believe an asset to be stable, which reinforces their belief that the asset is, indeed, stable (this is a perfect description of how risk parity investors behave in a given asset class). The cycle goes on until we reach the dreadful “Minsky moment”, where investors are forced to deleverage as the real risk of the asset reveals itself.
----------------------------------------------------------------------------------------------------------------------------------
4 This solution was discussed by the Fed in late 2010, and it has already been experimented withbetween 1942 and 1951.
Conclusion
Due to the fall in government yields over the last 30 years, risk parity strategies have had an easy time compared to traditional asset allocations. We should therefore disregard all the performance]based arguments that are often put forward by the proponents of risk parity.
From a conceptual standpoint, although it might seem unfair to make generalizations about a strategy that exists in many different variants and implementations, we believe that risk parity suffers from many structural flaws:
1) Risk parity requires to make choices between many different implementation options, asset selection, calculation parameters etc. These choices necessarily contain arbitrary components and will have a significant impact on the strategyfs performance under different scenarios.
2) By placing diversification above any other consideration, risk parity portfolios can hold assets at (or even move assets toward) uneconomic prices. This problem is magnified as risk parity - or other approaches focused on diversification - become increasingly popular.
3) After all, risk parity’s quest for diversification might prove fruitless, as risk parity portfolios end up harvesting the same basic risk premia as traditional asset allocation (mostly the equity premium and the term premium), albeit at different dosages.
4) The leverage used by risk parity strategies makes them prone to deleveraging and, therefore, to crystallization of losses.
5) Risk parity’s false premise that risk can be quantified as a single number exposes it to highly
asymmetric returns, which can happen to any asset class given the right set of circumstances.
If someone wants to run a passive asset allocation, we therefore believe that a market portfolio constitutes a better option from many perspectives: conceptual, foreseeable reward-to-risk and CYA.
For the same reasons, we strongly reject the idea that risk parity portfolios could represent an "all weather", quasi-absolute return strategy (we suspect marketing departments are the ones to blame for these outlandish claims).
There are certainly seasoned risk parity professionals out there who are able to mitigate risk parity's
numerous flaws. However, we have little doubt that when the next gblack swanh terrorizes the financial world (as seems to be the case on an increasingly frequent basis), we will witness the implosion of many risk parity strategies (those that are based on high leverage, overly simplistic assumptions on asset risks, and/or an unfortunate choice of underlying assets). Trusting risk parity to manage onefs life savings is therefore quite perilous, especially if it takes the form of a formula-based risk parity ETF - which should come out any day now.
That being said, the idea of a passive investment strategy that would be able to withstand any kind of financial weather is not unrealistic. However, its goal should be the long]term preservation of capital and not its theoretical maximization under a theoretical risk constraint. Additionally, the strategy should make very little use of leverage, and it should not make too many assumptions on the risk of a given asset (as risk becomes an unpredictable beast every now and then). In the end, we would probably end up with something quite similar to the Talmudic portfolio (N equally-weighted assets).
We realize that, without adhering completely to risk parityfs principles, many institutional investors
are implementing it as a part of their portfolio alongside other "absolute return" strategies. This approach is clearly less dangerous than an all]in commitment to risk parity. At a portfolio level, it simply results in overweighting low]volatility assets, which is obviously far-removed from the original purpose of risk parity.that is, true diversification at a portfolio level.
"Living at risk is jumping off the cliff and building your wings on the way down." - Ray Bradbury
Stay tuned!
Posts
No comments:
Post a Comment