GOLD ANALYSIS

Gold as a hedge against tail risk

Between 2007 and 2009, an 8.5% allocation to gold was able to reduce the total loss in a benchmark portfolio by almost 5% compared to a similar portfolio without gold

Author: Rhona O'Connell
Posted: Thursday , 14 Oct 2010

LONDON - 

The latest research piece from the World Gold Council discusses the use of gold in portfolio management and this time it looks at "tail risk". 

Tail risks are "infrequent or unlikely but consequential negative events" and gold can be an effective hedge against such developments.  The WGC starts by pointing out that the financial crisis of 2007-2009 has sharpened investors' minds over the merits of emphasising risk managing, rather than, as previously, looking for return at the expense of incurring higher risk.  The Council suggests that learning from the lessons of these in this difficult period may mean that investors are better prepared for new unforeseen eventualities.

It is not, the WGC argues, a question of being over cautious; such events may not be very likely, but they can have a substantial impact on investor capital and thus there should be a protective strategy in place.  Gold, the WGC opines, can be an integral part of cost-effective strategies (both short-term and long-term) that can provide this protection, without giving up return.  The Council's research has also shown that portfolios including gold decrease the Value at Risk (VaR) and that even relatively small allocations of gold, between 2.5% and 9.0%, can reduce the weekly 1% and 2.5% VaR of a portfolio by between 0.1% an 18.5% (this based on data from December 1987 to July 2010).

An analysis of events that could be deemed to be tail risks, including Black Monday, the LTCM (Long Term Capital Management hedge fund, which  sustained losses of almost $5 Bn in the space of four months after the Russian financial crisis), the recent 2007-2009 recession, etc, shows that in 75% of cases studied, portfolios that included gold outperformed those that did not.

The Council notes that most portfolio optimisers assume that returns from an asset are close to a normal distribution (i.e. they are symmetrical, and 95% of returns fall within two standard deviations), but that this is rarely the case.  Many assets returns have skewed distributions (commonly negatively skewed, i.e. with more outlives due to negative rather than positive returns), and also have heavy tails - i.e. more observations outside two standard deviations.  Furthermore, correlations between assets are not constant and although average correlations can be used to determine an optimal portfolio distribution, extreme conditions can change how assets interact with one another in unexpected and - typically - unwanted ways.

Gold differs from the assets in a number of ways, one being that it tends to exhibit lower volatility on negative returns than it does on positive returns.  Analysis of gold's historical performance from January 1987 to July 2010, in terms of the volatility of positive and negative weekly returns, shows that negative returns for gold tended to be less volatile than the overall volatility, while positive returns were slightly more volatile than the overall performance.  Over the same period, the S&P 500 was more volatile overall than gold, with a higher volatility of negative returns and a lower volatility of positive returns.  To put it another way, over this period gold was less likely to fall by twice the overall volatility than it was to rise by the same return.  Equities were the other way around.  This is largely a function of the concept of "flight to quality" when the markets are reacting to negative news, and tending to look for safe havens such as gold and Treasuries.

There is more to it than this, as the correlation structure between different assets is also important.  Gold's low correlation with the majority of asset classes highlights its role as a portfolio diversifier.  Furthermore, unlike most other typical diversifiers, its correlation to other assets tends to be beneficial in terms of portfolio returns.  Gold's correlation with US equities, for example, while low, has historically tended to decrease as US equities fall, and increase when they are in a bull phase.  Further mathematical analysis can be summed up to the effect that in economic and financial downturns, industrial-based commodities and equities tread similar paths, but gold's correlation to equities becomes more negative during these periods.

The paper goes on to demonstrate how gold can be used to manage risk effectively .  The analysis uses the common measure for "maximum expected loss", specifically, Value at Risk.  the WGC sensibly puts in the caveat that this is based on historical performance and that future uncertainty can of course affect results, but the data demonstrate that gold has shown itself, on multiple occasions, to be useful in protecting against systemic risk.  VaR is effectively a method of measuring how much an investor could expect to lose in a given portfolio in the event of an unlikely, sometimes infrequent, but possible external developments.  There are a number of ways of calculating it; WGC uses empirical distribution of returns to allow for skewness and kurtosis (i.e. heavy or light tails) that are typically found on financial data. 

Probably one of the key points is that VaR tends to be a function of volatility, while the "heaviness" of the tails also has an effect.  The greater the number of "unlikely" events more than two or three standard deviations to the left of zero, the higher the VaR.

The analysis covers the period from January 1987 to July 2010.  The Council would have preferred to have taken he series further back, but was constrained  by the proliferation of investment assets in the 1970s and 1980s, which meant that meaningful and comparable  series cannot be taken much further back than January 1987.  The paper contains matrices of results for different asset classes, looking at the correlations of weekly returns with gold in a variety of performance scenarios.  It looks also at how portfolios with /without gold behave in terms of volatility and expected annual returns for different gold portfolio weightings.

The paper demonstrates that not only does gold expand the efficient frontier (risk vs reward, essentially), but that portfolios including gold also have a lower VaR than those which don't.  Expected losses tend to diminish without necessarily sacrificing return.  The analysis goes on to look at six events in which the markets experienced unexpected and negative shock (e.g. Black Monday, October 1987, LTCM, the dot.com bubble bursting) and shows that in most periods of financial stress, portfolios including gold tend to outperform these without it.  In the case of Black Monday, gold would have improved a benchmark portfolio performance by 17%; in the "Great Recession" of 12^th October 2007 - 6^th March 2009, it would have improved performance by 8%.

The Council also points out that investors holding gold in the form of a commodity index are likely to be under-allocated, thus extending its argument that investors should maintain exposure to gold as an asset class in its own right and on its own merits.