Financial markets differ from real goods markets in one crucial aspect: under certain conditions higher asset prices increase the appetite for such assets (instead of restraining it), pushing an already high demand even higher. Similarly, in moments of panic, falling prices can easily trigger additional offer. This process is referred to as a positive feed-back. As George Cooper brilliantly exposed in “The Origins of Financial Crises”, a positive feed-back process is the perfect recipe for market instability. Hubris and panic in financial markets are a consequence of the ‘inverted’ laws of offer and demand.

The nature of financial markets will not change and will always differ from that of any other real goods market, where a lower price generates more demand and a higher price discourages buyers, moving the market towards a stable equilibrium.

What makes the problem worse is that in today’s financial markets we have two more elements that actually enhance the usual positive feed-back processes, structurally undermining the future stability of financial markets.

First, traditional relative investment strategies (anchored to a benchmark) have been increasingly losing ground in favour of absolute investment strategies. Hedge funds have started the process and now traditional investment houses have followed, shifting important amounts of money from relative to absolute investments.

The promise of absolute strategies is to make money in good times and to protect investors from losses in bad times. This is possible at an aggregated level when the relative percentage of investments in absolute strategies is small. But when the percentage becomes too big, the shift to absolute strategies is transformed into a mega-portfolio-insurance-scheme that becomes unsustainable. Some of you may remember where large portfolio insurance schemes led in the mid 1980s: to the disastrous Wall Street crash of 1987.

The second element relates to pro-cyclicality of VaR.
Recent high market volatility is sharpening attention towards risk and the need to measure and manage it. At the same time, rapid financial innovation has dramatically increased the ability to monitor and control risks, allowing measures like VaR to gain further ground. This would all be good news, if the markets were using the right type of VaR for establishing the risk limits.

Generally, when prices move down, VaR goes up, eventually triggering the risk limits and thus enlarging the troops of sellers. Symmetrically, the reduction of VaR in good times encourages traders and fund managers to pile up on risk, increasing their risk exposures when prices are already high and while demand is thriving. This ‘pro-cyclical’ behaviour can compound the aforementioned positive feed-back of financial markets. Furthermore, financial literature has recommended the practice of giving more weight to the most recent observations when computing a VaR measure (a practice called ‘exponential weighting’). The perverse and unwanted consequence is to increase the pro-cyclicality of VaR and the positive feedback process related to it, making the market environment even more unstable.

When looking at solutions, the second problem can be easily mitigated. VaR and exponential weighting can be improved upon or replaced by better measures and techniques. In my view, VaR should be blended with stress testing in single measures, applying anti-cyclical weights. I have proposed in a 2009 paper* a new measure of risk called Hybrid VaR. The measure weighs VaR and the worst outcome of a selection of historical stress test scenarios through an anti-cyclical mechanism: the more VaR declines, deviating from the worst historical outcome, the more the “memory” of the worst moments of financial stress increases.

In the hybrid VaR research, I measured the quality of this risk measure by performing a back-testing exercise during the five years between 1-1-2004 and 31-12-2008. I analysed the historical series of three indices, namely the S&P500, Eurostoxx50 and Nikkei225, back-testing a bi-weekly VaR 99%, under different models. The results show that Hybrid VaR has the lowest number of violations, performing far better than other models and in line with expectations. Table 1 summarizes the results.

Finding solutions for the first problem is harder. Investments anchored to benchmark are losing momentum and investors today ask for absolute strategies, perceiving them as a safer option. As we have explained, the commercial success of total return strategies can result in a lack of diversification of investment objectives that ultimately amplifies the positive feed-back effects in the financial markets.

If the traditional diversification of investment objectives by ‘benchmarks’ is not an option, what can we do to avoid uniformity of ‘views’ and of investment behaviour?

One option is to differentiate absolute investment strategies by ‘holding period’: certain vehicles would focus on very short term price movements; others would target medium term investment targets, while pension funds would probably remain the most appropriate long term investor. The absolute return investment funds would associate with their investment strategy a defined investment horizon. Subscribers can withdraw funds at any moment but they will suffer some sort of penalty if they withdraw their funds before the completion of the investment period, through a mechanism that I describe below.

Under this scheme, the investment fund is allowed to set aside a certain percentage of its profits to a reserve account. This percentage varies based on the investment horizon. Longer investment horizons allow for higher reserve percentages.

The proceeds allocated to the reserve are invested in very low-risk monetary instruments, to shield and protect them.

Each subscriber owns a portion of the reserve account proportional to his quote. However, if he withdraws his funds before the completion of the investment horizon, he will lose a portion of the reserve account directly proportional to the amount of time needed to fulfil the investment horizon. This “penalty” will be redistributed in the reserve account, benefiting proportionally the other subscribers: the penalty imposed to early withdrawals is transformed into an award to the investors who stick to the promised horizon.

If an investor maintains its commitment to the investment horizon, he will receive at the end of it a one-off dividend equal to the portion of reserve account he has accumulated at that point in time.

The mechanism of penalty/award described above leaves full freedom to investors for retrieving their money at their convenience. If they invest in a bad asset that loses money, they actually will be able to withdraw at NAV without any further penalty, as the fund has not been able to generate profits, and hence has not set aside the investor’s money in the reserve account.

However, the differentiation of investment horizon and in particular the targeting of a longer investment horizon can be attractive for an investor for two reasons:

• a longer term objective puts the fund in the position to benefit from short term volatility, profiting from overshooting;
• the transfer of a quote of profits into a reserve account ensures that those profits are not at risk anymore, and will be cashed in at the end of the investment horizon.

The idea that markets are not efficient and that they need to be governed is gaining ground. This is not necessarily a bad outcome if the regulators adopt measures that instil anti-cyclical incentives and introduce new types of investments that can preserve investment diversity. The current erosion of liquidity and market depth suggests that the time to act may be shorter than we think.

*Dario Cintioli – “The False Promises of Exponential-Weighting, Pro-Cyclicality and a New Measure of Risk: Hybrid VaR”, StatPro Research Papers, January 2009.