Quant Ideas: How VAR can add value to energy market analysis

Alessandro Mauro shows how using value-at-risk can improve market risk analysis in the energy sector

coins-finance

Oil market analysis is an important part of economic analysis, as it addresses the current situation and forecasted trends of a crucial sector of the world's economy. There is no lack of short-, middle- and long-term studies that inspect the fundamental aspects of this market: supply, demand, inventories, shipping and prices. This information is prepared by public agencies, oil companies, private firms and research institutes.

However, there is a shortage of data in relation to the measurement of market risk present in oil markets. Oil companies have invested considerable resources in shaping up their market risk management capabilities, in each of its different stages: identification, evaluation and treatment. But until today, public market analysis has neglected this fundamental aspect in explaining the oil markets.

Although it is known and understood in general that energy prices are volatile and subject to continuous change, currently available analysis does not propose any measure of market risk. This article demonstrates that simple techniques, based on modern quantitative risk measurement, can be applied successfully to oil markets. With this toolkit, additional information can enhance the understanding of energy market trends.

VAR applied to energy markets

VAR was made popular during the 1990s by JP Morgan and its RiskMetrics model. It proposed a quantitative measurement of market risk by applying basic statistical confidence levels theory.

RiskMetrics was targeted clearly towards financial institutions. In those years banks and the likes of did not hunt commodity markets as a source of profit from trading activities. It would not be a surprise if RiskMetrics' product coverage included only a handful of energy products, such as the energy financial futures traded on the New York Mercantile Exchange (Nymex), as reported in table A.

er0316-technical-table-a

In those years, energy firms lagged behind financial institutions in terms of the entire risk management process, for every type of risk. This neglect was unjustified, because these firms were already exposed to those risks. It was the case of market price risks – after two oil shocks in the 1970s and one counter shock in the 1980s – that demonstrated the vulnerability of firms and countries to sudden changes in prices and volatilities in the energy markets.

The application of modern risk management to energy trading firms started at the end of the 1990s, and it was not limited to the oil market. The ongoing liberalisation of power and natural gas markets was introducing more volatility. Under this setup, the need for proper risk measurement, as a fundamental step in modern risk management, was becoming more urgent. At the end of the 1990s, VAR was applied successfully by energy firms to gauge their exposure to market price risk. The application was possible for oil markets and for liberalised and non-liberalised gas and power markets.1

Today, VAR is included in the basic toolkits of large and mid-sized energy firms, which continually utilise this technique in order to perform risk measurement for market price risk. However, there are also many notable exceptions, as was recently discussed in the OW Bunker bankruptcy case.2

Although widespread, VAR is often criticised by many who point to its limitations. Such limitations have been well studied and documented. There is a general understanding that VAR needs to be handled with care; however, it is hard to accept the point of view of many in academia and industry, who suggest throwing away this approach altogether.

The VAR framework can be extended and made more complex in order to solve many of the theoretical issues that have been put forward. As far as practical application is concerned, it has been argued that VAR utilisation can make sense for financial institutions but is seriously flawed in the case of non-financial firms. These critiques are worth attention. In fact, non-financial firms, especially those operating in commodity markets, are subject to other sorts of risks, which go beyond pure market price risk.

For example, it is rare that the quantity of physical goods produced, exchanged and finally consumed is predetermined and fixed. What we call volumetric risk makes commodity markets crucially different from financial ones, where instead the quantities (shares, bonds, derivatives, etc) are certain most of the time. However, it has been demonstrated that there are other methodologies capable to take in to account the simultaneous uncertainty in prices and volumes, for example with the cashflow-at-risk approach or Monte Carlo simulations.

Commodity producing and trading companies, and especially those active in energy markets, have found suitable ways to overcome these obstacles and put VAR at the centre of their risk measurement efforts. It is not surprising these applications have been considered as a matter of intellectual property and only public listed companies systematically communicate VAR limits, VAR utilisations and the general hypothesis underpinning their models.

However, utilising VAR exclusively to measure the market risk of a specific firm is another way of downplaying the importance of the whole approach. In fact, VAR can also offer unique insights into trends developing in the energy and commodity markets. We look at two case studies to see how VAR is useful to detect, analyse and communicate oil market trends.

VAR and the evolution of the Brent-WTI spread

The oil market and its analysts were in turmoil at the beginning of this decade. The unthinkable was happening and it was a subject of intellectual speculation as to why that was taking place and when it would come to an end. It concerned the world's two major crude oil price benchmarks, namely WTI and Brent, the prices of which give global indications about oil market trends and are used to directly or indirectly value the multitude of other crude oil qualities around the world.

The price difference, or spread, between WTI and Brent has been historically narrow and stable for so long that it was frequent for traders to hedge a physical position on WTI by trading Brent futures and vice-versa.

The surprise came between the end of 2010 and the beginning of 2011, with the WTI price rapidly losing ground against Brent. That trend continued in the following months and years, and at some point the maximum difference came close to a whopping $30/bbl, as shown in figure 1. The causes for this price disconnection have usually been associated with the following factors:

  • a rapid increase of oil production in the US pushed by the ‘shale
    revolution';
  • logistical limitations in the supply chain, constraining new production to flow exclusively to US storages and refineries; and
  • the oil export ban in place in the US, allowing export only in North America. This compares with Brent crude, which is free from geographical trading restrictions.

er0316-technical-figure-1

er0316-technical-figure-2

Something else was less evident and essentially missing in the plethora of market analysis that came out. Which were the implications in terms of market price risk? Was there any implication for this relevant departure from the previous ‘market order'? To answer these questions, let us travel backwards and recalculate VAR for Brent and WTI starting from 2010. What would be the results? VAR figures would look similar to figure 2.4

Consider, for example, the situation in the middle of February, 2013. At that point, VAR for WTI was at $1.37/bbl. That meant if you held a position of one million barrels of WTI crude for one day, there was a probability of 5% that you would lose at least $1.37 million. The 5% probability looks low and negligible, but it actually corresponds to one day out of 20 – that is once in a month (oil markets are closed during weekends). Market price risk on Brent crude was even bigger than for WTI, as it stood at $1.52/bbl.

The big surprise would come after calculating the VAR of the Brent-WTI spread, ie, a portfolio where you go long Brent and short WTI (or vice versa) for same volumes. The VAR for this position was in fact $1.61/bbl. The interpretation is rather simple: holding a spread on Brent-WTI was riskier than holding a position on WTI standalone. This is pretty counterintuitive, as financial theory praises the benefits of diversification for the sake of risk reduction. In a trading environment this is normally vulgarised to the one-size-fits-all advice to ‘get rid of outright exposure', ie, eliminate directional risk and leave non-directional risk only (basis risk, in a trading environment). Well, here VAR was telling another story: it was actually less risky being just long (or short) Brent (or WTI) than holding a Brent-WTI spread.

After careful inspection, this was to be attributed to the very volatile, and often very low, correlation between Brent and WTI price returns, as demonstrated in figure 3.
The VAR figures discussed above came in correspondence of a Brent-WTI correlation that was lower than 40%. In such situations, VAR for a portfolio can be higher than VAR for the single components. As a corollary, at that point VAR was also communicating that a WTI physical exposure should be hedged with WTI-related derivatives. That was sanctioning, from a market risk point of view, the decoupling going on in the world's crude oil market.

It is worth mentioning that, without a VAR measurement system in place, this important and unusual market trend would have gone undetected.

VAR and the riskiness of oil refining margins

Let us now introduce the second case study to demonstrate further how VAR is useful to detect, analyse and communicate oil market trends.

The oil refining industry has a crucial role to play in the context of oil markets, as it represents the link between crude oil extraction and oil products ready for final consumption. Because of its importance, oil refining is constantly under the lens of analysts who try to explain or forecast the evolution of refining margins, ie, the difference between the value of oil-refined products and the cost of crude oils necessary to obtain that production. Those analyses are applied to different crude oils, world regions and refining technologies, as exemplified in figure 4.

er0316-technical-figure-3er0316-technical-figure-4

By looking at this graph one can appreciate that refining Brent crude oil in a refinery that uses a fluid catalytic cracking technology has delivered, during 2014 and 2015, a higher refining margin than by processing the same crude oil in a less technologically advanced hydroskimming refinery. That happens because the product's value of a cracking refinery is higher than a hydroskimming one.

The big absent from this commonly depicted picture is again market risk. One cannot compare two portfolios exclusively on the basis of past or forecasted performance, as risk is also a crucial part of the equation. Risky positions need to be compared on a risk-adjusted basis. That can be achieved by calculating and taking into consideration VAR, as exemplified in figure 5.5

Let us consider cracking refining technology. We observed in figure 4 that such a refinery is delivering a higher refining margin. However, data in figure 5 demonstrates that the same technology brings also a higher market risk to a refiner.

These findings confirm finance theory: if something has a higher expected return, then consequently it needs to be riskier. In fact, as refiners' main objective is to optimise the value of their refining margin, it is reasonable to assume that they position themselves on the efficient frontier, where an increase in the expected return can be achieved only by bearing additional risk.

So far we have ascertained that cracking refinery technology has a higher expected return and higher market risk than simpler hydroskimming technology. However, we should compare the two on a risk-adjusted basis. This is possible by dividing the refining margins by their respective VAR. The result is represented in figure 6.

It is true that a cracking refinery delivers a riskier position but, if compared against the margin it produces, the picture looks pretty different. In a less technologically advanced hydroskimming refinery, in May 2015 there was a 5% probability of losing close to 50% or more of the refining margin in a one-day time horizon (ie, one out of 20 business days). On the contrary, a cracking refinery can lose only 10% of its margin under the same conditions. Obviously, these values are subject to change every day and need constant monitoring through a sound VAR measuring setup. This practice would allow moving the refiners' objective out from pure refining margins optimisation to the maximisation of risk-adjusted refining margins. At the same time, studies produced by organisations and think-tanks should not neglect market price risk considerations when analysising past, present and future of the oil refining industry.

er0316-technical-figure-5er0316-technical-figure-6 Conclusions

VAR utilisation in the energy space is usually limited to companies that perform evaluation of market risk for their portfolios. Although widely used at the corporate level, there is a trend to downplay the importance of VAR. However, this instrument is very valuable, not only for the internal use of energy companies, but in the general energy markets analysis space.

By means of two real life case studies, we have demonstrated that VAR delivers important insights into the riskiness of energy portfolios and allows relevant improvements in the understanding of energy market trends. This aspect is largely neglected in currently available public analyses and comments, which focus on energy data without delivering a view on the riskiness of prices and of the values of assets.

Alessandro Mauro is an independent adviser in commodity risk management.
Email: [email protected]

Notes

1. The reader may refer to Mauro, A (1999).
2. See Mauro, A (2015).
3. Settlement prices of first maturity Futures on Brent and WTI.
4. VAR figures calculated in this article are based on a parametric normal VAR model with 95% confidence level and one day holding period.
5. These VARs are calculated based on typical refinery production with Brent crude as feedstock and using reference data published in IEA (2012).

References

Hyksos Commodities, 2015
Oil Value-at-Risk
www.hyksoscommodities.com, May 9

IEA (2012)
IEA Refinery margins. Methodology notes
International Energy Agency, September 2012

Mauro A, June 1999
Price risk management in the energy industry: the value at risk approach
Available at http://bit.ly/VaR_Energy

Mauro A, January 2015
OW Bunker: How one of the world’s largest marine fuel traders went from IPO to bankruptcy
Ship & Bunker. Available at http://bit.ly/AM_OWBUNKER_ARTICLE

RiskMetrics, December 1996
RiskMetrics – technical document
Fourth edition, page 205

  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: