A Short Note on Sovereign Commodity Risk Management* by Frank Lehrbass Working Paper, January 2015 JEL: C00, D7, F5, H3, H7, N4 Key words Sovereign risk management, rationalist explanations, expected utility maximization, commodity risk management, hedging Abstract I investigate sovereign risk management using expected utility theory. A proposition is derived under which conditions which degree of hedging is optimal. An application to the case of Russia shows that a risk-acceptant attitude can serve as an explanation of the decisions to bail out Rosneft and to leave some oil exposure unhedged. *I wish to thank Bruce Bueno de Mesquita from New York University, NY, U.S.A., for helpful comments and suggestions. A Short Note on Sovereign Commodity Risk Management Introduction Many countries in the world are exposed to commodity market price risk. It is an established element of corporate risk management to hedge these exposures using derivatives. Nonetheless the World Bank reports that "though well-established in the commercial sector, the use of market-based price risk management is not widespread in the public sector, particularly by sovereigns" (Dana and Sadler, 2012). However, there are some exceptions. "Sovereigns who have made public their hedging plans include Mexico, Panama, Ghana, and in the past Ecuador, for oil, and Chile for copper. The biggest oil exporter to publish its hedging program is Mexico, which gained acclaim in 2009 when its hedging program made a profit of approximately $ 5 billion" (Molloy, 2011). This example shows that the oil derivatives market is big enough to allow a sovereign to hedge its oil exposure – and that this can be a worthwhile action. However, recent research has elaborated that a group of sovereigns is currently running into problems with oil prices below $ 100 per barrel (bbl). The price of oil needed to balance their budgets is "one useful measure of their respective pain thresholds" (Deutsche Bank, 2014). The latest estimates per country are as follows: "Bahrain ($136bbl), Oman ($101bbl), Saudi Arabia ($99bbl), Nigeria ($126bbl), Russia ($100bbl), and Venezuela ($162bbl)" (Deutsche Bank, 2014). There is ample evidence that these countries are facing problems, which indicates that the oil production has not been hedged. Firstly, I will investigate the sovereign decision not to hegde on a general level by making use of expected utility theory as introduced by von Neumann and Morgenstern (1944). This presupposes that the sovereign is able to rank risky alternatives. But speaking of the risk preferences of a sovereign seems an unreasonable approach in light of the work by Arrow (1950). He has shown that – given only weak requirements for the ordering - there is no way to aggregate individual preferences into one single preference order. However, there are exceptions for countries which are ruled essentially by one or few, rather aligned individuals. The list of sovereigns in trouble might intersect to some degree with the exceptions. Secondly, for exposition I will apply the general model to the case of Russia to specify Russia's risk preferences. It will turn out that assuming a risk-acceptant attitude (i.e. a strictly convex utility function) conforms with the open position in oil and the bailout of Rosneft. A Simple Model of Sovereign Risk Management The application of expected utility theory to hedging decisions has a long tradition. One of the early publications is Ethier (1973). The history of the application to international politics is a bit shorter and starts with De Mesquita (1980). I take two of his original assumptions and apply them to the case of sovereign decision making. Sovereign decisions can be viewed "(a) as if they are the product of a single, all important decision maker [i.e. the leader]; (b) decision makers are rational expected utility maximizers". Furthermore let me assume as in De Mesquita (1980) that the "leader's welfare" is the argument of the utility function u(.) to be maximized. Since future oil prices are uncertain, the leader maximizes expected utility. The leader's welfare is certainly a function of governmental tax income, which is again a function of the revenues from selling oil. To keep things simple I denote the amount of the sovereign's oil production by x in units of bbl and the uncertain oil price by p in units of $. The leader maximizes the following expected value over a certain time horizon: Eu(xp) (1) E[.] denotes the expectation operator using the subjective probability distribution as seen by the leader. For exposition I assume a horizon of one year, which makes x the annual oil production. 2 A Short Note on Sovereign Commodity Risk Management As a representative of the hedging instruments available I introduce a one-year futures contract, which can be bought or sold at today's known futures price level of f in units of $/bbl – for instance at the Intercontinental Exchange (ICE). The leader certainly does not have a 'crystal ball' to foresee future oil prices. Many studies have investigated how far oil-futures prices can be treated as expected spot oil prices and concluded that "treating oil-futures prices as the expected future spot price is a good first approximation" (Alquist and Arbatli, 2010). Hence, I assume that the futures price f is an unbiased estimator of the future oil price p, i.e.: f Ep (2) The last bit of notation is the decision variable h, which is the amount of barrels sold forward at the current futures price f. For instance, if the leader chose to hedge fully, we would have x=h. What is chosen is the outcome of the following decision problem of the leader: max Eu(xp h( f p)) (3) h The only difference to equation (1) is the addition of the profit or loss term from hedging with futures. This simple model implies a proposition for leaders in general. Proposition (i) The leader will hedge fully if he or she is risk-averse. (ii) If he or she is risk-acceptant a full hedge is the worst decision. Hence, he or she will leave the oil exposure unhedged. (iii) If he or she is risk-neutral it does not matter whether a hedge is in place. Proof I start with (i). Risk-aversion means that the utility function is strictly concave. A full hedge reduces xp+x(f-p) to xf, which is non-random. Due to assumption (2) this is equal to xE[p]. With the help of Jensen's inequality from probability theory one sees that getting the expected welfare for sure is the best outcome for a risk-averse leader, because: Eu(xp) u(Exp) (4) The case of (ii) implies a strictly convex utility function. The inequality in (4) reverses. Thus getting the expected welfare for sure is the worst thing for a risk-acceptant leader, which is why he will avoid hedging. A risk-neutral decision maker maximizes E[xp+h(f-p)]. Insertion of (2) gives E[xp+h(E[p]-p)]. Since E is a linear operator the term following the control variable h vanishes. What remains is xE[p] for any choice of h. Hence, it does not matter. q.e.d. 3 A Short Note on Sovereign Commodity Risk Management Russia's Oil Exposure Management Sceptics might doubt whether the oil derivatives markets are liquid enough to allow Russia a sufficient degree of hedging. The answer is that hedging usually applies a whole range of instruments and that it is therefore insufficient to look only at the exchange traded futures markets. In the commercial sector exchange traded futures, OTC traded forwards and long term supply contracts are used in parallel. Hedges are built up over time and not on a single trading day. Hence, like Mexico or a commercial company, Russia could hedge its oil exposure if it wanted to1. The media report that there is a deep impact by the recent decrease in oil prices, which is reflected in falling Russian stock prices and a devaluing currency. More specifically the FT reports that Russia derived "more than half of its budget revenues from oil and gas extraction" in 2013 (Hille et al, 2014). Both commodities are closely related, because it is well-known that within Russian long term gas supply contracts gas is priced as a function of the oil price. If the oil price falls, so does the revenue from selling gas. "Falling oil prices were causing Russia economic damage of 'some $ 90 to $ 100 billion per year'” according to a statement from Russian Finance Minister Anton Siluanov (Khaleej Times, 2014). Obviously Russia has not hedged its oil price exposure to avoid these deep impacts. Application of the Model With respect to Russia the "he" formulation of the Proposition is relevant. The fact that there is no full hedge of the oil exposure implies a risk-acceptant attitude or a violation of the unbiasedness assumption as expressed in equation (2). Under risk-aversion a less than perfect hedge would be chosen if the futures price is below the expected oil price, i.e. f<E[p]. Therefore, I have to discuss the unbiasedness assumption. In theory the assumption could be checked by asking the Russian leader for his oil price expectations and comparing them to the current oil forward price curve. It is clear that this is out of question. As an available approximation I quote the Bank of Russia Governor. On 11 Dec 2014 the central bank expected "average oil prices to be $ 80 per barrel during the next three years. This average price results from consensus forecast of the leading analysts" (Nabiullina, 2014). Firstly, the quoted levels were close to the then current futures quotes (in the higher seventies).