There are two common rationales for the existence of forward commodity markets: that they allow buyers and sellers to hedge the risk of fluctuations in the spot price of a commodity, and that they aggregate information about the future values of spot prices across market participants.
There is ample empirical evidence in many industries and commodities that hedging is widespread.1 There is also growing empirical evidence that forward prices for a commodity provide important information about future spot prices.2
Many have argued that information aggregation can also yield quantifiable economic benefits to the producers and consumers of this commodity. For example, as long ago as 1953 Holbrook Working wrote that “[h]edging is commonly credited with reducing margins between the price paid by the producer and that paid by the consumer”.
Gray (1964) argued that the introduction of a futures market improved the profitability of Maine potato growers relative to those in other regions. And Charles Cox (1976) concluded that “[a] spot market is more efficient in the sense that price more fully reflects available market information when there is futures trading”.
In the more than 60 years since Working’s claim, however, there has been little empirical evidence that a liquid forward commodity market yields market efficiency benefits to producers or consumers of this commodity. Day-ahead forward and real-time locational marginal pricing wholesale electricity markets, introduced in many parts of the US in the early 2000s, are an ideal environment in which to investigate this claim.
Wholesale electricity markets
Exactly the same product – electrical energy delivered to a specific location in a specific hour of the day – is traded in both the day-ahead forward market and the real-time physical market. Also, purely financial trading is integrated into the physical market clearing process.
Purely financial participants (as well any other market participant) can take a position in the day-ahead market as a ‘virtual’ electricity supplier or demander by submitting an offer to sell or bid to buy energy at any location in the transmission network. If this offer or bid is accepted in the day-ahead market, then this position must be reversed in the real-time market as a price-taker. Selling one MWh of ‘virtual’ energy in the day-ahead market at a given location and buying this energy back in the real-time market pays the difference between the day-ahead and real-time price at the location. Therefore, purely financial participants can earn money on this price difference without supplying or consuming any actual energy.
The actions of financial market participants in wholesale electricity markets can have a direct effect on market prices, because the virtual bids and offers in the day-ahead market are treated exactly as bids and offers from physical electricity suppliers and demanders. For example, a financial market participant selling enough ‘virtual’ energy will reduce the day-ahead price at that location. The requirement that this ‘virtual’ energy is bought back in the real-time market is likely to increase the real-time price. This logic implies that the actions of a financial market participant to exploit expected differences between day-ahead and real-time prices at a location is likely to close, or even reverse, the gap between these two prices.
The effect of financial market participants
The ability of purely financial market participants to influence day-ahead and real-time prices through their behaviour can also reduce the cost of serving demand. Each day, the market operator computes prices and quantities for thousands of locations in the transmission network for each hour of the following day, using the offer and bid curves submitted by every supplier and demander (both physical and financial).
Specifically, the system operator minimises the as-offered cost of serving demand at all locations in the transmission network, subject to thousands of transmission network constraints, generation unit capacities and ramp rates, and many other constraints on the operation of the transmission network. This optimisation problem is non-convex for a variety of reasons, including generation unit start-up and no-load costs, ramping constraints, and minimum uptimes and downtimes. Consequently, this problem has many local optima, and the global optimum is unknown.
The expected profit-maximising actions of purely financial market participants can help market operators find lower-cost solutions to this optimisation problem in real time. To understand how, consider the case of a purely financial participant who notices that day-ahead prices are systematically lower than real-time prices at a location. To profit from this price difference, a financial participant would buy ‘virtual’ electricity in the day-ahead market and sell it back in the real-time market. This action increases demand at that location in the day-ahead market, which is likely to increase the day-ahead price, and increases supply in real time, which is likely to reduce the real-time price.
Suppose that the reason for the initial systematically higher real-time price is that the offer of a low-cost, long-start unit is not accepted in the day-ahead market. In real time a more expensive short-start unit must run to serve real-time demand at that location, because the long-start unit is unable to provide this energy with such short notice.
The ‘virtual’ demand bid of the purely financial player causes the lower marginal cost long-start unit to taken in the day-ahead market, which may raise the day-ahead price at that location, but the additional supply of energy from this unit reduces the real-time price and reduces the total cost of serving real-time demand at that location.
It is important to emphasise that the financial participant in this example is not taking these actions to reduce the cost of serving demand. The ‘virtual’ energy purchase and subsequent sale is made is solely to maximise profits from trading in day-ahead and real-time price spreads. Nevertheless, these actions result in day-ahead generation schedules, locational demands and day-ahead prices that are more representative of real-time conditions. In turn, this reduces the real-time cost of serving electricity demand.
Financial trading is most likely to reduce the cost of serving demand when the optimisation problem solved by the market operator has many potentially binding constraints. These physical operating constraints are most likely to bind when electricity demand is high, and the electrical grid separates into many spatially distinct markets due to transmission congestion. Under these conditions, the potential for finding lower-cost solutions is much greater than for lower-demand conditions, when computing prices and dispatch levels would be straightforward.
Is this empirically true?
We examined the empirical validity of these assertions using the introduction of purely financial participation in California’s wholesale electricity market (Jha and Wolak 2019). We were able to show that average price differences between day-ahead and real-time fell for the vast majority of locations in the grid, and the volatility of these price differences fell. This is consistent with the hypothesis that purely financial participants increase the information content of day-ahead prices and liquidity at each location in the grid.
During high demand hours, when we would expect the benefits from financial trading to be highest, we estimate that the introduction of financial trading has resulted in a 3.6% decrease in average fuel costs per MWh of electricity produced from natural gas-fired sources. The annual fuel-cost savings from financial trading in high-demand hours are $23 million.
Financial participants results in less natural gas being used to produce the same amount of electricity. Consequently, financial trading has substantial environmental benefits from reduced greenhouse gas emissions and local pollutants such as nitrous oxides. Our results indicate that the introduction of purely financial participants created price discovery and economic and environmental benefits that accrue both to producers and consumers.
Basu, D, and J Miffre (2013), “Capturing the risk premium of commodity futures: The role of hedging pressure”, Journal of Banking & Finance 37(7): 2652-2664.
Cheng, I-H, and W Xiong (2014), “Financialization of commodity markets”, Annual Review of Financial Economics 6(1): 419-441.
Cox, C C (1976), “Futures trading and market information”, Journal of Political Economy 84(6): 1215-1237.
Gray, R W (1964), “The attack upon potato futures trading in the United States”, Food Research Institute Studies 4.1387-2016-116207: 97-121.
Jha, A, and F A Wolak (2019), “Can Financial Participants Improve Price Discovery and Efficiency in Multi-Settlement Markets with Trading Costs?”, NBER working paper 25851.
Working, H (1953), “Hedging Reconsidered”, American Journal of Agricultural Economics 35(4): 544–561.
 Basu and Miffre (2013) show the existence of a forward market risk premium for 27 commodities spanning five categories: agriculture, energy, livestock, metals, and lumber.
 Cheng and Xiong (2014) for a survey.