Does arbitrage always improve market efficiency? Theory and evidence from sequential markets for electricity

Since the 1990s, many countries have deregulated their electricity markets. Electricity producers and distributors participate in auctions in forward and spot markets, which determine production allocation and wholesale prices.

A key policy question for the United States and the rest of the world is whether financial traders should be allowed to participate in the auctions to arbitrage differences between forward and spot prices. As it stands now, such traders – who do not have the physical capacity to generate or distribute electricity – are prohibited from many markets. However, that could change now that markets, such as those in California and New York, have begun to allow financial traders.

Does arbitrage benefit consumers? Does it lead to more efficient allocation of production resources?

In Ito and Reguant (2016), we address those questions from theoretical and empirical perspectives by examining the Iberian electricity market.

The study provides three key findings. First, in the presence of dominant players and limited participation in the arbitrage market, a forward-market price premium arises. By initially curtailing supply, large firms can flex their market power to drive up prices in the forward market.

Second, we find that some players, particularly wind farms, arbitrage some of the forward-market price premium, but they do not eliminate it entirely.

This leads to a policy-relevant question: Does creating full arbitrage improve social welfare? Our third finding shows that the answer is not straightforward for most markets, which tend to be under the sway of dominant players. Consumers do benefit, but not as much as when unfettered competition leads to the market clearing at the lowest price and highest production levels. Firms with market power anticipate arbitrage and reduce their output in equilibrium.

Introducing competitive arbitrage removes the price premium, but does not necessarily lead to competitive prices.

In other words, introducing competitive arbitrage removes the price premium, but does not necessarily lead to competitive prices. While price convergence between the forward and the spot market is a sign of a healthy arbitrage market, it should not be interpreted as a sign of a competitive product market.

Forward and spot markets

Many items such as treasury bonds, stocks, coal, electricity, natural gas, oil, and agricultural products are traded through sequential – forward and spot – markets. Sequential markets can promote more efficient allocation by allowing adjustments to changing demand and supply conditions. In an oligopolistic setting, sequential markets can also help reduce the extent of market power, which also improves efficiency. By creating multiple instances in which firms compete over the allocation of the same good, sequential markets reduce firms’ strategic incentives to withhold their output (Allaz and Vila 1993).

In the Iberian market, for example, electricity producers (e.g., thermal power plants) and consumers (e.g., retail electricity providers) first bid in a forward market that clears demand and supply of electricity for the following day. Intraday (spot) markets open after the day-ahead market, allowing firms to update their production decisions. For example, as shown in Figure 1, firms have seven intra-day markets to update their positions for electricity due for delivery in hour 21.

Figure 1: Forward and spot markets in the Iberian electricity market

Ito, chart

This table shows how market participants can update their positions in the forward and spot markets. The vertical axis delineates the transaction time of each market, and the horizontal axis the hour of energy delivery at day t. For example, the day-ahead market closes at 10 a.m. in day t-1 (the day before the energy delivery). Firms have multiple chances to update their positions through the subsequent intra-day markets before their final position is determined.

Standard theory vs. the real world

In an ideal environment, forward and the spot market prices should converge on average (Weber 1981). In the presence of systematic price premium, firms would have an incentive to arbitrage it by selling in the market with a high price and buying back in the market with a low price. This behavior would eventually close the gap between the prices in each market. In real-world situations, however, prices differences persist in sequential markets. In electricity markets, forward market prices tend to be higher.

In Figure 2, we use hourly data from 2010 to 2012 from the Iberian electricity market to show each market’s average price by hour. Indeed, we observe a systematic “day-ahead price premium” in the Iberian electricity market, particularly for those hours of high demand.

Figure 2: Forward-market price premiums in the data

Electricity 1

This graph shows average electricity prices at each hour of the day (1-24) for different markets: the day-ahead (forward) market and each of the subsequent intra-day (spot) markets.

Market power and limited arbitrage are key 

Why don’t the Iberian and other electricity market obey economic theory? Firms with sufficient market power can boost day-ahead prices by holding back on supply. They can sell some of their excess on the intra-day market, although prices then tend to be lower. But so long as they can make the bulk of their sales in the forward market, they can maximize profits.

Why don’t smaller firms exploit the price differential between the forward and the spot market, by offering more output in the forward market and buying it back later at lower prices?

Still, why don’t smaller firms exploit the price differential between the forward and the spot market, by offering more output in the forward market and buying it back later at lower prices? Such price premiums cannot be fully arbitraged because financial traders are not allowed to participate in the market, and arbitrage by existing market participants is constrained. For example, firms typically are restricted from selling more than their production capacity; therefore, they cannot oversell in the forward market, even if higher forward prices make such positions attractive. Meanwhile, smaller companies might not have the capacity to exploit arbitrage opportunities.

But do any players arbitrage the price premium? 

Even though financial traders (also known as virtual bidders) are not allowed to participate in the Iberian electricity market, some sellers and buyers might arbitrage. In our paper, we show that wind farms are particularly active arbitrageurs in this market; they often sell much more wind than forecasted in the day-ahead market, buying it back in the intra-day market.

Why are wind farms acting as de-facto arbitrageurs? First, wind farms have some technological advantages that make them quite attractive to arbitrage. They almost never use their maximum capacity because wind does not blow all the time. Second, because wind is difficult to predict, regulators find it tougher to determine when wind producers are overselling in the forward market. Thus, overcommitted wind producers can still meet demand – and make a profit − by buying in the lower-priced spot market.

Size still matters when it comes to wind farms. While smaller producers would have a strong incentive to capitalize on price differences between the forward and spot markets, those owned by larger companies would rather see high forward prices and forgo arbitrage opportunities.

Our empirical findings confirm these theoretical predictions.

To arbitrage or not to arbitrage 

Should arbitrage be encouraged in electricity markets? To shed light on this question, we complement the analysis described above with a structural model of the Iberian electricity market. We use this model to simulate market outcomes with and without arbitrage.

As shown in Table 1, we use as a benchmark the results for “first best,” namely the outcome from a perfectly competitive market. We then compare this benchmark to a few counterfactual situations.

First, we eliminate the forward market, meaning trading occurs only on the spot market. Average prices would be substantially higher: 46.5 euros per megawatt hour vs. the benchmark of 38.2 euros. That reflects dominant players exercising their market power by withholding supply, lowering their average production to 12.8 gigawatt hours, as compared with 15.3 GWh for first best.

Second, we introduce a forward market, but assume that there is no arbitrage. We observe that this change significantly cuts deadweight loss, by reducing firms’ market power. Our finding conforms with a theory developed by Allaz and Vila (1993) that posits that creating sequential markets reduces market power. Their theory predicts that having sequential markets makes dominant players in a forward market compete with themselves and others in a spot market.

In our case, as the table shows, firms end up selling in the spot market much of what they withheld in the forward market, bringing final output quite close to the first best scenario (14.9 GWh). Our intuition is that because the spot market is much smaller than the forward market, firms have less incentive to withhold remaining supply. Thus, the allocation of production between dominant and non-dominant firms is more efficient, and quite close to the first best. Consumer payments, however, are much larger than they were in the first-best due to higher prices in the forward market.

Arbitrage increases deadweight loss because firms that have market power produce less (14 GWh vs. 14.9 GWh). Anticipating the effects of arbitrage, dominant firms withhold their output more aggressively in the spot market. These simulations exemplify the trade-off between consumer payments (reduced by arbitrage) and deadweight-loss (increased by arbitrage).

Third, we introduce competitive arbitrage, which leads to full convergence of forward and spot prices. Compared to the no-arbitrage scenario, prices in the forward market fall from 45.1 euro/MWh to 42.5 Euro/MWh, while prices in the spot market increase. Such price convergence benefits consumers, as demand mostly pays the forward price. However, arbitrage increases deadweight loss because firms that have market power produce less (14 GWh vs. 14.9 GWh). Anticipating the effects of arbitrage, dominant firms withhold their output more aggressively in the spot market. These simulations exemplify the trade-off between consumer payments (reduced by arbitrage) and deadweight-loss (increased by arbitrage).

Table 1: Hourly welfare comparison across counterfactuals

ITO123

This table highlights some of the results of the structural analysis in Ito and Reguant (2016). We use data from the Iberian electricity market to build a structural model that enables us to compare counterfactual scenarios with and without arbitrage.

Our results indicate that arbitrage by financial traders is likely to benefit consumers by lowering electricity prices. However, market efficiency could suffer, as large companies reduce output to maximize profits. Additionally, while prices between markets converge due to liberalized arbitrage, the resulting prices are still substantially above the first best. Indeed, price convergence is not a sufficient statistic for market efficiency.

This cautionary tale should be considered when regulators allow arbitrage in a sequential market dominated by large firms, which is typically the case when electricity is the product.

Further reading

Ours is among several recent papers that investigate questions related to forward and spot markets for electricity around the globe. Saravia (2003) analyzes a policy change in the New York electricity market to show the effect of financial arbitrage. For the California electricity market, Borenstein, Bushnell, Knittel, and Wolfram (2008) examine price premiums and market inefficiency, and Jha and Wolak (2015) develop an econometric method to estimate transactions costs from observed price premiums. Birge, Hortaçsu, Mercadal, and Pavlin (2017) and Mercadal (2015) study the role of financial trading by examining the Midwest electricity market.

References:

Allaz, Blaise, and Jean-Luc Vila. 1993. “Cournot Competition, Forward Markets and Efficiency.” Journal of Economic Theory 59 (1): 1–16.

Birge, John, Ali Hortaçsu, Ignacia Mercadal, and Michael Pavlin. 2017. “Limits to Arbitrage in Electricity Markets: A Case Study of MISO.” https://www.dropbox.com/s/6nlpu8251wlzlgp/Virtuals_FTR.pdf?dl=0 (accessed March 22, 2017).

Borenstein, Severin, James Bushnell, Christopher R. Knittel, and Catherine Wolfram. 2008. “Inefficiencies and Market Power in Financial Arbitrage: A Study of California’s Electricity Markets.” Journal of Industrial Economics 56 (2): 347–78.

Mercadal, Ignacia. 2015. “Competition and Arbitrage in Electricity Markets: The role of financial traders” http://home.uchicago.edu/~ignaciamercadal/research.html (accessed January 19, 2017).

Ito, Koichiro and Mar Reguant. 2016. ”Sequential Markets, Market Power, and Arbitrage.” American Economic Review, 106(7): 1921-57.

Jha, Akshaya, and Frank A. Wolak. 2015. “Testing for Market Efficiency with Transactions Costs: An Application to Convergence Bidding in Wholesale Electricity Markets.” https://web.stanford.edu/group/fwolak/cgi-bin/sites/default/files/CAISO_VB_draft_VNBER_final_V2.pdf (accessed March 22, 2017).

Saravia, Celeste. 2003. “Speculative Trading and Market Performance: The Effect of Arbitrageurs on Efficiency and Market Power in the New York Electricity Market.” University of California, Berkeley CSEM Working Paper 121.

Weber, Robert J. 1981. “Multiple-Object Auctions.” In Auctions, Bidding, and Contracting: Uses and Theory, edited by Richard Englebrecht-Wiggans, Martin Shubik, and Robert M. Stark, 165–91. New York: New York University Press.