Tuesday, August 4, 2009

Electricity market trading strategies

The proposals by the two functioning power exchanges in India to offer term-ahead electricity contracts offers interesting possibilities. However, I am inclined to a model that combines features from both proposals. Designing a market by accounting for the reactions and counter-reactions of the various market participants being an extremely complicated task, this may be treated as an amateur attempt. So comments and counter-proposals are welcome.

Though exchange based trading of power has been in vogue in India for slightly more than a year, they constitute just 0.3% of the total consumed power in the country. Two power trading exchanges - Indian Energy Exchange (IEX) and Power Exchange of India (PXI) - are presently operational, and they trade only day-ahead contracts (for delivery next day). Further, of the total short term power trade, exchanges form just 9%, while bilateral trades (for terms upto one year) form 50% and Unscheduled Interchange (UI) (settlement of real time grid imbalances by frequency-based pricing mechanism) forms 41% (figures from April, 2009).

The fundamental requirements for the success of power trading exchanges, like that of any other trading exchanges, are
1. Adequate quantity of power to be traded (or liquidity)
2. Large enough number of sellers and buyers (so as to ensure competition)
3. Availability of transmission facilities so as to materialize the trades

Unfortunately, the Indian market suffers on all the three counts, with the liquidity problem being the most excruciating. The net result has been a marketplace which is illiquid, inefficient, and vulnerable to anti-competitive practices. Hitherto exchanges have been trading only day-ahead contracts, while the longer term contracts are executed over the counter through licenced traders. In order to improve liquidity in the market and increase its depth, both the power exchanges have put forward proposals for term-ahead forward contracts.

The proposals for price discovery for the Term Ahead Markets (TAM) being proposed by the two power exchanges are as follows

Proposes a two part trading - a double-sided, sealed bid auction followed by continuous matching trades. Every day, both buyers and sellers would simultaneously submit bids in closed covers during specified time period. The bids are then matched to calculate a uniform equilibrium price, which is the price of the last accepted bid. This is a form of Dutch auction. The uniform price nature of the auction would tend to encourage sellers to bid close to their marginal cost of production, and thereby facilitates more efficient price discovery. Once the market is cleared, all the remaining sellers and buyers are pitted in a continuous matching trade where the highest buy order is matched with the lowest sell order.

In the context of a deficit power environment and state utilities faced with demand that cannot be refused (due to political compulsions to provide assured supply), this market design does favor the seller in so far as it leaves the buyer with limited information to base his bid. Any high bid by a desperate buyer will profit the seller and create a "winners curse" for the buyer.

The continuous matching bid process, where price discovery is likely to be more efficient, is a welcome feature. However, this can also encourage both sellers and buyers to hold out during the double-sided closed-bid auction, in the hope of getting a better deal during the continuous matching. Safe in the assurance that they will have another opportunity, sellers will have no incentive to quote close to their marginal cost. Similarly, in their desperation to close their deficit and the uncertainty surrounding the availability of power after the first round auction, buyers may end up being forced into quoting higher prices to grab as much as they can.

Proposes to have separate sell and buy sessions. First, the sellers simultaneously place their closed bids during a specified time period. Then the sell orders are opened and disclosed and the buyers use this information to simultaneously place their bids in closed covers. The sell bids are then opened and the lowest priced sellers are matched with the highest priced buyers and the seller is paid the price bid by the buyer.

In regular pay-as-bid auctions, the lowest bidding seller is matched to the lowest bidding buyer at the market clearing price, and so-on. This incentivizes sellers to bid as close to the market clearing price as possible, irrespective of their marginal cost of production. The variation of price matching proposed by the IEX seeks to minimize this incentive distortion by reducing the incentive for sellers to guess the market clearing price and thereby bid accordingly (as in the regular pay-as-bid auctions). This cycle is then repeated in a second matching run where first the sellers and then the buyers modify their original bid prices.

This market design is more favorable for the buyers, in so far as they now have some information to base their bids. Sellers also have an incentive to quote close to their marginal cost given the fact that buyers can see all sale offer prices before making their own bids. This in turn dis-incentivizes competing sellers from pricing themselves out with very high bids and also encourages them to bid closer to their marginal cost to maximize their chances of winning the bid.

The biggest fault with this market design is that the entire market surplus is cornered by the sellers/producers. As can be seen from Figure 1, in a uniform, marginal pricing arrangement (as done by the auction in IEX), both the buyers (who were willing to pay higher than the market clearing price) and the sellers (who were willing to supply at lower than market clearing price) enjoy their respective surpluses as indicated in the graphic.
Figure 1

However, in the IEX model in Figure 2, by paying the lowest bidding seller the price agreed by the highest bidding buyer, the entire surplus gets transferred to the seller, and so on. In other words, the buyer/consumer surplus will also accrue to the sellers.
Figure 2

The two critical issues are the auction design which facilitates price discovery and the actual price fixation. So how about a market design that takes in the first from PXI and the second from the proposal of IEX? In other words, separate closed bid sell and buy sessions and the price being fixed at a uniform market clearing price. A continuous matching process may be a better alternative to clear the residual that does not get cleared in the first round of auction.

It however remains to be seen as to the extent of impact the deficits and the needs of the state utilities exerts on the market participants. If the market is illiquid and sellers perceive desperation on the part of buyers to get power at any cost to meet their local political compulsions, then any market design will always be vulnerable to being gamed. Such signals sent by the buyers and received by the sellers can have a feedback effect that amplifies the signals and leaves the market even more distorted in favor of sellers. Incidentally, given this eventuality, a simple uniform price bid with continuous matching trade to clear the residual is the best alternative.


Unschedule Interchange (UI) is a frequency-linked real-time pricing mechanism to operate and settle the electricity demand-supply imbalances in the grid at any point in time. The grid imbalances at any time are transacted as effective spot transactions for immediate delivery, with the prices determined by the grid frequency. This arrangement has been in successful operation for sometime now to settle the imbalances in the Central Generating Stations (CGS) pool. The frequency-linked rates are determined by the CERC.

No comments: