Competitive power market in Texas faces supply concerns. Now what?

Michael Giberson

The question troubling some folks in Texas’s competitive power market: Will Texas consumers want to consume more electric power than suppliers are able to supply? A resource adequacy review by ERCOT, the power system and market operator for most of the state, suggests that consumer demand may outstrip resources available as early as 2014. ERCOT officials have also warned that extreme temperatures this summer could result in reliability concerns, though the most recent review reveals resources will likely be adequate.

The longer-term resource review has attracted a number of media reports, including this morning’s story by Rebecca Smith in the Wall Street Journal, “Power Shortage Vexes Texas: Report Urges Price Increase to Spur Industry to Build More Generating Plants.” See links to other stories at the end of this post.

The “report urging price increases” is that of the Brattle Group, “ERCOT Investment Incentives and Resource Adequacy,” June 1, 2012. ERCOT asked Brattle to study generator investment criteria, the connections between incentives, investments, and resource adequacy, and policy options to support resource adequacy. The Brattle report will bear further study, but for now a few comments about it and the WSJ article.

The newspaper story, following the main thrust of ERCOT’s request and therefore the main part of Brattle’s response, is focused almost entirely on price incentives to potential investors in additional generation resources. The story mentions several of the relevant factors: demand growth, low power prices due to low natural gas prices, ERCOT’s “energy-only” market design, and the lack of significant connections to neighboring grids. The rest of the story plays out as expected: generators say the current offer cap is too low and consumer representatives express horror at the prospect of paying extreme prices to generators who might refuse to expand.  The story entirely misses the possibility that consumers are not complete idiots willing to sit idly by in their air-conditioned palaces and pay 100 times the usual power prices.

Consumers have two easy ways of avoiding any potential $9,000 MWh price: (1) have a fixed price contract with a retailer or (2) simply cut power consumption during pricing peaks. Few consumers actually paid $3,000 MWh last year during February 2011’s few hours of rolling blackouts or the summer’s infrequent emergency conditions. Instead what happened in February and summer 2011 is that retailers who did not secure all of the power their customers wanted by short- or long-term contracts ended up paying the $3,000 price (but just for the additional supplies they needed) AND power generators under contract to supply power who found themselves unable to meet their commitments also ended up paying the $3,000 price (for any committed capacity that they could not deliver). The market risks are divided up between retailers and generators and very little of it is pushed out directly onto the consumer.

Obviously, whatever risks generators take on will be reflected in the prices they’ll seek in contracts with retailers, and whatever risks retailers take on will be reflected in the prices that retailers offer to consumers. But competition among generators to contract with retailers and competition among retailers to sell to consumers should work to do well one thing that the usual rate-regulated monopoly power systems do poorly: competition should shift risks onto the market participant who can most efficiently manage the risks. Consumers typically are not the best able to handle the risks, so competitive markets usually won’t stick them with the risks.

The Brattle report makes a couple of additional valuable points. First, the study assumes only the current level of demand response activity, but additional price-responsiveness on the consumer side of the market would provide additional resource adequacy support. Second, the “1-in-10” reliability standard typically employed in power systems reliability analyses has rarely been studied from an economic standpoint. The report suggests that overall reliability of delivered power to consumers could be improved and costs reduced by shifting some of the expense away from the bulk power system and toward distribution systems.

So far as I have noticed, the report itself doesn’t recommend a particular policy course, but simply reports on some of the likely advantages and disadvantages of several resource adequacy policy options. The Brattle press release accompanying the report does, however, indicate a clear preference for adding a centralized forward capacity market (similar to that employed by PJM; though note not everyone is happy with PJM’s capacity market).

One last bit of perspective. It is the goal of a resource adequacy study to be excessively cautious. Things probably will not turn out as bad as projected, in part because suppliers, retailers, and consumers will continue to adjust to changing conditions.  But things could be as bad as projected, and that is exactly what the study is intended to highlight.


NOTE: Prices above are all quoted in $ per Megawatt Hour (MWh), a typical price metric for wholesale markets, but consumer bills are usually quoted in cents per kilowatt hour (kwh). Typical wholesale prices in ERCOT have been running between $20 and $50 MWh, the equivalent of between 2 and 5 cents kwh. Typical consumer prices in ERCOT range between 8 and 14 cents kwh. The $3,000 MWh price cap is equal to $3 kwh (so $9,000 MWh is the same as $9 kwh or about 100 times  typical retail prices).


5 thoughts on “Competitive power market in Texas faces supply concerns. Now what?

  1. Consumers have two easy ways of avoiding any potential $9,000 MWh price: (1) have a fixed price contract with a retailer or (2) simply cut power consumption during pricing peaks

    As far as I am aware, no available contracts based on ERCOT 15-min or 1-hr LMP pricing have
    been offered by REPs on Very few block type pricing ( peak summer hours
    and then all other times ) contracts are even offered. Thus consumers are never incentivized to cut
    power as in (2),

    Obviously, whatever risks generators take on will be reflected in the prices they’ll seek in contracts with retailers, and whatever risks retailers take on will be reflected in the prices that retailers offer to consumers.

    This risk premium can be large as the risk mindset for marketers/traders is ‘better to be long summer
    peak power and be wrong, then short summer peak power and be fired’

    Competition should shift risks onto the market participant who can most efficiently manage the risks.

    In the aggregate and long-run, the above is true. In the short-run and at the margin, competition
    shifts risk onto the market participant who underprices the risk the most ( or charges the smallest
    excessive premium ). With low barriers to entry in the REP market in ERCOT, one could argue that
    the market clearing price is continually depressed by the imprudence of soon-to-blow-up entities.

  2. True that entrepreneurial errors, or simple gambles, could be depressing retail prices. As long as the retailer bears most of the costs of the risks taken on, then over time they should get better and the costs to others are minimized. (If nothing else, new imprudent potential retailers would be chastened by bankers when they showed up to borrow money on an under-capitalized risky business plan.)

  3. Ah, red meat!  

    “1 day in 10 years” is a reliability criterion the same way “9-9-9” was a tax plan. They’re both catchy slogans, easy to remember. 

    1d/10y originated in the 40s, when statistical analysis of generator availability was first developed. It was based on a single peak hour of every day. The calculations were done on a system that had exhibited what I’ll call “pretty good reliability” (PGR), and the LOLP came out to roughly an average rate of 1 day in 10 years. Ergo, PGR = 1d/10y. 

    Of course, by today’s standards, the methods in the 40s were quite primitive. Today’s models are chronological, they respect unit commitment constraints, and some of them have proper respect for imperfect foreknowledge, the dispatch of energy-limited resources, and the possibility of dispatcher and/or forecaster error. They are beginning to be combined with transmission models that respect real transmission constraints. There are sophisticated methods for dealing with weather uncertainty, but note the plural.

    There are no standard methods, no standard model, but the standard bogey is still 1d/10y. I wonder what today’s model would make of the 1940s system that exhibited PGR = 1d/10y by primitive methods. It’s laughable that the slogan has had such staying power, while we have no idea how different today’s models evaluate reliability compared with those of yesteryear. I think it would be very useful to have a complex test system with 30 years worth of load and weather data, against which to compare models and methods. Those are different things… models and methods. 

    An economic evaluation of 1d/10y is pointless. Rather, there have been analyses to establish reliability levels that are in economic balance, according to a list of assumptions, methods, models, etc. They are also particular to specific systems with their specific resources and load shapes. They generally render reasonable reserve margins. They are not anchored to 1d/10y! 

    I’ve probably written enough to blow my cover here.  😉 

  4. D.O.U.G., sloganeering aside, the Brattle report suggests that reliability to consumers should be improved by a shift of resources away from central dispatch/resource adequacy type concerns and toward distribution system concerns. Do you agree or not?

  5. The suggestion is that there is over-investment in generation and under-investment in distribution with respect to the cost of interruptions. Certainly there are places on distribution systems where the marginal value of a dollar invested in the distribution system is greater than the marginal value of the dollar otherwise invested in generation. It is important to estimate the costs and values in either case, and both are dependent on the state of over/under investment in the given system. Of course, the value of distribution investments is very much a local matter, and is much more case specific than for investments at the generation level (not that they are substitutes).

Comments are closed.