Alternating current electricity systems require that demand equals supply in real-time. Any supply-demand imbalance must be remedied in minutes to avoid collapse of the system that would take weeks to repair. And the Texas system was very close to collapse.

So what went wrong in Texas? The short answer is that demand was about 69,000 megawatts (and estimates of what demand would have been had supply been available were 74,000 MW) while available supply was about 46,000 megawatts. When supply is unexpectedly reduced, the remaining generators must work harder. If demand is not reduced within minutes the remaining generators will fail. The first demand reduction occurs through contract. Large commercial customers pay less in return for giving the system operator (in this case a non-profit corporation called the Energy Reliability Council of Texas) the right to terminate their electricity use. When this reduction proved insufficient, ERCOT instituted rolling blackouts to remaining customers who had contracted for normal service. This last resort was instituted with less than 5 minutes from the collapse of the Texas electricity system.

Why was available supply so short? The Texas system has two unique features that many have argued were responsible. The first is the lack of alternating current transmission connections to other states to avoid federal regulation. The result is that Texas is (almost) totally reliant on electric generation located within the state. Because Texas is a large state this is not an obvious defect. The New York and New England grid systems have less than half the peak demand of Texas while the Mid-Atlantic and Midwest systems have double and 1.75 times, respectively, the peak demand of Texas.

While I have not yet seen a definitive analysis of the availability of generation reserves in nearby states, interstate connections would have prevented rolling blackouts only if reserves in other states were large enough. Given that nearby states also were cold and had high generation utilization, and the generation shortfall in Texas was so large, interstate connections may not have altered the outcome.

The second unique feature of the Texas electricity system is the absence of governmentally imposed generation capacity requirements. In all states, except Texas, the electricity regulatory agencies require generation capacity to exceed peak demand use (in the previous few years) by an arbitrary administratively determined percentage. Instead, Texas relies only on an auction market in which generators offer power and ERCOT accepts bids until supply equals anticipated demand. The highest price offered that clears the market sets the price for all generators.

At times of peak demand (usually in the summer) prices received by generators can exceed normal prices by a factor of more than 100. These high prices are the sole mechanism Texas uses to induce supply that is only utilized a few weeks a year during the summer peak. The intellectual debate about the adequacy of the Texas system to induce sufficient supply has existed since its inception and many blame the lack of explicit capacity requirements for the Texas meltdown.

What are the comparative results of these two methods of procuring adequate supply? The estimated reserve margin (Table 3) in Texas is the lowest among electricity regions. But this comparison is misleading because the reserves in other regions with explicit excess capacity requirements greatly exceed their own requirements (p. 44). “The three east coast grid operators have set a goal of procuring 13.5% reserve margins, yet the New York, New England, and Pennsylvania-New Jersey-Maryland system operators each have reserve margins that hover around thirty percent.” The 2020 reserve margin (p. x) in Texas was 12.6%, not that different from the planned reserve capacity of the regions that have explicit capacity markets.

If Texas had explicit capacity requirements, they would have been on the order of 15% rather than more than a third. Capacity markets in other regions have been the subject of much criticism for defects in design that result in excess capacity in the 30 percent range, more than double their intention, so one cannot argue a well-designed capacity market would have prevented the blackout because the capacity goals of such systems are much less than the generation loss that occurred in Texas.

So, Texas had available generation capacity of over 83,000 megawatts (Table 3) but only 46,000 was operational. So why was so much Texas supply unavailable? According to a report on a similar winter blackout in 2011, generators were unavailable because of failure to “winterize” their operations. In plain English, the boilers and turbines in Texas are outdoors to avoid having to dissipate the heat during the summer (p. 142). Thus, during infrequent cold snaps, important components freeze and automatically trip sensors that force generating units offline.

Why don’t Texas generators winterize their operations? The 2011 report (pp.179–180) concluded that the costs of winterization (insulation and heat tape) were not large ($50,000 to $500,000), yet generators did not heed the winterization recommendations of the 1989 report following the freeze in that year.

One claim is that the lack of capacity requirements in Texas (and the lower prices that result) serves customers well most of the time with low prices but reduces incentives for generators to invest in precautions against extreme events. But in 2019 in the Pennsylvania New Jersey Maryland region, wholesale electricity prices (p.102) averaged 2.7 cents per kWh and 2.1 cents in the first 9 months of 2020 while in Texas wholesale prices (p.ii) averaged 4.7 cents per kWh in 2019. And retail residential prices in 2019 were 13.12 cents per kWh in Maryland and 11.76 cents per kWh in Texas.

So wholesale prices are actually higher in Texas because of the lack of subsidized capacity, but are the decreased retail residential prices in Texas enough lower to make blackouts a rare but rational cost-benefit outcome? ERCOT commissioned an analysis of the value of lost electricity to ERCOT customers in 2013. It concluded (p.66) the value was $6 a kWh for commercial and industrial customers but only 11 cents per kWh for residential, more or less the current retail price they pay.

Some residential customers in Texas (about 29,000) contracted to pay wholesale prices (5 cents per kWh on average) but up to $9 a kWh during the blackout. Those high prices induced 2/3 of generators to winterize, stay operational, and allow those customers to have their electricity. But the negative reaction of the customers and the Governor of Texas (as well as the time-inconsistent behavior of the retail provider, which told its customers to switch to fixed-rate retail service just before the storm) suggests the limits of exposing residential customers to low prices most of the time and very high prices some of the time in return for uninterrupted electricity.

So the decentralized decisions of more than a third of independent generators in Texas appear to have resulted in inadequate winterization investments. But there is another possibility, the susceptibility of natural gas supply itself to freezing. Water is a byproduct of natural gas production and small amounts of ice can shut down a well in what is termed a “freeze off.” Water from production is also stored at natural gas wells and must be hauled away. Bad weather can stop that. Once water storage capacity is reached, sensors shut down production (p. 9). The analysis of the 2011 cold weather electricity outage (pp. 140–141) concluded that while the majority of generator failures were related to their lack of weatherization efforts, 10 percent of the generator failures and kilowatt-hour reductions were the result of natural gas delivery reductions because of freezing water. It is possible that natural gas supply reductions rather than the lack of generator winterization played a larger role this year.

So in my view three questions require answers. First, were generation reserves large enough in neighboring states so that conventional transmission interconnections available elsewhere in the U.S. would have provided sufficient supply to avoid blackouts? Second, why do a third of Texas generators repeatedly fail to winterize their operations while two-thirds remain operational during cold snaps given that they all operate under the same system of payments for energy produced? Finally, did natural gas supply failures play a larger role in 2021 than in 2011.