In February of 2021, a winter storm brought historically extreme conditions to the state of Texas, in the southern US. The cold and ice had two initial effects: first, it caused an increase in electricity demand for heating, and second, the conditions took a significant fraction of Texas electricity supply, most importantly natural gas, offline.
These first-order effects were already devastating: at least 210 lives were lost. But their consequences then played out through the state's electricity market, which turned out to be ill-adapted to the extreme conditions. Rather than smoothing out the disruption, markets amplified it.
There are some helpful lessons here for thinking about market-making and the role of prices during extreme events.
What happened
Texas's electrical grid is operated by ERCOT, which acts as a market-maker between generators of electricity—power plants, wind farms and solar fields—and retailers, who buy from generators and sell to households. The market is built around highly flexible prices, both wholesale and retail. When demand is high relative to supply, prices rise. This graph of monthly retail electricity prices shows the February freeze:
The wholesale price gives a better sense of the severity of the price spike. This graph shows a 24-hour moving average of the hourly ERCOT wholesale price of electricity at the southern hub. (Similar spikes occurred at all hubs.) The normal price is around $25 per MWh. During the February freeze, the price peaked at $9,000 per MWh, the regulated maximum price.
The graph shows that high prices remained in effect for several days. During that time, some $50 billion in energy purchases were made in the wholesale market, more than a year's worth at usual prices. Household electric bills were similarly high. This was enough to prompt calls to unwind many of the trades.
Adjustment at the outside spread
ERCOT, acting as a market-maker, turned supply and demand shocks relating to physical conditions—cold weather—into prices. That's what the mechanism is for: higher prices, the theory goes, will bring more supply online and reduce demand.
That idea, however, assumes that supply and demand are flexible in a way that was not true in February. Demand was not sensitive to prices, because it was dangerously cold and so people had no choice but to turn on their heating systems. Supply, meanwhile, did not increase in response to high prices. In part, the weather conditions had actually disabled generating capacity: fossil-fuel systems need to be winterized to operate at low temperatures. Where this was not done, the equipment was just turned off instead.
For rising prices to stabilize energy supply, there would have to have been big additional sources that would come online not too far above the normal price—access to neighboring states' grids, for example. In the terms of dealer function, we could say that the market was designed using inside-spread logic, on the idea that price movements would be sufficient to balance flow supply and demand. February 2021 brought outside-spread conditions. But no one was making an outside spread, so the price tried to rise to infinity.
The figure below, which is an illustration and not based on actual data, captures this idea in a simple model. In both panels, prices rise as spare capacity falls. On the left, those price rises are capped at the price at which some other supply becomes available. On the right, there is no such other supply, so prices can rise without limit.
Implication
The price mechanism is only one piece of the Texas electricity story. A full account would also need to address the role of climate change, the political economy of energy, and the regulation of energy production beyond the price mechanism.
But I think there is at least one clear message, which is that counting on prices, and prices alone, to clear markets is dodgy in theory, and poses serious risks in practice.