Winter storm Uri crushed the Lone Star State a few weeks ago and brought chaos, destruction, and senseless loss of life. In its aftermath the state is trying to discern precisely why it occurred and how to prevent such a disaster from repeating.
One way to interpret what happened in Texas is that it was the result of a not-uncommon statistical mistake.
Entities must decide whether to take precautions against extreme and—presumably—rare events, but they need to get a sense for the chance that such extreme events may occur. When people try to forecast dangerous precipitation and temperature extremes in future years they often use historical data. A plot of weather data—whether it is annual rainfall, temperature highs or lows or snowfall—invariably forms a bell-shaped distribution. Because normal distributions appear to be bell shaped and are also very tractable, people have described these weather phenomena as following normal distributions.
The problem is that weather does not actually conform to a normal distribution: it tends to be lepto-kurtotic, which means that extreme events happen more often than a normal distribution would account for.
At one time Egypt’s government assumed that a rainfall event four standard deviations away from the mean had a probability of occurring of just .01 percent—once in 10,000 observations—and dismissed that as being unlikely to occur. But the actual tail-event probability was, in fact, much higher: The assumption of normality created a sense of complacency, and people died because of the resulting inaction.
Likewise, Texas authorities had assumed that a 50 inch rainfall in Houston or sustained temperatures approaching zero fahrenheit across the state for several days were improbable as well.
Who takes precautions against black swan events? Ideally, it should be entities that plan on being around for an extended period of time, like governments. However, these often fail us too: Governments may be long-lasting but politicians are not, and that they often have an incentive to make decisions that are expedient for the next election.
In the aftermath of the crisis I observed that Texas is—or was—a pure power market, and utilities buy their power on the open market and produce little themselves. The rationale for this was that because prices spike on days of high demand, this creates an incentive to ensure that producers act so that there will be sufficient supply. And to a large extent that has been true.
But the market did not provide sufficient incentive to winterize gas pipelines, windmills, and the like, and with several producers offline to do seasonal maintenance to prepare for what is normally the high-demand season in Texas—summer, with its attendant air conditioning needs—the situation was ripe for just such a disaster to occur.
In the aftermath of the crisis a variety of proposals are being discussed to ensure that such a disaster never occurs again, and it is likely that whatever does come out of this political and economic debate, a repeat of this particular crisis will not occur again.
However, no post-mortem will convey perfect foresight for either the government or any of the market participants, so it’s conceivable that there will be other currently unforeseen crises that may lead to other problems in the future.
The outcome of any legislation passed by Texas should give the market participants the incentive to conceive of and prepare for future crises and ensure that extended power losses become less likely to occur.
One way to be certain that these tail risks get acknowledged by the market would be to have some degree of vertical integration so that the utilities that deliver the energy are also providing a good deal of the energy as well. This would lessen the moral hazard problem present in the current arrangement and lead to utilities internalizing more of the risk of future weather exigencies and taking appropriate precautions. Such an arrangement would also undoubtedly result in utilities paying producers not just to provide energy but also to maintain excess capacity as well.
The Texas debacle should give pause to other electricity markets that are considering the adoption of a model that resembles the current Texas one that completely separates utilities and energy producers.
Everyone in Texas agrees that steps should be made to prevent another widespread, extended power disruption, but those actions should result in creating a market where everyone has an incentive to take appropriate action, and not just direct producers to winterize production or coordinate shutdown activities across one another.
A modicum of vertical integration would create the environment to ensure that utilities approach the market with appropriate foresight.