On October 1, the federal government shut down—halting non-essential services after the failure of Republicans and Democrats to pass an appropriations bill. Even after becoming the longest government shutdown in American history, the end was nowhere in sight. It may seem that both parties would want to avoid shutdowns, especially ones lasting that long, to accomplish at least some elements of their agendas rather than waste weeks of a legislative session. However, this is not the case when certain incentives are at play, including the incentives to pressure, misrepresent, and engage in costly signaling of resolve.
The latest government shutdown can be analyzed using game theory, specifically by conceptualizing it as a war of attrition game over the extension or non-extension of Affordable Care Act (ACA) subsidies. (I am indebted to Dr. Lucas Engelhardt for suggesting this interpretation.) Republicans wanted to end the shutdown without extending ACA subsidies, while Democrats wanted to end the shutdown with the subsidies extended. Both parties wanted to incur the lowest possible political cost from waiting for the other party to make an acceptable offer to end the shutdown. This political cost can be thought of as primarily the expected loss of support for a party from voters and donors.
Government shutdowns, like wars, occur when neither party agrees on what the outcome of the conflict will be. If both parties had agreed before October 1 on the number of ACA subsidy extensions that would ultimately resolve the shutdown, they would have immediately made a deal to achieve that outcome while avoiding any of the costs of fighting. For such an immediate resolution to have occurred, both parties would have needed to reveal everything they knew and agree on the deal they genuinely believed would have ended the shutdown. Or, one party would have needed to be so effective at misrepresenting its resolve as stronger than the other’s, that the other party immediately conceded.
Parties may distrust each other, making agreement more difficult, since each may believe that when the other party is claiming stronger resolve, this is a calculated exaggeration. The Republicans may recognize that if they can convince the Democrats that their resolve is stronger, the Democrats will be persuaded to concede a better outcome to them. The Democrats may anticipate this attempted persuasion by Republicans, and seek to do the same. When neither party can agree on what the outcome of the shutdown would be, each is incentivized to prove their resolve by enduring an actual shutdown.
This shutdown began with both parties expecting to have the stronger resolve, and lasted until both parties were convinced that the Republicans had it. Each party was trying to weaken the other’s resolve while convincing them that its own was stronger. Both parties tried to build a reputation for stubbornness, with Republicans claiming that any extension to ACA subsidies was unacceptable, and Democrats claiming that any refusal to extend the subsidies was unacceptable. The more that one party can convince the other of its willingness to suffer an indefinite shutdown rather than compromise, the better the concession it can extract, and the sooner it can end a shutdown.
Beneath the mutual deception, the actual resolve of each party still matters. Presumably, a party’s resolve is largely based on whether it expects that prolonging the shutdown will harm support primarily for itself or the other party. Conditions may constantly change such that one day, prolonging costs Republicans more, and the next day, prolonging costs Democrats more. During a shutdown, it is not particularly useful for us to attempt to predict either party’s resolve, since we cannot know exactly how they perceive their support or how they compare the other party’s resolve to their own. We cannot know which party views the other as more stubborn, or whether either is actually stubborn. We cannot even know whether both parties are fighting for a better outcome than total failure. For instance, in the latest shutdown, the Democrats may have believed from the beginning that they would eventually have to agree to fund the government without any extension of ACA subsidies, and were merely trying to save face.
We are not alone in this poverty of prediction. Even though it would be useful for Democrats and Republicans to accurately estimate each others’ resolve, they cannot. During the latest shutdown, neither party was expertly predicting each others’ strategies based on facts, both were making bets based on incomplete data. For instance, polls consistently showed that the general public blamed the Republicans more for the shutdown. This suggested that the cost of prolonging the shutdown was higher for the Republicans than for the Democrats, presumably weakening Republican resolve. Another poll showed that a much higher percentage of adults identifying as Democrats (92 percent) supported the extension of the ACA subsidies than the percentage of adults identifying as MAGA Republicans (43 percent) who opposed their extension. This suggested that the benefit to the Democrats of prolonging the shutdown was higher than the benefit to the Republicans, presumably strengthening Democratic resolve.
At the same time, Democrats seemed to be more concerned than Republicans with the furloughing of non-essential federal workers and the possible delays in SNAP benefits, being lobbied both by the largest federal worker union and by nutritionists to end the shutdown. That suggested the cost of prolonging the shutdown was higher for Democrats, presumably weakening their resolve. Both parties were arbitrarily weighing these conflicting factors to assess their own support and the support for the other party. They were making similar judgments to gauge the other party’s stubbornness relative to their own.
What the most recent shutdown showed, and what the theory of war-of-attrition games predicted, is that when both parties are unsure of which has the stronger resolve, each is incentivized to initiate a shutdown to engage in the costly process of finding out. By prolonging the shutdown, both parties gained valuable information on how long the other party could endure while facing numerous political costs. In this absurd yet usual state of affairs, the government is incentivized to go on strike against itself.
Meanwhile, the American people bear the costs. The government gambles with taxpayer money on political victories while denying citizens a slew of services that the free market arguably could have provided, such as charitable food assistance, grants for research, healthcare, and education, and paid air traffic controllers. It is a special cruelty to force taxpayers to fund government services, and then refuse to provide those services, all while insisting that the market would not have provided enough of them to benefit society. These services are likely provided more reliably when subject to consumer demand rather than petty politics. When the government becomes schizophrenic, it is difficult to envision it advancing orderliness.
The government failed to govern, but not by mistake, it was fully incentivized to do so. It strategized against itself, and did so with our involuntary financial support. To politicians, this was all just a game, but we were not players.
