President Trump congratulates Senate Majority Leader Mitch McConnell, while House Speaker Paul D. Ryan watches, after the final passage of tax overhaul legislation. (Manuel Balce Ceneta/AP)

The Republican Congress’s 2017 agenda was remarkably unpopular. By the time Congress took its final votes on the Affordable Care Act repeal, which failed, and its tax reform bill, which passed, Americans’ opinions of both pieces of legislation were at historic lows.

Why would the party pursue such unpopular policies? One reason is that Republican lawmakers believed in the policies, however unpopular they might have been. Another seems to be that they were worried about backlash from conservative donors if they failed.

But in comments to GOP senators in July, President Trump suggested another motivation — that Republicans would look like “dopes” if they failed to repeal Obamacare. “How can we not do this after promising it for years?” the president reportedly said.

Throwing good money after bad, and so on

In those remarks, Trump seemed to be showing a decision-making bias known as “the sunk cost fallacy.” This occurs when people pursue a questionable course of action not because it’s the prudent thing to do, but simply because they’ve already invested time, resources or energy into it. It’s like sitting through a terrible two-hour movie just because you paid for the ticket.

In a paper published in the American Political Science Review, we (along with Stefaan Walgrave, Stuart Soroka and Tamir Sheafer) show that politicians are just as susceptible to the sunk-cost fallacy and other decision-making biases as are regular folks — and sometimes even more so. That finding can improve our understanding of why governments sometimes make questionable decisions, such as pursuing infrastructure projects long after they are exposed as obsolete or escalating commitment to armed conflicts in which they have suffered many casualties.

How we did our research

We drew this conclusion from face-to-face interviews we conducted with several hundred incumbent lawmakers in Belgium, Canada and Israel as part of a project called Infopol. During these hour-long sessions, we asked the participating politicians to fill out surveys on tablets and laptops that we handed to them, presenting them with decision-making experiments. At the same time, we also had several thousand citizens in these countries answer the same questions, serving as a comparison group. All of the questions we posed to politicians in these experiments were about policy. They dealt with budgetary decisions, government loans, building community centers and responding to public health crises. In short, they looked a lot like the hundreds and thousands of issues on which politicians make judgments each year.

To evaluate how these politicians deal with sunk costs, we presented them with a scenario in which an underperforming small business loan program run by the government is being considered for an extension and further budgetary commitment, although it hasn’t recovered its original investment. We experimentally varied the size of the budgetary shortfall — the decision’s sunk cost. We also varied the implied level of political accountability for the task — meaning that we either mentioned or didn’t mention that the media is asking to know the respondent’s decision, and that this is happening a month before an upcoming election. We then asked respondents to indicate, given what they knew, whether they would vote for the proposed extension and the additional investment.

Politicians are as susceptible to cognitive bias as ordinary citizens — and sometimes more 

We find that politicians show a significantly stronger pull toward the sunk-cost fallacy than do ordinary citizens. They agree to extend programs that have incurred such costs far more often than citizens do; larger sunk costs do not deter them from doing so.

We also find that politicians are no better than citizens in other ways. For example, they adhere more to policy choices that are presented as the status quo. They are equally susceptible to “framing effects,” meaning that they change their  policy choices depending on whether the question is phrased in ways that emphasize the potential gains or the potential losses, even if the choices don’t change. And they discount the future at least as steeply as citizens. In other words, given similar policy outcomes or budgetary payoffs, they strongly prefer those that are offered now over those that require waiting, especially if they are told that there will be elections during the wait.

In many ways, these patterns are surprising. Politicians have expert advisers who can help them direct their thinking. They regularly evaluate ongoing and prospective programs and are encouraged to consider cost-benefit analysis. They are, in theory at least, sanctioned for making poor decisions. But this does not seem to prevent them from stumbling into the same kinds of cognitive biases as the rest of us.

Our findings are part of a growing body of experimental work that focuses on politically critical judgment errors by elites. These include work such as Dan Butler and Adam Dynes’s 2016 study of American state and local politicians; they find that representatives systematically discount the opinions of constituents with whom they disagree, even though they should be strongly motivated to avoid such bias, as those citizens will have a say in the politicians’ performance review at the polls.

What does this mean for the current Republican Congress?

Does this tell us anything us about 2017’s congressional politics? It might be that elected officials are simply less likely than non-politicians to back away from a course of action that they have already invested time and resources in, even if additional information emerges that should give them pause.

Yes, other reasons lay behind the GOP’s legislative agenda choices in 2017; but the tendency to chase “sunk costs” illustrates a more complex story on the role that psychological drivers have in outcomes that we normally ascribe to cold interest-based calculus.

Perhaps when citizens elect representatives we should ask ourselves what reasoning skills those candidates bring with them to the table, and what kind of outcomes they are likely to produce as a result.

Lior Sheffer is a PhD candidate in political science at the University of Toronto.

Peter John Loewen is the director of the School of Public Policy and Governance and an associate professor of political science at the University of Toronto.