People respond to incentives. With some generous interpretation, this simple sentence summarizes the bulk of what you’ll learn in any good undergraduate economics course. The challenge, of course, is to realize that this is always true even when we might want it to not be.

For example, imagine for a minute that you’re in a position where you have to make decisions. For your decisions, there are three options: you can get the decision right or you can get it wrong. “But Dave,” you say, “that’s only two options!” In the following nuance lies the heart of this piece: there are two ways in which you can be wrong. You can act when you shouldn’t have or you can fail to act when you should have.

Getting things right all the time is not possible. So which type of mistake are you more likely to guard against? It depends on which one will get you fired.

In most government settings, the answer is biased in particular (and predictable) ways. This is what economists call asymmetric accountability and it occurs when the consequences of one type of error are visible, attributable, and (potentially) career-ending, while the consequences of the other are diffuse, invisible, and nobody’s fault in particular. These incentives lead government agencies to guard against acting wrongly more strongly than they do against failing to act when action could help. The resulting problems aren’t the result of incompetence, but of rational self-protection. We see this routinely with the FDA, the TSA, and the seemingly ever-growing national debt.

Take, for example, the FDA. They can approve a treatment that turns out to be unsafe or they can delay (or deny) one that would have saved lives. Approving an unsafe treatment leads to identifiable patients, newspaper headlines, and potentially congressional hearings. A failure to approve a treatment usually has none of these effects. The continued suffering (sometimes even deaths) of patients who don’t receive treatment that’s never gained access to the market are rarely linked to regulatory bottlenecks.

The incentives this creates are obvious. An FDA reviewer who sits on an application for two more years faces no accountability for the deaths the delay may have caused. A reviewer who approves something that later causes harm potentially faces career catastrophe. The result? An institution that tilts toward cautious delay, requiring more studies and more clinical trials before approval.

The same pattern explains why the TSA confiscates your water bottle. If a security screener waves through a passenger who later commits an attack, the consequences would be catastrophic and, importantly for the screener, traceable. If they instead inconvenience 10,000 passengers with “security delays,” missed flights, and ritual humiliation, the cost is still very real but nobody gets fired. Travelers grumble but ultimately move on.

The screener, the supervisor, and the agency all face serious accountability when they let a real threat through, but essentially none when they treat everyone and everything like a threat. The predictable result is one of security theater that is not designed to minimize total harm, but to minimize a particular kind of harm that could show up in a congressional investigation.

Asymmetric accountability also helps explain the problem of persistent government deficits. Balancing a budget isn’t difficult. Most people, including elected officials, manage to do it in their own lives with such regularity that we barely pay any attention. But doing so with actual government spending requires every policymaker to agree to restrain spending, including on their own district.  This is where it falls apart. Restraint only works if everyone goes along with it.

Spending, by comparison, requires cooperation too, but this cooperation is much easier to secure than mutual restraint.  Every politician knows that spending today gets immediate, attributable credit. They’re branded as someone who “gets results” and is rewarded with ribbon cuttings, press releases, and grateful constituents. The bill, however, arrives much later, often paid for by taxpayers who weren’t even old enough to vote when the spending was approved in the first place. And herein lies the cooperation: politicians can trade votes in a process known as “logrolling.”  That way, everyone gets a ribbon cutting, press releases, and grateful constituents. Nobody campaigns on sparing our grandchildren an oversized tax bill.

These examples all look like failures of different kinds, but they’re all the result of the same problem: asymmetric accountability. In each case, from the perspective of the decision-maker, one is visible, attributable, and personally costly. The other is invisible, diffuse, and consequence-free. There might be good reasons to be frustrated by this reality, but the fix isn’t to find better people or to improve training. It’s to build better accountability structures.

After all: people respond to incentives.