Government as a principal-agent problem

I’m a bit surprised this hadn’t occurred to me sooner.

Wikipedia defines the principal-agent problem as ” the problem of motivating a party to act on behalf of another […] when a principal compensates an agent for performing certain acts that are useful to the principal and costly to the agent, and where there are elements of the performance that are costly to observe”.  Most of the article’s examples involve the private sector: contract workers, corporate shareholders hiring executives, and so on.  But the biggest example of a principal-agent problem is elected government: We The People elect politicians to run the government on our behalf, and end up putting them in a position to run the government on their behalf instead.  Things get even worse when those politicians establish appointed institutions and bureaucracies — there the agents are hired by the politicians (more or less, often less) to work on the people’s behalf.

Which is to say: No wonder things are fucked up.

The trigger for this observation is a post by Bruce Schneier on a WSJ editorial by his long-time airport-security sparring partner Kip Hawley:

• Hawley channels his inner Schneier (Schneier on Security)

In particular, this bit at the end (though the whole thing’s worth a read):

There’s one point Hawley makes, but I don’t think he makes it strongly enough. He says:

I wanted to reduce the amount of time that officers spent searching for low-risk objects, but politics intervened at every turn. Lighters were untouchable, having been banned by an act of Congress. And despite the radically reduced risk that knives and box cutters presented in the post-9/11 world, allowing them back on board was considered too emotionally charged for the American public.

This is the fundamental political problem of airport security: it’s in nobody’s political self-interest to take a stand for what might appear to be reduced security. Imagine that the TSA management announces a new rule that box cutters are now okay, and that they respond to critics by explaining that the current risks to airplanes don’t warrant prohibiting them. Even if they’re right, they’re open to attacks from political opponents that they’re not taking terrorism seriously enough. And if they’re wrong, their careers are over.

It’s even worse when it’s elected officials who have to make the decision. Which congressman is going to risk his political career by standing up and saying that the cigarette lighter ban is stupid and should be repealed? It’s all political risk, and no political gain.

Airport security is intended to keep the people safe(r), but the incentives are aligned mostly to keep the careers of government agents and politicians safe.  So while most of the country — Kevin Drum excepted — thinks it’s pretty horrible that TSA’s blue-gloved agents feel up children and herd rape survivors through pornoscanners, the TSA and its bureaucratic and political masters operate under far different incentives and end up giving us a lot more airport security theatre than we’re asking for.

Similarly, law enforcement agencies have used asset-forfeiture laws to steal eleven thousand dollars from a hard-up waitress, unreasonable interpretations of interstate anti-drug laws to break up families over a pack of Sudafed, and the nebulous concept of Homeland Security to militarize a small New Hampshire town.  This is manifestly not the kind of policing that the people who vote for the politicians who run the governments that oversee the bureaucracies that pay the police in question want — but due to the iterated principle-agent problem it’s the kind of policing the people get.  Edit: It seems to me that this story about the NYPD ignoring serious crimes and generating spurious drug-possession arrests via stop-and-frisk is about as good an example as it gets of the principal-agent-agent-agent problem in policing.

So how do you fix a principal-agent problem?  You change the incentives.  (There’s that word again.)