Principal-Agent Problems and the Structure of Government

This post on Best Of A Great Lot is a part of a series on the subject of designing a new form of governance. Each piece aims to stand alone, but fits together on the Table of Contents.

Updates: This is the last in the prework section of articles about problems and difficulties, after this will come the core system descriptions. I am now cross-posting these articles to LessWrong in an attempt to expand the audience of who’s seeing them. If you are finding them interesting and have an account over there, please upvote them so that more can see them. -b

Every system fails. Some failures can be thought of as a tradeoff between consistency and flexibility, while others are unexpected features we discover once the system is widely used. Corporations hire armies of lawyers to think deeply about tax law, looking for loopholes. Political consultants examine election law for similar reasons.

When lawyers find loopholes that aren't good for the citizenry at large, it's supposed to be government's job to change the law. But when the loophole or flaw isn't in the law, but in the structure of the government itself, fixing it becomes a much more interesting and difficult question. With the Constitution, the founders of the US answered that question with an amendment process that requires much more agreement among the state governments and Congress than is needed to pass a law. It's sufficiently difficult that in 2 centuries there have only been 15¹ or 16 amendments that didn’t happen within the first decade. The most recent amendment to be newly proposed and ratified was 50 years ago. It’s common to conclude that the amendment process has largely run out of steam in terms of its ability to improve our system. Particularly notable is that unlike with law, where it’s normal if not common for us to pass a law and change or revert it later, we’ve only reverted one amendment: when we prohibited alcohol. This is not the picture of a system we are using much to improve our government.

Requiring more agreement among the many politicians involved in the system is a deeply imperfect band-aide on the underlying problem, which is simply that it's hard to trust those who have power — those who are currently selected to run the government — with the power to change the structure of the government. The incentive for them to make the governmental structure better for everyone and not just for themselves is too weak.

To snag a term from economics, this is a minefield of principal-agent problems. When you hire a lawyer to represent you, or when you're a shareholder of a company and the CEO represents you in the day-to-day management of the company, you are a principal and you are hiring an agent (lawyer, CEO). Their job is intended to be to look out for your interests. The lawyer is supposed to care about getting you a good deal. The CEO is supposed to care about making you a profit. But the reality is that the agent has their own interests. The lawyer may know the lawyer on the other side and wants to maintain a good working relationship with them. The CEO wants more money and a bigger office and to have more power. Even a great lawyer or CEO won't screw up their career to get you a better deal.

When agents can choose things that don't hurt their own interests and benefit the principal's, the relationship works well. When those interests diverge, the agent tends to pursue their own and the principal may not have the ability to prevent them.

Allowing those who run the government — such as legislators, governors and presidents — to change the structure of government runs directly into principal-agent problems. We citizens want the structure of government to help us get better outcomes, including elected officials we’re happier with. Elected officials, even the well-meaning ones, want to stay in power. The power-driven ones want power for its own sake. The rest view power as necessary to pursue their agenda so they can make the world better. It's the rare idealistic politician who thinks that their opponent will do better for the people than they will, and among politicians who enjoy power for its own sake this viewpoint is even rarer. Allowing representatives to entrench their own power through structural changes is the path of banana republics, ever-updating Constitutions to keep dictators in power and single-ideology rule.

We see a version of this with gerrymandering: entrenching structural advantage by letting politicians carefully choose their voters. Politicians drawing the district lines to help themselves dates back to the earliest days of the US.

So we can see the danger of letting politicians control the structure of government, and the reasons why the Founders wanted to make it harder. But making it harder produces stagnation and other less visible problems.

Stagnation

Consider the challenge of putting corrupt politicians in jail. Given the current structure of government, this is a difficult and touchy thing to do, and many avoid real consequences. The FBI does its best to claim to be neutral, but the head of the FBI reports to the Attorney General, a key political appointee. When the FBI pursues members of the President's party, they have to be careful not to piss off their grandboss. When they pursue members of the opposition party, they have to be careful not to look like they're serving the President's political interests. It's a tremendously tough balancing act created by the fact that they work for an agent, not the principal (us).

Occasionally you run across proposals to make the FBI truly independent, such as this proposal to elect the federal Attorney General directly, the way most states do. Would this improve the situation? Attorneys General who are elected will have a different principal-agent problem than appointed ones: they'll want to ensure their re-election more than they'll want to put the corrupt and powerful behind bars. Perhaps elected Attorneys General will prosecute more corruption, but only the unimportant cases. Or perhaps prosecuting the powerful will be what impresses the people the most, and we'd get some real cage matches.

It's unlikely we'll find out, given our current system. If we had some truly independent group of thinkers who we could trust with changing the Constitution and not warping it to their own interests, this seems like the kind of change we would try. If that were true, we might see frequent changes to the structure of government, including reverting mistakes, the way we see frequent changes to tax law. Instead, the difficulty of amending the Constitution means we mostly live with the structural problems we have, since the process gives cover to politicians who don't want to change the playing field that they are winning on.

Borders

Consider the question of changing jurisdictional boundaries. Most of our city, county and state borders were made a long time ago and are vastly different than the borders we would pick now if we had the ability to redraw them easily. Recent proposals that the rural parts of Oregon to become part of Idaho and to break California into 3 states are the closest anyone’s come to serious, and they’ve largely been met with laughter by political insiders. Borders are something that we basically don’t change. Major cities often have metropolitan areas that extend into other states, resulting in deeply problematic and confusing jurisdictional problems. There are at least 8 major cities, and many more minor ones, that can be referred to as "the tri-state area." Public transit, ecology management, urban planning and taxation structures are all rife with cross-border problems in many of these.

We have set things up so that any jurisdictional change would require both jurisdictions to approve it (at minimum - if it required changing state boundaries, Congress would likely get involved). But jurisdictional changes — even ones that make the citizens better off — will always have winners and losers among the leaders of the jurisdictions. The status quo benefits the leaders of the status quo, and those who will stand to be worse off after a change have the power to block it.

Parties

Political parties exacerbate principal-agent problems: the agents all join together into a small number of mutual aid societies. It would be like if all of the lawyers worked together in a single law practice where they were told every day that they should help each other out if they want to succeed at all. The specifics of the deal you want fades away into the background of their clubhouse relationships.

Within a partisan system, electing an Attorney General directly only marginally reduces the principal-agent problems compared to having them appointed. Either they’ll work for the President who’s a member of their party or against them because they’re not. When we have divided government, Congress tends to constantly be investigating the executive branch to make electoral hay.

The Founders were fearful of parties, and yet did not know how to prevent them. Washington attempted to warn against them in his farewell address:

The alternate domination of one faction over another, sharpened by the spirit of revenge, natural to party dissension, which in different ages and countries has perpetrated the most horrid enormities, is itself a frightful despotism. - George Washington, Farewell Address

John Adams was even more succinct:

There is nothing which I dread so much as a division of the republic into two great parties ... This, in my humble apprehension, is to be dreaded as the great political evil. - John Adams, letter to Jonathan Jackson, 1780

Are parties just an essential feature of self-government? Madison seemed to think so. He argued that factions were an essential outgrowth of the freedom to have different opinions and the inherent reality that people are different.

There are two methods of curing the mischiefs of faction: the one, by removing its causes; the other, by controlling its effects. There are again two methods of removing the causes of faction: the one, by destroying the liberty which is essential to its existence; the other, by giving to every citizen the same opinions, the same passions, and the same interests. Liberty is to faction what air is to fire, an aliment without which it instantly expires. - Federalist Papers, No 10. (emphasis added)

Sure, different opinions lead to different interest groups, but do they have to coalesce into monopolistic political parties? Governance structure matters also. Some evidence for this is that while the US has only had two major political parties for nearly all of its history, parliamentary democracies often have more. The UK has often had between 3 and 5 major political parties, and has more change in which parties are considered major than is common in the US. Israel is even more extreme, with 13 political parties that have seats in the Knesset (in 2024) and the top two parties having less than half the seats between them.

Principal-agent problems, are exacerbated by scale. When the agent is the CEO of a small company with a dozen investors, those investors have much more influence than when they're the CEO of a giant multinational and thousands of investors don't even bother showing up to shareholder votes. Political parties are worse for the citizens when there's 2 giant ones that are the only choices in town.

In regular life, we join more than one group, whether we're talking book clubs, volunteering for charities, or contributing to interest groups that will lobby for laws. It's only in the very specific realm of selecting representatives that we only join one. You simply can't join multiple political parties, even in Israel where it's very possible for there to be more than one that might represent you.

This is not an inevitable result of self-government, it's a result of voting and bundled governance. You can't be a member of more than one party for the same reason you can't root for both teams when watching a football game. They can't both win. Because of bundled governance, those who win make choices about everything, so you can't be a social liberal and a fiscal conservative (or the reverse, or any number of different combinations of other philosophies) and vote for different parties for the different aspects of society.

It's possible that parties are an inevitable evil that must be managed. But it's also possible that if we didn’t have bundled governance, a thousand nonrivalrous parties would bloom. A thousand small-scale parties each focused on a smaller set of concerns are much less likely to be the kind of mutual aid society that prevents politicians from working for us.

Time’s article on how to solve gerrymandering proposes independent nonpartisan or bipartisan commissions. All attempts to get nonpartisanship in our partisan system should be viewed with caution. California has 10m registered Democrats, 5m registered Republicans, and 5m registered Independents. If the commissioners reflect the population, they'll be likely to want to draw lines that benefit Democrats. If they reflect parties, Republicans will be overpowered in the commission compared to their popular representation. If only registered independents are allowed, either most people are excluded from an essential process of the system or they have to choose between participating in selecting candidates or selecting district lines. They’ll likely still be partisans, just pretending otherwise. We don't have a way to tell who is a loyal partisan and who isn't, especially among the class of "people interested in participating in a charged political question", so we're unlikely to get true independents.

In an ideal world, we'd have independent thinkers working on the structure of government whose only incentive was to make it better at doing its job, which wouldn't be about who gets power as much as it was about how effective it was at solving the problems that we, the citizens, care about. I’ll be describing how belocracy can get us there.

If you find my work interesting, please share and subscribe. It helps tremendously.

1. 15 if you don't count the 27th amendment, which was proposed in 1789 and came about two centuries later when a college student got a C in history for arguing that it was still live and could be ratified at any time. His quest to prove his teacher wrong was remarkably successful and should inspire every antiauthoritarian student, but certainly hinged in part on the fact that it was not his idea, but original to the founders.