Skip to content

Principled principals and ethical agents

May 16, 2015

He tasks me; he heaps me; I see in him outrageous strength, with an inscrutable malice sinewing it. That inscrutable thing is chiefly what I hate; and be the white whale agent, or be the white whale principal, I will wreak that hate upon him.

–Captain Ahab, in Chapter 36, “The Quarterdeck”

Ahab doesn’t care which one the white whale is. Should we?


So strongly and metaphysically did I conceive of my situation then, that while earnestly watching his motions, I seemed distinctly to perceive that my own individuality was now merged in a joint stock company of two; that my free will had received a mortal wound; and that another’s mistake or misfortune might plunge innocent me into unmerited disaster and death. Therefore, I saw that here was a sort of interregnum in Providence; for its even-handed equity never could have so gross an injustice. And yet still further pondering—while I jerked him now and then from between the whale and ship, which would threaten to jam him—still further pondering, I say, I saw that this situation of mine was the precise situation of every mortal that breathes; only, in most cases, he, one way or other, has this Siamese connexion with a plurality of other mortals. If your banker breaks, you snap; if your apothecary by mistake sends you poison in your pills, you die. True, you may say that, by exceeding caution, you may possibly escape these and the multitudinous other evil chances of life. But handle Queequeg’s monkey-rope heedfully as I would, sometimes he jerked it so, that I came very near sliding overboard. Nor could I possibly forget that, do what I would, I only had the management of one end of it.

–Ishmael, in Chapter 72, “The Monkey-Rope”

As Ishmael says, we are all principals, and we are all agents, for all of us; which, if true, makes the principal-agent problem the central problem of ethico-political existence.


The principal-agent problem, in a nutshell: One person, call him the principal, wants something to happen, but he is unable to do it himself, and so he hires someone else, the agent, to do it for him. But he will often find it difficult, even after the fact, to ascertain whether or not the agent has actually done what he was hired to do. This gives the agent a reduced incentive to actually do his task, since the principal might never find out whether he did it or not. The principal-agent problem is the cause of the distrust we feel for experts of every stripe: jewelers, lawyers, doctors, plumbers, contractors; we just can’t tell if they’re doing a good job. We have to take their word for it.

This problem was also, of course, the downfall of communism. Agents went rogue at every possible level: the workers, the managers, and the central planners all had no real incentive to accomplish the tasks for which the managers, the central planners, and History, respectively, had engaged them.

Capitalism solves the principal-agent problem through the joint-stock company. When economic enterprises are small, economies can work with everyone just looking out for their own interest, but when enterprises grow more complex, it becomes necessary to ensure that different people cooperating on an enterprise are actually, well, cooperating. Investors, who have capital but not expertise, don’t just want high profits this quarter, they want the company to do well in the long term. They hire managers, who have expertise but not capital, whose job it is to maximize the value of the stock, which is used as a proxy for the company doing well. Unless they control more resources than the company is worth, which almost by definition they do not, the managers will find it difficult to manipulate the stock price directly; while to manipulate it indirectly, they will have to systematically deceive the entire market, which it’s difficult to do over the long term. So the investors have an easy way to see if the company is doing well, and can always fire the managers if they think it isn’t, which gives the managers a strong incentive to make sure it does.


This solution, however, is of extremely limited scope. It promises that your every desire can be fulfilled by an external agent–so long as you agree that you desire long term financial success.

Money is, of course, more fungible than almost anything else, so the solution is still attractive. Maybe too attractive. It gives us the illusion that similar strategies can solve the principal-agent problem in other domains. Worried about whether teachers are really teaching students anything? Impose standardized tests. Worried about whether the money spent on health care is really healing anyone? Measure procedures in terms of QALYs (quality-adjusted life years).

These sound like reasonable solutions because, even if test scores and QALY counts aren’t exactly what we, the principal, want, they’re at least correlated with our desires. But mere correlation is exactly what causes principal-agent problems in the first place! In the days before joint-stock companies, the owner of a firm would see that his manager was bringing in profits; he would know that current profits were correlated with long-term financial success; but he would have no way of knowing if these profits were signs of long-term health, or if they were merely fabricated to look like such signs. The exact same thing happens with test scores and QALY counts. Teachers teach to the test, and students learn nothing but the test; insurance companies provide coverage based on what they’ve decided you want, and other important components of health get left by the wayside. The incentive structures we create to solve principal-agent problems create new ones.


This happens frequently when the goods we desire are difficult to verify, as with jewelry appraisals. It happens perhaps inevitably whenever the goods we desire are not quantitative, like money, but qualitative, like health and education. If we can’t explain to our own satisfaction exactly in what they consist, how can we possibly verify whether the agent whom we’ve contracted has sought them adequately? The traditional solution, in such cases, is ethical. Lawyers and doctors and teachers are all held to high ethical standards.

At the same time, they’re all routinely distrusted, accused variously of dishonesty, malice, and incompetence. For they cannot, of course, actually be held to such standards: there is no one with the knowledge necessary to do the holding. Rather, we expect them to hold themselves to such standards. We expect them to embody our ideals, of the rule of law, of the balanced life, of the quest for wisdom. This gives them an incentive to adopt the appearance of embodying such ideals; it gives them no incentive not to be hypocritical.

Hypocrisy, then, is a special case of the principal-agent problem. Is it better, or worse, than other principal-agent failures? I’d argue that it’s better. Ethical standards demand real obedience, not just lip-service. This means that, if we notice that someone is being hypocritical, we can call him out on his hypocrisy. When we attempt to replace ethics with a non-ethical solution to the problem we block off our recourse to this calling out. If we’re evaluating our teachers based on their students’ standardized test scores, it rings hollow to accuse them of teaching to the test: that’s what we asked them to do. The principal cannot in good conscience evaluate the agent’s actions other than by seeing whether the agent did what he was told. That our entire country does so, incessantly, is yet another instance of incessant hypocrisy in what academics love to call the “modern liberal state”; what an earlier thinker would have called Leviathan.


So be the white whale agent, or be the white whale principal? If principal only, he has no agents, and we have no commerce with him. Ahab knows this to be false:

Talk not to me of blasphemy, man; I’d strike the sun if it insulted me. For could the sun do that, then could I do the other; since there is ever a sort of fair play herein, jealousy presiding over all creations. But not my master, man, is even that fair play.

Here Ahab’s argument undoes itself. If we have commerce with the whale, whose agent is he if not ours, the outrageous strength and inscrutable malice revealed as the strength and malice of the body politic, inscrutable because it refuses to scrutinize itself?

Then Ahab strikes through the pasteboard mask at himself? Well, we did know all along that he was suicidal, which raging against the modern liberal state always is. The principal cannot in good conscience evaluate the agent’s actions other than by seeing whether the agent did what he was told. Which, like a demon granting wishes, it always does.

8 Comments leave one →
  1. CJ Wolfe permalink
    May 28, 2015 9:44 am

    Joseph- in the case of teachers, where standardized tests are an “incentive-based” non-ethical check on their expertise and performance, I think there may be slightly more to the issue. Standardized test scores in Mathematics, for example, are compared between countries. The performance of American teachers can be verified that way for the American voters and taxpayers (in this case, state property taxpayers), not merely correlated. Unless you would argue that Mathematics skills worldwide are somehow all down.

  2. May 28, 2015 10:44 am

    Yeah, I definitely overstated my case for rhetorical effect. I have a problem with how they’re often used, but standardized tests aren’t bad per se—they do measure something, after all, even if part of that something is the ability to sit quietly for three hours and answering multiple choice questions (something I am *very* good at, a skill which hasn’t helped me much since taking the GRE five years ago). And you’re also right that this is less of a problem with more objective disciplines like math (though we shouldn’t imagine that even mathematical ability can be entirely captured by a standardized test).

    To make explicit the comparison I drew between standardized tests and short-term profits: of course companies should publish their financial data, and of course we should make some attempt to measure our students’ progress. y point is that there shouldn’t be any incentives *systematically* tied to these measurements, because that creates moral hazard: CEOs aim for short-term profits, teachers teach to the test. We should use the data to get a picture of what’s going on, but we shouldn’t act as if the data can interpret itself. We need to use our judgment.

    This is pretty clear, I think, when it comes to the use of tests to evaluate individual teachers. It doesn’t apply in quite the same way when tests are used to measure performance on a city- or state- or country-wide scale. Such measurements create possibly dangerous incentives, not for particular teachers, but for administrators. We might think of the task of administration as one of ensuring those measurements are taking into account while the perverse incentives they can create are kept at bay. Something is definitely going wrong if the administrator just passes the measurement-based incentive directly to teachers.

    In retrospect, it sounds like my argument is basically about subsidiarity, i.e. the importance of letting the people who understand the situation determine how the job ought to be done. This is only possible, I’m saying, if we don’t micromanage the incentives surrounding their performance. The strongest incentive should be the incentive to do a good job for its own sake.

    My main target is people who blame the teachers, not the test, for the results of the moral hazard the tests create. Since a few weeks ago we were talking about double effect, I’ll compare it to that. Some people argue that, because abortion is *intrinsically* the killing of a child, you can’t invoke the principle of double effect to defend performing an abortion to save the life of the mother, for it would be false to call the child’s death an unintended *side* effect. Similarly, I’d say, standardized tests used in a way that violates the principle of subsidiarity create moral hazard intrinsically, not just incidentally.

  3. CJ Wolfe permalink
    May 28, 2015 2:02 pm

    Absolutely. And at the University level instead of standardized tests, professors have to do all sorts “Learning Objective” reports for our managers at the accreditation organizations. Those would probably be an even worse violation of subsidiarity than High School standardized tests, except for the fact that their ways of gathering data are so abstruse. The learning objective reports really aren’t testing the “ethical standards” of teachers even, their goal is to dictate the curriculum of the course.

    Using the “double effect” language, I guess you’d have to say this. Those learning objective reports are direct violations of the subsidiarity that should be allowed to teachers; standardized tests are indirect though still intrinsic violations of that subsidiarity


  1. The custom of our times | Ironical Coincidings
  2. A charity that would go of itself | Ironical Coincidings
  3. The tragedy of the algorithm | Ironical Coincidings
  4. What is Caesar’s? | Ironical Coincidings
  5. Divagations on the impending apocalypse | Ironical Coincidings

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: