Tuesday, January 24, 2017

The Inevitability of Principal-Agent Theory

My resistance seems to have dominated my destiny.  When I was in high school, my worst class was French, and it was not even close.  So, of course, I end up in Quebec for ten years.  In grad school, I avoided the Americanist classes and especially those taught by a particular one.*  As a result, I did not drink deeply from the dominant thinking in that department at the time--principal-agent theory.  I dodged it for more than a decade, but when it came time to study how and why countries controlled their troops in puzzling ways in Afghanistan, it was the obvious way to frame the project.  And now I can't stop seeing it everywhere, from the Avengers to Breaking Bad to Star Wars, yes, contemporary politics (not to mention the new project).

* Experts on this stuff might cringe at how I approach P-A precisely because my training on this is so incomplete. All my career I have wondered into literatures that are outside my area of expertise.  Oops.

The basic idea is this: anytime someone (the principal) hires another person (the agent) and gives them some responsibility, the agent will know more about that responsibility than the principal.  The agent is simply closer to the matter at hand and focused on it, while the principal delegated responsibility precisely because they did not want to be doing all the work to make sure the particular work was being carried out.  There is inherently slippage between what the principal knows and what the agent knows WITHOUT presuming that the agent has different beliefs about the task or is deliberately acting against the principal's wishes.

Agents may simply not a perfect understanding of the intent of the principal.  As a professor hiring research and teaching assistants, I have often given lousy instructions which then leads to either the assistants asking for more information about what they are supposed to do or they try to figure it out.  I suck at delegation.

Why am I thinking about this now?  Besides the daily work on the comparative legislatures project (how/why do legislatures engage in oversight over their armed forces in varying ways) and the big grant project where P-A is the theoretical glue, I think we are headed towards peak Principal-Agent dynamics in the years ahead in the US.

For less conflictual P-A relations, principals ought to pick agents whose views of the problem and of the solutions are highly congruent.  Given Trump's uncertainty and conflicting views, this is going to be mighty hard for those he selects--the White House staff and the top political appointees.  The gap is likely to be far wider with the civil servants in the bureaucracy because Trump is the least conventional thinker (thinker?) they have ever experienced.  The latest example--using a spot that is meaningful to the members of the CIA as a bully pullpit to rail against the media.

The second step in the principal-agent relationship is to establish how much discretion is to be delegated.  Trump has indicate that he would delegate a great deal, except for when it comes to a few key issues.  More delegation is inherently neither good nor bad, but more does open up more room for more slippage.

So, the key would be the third step: oversight.  Figure out ways to keep an eye on the subordinates so that they know they are being watched, and, thus, less likely to use their authority in ways that are undesired.  The problem is that this requires .... expertise.  The more one knows about the stuff, the more one can detect shirking (deviations from intent).  Outside of DoD with Mattis and a few other spots, few of Trump's appointees have any familiarity with government and some have an appalling lack of expertise on the issue at hand (Carson is the extreme example).

The Trump folks have focused on the fourth part of P-A management--incentives.  They are essentially threatening to fire people who deviate.  All those requests for lists of names of people who have taken stances the Trumpsters don't like are efforts at intimidation.  Threats may deter some agents, but are likely to antagonize others.

By generating hostility with the agents, such as criticizing the CIA and their work and by putting lousy stewards in place and by being an uncertainty engine, Trump is going to encourage shirking on a massive scale.  Shirking, to be clear, is when the agents do not do as intended.  They can do more and they can do less and they can do differently.  They can shirk because their standard operating procedures tell them to do stuff that is not what principal wants, they can shirt because they think they have a clearer idea of what they should do, they can shirt because they think that the policy from on high is misguided.  They can give less information back to the principals.  They can selectively implement complex laws and procedures.  And, yes, they can leak.  They can share with the media tales of the principals' bad leadership, they can share stories of how the new policies are likely to hurt Americans and risk wars.  And on and on.

So, for the US media, the fire alarm in this model--the folks who point a spotlight at bad government policy and yell about it until the Congressfolk and voters pay attention, are in for some fun days ahead.

And cracking down on leakers rarely leads to good things (Watergate, some of Obama's troubles).  As always:

No comments: