Triangle Principal-agent Model
It wasn't particularly considerate of me to use the only personal experience even remotely related to the principal-agent model for the previous post, which could have been written using an entirely different exemplary story. I want to note also that I didn't immediately grasp the concept of the triangle principal-agent model correctly, and the initial misconception was interesting: I defaulted to thinking that rather than being triangular, the relationship was transitive and vertical, as if the principal's agent was herself a principal to another agent. That model is vertical simply by its nature and is transitive due to the agent passing the top principal's objectives down to the bottommost agent. Now I wonder if that non-triangle model has a name of its own.
Reading more on the triangle principal-agent model that is generalized to multiple principal problem, I sensed that it is most ubiquitous in governance mechanisms, like that of a firm's board of directors (each being a principal whose agent is the leader of the company) or of a democratic state's legislative body (each legislator being an agent for the electorate). Clearly, if there is no coordination between the principals with regards to the objective for the agent's performance, the agent becomes a moral hazard due to limited accountability—there's no cost for the agent to act opportunistically and favor the principal whose interests are most aligned with her own. Importantly, even a benevolent agent might still face a conflict of interest under the model; stalemated by multiple non-cooperating principals, the agent will have no choice but to disappoint some of them. In politics, the aforementioned coordination is maintained through public discourse, most known instance of which is the media and independent journalism. Coming from another country, I want to assert that lack of independent journalism—and, thereof, public discourse on its own—only worsens the incumbent situation of limited governmental accountability.
The immediate solution to the multiple principal problem is to not allow an agent to act on behalf of multiple principals. One will soon realize, however, that some mechanisms can't exist otherwise and are naturally subject to the conflict (say, elected U.S. officials). If there is no way to avoid the problem, what are the ways to solve it?
There are two solutions that come to mind, both converging to the same idea: the agent needs a single objective. That means that either the principals have to communicate and work on having a common objective or that the principals have to elect a single principal representing the median of the population.
Reading more on the triangle principal-agent model that is generalized to multiple principal problem, I sensed that it is most ubiquitous in governance mechanisms, like that of a firm's board of directors (each being a principal whose agent is the leader of the company) or of a democratic state's legislative body (each legislator being an agent for the electorate). Clearly, if there is no coordination between the principals with regards to the objective for the agent's performance, the agent becomes a moral hazard due to limited accountability—there's no cost for the agent to act opportunistically and favor the principal whose interests are most aligned with her own. Importantly, even a benevolent agent might still face a conflict of interest under the model; stalemated by multiple non-cooperating principals, the agent will have no choice but to disappoint some of them. In politics, the aforementioned coordination is maintained through public discourse, most known instance of which is the media and independent journalism. Coming from another country, I want to assert that lack of independent journalism—and, thereof, public discourse on its own—only worsens the incumbent situation of limited governmental accountability.
The immediate solution to the multiple principal problem is to not allow an agent to act on behalf of multiple principals. One will soon realize, however, that some mechanisms can't exist otherwise and are naturally subject to the conflict (say, elected U.S. officials). If there is no way to avoid the problem, what are the ways to solve it?
There are two solutions that come to mind, both converging to the same idea: the agent needs a single objective. That means that either the principals have to communicate and work on having a common objective or that the principals have to elect a single principal representing the median of the population.
I'm going to avoid talking about politics in this context, because it is quite distressing to me now. So let me give a theory response first and then give the applied response after.
ReplyDeleteIf the two principals and the agent had a discussion about what the preferred agent behavior was, and if they ultimately could agree even though the two principals have preferences that are not aligned, then the solution would be a compromise of sorts, with the agent instructed to take some middle position.
In practice this conversation rarely if ever happens. In the real situation the agent will find that the expectations of the two principals are mutually inconsistent. If the agent performs the arbitration alone without consulting either principal, so finds that middle position, the agent runs the risk that both principals will be unhappy with the outcome, because they didn't see the need to compromise. In this case, it might actually be safer for the agent to please one principal and take his chances with the other. Which principal to please might then be determined by non-economic criteria, e.g., the agent likes one of the principals more than the other, or by extra-economic criteria, e.g., one principal bribes the agent and the agent accepts the bribe.
In a larger organization, once you rise in the hierarchy some, you start to have interactions with other parts of the organization. If you feel your role is as a go between and a coordinator, you make get rewards at first from doing that job, but then you might get a lot of push back later as people don't like the preferred solution. That was my experience as an administrator. I lived this triangle problem then.
Possibly, it is best if a said conversation happens - a) with all the principals and the agent physically present in the same room, - b) with the final decisions on agent's objectives made based on voting. Seems like that is how boards of directors tend to work. I'm now asking myself why is it that in most of the stories I empathize with CEOs more than with the board. In that particular situation, are they more than just agents? Many CEOs are unique in their strategic capabilities, making them irreplaceable; is that true for a general agent in the model? In reality, agents might have more insight into how theoretical objectives translate into practice and might have a hard time conveying reality-induced obstacles to the principals. Does that justify agent's arbitration? I don't know. I'll think about this more.
ReplyDeleteWhat you said about a larger organization—does that mean that an agent's job is not only to deliver a solution but also to deal with push back? Is it an agent's responsibility to enforce changes?