Agency problem: Difference between revisions

The [[agency problem]] addresses the intrinsic [[conflict of interest]] any [[agent]] working on a [[commission]] faces — any [[introducing broker]], [[broker/dealer]], [[asset manager]], [[architect]], building contractor — and that is that once it have received its [[commission]], it doesn’t really care a hill of beans what its [[principal]] gets, however much it much protest to the contrary. In a sense this is a basic articulation of the [[prisoner’s dilemma]] and so shouldn’t surprise anyone — and should be cured by repeat interactions — your clients have memories and will remember when you ripped them off.  

But the [[iterated prisoner’s dilemma]] has a couple of natural limits. One is that it relies on repeated interactions with an indeterminate end. When the sky is falling on your head, it looks like a final interaction, and the calculus is utterly different. Second, it takes no account of convexity effects. I can build up my reputation incrementally with thousands of small transactions — I can look like a five star collaborator — only to bit on a big position and defect. This is what {{author|Nassim Nicholas Taleb}} calls the “Rubin Trade”.