Performance Attribution, Knowledge Transfer, and Team Formation

How do evaluation systems shape knowledge transfer within teams? We model knowledge transfer from a high-skill worker to a low-skill coworker as a bilateral activity that improves the recipient's skill but consumes both workers' production time, under a performance-attribution rule that credits output partly to individuals and partly to the team. Individual attribution creates a friction: teaching narrows the output gap and thereby reduces the high-skill worker's attribution premium, so she chooses less transfer than is team-efficient, even though no job, prize, or future rent is at risk. Stronger individual attribution widens the gap between implemented and efficient transfer and, beyond a threshold, halts transfer completely; the efficient level is independent of the attribution rule. We then embed teams in an organization that chooses their composition. Individual attribution erodes the value of mixing high- and low-skill workers, and the organization switches to assortative teams strictly before transfer would collapse within a mixed team. Finally, we endogenize the attribution rule when high-skill workers have attractive outside opportunities. When individualized retention pay is constrained, the organization uses individual attribution as a second-best retention instrument: although team attribution maximizes current output, assigning more individual credit helps retain high-skill workers. Compensation rigidity can therefore induce stronger individual attribution at the cost of knowledge transfer, with sufficiently strong retention pressure eventually leading to assortative team formation.