Dynamic coordination with payoff and informational externalities

I study a two -player continuous -time dynamic coordination game with observational learning. Each player has one opportunity to make a reversible investment with an uncertain return that is realized only when both players invest. Each player learns about the potential return by observing a private signal and the actions of the other player. In equilibrium, players' roles as leader and follower are endogenously determined. Information aggregates in a single burst initially, then gradually through delayed investment and disinvestment over time. More precise signals lead to faster coordination conditional on initial disagreement, but might also increase the probability of initial disagreement.