We investigate deterministic and stochastic repeated rent-seeking contests. In the deterministic contest a contestant receives a share of the rent equal to her share of rent-seeking expenditures. When subjects get feedback about the choices and earnings of all contestants we find that changes in behavior can be explained partially by adjustments in the direction of best response to rivals’ choices in the previous round and partially by adjustments in the direction of imitating the most successful contestant. The implication of imitating the most successful contestant is over-dissipation relative to Nash equilibrium expenditures levels. When subjects are only informed of own earnings imitating the most successful player is no longer possible and average behavior converges to Nash equilibrium levels. In the stochastic contest a contestant wins the entire rent with probability equal to her share of rent-seeking expenditures. Here we also see over-dissipation relative to Nash equilibrium in a full information feedback treatment. However, when we restrict information feedback to own earnings we observe the highest over-dissipation rates of all.
http://www.creedexperiment.nl/creed/seminars.php