This paper provides experimental evidence of the economic impact from shifting bargaining power in relational contracts. I implement an experimental design that adjusts the bargaining power of sellers (agents) and the enforceability of the contract. I find that the vast majority of contracts take the form of efficiency wage contracts instead of contingent performance contracts when enforcement is partially incomplete and sellers have more bargaining power than buyers. The total contracted and actual compensation increase with the bargaining power of the sellers. However, sellers' profits are found to increase only if a part of the total payment is third-party enforceable. In this case, observed surplus and efficiency are lower than predictions. When no part of a contract is third-party enforceable, more cooperative relationships emerge, exhibiting higher quality provision resulting in higher surplus and efficiency while rent sharing is lower. The result is explained by the stronger buyer's deviation, confirming predictions derived in Cordero Salas (2010). The results here provide insight into the economic consequences of enacting policies that improve the bargaining conditions of weaker parties in market settings relying on self-enforcement from underdeveloped legal institutions.