In this study, I investigate supply chain contracts in a setting where a supplier uses its inventory to directly satisfy a retailer's demand. These “pull” contracts have increased in popularity in practice but have not been studied experimentally. In a controlled laboratory setting, I evaluate a wholesale price contract and two coordinating contracts. The data suggest that the benefit of the two coordinating contracts over the wholesale price contract is less than the standard theory predicts, and that retailers, in the two coordinating contracts, exhibit a systematic bias of setting the coordinating parameter too low, and the wholesale price too high, relative to the normative benchmarks. In an effort to explain this deviation, I explore three behavioral models and find that loss aversion and reference dependence fit the data well. I empirically test this result in a follow-up experiment, which directly controls for loss aversion and reference dependence, and observe that retailers make significantly better decisions. Lastly, I administer a number of experiments which reduce the complexity of the problem, curtail the amount of risk, and increase the level of decision support, and find that none improve decisions relative to the treatment that controls for loss aversion and reference dependence.