I explore the implications of reference-dependent preferences in sealed-bid auctions. In the first part, I develop a Prospect theory based model to explain bidding in first-price auctions. I show that bidding in induced-value first-price sealed-bid auctions can be rationalized as a combination of reactions to underlying ambiguity and anticipated loss aversion. Using data from experimental auctions, I provide evidence that in induced-value auctions with human bidders, this approach works well. In auctions with prior experience and /or against risk-neutral Nash rivals where ambiguity effects could be altogether irrelevant, anticipated loss aversion by itself can explain aggressive bidding. This is a novel result in the literature. Using data from experiments, I find that ambiguity effects become negligible in auctions with experienced human bidders against (i) experienced human rivals and (ii) Nash computer rivals, when loss aversion is taken in consideration. The estimates for loss aversion are similar in auctions with human bidders (with or without experience).

     Next, I extend my approach of anticipated loss aversion to address bidding outcomes in first- and second-price sealed-bid auctions. As shown in first part, the model predicts overbidding in first-price induced-value auctions consistent with evidence from most laboratory experiments. However, substantially different bidding behavior could result in commodity auctions where money and auction item are consumed along different dimensions of the consumption space. Differences also result in second-price auctions. The study thereby indicates that transferring qualitative behavioral findings from induced-value laboratory experiments to the field may be problematic if subjects are loss averse and anticipate such losses at the time of bidding.

                 Finally, I explore the effect of resale or procurement opportunities, to which bidders have heterogeneous market access, on bidding in first- price sealed-bid auctions. My models suggest that in auctions with resale, loss aversion causes underbidding with respect to the risk-neutral-Nash prediction. Bidders with greatest level of market access are least affected by loss aversion and therefore bid closer to the risk-neutral-Nash than bidders with smaller market access. In auctions with procurement, the effect of loss aversion is such that it causes overbidding (underbidding) for bidders with respect to the risk-neutral-Nash. Bidders with greatest level of market access are again least affected by loss aversion and therefore bid much conservatively and closer to the risk-neutral-Nash than bidders with very low market access. If market access is interpreted as a proxy for experience, the predictions of my model are qualitatively similar to the findings in List (2003, 2004).  Since these indirect effects are obtained without altering reference-dependent preferences, it raises the possibility that the effects obtained in List (2003, 2004) in field settings may not arise entirely due to the direct effect of experience on reference-dependent preferences. This calls for a more careful reexamination of the underlying issues.