Agricultural subsidy policies in several devel oped countries are moving from price support to decoupled payments-subsidies that are not linked to production and instead are enti tlements based on, for example, historical land use. Since these entitlements are tradable and can only be used in connection with land, es pecially in EU countries, this shift in policies basically generates a new good traded in new markets: retiring farmers have an incentive to sell their entitlements, while farmers with expanding land but without the correspond ing entitlements have an incentive to buy. The determination of market prices, however, is complicated by uncertainties about the future legal status of entitlements and therefore their values, by their different forms and sizes, the uncertainty about whether entitlements are in excess supply or in excess demand, and the lack of existing market forms. How, and how well, the markets for enti tlements will work is of interest to farmers and policy makers who are dealing with chal lenges to traditional agricultural policies and must consider the effects of alternative poli cies. Since policies changed only recently, and the first attempts of trading have just started, real-world market evidence is spurious and anecdotal. Thus, controlled laboratory experi ments can give a first insight into potential mar ket outcomes and how they are influenced by the institutional rules. In two previous projects (Bahrs, Kroll, and Sutter 2008a, 2008b) we designed experimen tal markets that were based on institutional changes of agricultural entitlement policies in Germany. In Bahrs, Kroll, and Sutter (2008a), we studied modified posted-bid and posted ask markets, while in the second study (2008b) we focused on eBay-like markets. In all of our treatments, including those in the current paper, market participants did not know the true value of the entitlements but received in dependent private signals of the value; there were two different kinds of entitlements high-value and low-value, and there was ei ther an excess supply or an excess demand of entitlements. Results from the previous projects indicate that the institutional effect from posted-bid or -ask markets seems to outweigh the effect of the excess-supply or -demand condition (2008a). The eBay-like market institution, however, generated signif icantly higher prices and lower efficiency un der excess demand compared to excess supply (2008b). In this paper we study another real world market institution-decentralized bi lateral negotiations. The previous projects assumed that all participants had the op portunity to trade with everybody else in a centralized market. There is a good chance, however, that most trades, at least in early stages, will be conducted in a more private and therefore decentralized setting-for example, people in the same village trying to sell and buy entitlements to and from their neighbors. Our experiment reflects this situation by allowing subjects to trade with only one other subject at a time. The interesting efficiency problem with such an institution is, of course, based on a matching predicament-what if potential buy ers have problems finding potential sellers and vice versa? How large will the efficiency loss be compared to the previously examined in stitutions and will the negotiated prices reflect