We present a systematic robustness test of the persistence of gift-exchanges in the laboratory. Our data clearly establish that the effect of social forces is dramatically crowded out by loss aversion. This was not observed before, as in other studies that allow for nominal losses participants were endowed with a substantial lump sum payment. We did not endow our participants with some initial wealth (they also got no show-up fee). Instead, participants were required to sign an agreement before the start of the experimental session in which they agreed to cover losses by either incomes from future participation in experimental sessions or by their own money. We conjecture that by providing some initial endowment to their participants, previous experimental studies have clearly failed to investigate the impact of losses on the level of gift exchange reported. Further, we observe a considerable between treatment variability in the effort-wage relation. Small lump-sum payments to the first-mover reduce the effort-wage slope significantly. A reduction in the profitability of effort increases the effort-wage slope