Sunk Costs and Escalation of Commitment

One important way in which a series of decisions over time can be linked is when nonrecoverable costs incurred at an earlier stage influence decisions at a later stage. The prescrip­tive advice on such matters is clear: The costs are sunk and should play no part in the later decisions. Equally clearly, many of us violate such advice. We finish the indifferent restaurant meal, sit through to the end of bad movies, and re­main in failed relationships so as not to "waste" the money spent on the restaurant bill or movie ticket or the time "in­vested" in the relationship. We fall, in short, into the sunk cost trap.

Arkes and Blumer (1985) reported 10 small experiments in which sunk cost effects were demonstrated. Although most used a scenario format (and are thus open to the criticism that they involved the subjects in no real decisions). Experi­ment 2 made clever use of actual theater-ticket buying deci­sions to investigate sunk cost effects. Of patrons buying season tickets for a university theater, one third were given a modest discount, and one third a substantial discount, from the normal price. Patrons paying full price subsequently at­tended significantly more of the performances than did those who received discounts, although the effect faded later in the theater season. Arkes and Blumer interpreted this as evi­dence that the larger sunk costs of the full-price patrons influenced their later attendance decisions.

Similar effects have been reported in organizational (e.g., Staw & Ross, 1989) and other (Brockner, Shaw, & Rubin, 1979) contexts. In a typical organizational study, Staw, Barsade, and Koput (1997) found that loan officers at banks were more likely to continue funding and extending problem loans when they had been responsible for the initial decision than when they took over responsibility for the loan after its initiation. A related effect in the persuasion literature, the foot-in-the-door technique, involves winning compliance to a large request by first obtaining compliance to a smaller one (Freedman & Fraser, 1966). More subjects agreed to put up a large lawn sign when they had earlier been asked to sign a

508 Judgment and Decision Making

petition on the same subject than when subjects were ap­proached directly with the large request.

Despite such apparently robust demonstrations, there is some confusion as to what phenomena are appropriately in­cluded in sunk cost effects, and an embarrassing range of par­tially conflicting explanations has been offered. One setting in which escalating commitment has been demonstrated in scenario studies is in continuing to fund partially completed projects (e.g., Staw, 1976). However, when degree of project completion and expenditure are independently manipulated (Garland, 1990; Conlon & Garland, 1993), only the former factor shows an effect. Public use of sunk cost arguments by public officials may reflect either the entrapment of the speaker or the calculation that sunk cost arguments will per­suade the audience. Staw and Hoang (1995) claim to have demonstrated sunk cost effects in their finding that basketball players drafted early (and expensively) into the NBA there­after are played more and traded at higher prices than their performance appears to justify. The result could, however, simply reflect the failure of their performance model to cap­ture what a player is worth to a team. It is thus somewhat un­clear just what is to be included as a sunk cost effect, or how reliably such effects can be reproduced.

One account of the sunk cost effect has been offered in terms of prospect theory's loss function. The initial cost is taken as a loss (below the reference point), thus putting the de­cision maker into a region of risk seeking. Continuing the proj­ect now offers a risky project with some hope of gain, while abandonment forces acceptance of a certain loss (Thaler, 1980). Arkes (1996) argued instead for a quite general aver­sion to waste, a category mistakenly expanded to include par­tially completed projects or previously incurred costs. Staw (1976) and Aronson (1984) offered accounts based on self-justification, whereas Kiesler (1971) saw behavioral commit­ment as the central mechanism. Brockner (1992) presented a multitheoretical perspective.

Overall, then, the sunk cost effect and its relatives seem obviously worrying and possibly widespread. There is, how­ever, a suggestion that we may be lumping together several rather different effects, each driven by a complex psychology of its own.

Date: 2016-03-03; view: 789