Anchor Points, Reference Points, and Counteroffers in Negotiations Henrik Kristensen Tommy Gärling Department of Psychology Department of Psychology Göteborg University, and Göteborg University Department of Business Administration Göteborg School of Economics and Commercial Law Kristensen, H. & Gärling, T. Anchor points, reference points, and counteroffers in negotiations. Göteborg Psychological Reports, 1997, 27, No. 7. The aim was to test the hypotheses that in negotiations counteroffers are generated through an anchoring-and-adjustment process leading to an effect of the anchor point, and that counteroffers are influenced by changes in reference point which in turn determine whether the anchor point is perceived as a gain or a loss. In a simulated price negotiation 96 undergraduate students of business administration playing the role of buyers were presented a proposed selling price and were requested to generate a counteroffer which the seller would accept or reject. The reference point was manipulated by making an initial offer higher, lower, or equal to the subsequent proposed selling price and by providing an estimated market price which was above or below the proposed selling price. The results showed that the proposed selling price operated as an anchor point in that it affected the counteroffers. In support of the hypothesis that anchor and reference points jointly influence counteroffers, it was also shown that adjustment from the anchor point (the proposed selling price) was larger, in both monetary units and with respect to rated satisfaction, when the anchor point was perceived as a loss than when it was perceived as a gain.
Key words: Negotiation, consumer choice, anchoring. In negotiations high claims and low offers are made in attempts to mislead opponents. Such attempts can be effective even though the deceptive message is neither accepted nor even believed. As suggested by Kahneman (1992), an anchoring-and-adjustment process may be involved. It has been demonstrated that an arbitrarily chosen anchor point will significantly influence a value estimate because the latter is insufficiently adjusted away from the anchor
Author note: This research was financially supported by grant #F77/95 to the second author from the Swedish Council for Research in the Humanities and Social Sciences. The authors thank Eric Johnson and David Messick for valuable comments.
No. 7:27, 2
point toward the true value (Slovic & Lichtenstein, 1971; Tversky & Kahneman, 1974). Significant qualifications are, however, that effects of anchor points are likely to be weaker in the presence of other information, and that anchor points far outside the distribution of possibilities indicated by other evidence are likely to have reduced impact (Chapman & Johnson, 1994; Markovsky, 1988). Some research has explicitly investigated effects of anchor points in negotiations (Kahneman, 1992; Neale & Bazerman, 1991). For instance, Northcraft and Neale (1987) demonstrated that estimates of the value of a house made by real estate agents and students depended on information about the list prices. Both groups were equally affected by this information although experts denied the influence. In fact, susceptibility to the anchor-point effect may influence a negotiation process in a number of ways (Neale & Bazerman, 1991), perhaps providing a partial explanation for the recurrent finding that final agreements in negotiations are more strongly influenced by initial offers than by subsequent concessions (Liebert, Smith, Hill, & Keiffer, 1968; Rubin & Brown, 1975; Yukl, 1974). In price negotiations, it is easy to imagine that a proposed selling price is an anchor point from which a counteroffer is generated. In accordance with the anchor-point effect (e.g., Tversky & Kahneman, 1974), the proposed selling price should then be expected to influence a counteroffer. However, other information may exist, such as an estimated market price, that also influences a counteroffer. In line with the theory of Hogarth and Einhorn (1990), the net effect of adjustments from an anchor point may reflect the relative weight given in imagination to values above or below the anchor point. Individual factors, situational factors, and, in particular, the sign and size of payoffs constitute potential influences on the direction and magnitude of adjustments. The distinction between anchor and reference points should be noted (Kahneman, 1992). The concept of reference point was introduced in prospect theory (Kahneman & Tversky, 1979; Tversky, & Kahneman, 1991; Tversky & Kahneman, 1992). In this theory unidimensional (e.g., monetary) decision outcomes are coded as gains or losses relative to a reference point. Thus, if negotiators adopt different reference points they may frame offers differently (i.e., as gains or losses). Adoption of a gain or loss frame has empirically been found to affect the outcome of negotiations (Bazerman et al., 1985; Neale & Bazerman, 1985; Neale et al., 1987). In this study it is assumed that an anchor point affects the counteroffer a negotiator makes (Northcraft & Neale, 1987), whereas a reference point determines how an offer is perceived. A primary aim was to investigate the validity of this assumption, which is a first step towards an understanding of the observed effects of anchor and reference points on negotiation processes and outcomes. It has been suggested that negotiators evaluate offers relative to multiple reference points (Kahneman, 1992; Neale & Bazerman, 1991; Neale, Huber, & Northcraft, 1987). However, Kahneman (1992) and White, Valley, Bazerman, Neale, and Peck (1994) argued that negotiators faced with multiple potential reference points simplify and allow only one to dominate. In their study of dyadic bargaining about house prices, White et al. (1994) found that
No. 7:27, 3 reservation price or resistance point (a buyer’s highest acceptable price) was a dominant reference point. In the first of two experiments, Kristensen and Gärling (in press a) found that both an initial offer (an initially proposed price which was later changed) above and an estimated market price below the buyer’s reservation price were adopted as reference points, whereas the buyer´s reservation price did not seem to play any role. However, in a second experiment an induced reservation price was adopted as reference point. At the same time it was shown that the initial offer affected the reference point. Kristensen and Gärling (in press a) therefore suggested that the adopted reference point is the reservation price which changes as a funtion of estimated market price and initial offer. This suggestion was confirmed in an experiment in which subjects indicated their reservation price (Kristensen & Gärling, in press b). Consistent with research showing anchor-point effects that counteroffers are generated through an anchoring-and-adjustment process (Kahneman, 1992; Neale & Bazerman, 1991). Since adjustments in general are insufficient, the anchor point will affect counteroffers. However, counteroffers may also be affected by a reference point which determines whether the anchor point is perceived as a gain or loss. If anchor and reference points play different roles, a testable hypothesis is that adjustments are larger when an anchor point is perceived as a loss than when it is perceived as a gain. As shown in Figure 1, this follows if it is further assumed that adjustments should result in a counteroffer which is perceived as a gain (positive utility uc). In the experiment to be reported, subjects playing the role of buyers in a simulated price negotiation (cf. Kristensen & Gärling, in press a; b) were presented with a proposed selling price and asked to generate a counteroffer which the seller had to accept or reject. In this simulation, the proposed selling price was expected to be adopted as an anchor point. The proposed selling price was varied to test whether the counteroffers would change in the expected directions. As in the previous studies (Kristensen & Gärling, in press a & b), two methods were used to affect the reference point (reservation price). First, the reference point was manipulated by presenting subjects with an initially advertised price (an initial offer) which was higher, lower, or equal to the subsequently proposed selling price. Secondly, the reference point was manipulated by presenting subjects with an estimated market price which was higher or lower than the proposed selling price. It was expected that if a change of reference point made subjects perceive the anchor point as a gain, their counteroffers would be higher than if it made them perceive the anchor point as a loss. As a check of the manipulation of the reference point, subjects were also asked to rate how satisfied they were with the proposed selling price on a scale ranging from dissatisfied (loss) to satisfied (gain).
No. 7:27, 4
(a) Anchor point perceived as a gain Utility
+
uc Ca
Anchor point
$
-
(b) Anchor point perceived as a loss Utility
uc
+
Anchor point
$
-
Cb Figure 1. An illustration of counteroffers (Ca and Cb, both with utility uc) adjusted from an anchor point perceived as either a gain (a) or a loss (b).
No. 7:27, 5
Method Subjects Ninety-six (42 men and 54 women) undergraduate students of business administration participated on a voluntary basis. The subjects’mean age was 22.3 years (SD = 3.8).
Procedure Subjects answered a brief questionnaire in class. On the first page of the questionnaire, subjects were asked to imagine that they intended to buy a condominium. They were then presented some market information indicating the actual price ranges of condominiums in the metropolitan area where they lived. On the basis of this information, subjects indicated the highest price (reservation price) they would pay for a particular condominium described to them. After having indicated their reservation prices, subjects were presented 18 descriptions of different condominiums, each description was given a single page in the questionnaire. In each situation they were asked to imagine that the seller had first advertised the condominium at one price (initial offer) and that the seller then asked another price (proposed selling price) when later contacted by the subject. Subjects were also informed about a market price estimated by an expert. Subjects were randomly assigned to four equally large groups with sex approximately balanced1. In the first group (Group I), the proposed selling price was equal to the initial offer (the advertised price) and higher than the estimated market price. In the second group (Group II), the proposed selling price was equal to the initial offer and lower than the estimated market price. In the third group (Group III), the proposed selling price was lower than the initial offer and higher than the estimated market price. Finally, in the fourth group (Group IV), the proposed selling price was higher than the initial offer and lower than the estimated market price. As exemplified in Table 1, all subject groups were presented a low, medium, or high proposed selling price. For Groups I and III, the proposed selling price was SEK 20,000, SEK 40,000, or SEK 60,000 higher than the estimated market price, whereas for Groups II and IV, the proposed selling price was SEK 20,000, SEK 40,000, or SEK 60,000 lower than the estimated market price. For Group III, the proposed selling price was SEK 20,000, SEK 40,000, or SEK 60,000 lower than the initial offer, whereas for Group IV, the proposed selling price was SEK 20,000, SEK 40,000, or SEK 60,000 higher than the initial offer. 1
There were four instead of six groups because an initial offer and an estimated market price both higher than the proposed selling price or an initial offer and an estimated market price both lower than the proposed selling price would give no information about the relative importance of the initial offer and the estimated market price.
No. 7:27, 6
Table 1 Examples of Low, Medium, and High Proposed Selling Prices (SEK103)a in the Different Subject Groups Initial offer equal Initial offer Initial offer to proposed higher than proposed lower than proposed selling price selling price (320) selling price (240) Estimated Low Medium High Low Medium High Low Medium High market price
Lower (240) Higher (320) a
260
Group I 280
260
Group II 280
300
Group III 260 280 300 Group IV 260 280
300
300
1 SEK (Swedish Crown) is approximately equivalent to 0.13 USD.
Each subject was presented six replicates of the low, medium, and high proposed selling price. In Table 1 one replicate is shown for each group. The other replicates were obtained by incrementing the values in steps of SEK 10,000. The order of presentation of the descriptions in the questionnaire was randomized for each subject. For each of the 18 descriptions subjects first made a rating of how satisfactory or unsatisfactory they considered the proposed selling price to be, then they generated a counteroffer in Swedish crowns which the hypothesized seller would accept or reject. The rating of satisfaction with the proposed selling price was made on a scale ranging from 10 (very unsatisfactory) to 90 (very satisfactory) through 50 (neither satisfactory nor unsatisfactory).
Results The ratings of satisfaction were transformed by subtracting 50 from the scale values. A positive value therefore represents a positive evaluation (gain) and a negative value a negative evaluation (loss). Statistical analyses were performed on averages across the six replicated price levels. Table 2 shows that subjects´ mean ratings of satisfaction decreased with the proposed selling price. Furthermore, indicating an influence on the reference point (reservation price), when the initial offer was equal to the proposed selling price, subjects were more satisfied when the estimated market price was higher and less satisfied when it was lower than the proposed selling price. In further support of the effect on the reference point, the ratings of satisfaction were higher when the initial offer was above the proposed selling price than when they were equal. Conversely, ratings of satisfaction were lower when the
No. 7:27, 7 initial offer was below the proposed selling price than when they were equal. These observations were substantiated by a 2 (estimated market price: lower vs. higher than the proposed selling price) by 2 (initial offer: equal vs. higher/lower than the proposed selling price) by 3 (proposed selling price: low vs. medium vs. high) mixed ANOVA with repeated measures on the last factor. This analysis yielded a highly significant effect of estimated market price, F(1, 92) = 22.36, p < .001, MSe = 796.53, of proposed selling price, F(2, 184) = 221.0, p < .001, MSe = 144.57, of the interaction between estimated market price and initial offer, F(1, 92) = 18.65, p < .001, MSe = 796.53, and of the interaction between estimated market price and proposed selling price, F(2, 184) = 4.83, p < .01, MSe = 144.57. Bonferroni-corrected separate t tests showed that all pairwise mean differences between the proposed selling prices were significant at p = .01. The interaction involving estimated market price and proposed selling price was not significant due to one deviating mean in the condition where the estimated market price was higher than the low proposed selling price. The effect of the initial offer was substantiated by further post hoc tests (p = .05), which showed that the ratings of satisfaction were reliably higher when the initial offer was above the proposed selling price than when they were equal and reliably lower when the initial offer was lower than the proposed selling price. The effect of estimated market price was only significant when the initial offer was equal to the proposed selling price. Subjects´ mean counteroffers are displayed in Table 3. As may be seen, the counteroffers increased with the proposed selling price. Thus, they appeared to be influenced by the anchor point (proposed selling price). When the initial offer was equal to the proposed selling price, the counteroffers were higher for an estimated market price above the proposed selling price and lower for an estimated market price below the proposed selling price. This indicates that the counteroffers were influenced by the reference point. Also supporting an influence of the reference point on the counteroffers, when the initial offer was above the proposed selling price the counteroffers were higher than when the initial offer was equal to the proposed selling price. Conversely, the counteroffers were lower when the initial offer was below the proposed selling price than when they were equal. These effects were statistically confirmed in a 2 (estimated market price: lower vs. higher than the proposed selling price) by 2 (initial offer: equal vs. higher/lower than the proposed selling price) by 3 (proposed selling price: low vs. medium vs. high) mixed ANOVA with repeated measures on the last factor. Significant effects were obtained for estimated market price, F(1, 92) = 14.24, p < .001, MSe = 839.05, proposed selling price, F(2, 184) = 21.83, p < .001, MSe = 322.41, the interaction between estimated market price and proposed selling price, F(2, 184) = 5.36, p < .01, MSe = 322.41, and the interaction between estimated market price and initial offer, F(1, 92) = 3.91, p < .01, MSe = 839.05. In Bonferonni-corrected separate t tests the pairwise mean differences between proposed selling prices were significant at p = .05 for the differences between low and medium and low and high proposed
No. 7:27, 8 Table 2 Ratings of Satisfaction with Low, Medium, and High Proposed Selling Prices which were Higher or Lower than the Estimated Market Price for an Initial Offer Equal to, Above, or Below the Proposed Selling Price Initial offer equal Initial offer Initial offer to proposed higher than proposed lower than proposed selling price selling price selling price Estimated Low Medium High Low Medium High Low Medium High market price Group I Group III Lower 7.1 -8.4 -21.3 14.1 -3.4 -15.6
Higher
25.8
Group II 15.2
0.2
8.9
Group IV -0.2 -10.6
selling prices, respectively, when the estimated market price was above the proposed selling price and the initial offer was equal to the proposed selling price. The pairwise mean difference between low and high proposed selling prices was also significant when the initial offer was below the proposed selling price. The effect of estimated market price was substantiated by further post hoc tests (p = .05), which showed that for the condition where the initial offer was equal to the proposed selling price, the counteroffers were reliably higher when the estimated market price was above the proposed selling price than when it was below. In Figure 2 the mean counteroffers are plotted against the ratings of satisfaction with proposed selling prices (the anchor points) for low, medium, and high anchor points. As may be seen, mean adjustments (indicated by horizontal arrows) away from medium and high anchor points are insufficient. Moreover, the differences in slopes indicate that adjustments are on average larger, in monetary units, when the anchor points are perceived as losses than when they are perceived as gains. Table 3 Mean Counteroffers (SEK103) for Proposed Selling Prices which were Higher or Lower than the Estimated Market Price for an Initial Offer Equal to, Above, or Below the Proposed Selling Price Initial offer equal Initial offer Initial offer to proposed higher than proposed lower than proposed selling price selling price selling price Estimated Low Medium High Low Medium High Low Medium High market price Group I Group III Lower 270.4 274.8 274.8 277.4 281.7 284.8
Higher
277.6
Group II 293.5 300.7
275.3
Group IV 283.6 287.6
No. 7:27, 9
Figure 2. Mean indicated counteroffers plotted against the ratings of satisfaction with proposed selling prices (anchor points). The horizontal arrows represent the adjustments away from the anchor points.
No. 7:27, 10
Discussion In line with the hypothesis, a main finding of the present experiment was that an anchor point influences counteroffers in a simulated negotiation. On average the adjustment effect was quite substantial (Figure 2), varying in monetary units from SEK 7,400 to SEK 50,200. This supports previous results demonstrating that anchor points appreciably influence value estimates (Northcraft & Neale, 1987). Extending such results, the unique finding of the present study was to show that the adjustments from an anchor point were larger, in both monetary units and with respect to rated satisfaction, when subjects adopted a reference point (reservation price) that made them perceive the anchor points as losses instead of gains. In addition to providing support for the conceptual distinction between an anchor and reference point (Kahneman, 1992), the results thereby confirm the present assumption that anchor points affect counteroffers whereas reference points affect how offers are perceived. As predicted (Figure 1), this effect resulted in substantially larger adjustments (in the order of SEK 34,000) when the anchor point was perceived as a loss than when it was perceived as a gain (in the order of SEK 12,000). In line with the results of previous studies (Kristensen & Gärling, in press a; b), it was also shown that subjects´ ratings of satisfaction with the proposed selling prices were influenced both by an initial offer and an estimated market price which were lower or higher, respective, than the proposed selling price. Exactly as would be the case if subjects changed their reference point, the ratings of satisfaction indicated that they were either satisfied (perceived as a gain) or dissatisfied (perceived as a loss) with the same proposed selling price. However, as discussed in Kristensen and Gärling (in press a; b), this does not necessarily mean that the estimated market price or initial offer were adopted as reference points. A reservation price may still be the reference point (White et al., 1994) if it is assumed that the reservation price is influenced by an initial offer and an estimated market price. Although the present results suggest what roles anchor and reference points may play in negotiations, it is important to realize that the results may only apply to negotiations in which there is a single outcome attribute. Furthermore, it is difficult to know how the results generalize to negotiations in which bidding continues beyond the first stage where a negotiator evaluates an offer and generates a counteroffer. Studies of actual negotiations demonstrating framing effects (Bazerman et al., 1985; Neale & Bazerman, 1985; Neale et al., 1987) as well as impacts of initial offers on concession rates (Liebert et al., 1968; Yukl, 1974; Rubin & Brown, 1975) are consistent with the present findings. However, more detailed theoretical and empirical analyses of actual negotiations are needed. The results of the present study constitute an important starting point for such analyses.
No. 7:27, 11
References Bazerman, M. H., Magliozzi, T., & Neale, M. A. (1985). Integrative bargaining in a competitive market. Organizational Behavior and Human Decision Processes, 34, 294-313. Chapman, G. B., & Johnson, E. J. (1994). The limits of anchoring. Journal of Behavioral Decision Making, 7, 223-242. Hogarth, R. M., & Einhorn, H. J. (1990). Venture theory: a model of decision weights. Management Science, 36, 780-803. Kahneman, D. (1992). Reference points, anchors, norms, and mixed feelings. Organizational Behavior and Human Decision Processes, 51, 296-312. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-291. Kristensen, H., & Gärling, T. (in press a). Adoption of cognitive reference point in negotiations. Acta Psychologica. Kristensen, H., & Gärling, T. (in press b). Determinants of buyer’s aspiration and reservation price. Journal of Economic Psychology. Liebert, R. M., Smith, W. P., Hill, J. H., & Keiffer, M. (1968). The effects of information and magnitude of initial offer on interpersonal negotiation. Journal of Experimental Social Psychology, 4, 431-441. Markovsky, D. M., (1988). Anchoring justice. Social Psychology Quarterly, 51, 213-224. Neale, M. A., & Bazerman, M. H. (1985). The effects of framing and negotiation overconfidence on bargaining behaviors and outcomes. Academy of Management Journal, 28, 34-49. Neale, M. A., & Bazerman, M. H. (1991). Cognition and rationality in negotiation. New York: The Free Press. Neale, M. A., Huber, V. L., & Northcraft, G. B. (1987). The framing of negotiations: Contextual versus task frames. Organizational Behavior and Human Decision Processes, 39, 228-241. Northcraft, G. B., & Neale, M. A. (1987). Expert, amateurs, and real estate: An anchoring-and-adjustment perpective on property pricing decisions. Organizational Behavior and Human Decision Processes, 39, 228-241. Rubin, J., & Brown, B. (1975). The social psychology of bargaining and negotiation. New York: Academic Press. Slovic, P., & Lichtenstein, S. (1971). Comparison of Baysian and regression approaches to the study of information processing in judgement. Organizational Behavior and Human Performance, 6, 649-744. Tversky, A., & Kahneman, D. (1974). Judgement under uncertainty: Heuristics and biases. Science, 185, 1124-1131. Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: A reference-dependent model. The Quarterly Journal of Economics, 106, 10391061. Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5, 297-323. White, S. B., Valley, K. L., Bazerman, M. H., Neale, M. A., & Peck, S. R. (1994). Alternative models of price behavior in dyadic negotiations: Market prices,
No. 7:27, 12 reservation prices, and negotiator aspirations. Organizational Behavior and Human Decision Processes, 57, 430-447. Yukl, G. A. (1974). Effects of situational variables and opponent concessions on a bargainer’s perceptions, aspirations, and concessions. Journal of Personality and Social Psychology, 29, 227-236.
