Management Accounting Research 25 (2014) 223–229 Contents lists available at ScienceDirect Management Accounting Research journal homepage: www.elsevier.com/locate/mar Voluntary disclosure in a bargaining setting: A research note夽 Ulrike Denker a , Steven Schwartz b,c,∗ , Christopher Ward a , Richard Young d a b c d Ernst and Young, New York, NY, United States School of Management, Binghamton University, United States Ernst and Young, Chicago, IL, United States Fisher College of Business, The Ohio State University, United States a r t i c l e i n f o Keywords: Disclosure Fairness Mini-ultimatum game a b s t r a c t We conduct an experiment on voluntary disclosure within a simple bargaining setting wherein a proposer must choose one of two possible offers and a responder chooses whether to reject or accept that offer. In one treatment the proposer has the option to disclose whether a fairer (more equal) offer was available relative to the one chosen. Under standard economic theory, a responder will interpret no disclosure to mean the proposer’s offer was the less fair alternative, and so a proposer who is making the fairer offer will disclose. In consequence, voluntary disclosure should perform as well as mandatory disclosure in motivating proposers to make fair offers. Given their rejection rates, we find responders properly infer the meaning of non-disclosure. However, despite the correct inferences made by responders, proposers submit twice as many fair offers with mandatory disclosure than with voluntary disclosure. Our results suggest that the choice of voluntary versus mandatory disclosure has consequences for resource allocation within the firm even though under standard assumptions about preferences it should not. © 2013 Elsevier Ltd. All rights reserved. 1. Introduction In this note we present a bargaining experiment with implications for intra-firm resource allocations. Intra-firm allocations are often the result of bargaining outcomes; examples include budgeting, transfer pricing and salary adjustments. Bargaining becomes especially important when contracts are incomplete or implicit.1 Executive 夽 We thank participants at the 2012 annual meetings of the American Accounting Association and the Economic Science Association for helpful comments. The authors are grateful for the financial support of The Ohio State University Departments of Psychology and Accounting & MIS, and research assistance from Ting Qu. ∗ Corresponding author at: School of Management, Binghamton University, United States. Tel.: +1 6077772102. E-mail address: sschwart@binghamton.edu (S. Schwartz). 1 There are efficiency reasons to make contracts implicit rather than explicit; some researchers have argued that organizations have competitive advantages over markets and courts in keeping track of contextual details and history which are relevant in determining how to effectively settle disputes (Glover, 2012). 1044-5005/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mar.2013.10.003 compensation is a ready example; corporate Boards of Directors often have significant discretion over the allocation of the bonus pool among the firm’s various executives (Murphy and Oyer, 2003). Standard economic analyses of the bargaining process usually assume individuals only care about their own wealth. However, a robust set of observations has shown that individuals care not only for their own wealth but have preferences for distributional fairness (Fehr and Schmidt, 1999), honesty (Gneezy, 2005), and reciprocity (Fehr and Gächter, 2000). Evidence from bargaining experiments in an institutionally richer setting reveals similar results. For example, Kachelmeier and Towry (2002) found that equity concerns affected transfer prices in face-to-face negotiations. Also, Evans et al. (2001) and Zhang (2008) found that an agent’s willingness to be honest was partially a function of how reporting would affect the allocation of wealth between principal and agent. Finally, Charness (2004) found that the extent to which an employee was willing to work hard when offered a generous wage was affected by whether the employer chose the wage or whether some 224 U. Denker et al. / Management Accounting Research 25 (2014) 223–229 outside agency (such as a bingo draw) determined the wage. Because individuals care not only about their own wealth but also the wealth of others, there are situations where information about the allocation of wealth can play a crucial role in bargaining outcomes. Notably, consider a bargaining game where an individual (responder) knows the amount offered but not the amount retained by the person making the offer (proposer). Under standard theoretical assumptions, the size of the retained amount should be irrelevant. However, research has shown that in private information bargaining games the less informed party does worse than in the full information version (Rapoport et al., 1996). Presumably, had lesser-informed responders known how relatively small their shares were, they might have rejected some offers. Further, fear of rejection would have caused the better-informed proposers to increase their offers. Consider also the experiment by Güth et al. (2001), who find that principals give agents with different skill endowments more similar contracts when agents could observe each other’s contracts than when contracts were private information. Their interpretation is principals fear differences in pay will cause costly resentment from the lower-paid agent, even if that agent is less skilled. We extend prior research on bargaining and information structure by allowing one bargainer to make his private information known to another bargainer. That is, we study the role of voluntary disclosure. We assume that any disclosures made must be truthful, perhaps due to the presence of internal auditing, but that there is discretion regarding whether to make a disclosure.2 Under the typical assumptions about markets, (credible) voluntary disclosure should be as effective as mandatory disclosure in informing market participants. The reasoning is straightforward. If a seller does not disclose his private information, potential buyers will assume the worst possible state (Grossman, 1981; Milgrom, 1981). Therefore, in equilibrium only the worst information is withheld, and market participants’ beliefs are consistent.3 Extending this thinking to bargaining settings, where fairness considerations often come to the foreground, voluntary disclosure should also be effective as mandatory disclosure in conveying information – lack of disclosure would be properly interpreted as an unfair offer. As a practical example of the role of voluntary disclosure in bargaining settings, consider the recent labor dispute between the National Football League and its players’ union. The union requested that team owners provide audited financial statements to prove they needed financial relief. When the owners were reluctant to comply, an NFL players’ union representative was quoted as saying “There’s a level of distrust until they prove it – until they show us the books” (La 2 Our assumptions are similar to those in Penno (1990), who assumes managers have a wide degree of discretion in making intra-firm reports, but that reports are constrained to be truthful due to the internal audit function. 3 Complications can arise in capital markets if the information is proprietary or if market participants are uncertain as to whether the firm has received information (Dye, 1985, 1986; Wagenhofer, 1990). See King and Wallin (1991a,b, 1995) for experimental evidence. Conforna, 2011). The idea behind the players’ union position is simple: had the audited information indicated that the owners’ offer was fair, it would have been disclosed.4 With respect to a richer institutional setting, consider the study on fairness and transfer pricing by Luft and Libby (1997). Respondents to a questionnaire indicated that if they were managing the purchasing division they would not expect the transferring division to make a large profit at their expense. For example, if an item had an outside price of $500 the purchasing division manager should be willing to pay up to $500, regardless of the transferring division’s marginal cost to manufacture. However, if the transferring division had a marginal cost to manufacture of $300, the purchasing division manager might expect to only pay $450, sharing some in the cost savings from internal manufacture with the transferring division. Kachelmeier and Towry (2002) produce generally the same findings using monetary incentives. In practice, a manufacturing division surely collects marginal cost information (or should), so if a manufacturing division manager decides not to make this information available to the purchasing division, the purchasing division manager might assume that an “unfair” decision is being hidden. Our hypotheses rely on similar reasoning. The basis for our investigation is an experiment using the mini-ultimatum game (Bolton and Zwick, 1995), which is very simple and has been employed in previous experiments to manipulate perceptions of fairness. A proposer has two choices on how to allocate a sum of money. After the proposer chooses one of the options the responder may accept or reject the offer. Prior experiments provide evidence that responders consider the fairness of an offer in terms of (1) how it allocates the surplus and (2) the alternative allocation that could have been offered (Brandts and Sola, 2001; Falk et al., 2003). We modify this game by introducing information asymmetry regarding the potential offers available to the proposer. In each of two possible mini-ultimatum games the proposer can offer (8,2), keeping 8 for herself and giving 2 to the responder. The alternative offer, chosen by nature and observed only by the proposer, is either (5,5) or (10,0). Conditional on the proposer offering (8,2), if the proposer is able to communicate her alternative offer but chooses not to, the responder should conclude it was (5,5). The reasoning is that had the alternative been (10,0) the proposer surely would have made this known so as to demonstrate that (8,2) was the “fairer” of the two possible offers, and hence increase the chances the responder will accept the offer. In our control treatment, MAN (Mandatory Disclosure), the responder always knows the alternative offer. That is, there is no information asymmetry. In our manipulation, VOL (Voluntary Disclosure), the proposer has the option, but not the obligation, to disclose the alternative offer, 4 This situation generalizes to many employer–employee salary negotiations where the employer can make claims regarding the overall profitability of company and how this impacts potential salary offers. The employee would presumably want a verification of the employer claims to judge the fairness of any salary offer. U. Denker et al. / Management Accounting Research 25 (2014) 223–229 where disclosure is constrained to be truthful. To the extent that responder behavior creates incentives for proposers to offer fairer allocations, offers should be similar with voluntary and mandatory disclosure, because responders will treat a choice of non-disclosure and an (8,2) offer as if the alternative was a more equal allocation (5,5). Our results, however, differ. While we do find that responders in VOL on average properly interpret the choice of non-disclosure, proposers in VOL make fairer offers much less frequently than in MAN. Our results suggest that fairness may not be as well served by voluntary disclosure as would be predicted by the standard theory. 2. Design and hypotheses Milgrom (1981) and Grossman (1981) theorize that in market settings where credible disclosure is available but not mandatory, all but the worst types have strict incentives to disclose because non-disclosure would be inferred to mean the worst type. Therefore, the meaning of nondisclosure can be perfectly deduced. Our experiment is designed to extend the Milgrom (1981) and Grossman (1981) reasoning to fairness attributes in a bargaining setting. The bargaining literature indicates that the fairness construct is complicated. What is fair is context specific. Fehr and Schmidt (1999) construct a model wherein the only attribute of relevance in bargaining outcomes is the final distribution of wealth. In the model individuals have positive marginal utility for their own wealth and also for an equal distribution of wealth, the latter being a representation of fairness. In contrast, Falk et al. (2008) go farther in arguing that the distribution of wealth is not all that matters. They posit that an offer that is lopsided in favor of the proposer may be more palatable if the responder understands it is the fairest offer that could have feasibly been made. Experimental studies, such as Brandts and Sola (2001) and Falk et al. (2003) have found support for this theory. The underlying rationale behind these studies is intentions matter. An offer that is as fair as possible is evidence of an intent to be fair, whereas an identical offer that is not as fair as possible is evidence of intent to be unfair. The analog to Milgrom (1981) and Grossman (1981) in a bargaining setting where credible disclosure is available is all but those making the least fair offer have incentives to disclose this. Therefore, as in the market settings, the meaning of non-disclosure can be perfectly disclosed. Although there is a significant and growing body of accounting related research on fairness, most of that literature treats the information regime as exogenous. Examples include Kim et al. (2005), Libby (2001) and Matuszewski (2010). We endogenize disclosure and examine how disclosure decisions about alternative feasible offers affects bargaining outcomes. In order to test our theory we use the mini-ultimatum game where there are two potential offers. We administer two treatments of our experiment, MAN and VOL, in which there is a sum of $10 US to be distributed between a proposer and a responder. In both treatments one of the proposer’s options is to offer an allocation of (8,2), and the alternative offer is either (10,0) or (5,5), equally likely. If 225 the responder accepts, the experimenter implements the distribution; if the responder rejects both proposer and responder receive zero. In the MAN treatment, the responder is informed about the proposer’s alternative offer. In the VOL treatment, proposers have the option of whether to disclose the available alternative offer. Disclosures must be truthful. The experiment was conducted at a large Midwestern university with student volunteers. Our use of student volunteers seems appropriate in that the students were truly volunteers and the theory tested does not rely upon industry-specific expertise (Croson, 2005). Two sessions of each treatment were run, with either 18 or 20 students participating in a session. The total number of participants in each treatment was 38: 19 proposers and 19 responders. Participants were presented with written instructions upon arrival. Instructions were then read aloud by the experimenters; a quiz was then administered to ensure understanding. Participants kept their role as proposer or responder throughout the experiment. Five rounds of the game were played, using a turnpike design to mitigate contagion. Participants recorded their choices on paper and were informed of the pair-members choice before moving to the next round. Cash payments, based on accumulated points, were made privately at the end of the session, and averaged approximately $18. We formulate six specific hypotheses regarding our experiment. We begin with expectations on responder behavior, because in ultimatum settings responder behavior is an important driver of proposer behavior, more so than the reverse (Winter and Zamir, 2005). As discussed, (8,2) offers should be more appealing to the responder when accompanied by a disclosure that the alternative was (10,0) than when accompanied by a disclosure that the alternative was (5,5). Simply put, an offer that is as fair as possible is more appealing than offer that is not as fair as possible, even if the offers are nominally identical. Hypothesis 1. Responders in both treatments are more likely to accept an offer of (8,2) when the alternative is disclosed to be (10,0) than when it has been disclosed to be (5,5). Continuing with responder behavior, any (8,2) offer made without a disclosure should be rationally viewed as coming from an alternative of (5,5), for had the alternative been (10,0) the proposer would have been inclined to disclose it. Therefore, we would expect a responder to treat an offer of (8,2) without a disclosure the same as an offer of (8,2) with a known alternative of (5,5). Hypothesis 2. Within VOL, responders are as likely to accept an offer of (8,2) when the alternative is disclosed to be (5,5) than when it is not disclosed. As stated in Hypothesis 1, we expect offers of (8,2) to be accepted less often when the alternative is disclosed to be (5,5) than when the alternative is disclosed to be (10,0). Further, if offers of (8,2) without disclosed alternatives are treated identically to offers of (8,2) with disclosed alternatives of (5,5), then a direct implication is that an offer of (8,2) with a disclosed alternative of (10,0) will be more 226 U. Denker et al. / Management Accounting Research 25 (2014) 223–229 likely to be accepted than an offer of (8,2) without a disclosure. We make this an explicit hypothesis because it is possible (for reasons unforeseen) that either of the first two hypotheses may not be supported. Finally, we expect this relation to hold whether the disclosure of an alternative is voluntary or mandatory. Hypothesis 3a. Within VOL, responders are more likely to accept an offer of (8,2) when the alternative is disclosed to be (10,0) than when it is not disclosed. Hypothesis 3b. Responders in MAN are more likely to accept an offer of (8,2) when the alternative is (10,0) than responders in VOL to whom the alternative has not been disclosed. The next hypothesis pertains to the proposer’s willingness to disclose the alternative offer. As noted above, the responder should view an offer of (8,2) as fairer if the proposer’s alternative was (10,0), than if the alternative was (5,5). If the proposer anticipates this “framing effect”, we would expect the proposer to choose to disclose with greater relative frequency when the alternative was (10,0) than when it was (5,5). Hypothesis 4. Within VOL, conditional on an offer of (8,2) proposers are more likely to disclose the alternative of (10,0) than (5,5). We can summarize the first four hypotheses as follows. Responders are more likely to accept offers of (8,2) with a disclosed of an alternative of (10,0) then either offers of (8,2) with a disclosed alternative of (5,5) or offers of (8,2) without a disclosed alternative (because the implied alternative will be (5,5)). Proposer disclosure behavior will be consistent with these expectations. If (8,2) offers are only viewed favorably by responders when the alternative is disclosed to be (10,0) and responders can infer undisclosed alternatives are (5,5), then we expect more offers of (8,2) when the alternative is (10,0) than when the alternative is (5,5). Hypothesis 5. Proposers are more likely to offer (8,2) if the alternative is (10,0) than if it is (5,5). Finally, given the above reasoning that a responder should be able to infer the alternative offer when disclosure is voluntary, and that responder treatment of undisclosed alternative will be that same as disclosed alternatives of (5,5), whether voluntary or mandatory, we expect no differences in the frequency of (5,5) offers between the treatments. Hypothesis 6. The relative frequency of (5,5) offers will be equal in VOL and MAN. Although the rationale for our expectations is straightforward, behavior in bargaining settings is not always predictable, as seen for example with the claiming effect (Larrick and Blount, 1997) and the effect of earned property rights (Hoffman et al., 1994). Therefore, empirical validation is important. Table 1 Results from the experiment. Panel A: frequency of (8,2) offers Treatment VOL Alternative (10,0) (5,5) (10,0) MAN (5,5) Relative frequency of offer of (8,2) 1.00 n = 47 .71 n = 48 .96 n = 46 .37 n = 49 Panel B: disclosure choices in voluntary disclosure when (8,2) is offered Alternative (10,0) (5,5) Relative frequency of disclosure .98 n = 47 .21 n = 34 Panel C: acceptance rates of (8,2) offers Treatment VOL Disclosure None (10,0) (5,5) (10,0) MAN (5,5) Relative frequency of acceptance .68 n = 28 .94 n = 46 1.00 n=7 .86 n = 44 .72 n = 18 Panel A: The relative frequency of (8,2) offers is partitioned by treat# of (8, 2) offers made/total # of offers made, ment and alternative offer, with n = total # of offers made. Panel B: The relative frequency that the alternative was disclosed in VOL is # of disclosed alternatives/# of (8, 2) offers, partitioned by alternative offer, with n = total # of (8,2) offers. Panel C: The relative frequency of accepted (8,2) offers is # of accepted (8, 2) offers/total # of (8, 2) offers, partitioned by treatment and alternative offer, with n = total # of (8,2) offers. 3. Results Hypothesis 1 predicts that in both VOL and MAN the acceptance of offers of (8,2) would be greater when the disclosed alternative offer is (10,0) than when it is (5,5). Table 1 Panel C indicates that in VOL we do not find support for Hypothesis 1. All seven of the (8,2) offers where the alternative was disclosed to be (5,5) were accepted, whereas 94% were accepted when it was disclosed to be (10,0). However, in MAN we do find support for Hypothesis 1: 86% of the (8,2) offers were accepted conditional on an alternative of (10,0), while 72% were accepted conditional on an alternative of (5,5), which is significant at p = .07, using generalized estimating equations to control for lack of independence for observations from the same participant.5 Hypothesis 2 addresses the response to non-disclosure. Specifically, an offer of (8,2) combined with non-disclosure should be an indicator that the alternative was (5,5), implying the proposer could have been fairer. However, Table 1 5 For within-treatment hypotheses, if there is a full set (or almost a full set) of matched pairs available, the tests are paired t-tests where each participant counts as a single independent observation. However, if there are a significant number of incomplete pairs, the generalized estimating equations of approach of Zeger and Liang (1986) is used to control for multiple observations from the same participant, operationalized using the GENMOD procedure in SAS. For between treatment hypotheses, a two-sample t-test is used where each participant counts as a single independent observation. p-Values for all hypothesis tests are one-sided, except where noted. U. Denker et al. / Management Accounting Research 25 (2014) 223–229 Panel C shows that acceptance of offers of (8,2) without a disclosure in VOL seems considerably lower (68%), than when an alternative of (5,5) is disclosed (100%).6 Given the small sample size of voluntarily disclosed (5,5) alternatives when (8,2) is offered (n = 7), it is difficult to draw any conclusions.7 Perhaps responders in VOL were rewarding proposers for their honesty, in the same spirit as Forehand and Grier (2003). Further experimentation would be necessary to shed more light on this unexpected result. Hypotheses 3a and 3b further address the response to non-disclosure. We expected non-disclosure to be treated by responders as a disclosure of an alternative of (5,5). Indeed, of the 28 undisclosed alternatives, 27 of them were (5,5). The difference between acceptance rates for voluntary disclosure of (10,0) in VOL, 94%, and non-disclosure in VOL, 68%, is significant at p < .01, using generalized estimating equations. Similarly, the difference between acceptance rates for non-disclosure in VOL, 68%, and mandatory disclosure of (10,0) in MAN, 86%, is significant at p = .02 using a two-sample t-test. Overall, these results indicate that responders believe that if a proposer offers (8,2) and had an alternative of (10,0), he would have disclosed this, presumably to demonstrate he was being as fair as possible. Hypothesis 4 deals with proposers’ disclosure choices. It predicts that conditional on an offer of (8,2), the relative frequency of disclosure in VOL is higher when the alternative is (10,0) than when the alternative is (5,5). Table 1 Panel B displays the disclosure behavior of proposers in VOL. The difference, 98% for an alternative of (10,0) versus 21% for an alternative of (5,5), is significant at p < .01 using a paired t-test, providing support for Hypothesis 4. Hypothesis 5 addresses with proposers’ offers. Table 1 Panel A illustrates that, as predicted, in each treatment (8,2) is offered more frequently when the alternative is (10,0) than when the alternative is (5,5). For example, in MAN an offer of (8,2) was made in 96% of the cases where the alternative was (10,0) but in only 37% percent of the cases where the alternative was (5,5). The difference in frequency is significant (p < .01) using a paired t-test, providing support for Hypothesis 1.8 The above results regarding Hypotheses 1–5 are generally consistent with our expectations, with the exception of the high acceptance rate of (8,2) offers when an alternative of (5,5) is voluntarily disclosed. More importantly, the acceptance rate for (8,2) offers (68%) is almost identical when proposers choose not to disclose as compared to when they are forced to disclose the alternative is (5,5) (72%). This suggests that responders properly infer the 6 Both Brandts and Sola (2001) and Falk et al. (2003) have full disclosure settings with payoffs equivalent to the (5,5) sub-game. Brandts and Sola report frequency of rejection of the (8,2) equal to 17% while Falk et al. report frequency of the rejection of the (8,2) offer equal to 44%. Therefore our observed rejection rate of 28% in the Mandatory Disclosure treatment appears reasonable. Interestingly the observed frequency of rejection of the (8,2) offer in the (5,5) sub-game in both the Brandts and Sola (2001) experiment and our experiment indicates that the proposer would maximize earnings with an (8,2) offer rather than a (5,5) offer. 7 The GENMOD procedure described in footnote 5 failed to produce a p-value. 8 The difference within VOL alone is also significant at p < 0.01 using a paired t-test. 227 meaning of non-disclosure, which is the transcending idea in the theoretical literature (Grossman, 1981; Milgrom, 1981) as applied to our bargaining setting. Therefore, we would expect voluntary disclosure is as effective as mandatory disclosure in promoting fairness. In fact, this is our rationale for Hypothesis 6, which conjectures that the frequency of (5,5) offers would be the same in MAN and VOL. It is rather surprising then that (5,5) offers are made more than twice as frequently in MAN as in VOL – 65% versus 29%. This difference is significant at p = .02 (two-tailed), using a two-sample t-test. In order to put our finding regarding the frequency of fair offers into perspective, we compare our results to the reported results of similar experiments. Both Brandts and Sola (2001) and Falk et al. (2003) implemented a full disclosure version of our game. Brandts and Sola found that (5,5) offers were made in 54% of the their observations, while Falk et al. found that (5,5) offers were made in 69% of their observations. Our result within the mandatory disclosure treatment is similar: when possible, 65% of offers were fair. However, when possible, only 29% of offers were fair in voluntary disclosure. Given our replication of previous results that imposed disclosure, we are more confident that it is the presence of voluntary disclosure that is causing some of the deviations from prior experimentation. We discuss a possible reason for the effect of voluntary disclosure on fair offers in the next section. 4. Discussion and conclusion Experiments on managerial accounting and intra-firm resource allocation have tended to manipulate either incentive structures (including negotiation protocol and bargaining power) or information regime. Some experiments falling in the former classification include Waller (1988), Fisher et al. (2000), Evans et al. (2001), Coletti et al. (2005) and Rankin et al. (2008). Some experiments in the latter category include Young (1985), Stevens (2002), Hannan et al. (2006) and Nikias et al. (2009). Our experiment falls in that part of the literature that studies the effect of information regime. We manipulate whether proposers, who offer a split of wealth, must disclose the alternative split not offered, or have a choice on disclosure. Hence, within a bargaining setting, we study, two information regimes; mandatory disclosure and voluntary disclosure. We find that responders (correctly) see through the absence of disclosure and treat it as if a less fair offer than possible was made. Despite this, proposers, when given the choice on disclosure, make fewer fair offers than when disclosures are mandatory, and accompany the less fair offers with lack of disclosure. There are two possible explanations. The first is that with enough experience, proposers will learn that the meaning of nondisclosure is perfectly inferred by responders and act like proposers in the mandatory disclosure treatment. Even if this explanation holds, many intra-firm interactions are of limited durations and so the study of early repetitions is interesting. The second explanation, discussed below, is related to the idea of image management. One of the more robust findings from experiments on information regime is that individuals are less willing to lie 228 U. Denker et al. / Management Accounting Research 25 (2014) 223–229 (up to a point), the more obvious their lies become (Young, 1985; Chow et al., 1988; Hannan et al., 2006). A related result is found in Irlenbusch and Sliwka (2005), wherein trustees in a gift-exchange game condition their responses less on the size of the gift when the trustor cannot perfectly infer their action than when they can. The idea linking these studies is that the less an individual’s actions are observable, the less that individual adheres to social norms. Hannan et al. (2006) consider this a form of image management, where individuals do not want to be detected failing to adhere to social norms and will take advantage of noisy disclosure to hide their actions. Conceivably, proposers using voluntary disclosure may view non-disclosure as enough of a veil to hide their unfair action. The strength of our design resides in (1) the comparability to earlier experiments, namely Brandts and Sola (2001) and Falk et al. (2003), (2) the straight-forward predictions derived from standard agency analysis, as recommended in Brown et al. (2009), and (3) the transparency to studentparticipants. However there are also important limitations. We only consider one-sided information asymmetry. In many intra-firm bargaining situations both parties have private information. Also, we consider a very limited negotiation protocol, with only a single, take-it-or-leave-it offer. Finally, we implicitly consider only perfect, costless verification. Future studies, which both increase the length of the experiment and add post experiment questionnaires, may help distinguish between our proposed explanations for the decrease in fair offers found with voluntary disclosure. Future experimentation may also explore each of the limitations described above by expanding our basic setting. For example, an experiment may feature a responder with an outside option that is private information as well as a proposer with private information on proposal alternatives. Should our findings be robust to future experimentation, it may speak to why some disclosures are mandated. Lack of fairness, even when negotiations are viewed as a zero sum game, might in some circumstances have negative repercussions, such as difficult-to-detect retaliation against the firm.9 In such cases unfair outcomes should be avoided, at least if the cost of doing so is low. References Bolton, G., Zwick, R., 1995. Anonymity versus punishment in ultimatum bargaining. Games and Economic Behavior 10 (1), 95–121. Bewley, T., 1999. Why Wages Don’t Fall During A Recession. Harvard University Press, Cambridge, MA. Brandts, J., Sola, C., 2001. Reference points and negative reciprocity in simple sequential games. Games and Economic Behavior 36 (2), 138–157. Brown, J., Evans III, J.H., Moser, D., 2009. Agency theory and participative budgeting experiments. Journal of Management Accounting Research 21 (1), 317–345. Charness, G., 2004. Attribution and reciprocity in labor economics. Journal of Labor Economics 22 (3), 665–688. 9 The need for internal equity within the firm is considered vital for morale and productivity. This helps to explain the results earlier cited in Güth et al. (2001). Bewley (1999, p. 477) conducts interviews with managers and states, “Managers regarded any violation of internal equity as potentially disruptive. Lack of equity spawns jealousies, resentments, and perceptions of unjust treatment.” Chow, C., Cooper, J., Waller, W., 1988. Participative budgeting: effects of truth-inducing pay scheme and informational asymmetry on slack and performance. The Accounting Review 63 (1), 111–122. Coletti, A., Sedatole, K., Towry, K., 2005. The effect of control systems on trust and cooperation in collaborative environments. The Accounting Review 80 (2), 477–500. Croson, R., 2005. The method of experimental economics. International Negotiation 10 (1), 131–148. Dye, R., 1985. Disclosure of nonproprietary information. Journal of Accounting Research 23 (1), 123–145. Dye, R., 1986. Proprietary and nonproprietary disclosures. Journal of Business 59 (2), 331–336. Evans III, J.H., Hannan, R.L., Krishnan, R., Moser, D., 2001. Honesty in managerial reporting. The Accounting Review 76 (4), 537–559. Falk, A., Fehr, E., Fischbacher, U., 2003. On the nature of fair behavior. Economic Inquiry 41 (1), 20–26. Falk, A., Fehr, E., Fischbacher, U., 2008. Testing theories of fairness – intentions matter. Games and Economic Behavior 62 (1), 287–303. Fehr, E., Gächter, S., 2000. Fairness and retaliation: the economics of reciprocity. Journal of Economic Perspectives 14 (3), 159–181. Fehr, E., Schmidt, K., 1999. A theory of fairness, competition and cooperation. Quarterly Journal of Economics 114 (3), 817–868. Fisher, J., Frederickson, J., Peffer, S., 2000. Budgeting: an experimental investigation of the effects of negotiation. The Accounting Review 75 (1), 93–114. Forehand, M., Grier, S., 2003. When is honesty the best policy? The effect of stated company intent on consumer skepticism. Journal of Consumer Psychology 13 (3), 349–357. Glover, J., 2012. Explicit and implicit incentives for multiple agents. Working Paper. Carnie Mellon University. Gneezy, U., 2005. Deception: the role of consequences. American Economic Review 95 (1), 384–394. Grossman, S., 1981. The informational role of warranties and private disclosure about product quality. Journal of Law and Economics 24 (3), 461–483. Güth, W., Königstein, M., Kovács, J., Zala-Mezõ, E., 2001. Fairness within firms: the case of one principal and many agents. Schmalenbach Business Review 53 (April), 82–101. Hannan, R.L., Rankin, F.W., Towry, K., 2006. The effect of information systems on honesty in managerial reporting: a behavioral perspective. Contemporary Accounting Research 23 (4), 885–918. Hoffman, E., McCabe, K., Shachat, K., Smith, V., 1994. Preferences, property rights and anonymity in bargaining games. Games and Economic Behavior 7, 346–380. Irlenbusch, B., Sliwka, D., 2005. Transparency and reciprocal behavior in employment relations. Journal of Economic Behavior in Organization 56 (3), 383–403. Kachelmeier, S., Towry, K., 2002. Negotiated transfer pricing: is fairness easier said than done? The Accounting Review 77 (3), 571–593. Kim, C., Evans, J., Moser, D., 2005. Economic and equity effects on tax reporting decisions. Accounting Organizations and Society 30 (7–8), 609–625. King, R., Wallin, D., 1991a. Market-induced information disclosures: an experimental markets investigation. Contemporary Accounting Research 8 (1), 170–197. King, R., Wallin, D., 1991b. Voluntary disclosures when seller’s level of information is unknown. Journal of Accounting Research 29 (1), 96–108. King, R., Wallin, D., 1995. Experimental tests of disclosure with an opponent. Journal of Accounting and Economics 19 (1), 139–168. La Conforna, J., 2011. Players pledge unity, call for financial data at annual meetings. NFL.com, Available at: http://www.nfl.com/news/story/ 09000d5d81ece0d8/printable/players-pledge-unity-call-forfinancial-data-at-annual-meetings Larrick, R., Blount, S., 1997. The claiming effect: why players are more generous in social dilemmas than in ultimatum games. Journal of Personality and Social Psychology 72 (4), 810–825. Libby, T., 2001. Referent cognitions and budgetary fairness: a research note. Journal of Management Accounting Research 13, 91–105. Luft, J., Libby, R., 1997. Profit comparison, market prices and manager’s judgments’ about negotiated transfer prices. The Accounting Review 72 (2), 217–229. Matuszewski, L., 2010. Honesty in managerial reporting: Is it affected by perceptions of horizontal equity? Journal of Management Accounting Research 22 (1), 233–250. Milgrom, P., 1981. Good news and bad news: representation theorems and applications. Bell Journal of Economics 12 (2), 380–391. Murphy, K., Oyer, P., 2003. Discretion in executive incentive contracts. Working Paper. University of Southern California. U. Denker et al. / Management Accounting Research 25 (2014) 223–229 Nikias, A., Schwartz, S., Spires, E., Wollscheid, J., Young, R., 2009. The effects of aggregation and timing on budgeting: an experiment. Behavioral Research in Accounting 22 (1), 67–83. Penno, M., 1990. Auditing for performance evaluation. The Accounting Review 65 (3), 520–536. Rankin, F., Schwartz, S., Young, R., 2008. The effect of honesty and superior authority on budget proposals. The Accounting Review 83 (4), 1083–1099. Rapoport, A., Sundali, J., Seale, D., 1996. Ultimatums in two-person bargaining with one-sided uncertainty: demand games. Journal of Economic Behavior and Organization 30 (2), 173–196. Stevens, D., 2002. The effects of reputation and ethics on budgetary slack. Journal of Management Accounting Research 14 (1), 153–171. Wagenhofer, A., 1990. Voluntary disclosure with a strategic opponent. Journal of Accounting and Economics 12 (4), 341–363. 229 Waller, W., 1988. Slack in participative budgeting: the joint effect of a truth-inducing pay scheme and risk preferences. Accounting, Organizations and Society 13 (1), 87–98. Winter, E., Zamir, S., 2005. An experiment on the ultimatum bargaining in a changing environment. Japanese Economic Review 56 (3), 363–385. Young, S., 1985. Participative budgeting: the effects of risk-aversion and asymmetric information on budgetary slack. Journal of Accounting Research 23 (2), 829–842. Zeger, S., Liang, K., 1986. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 42 (1), 121–130. Zhang, Y., 2008. The effects of perceived fairness and communication on honesty and collusion in a multi-agent setting. Accounting Review 83 (4), 1125–1146.