1 Introduction

Negotiators often attempt to build trustful relationships with counterparts through the negotiation, because trust can help them implement the agreement terms (Campagna et al. 2016), reduce transaction costs (Connelly et al. 2018), and strengthen future cooperation (Krishnan et al. 2006). Famed investor Warren Buffett once said: “we only want to link up with people who we like, admire, and trust” (Cunningham and Buffett 2013, p. 408) because “we have never succeeded in making a good deal with a bad person” (Cunningham and Buffett 2013, p. 348). Despite the well-acknowledged importance of trust building in negotiations, the empirical studies on trust and negotiation have disproportionally focused on how trust affects the negotiation processes and outcomes (see a review by Kong et al. 2014), and not so much on examining how the focal negotiation influences trust as the outcome, let alone whether and how long trust will last beyond one negotiation. To better understand trust as the consequence in negotiations, we pose two interrelated questions: (1) in the short run, what will affect trust building immediately, and (2) in the long run, will trust built in one negotiation continue to increase, maintain, or decrease over time afterward?

The current literature is inadequate to offer a clear answer to the above questions for two reasons. First, previous studies often adopted a static rather than a dynamic approach. Trust is defined as a psychological state that embodies a willingness to accept vulnerability (Rousseau et al. 1998). But trust building is different from trust because it refers to a dynamic change of trust within a certain period. Lewicki et al. (2006) argue that most trust development research has taken a static, “snapshot” view. To overcome this limitation, recent reviews explicitly call for more future studies “to systematically examine sustained changes or trajectories over time in relation to key trust variables (Costa et al. 2018, p. 10)” and to understand “the pattern, trajectory, or mechanism of trust development in negotiations (Kong and Yao 2019, p. 5).”

Second, previous studies often adopted a narrow rather than a comprehensive approach. The narrow approach focuses only on the difference between pre- and post-negotiation trust. Studies using this approach often found a higher level of trust after as opposed to before the negotiation (e.g., Naquin and Paulson 2003; Yao et al. 2017). However, trust building via negotiation is future-oriented, so it is important to examine whether trust built in one negotiation continues to increase, maintain, or decrease in the long run. This more comprehensive approach could offer insights into how trust evolves from one negotiation to future interactions, and it would also contribute to the emerging perspective that negotiation has profound, future-oriented consequences (Curhan et al. 2010; Friedman et al. 2020; Hart and Schweitzer 2020).

In this study, we intend to address the above two limitations and offer an answer to the two research questions. To do so, we draw on social exchange theory and examine the effects of two categories of predictor variables—general trust and negotiator satisfaction. General trust is a relatively stable individual difference that affects the likelihood that a person will trust (Colquitt et al. 2007). Some recent studies show a link between general trust and trust building in one negotiation (e.g., Yao et al. 2017), while it is unclear whether this effect will last over time. Negotiator satisfaction is a negotiator’s overall subjective feeling of a negotiation (Curhan et al. 2006; Geiger 2014). Negotiator satisfaction consists of four different types: satisfaction with the instrumental outcome, of themselves, of the process, and of the relationship (Curhan et al. 2006). However, little is known about which type of satisfaction among the four will influence trust immediately versus over time.

We conducted face-to-face simulation research with 260 participants. We found that general trust predicts negotiators’ trust change immediately at the end of one negotiation, but it does not predict their trust change over the week afterward. In addition, we found that negotiators’ satisfaction with the outcome predicts their trust change immediately at the end of one negotiation, while their satisfaction with the relationship predicts their trust change over the week afterward. More precisely, over this one week, negotiators’ trust will maintain unchanged for those who have a low level of relationship satisfaction, while their trust will continue to increase for those who have a high level of relationship satisfaction.

Our research findings attempt to contribute to the literature in three major ways. First, they will enrich the negotiation literature by documenting the theoretically important yet empirically overlooked post-negotiation trust dynamics. Second, they will add insights into the trust literature by documenting both immediate trust building and lingering trust change with their respective predictors. This empirically tests and further extends the traditional trust development model. Third, they will extend social exchange theory by incorporating it into the negotiation—a context essentially entailing both economic and social exchanges, beyond the well-studied organizational trust context in which social exchange dominates.

2 Background and Hypothesis

2.1 Social Exchange Theory

Social exchange theory is the dominant framework to study trust building in social interactions (Costa et al. 2018). It posits that the exchange of social and economic resources is the fundamental form of human interactions and that the exchange quality affects how relationships evolve into trusting, loyal, and mutual commitments (Blau 1964). Following the theory, to identify predictors of trust building in negotiations, we select indicators that capture negotiators’ economic and social exchange quality. Accordingly, we build negotiator satisfaction—the most crucial indicator of people’s overall feeling of the exchange quality in negotiation—into our model as the predictor variables.

Negotiators experience complex internal and external social comparisons to generate their feelings of satisfaction in the negotiation context (Novemsky and Schweitzer 2004). Curhan et al. (2006) devised a valid, highly tractable measure of negotiator satisfaction and validated the inventory to comprehensively capture four types of negotiator satisfaction: outcome satisfaction (e.g., be satisfied with the agreement terms), self-image satisfaction (e.g., not losing “face”), process satisfaction (e.g., perceiving the process as easy and fair), and relationship satisfaction (e.g., having a positive impression of the counterpart). The inventory has then been widely adopted to measure negotiator satisfaction (Mueller and Curhan 2006; Amanatullah et al. 2008), with some interchangeable terms such as subjective value (Curhan et al. 2009) and subjective utility (Olekalns and Smith 2018). The experimental negotiation research overwhelmingly focused on examining the calculable instrumental outcome regulated by the negotiation simulation. Curhan et al. (2006) theorized this type of outcome as the objective value of a negotiation, and they contrasted it with what they called subjective value, which is negotiator satisfaction with the outcome, themselves, the process, and the relationship. In subsequent empirical studies, scholars often hypothesized and examined only outcome satisfaction and/or relationship satisfaction out of the four types of negotiator satisfaction (e.g., Mueller and Curhan, 2006; Amanatullah et al. 2008). We argue that this is because these two types of satisfaction well represent the overall quality of two primary exchanges—economic exchange and social exchange, respectively.

Social exchange theory allows us to hypothesize the relationship between the two types of negotiator satisfaction and trust change in negotiations. The theory suggests that the quality of exchange relationships of both economic resources (e.g., monetary rewards) and social resources (e.g., respect) affects exchange outcomes (Blau 1964). The key exchange rule in social exchange is reciprocity: social actors expect from each other that they will respond in similar ways (Cook et al. 2013). Cropanzano and Mitchell (2005) suggest that Blau (1964) had been treating social and economic exchanges as types of transactions, rather than as types of relationships. Instead, they suggest that in the economic transaction context, people can still have both economic exchange and social exchange in terms of the relationship. Based on this theorizing, we argue that both types of exchange may intertwine in the negotiation, but that their relative contribution to trust building may be different immediately versus over time.

2.2 Trust Building in Negotiation

What causes trust to change? The trust literature has well documented two primary categories of antecedents of trust—factors related to dispositions and factors related to interactions (Colquitt et al. 2007; Mayer et al. 1995). General trust is one important dispositional antecedent of trust because it reflects individuals’ generalized expectancy of others. It is often used interchangeably with other terms such as trust propensity (Ferguson and Peterson 2015), dispositional trust (Kramer 1999), or generalized trust (Rotter 1971), to highlight individuals’ baseline level of trust in general others. In addition, the characteristics of the social interaction also contribute to trust, such as trustworthiness of the counterpart (Colquitt et al. 2007; Mayer et al. 1995), the content of the interaction (Levine et al. 2018), and the interpretation of the interaction (Dunning et al. 2014). Taken together, these factors represent how people evaluate the other’s trustworthiness in a particular interaction context. In sum, based on the two primary categories of trust antecedents, we argue that negotiators’ own general trust and negotiators’ overall evaluation of the negotiation would influence trust in the negotiation context.

Moving from trust to trust building, we consider the two above factors but also adopt a more dynamic perspective. Lewicki and Stevenson (1997) proposed a theoretical model suggesting that the nature of trust would evolve over time. According to the model, trust emerges, in the beginning, as a calculus-based, instrumental reaction of a cost-and-benefit analysis, but it then develops into knowledge-based trust on other’s predictable behavior, and even later into identification-based trust on mutual affective attachment. In the same vein, Rousseau et al. (1998) suggest that during the trust-building period, calculative trust comes first while relational trust emerges later. In other words, in the beginning, trust building is usually influenced by instrumental or calculative analysis, but in the long run, trust evolves into a relational cognition and affection based on social interactions with the trustee. This key contrast can be generalized to trust building in various social settings beyond negotiations (Lewicki et al. 2006).

2.3 Immediate Trust Change

Negotiators with higher levels of general trust will be more likely to develop trust immediately in the negotiation. The underlying rationale is that a higher level of general trust translates into more positive expectations of others in general, irrespective of the ability to monitor or control the other party, thereby encouraging people to display higher levels of intention to accept vulnerability from others (Colquitt et al. 2007; Mayer et al. 1995). Some scholars conceptualized this phenomenon as swift trust (Meyerson et al. 1996; Robert et al. 2009). People are more likely to build swift trust when they are engaged in interdependent tasks (Meyerson et al. 1996). Therefore, some empirical studies have shown that, as a typical interdependent task, negotiation allows negotiators with higher pre-negotiation general trust to build higher post-negotiation trust in counterparts (Yao et al. 2017). Following previous studies, we also predict that,

H1

Negotiators’ general trust will be positively associated with their immediate trust change.

In addition to relying on their disposition, negotiators also rely on the information that they obtained in the negotiation to make trust decisions. Social exchange theory implies that in a social interaction characterized by economic exchanges, people’s trust perception is more influenced by the economic exchange quality (Cropanzano and Mitchell 2005). Since negotiation is an inherently competitive social setting where negotiators have to claim economic value from their counterpart (Thompson et al. 2010), the unacquaintance between two parties naturally encourages negotiators to focus more on short-term economic gains than on long-term social gains (Brown and Curhan 2012). That is to say, in the short term, negotiators draw more on the quality of economic exchange than on the quality of social exchange to shape their attitudes and behaviors.

Outcome satisfaction captures the extent to which negotiators are satisfied with the economic outcome (Curhan et al. 2006). It is the only dimension among the four that reflects negotiators’ general feelings about the economic exchange quality of the negotiation. Empirical studies have shown that outcome satisfaction is the only dimension that is associated with negotiators’ actual economic gains in negotiations (Curhan et al. 2006, 2010). Similarly, some empirical studies identified negotiation economic outcome as the primary predictor of negotiators’ post-negotiation trust (Barry and Oliver 1996; Mislin et al. 2011). These findings essentially support our rationale that negotiators’ interpretation of the economic exchange quality—outcome satisfaction—would influence their trust immediately in the negotiation. We propose that,

H2

Negotiators’ outcome satisfaction will be positively associated with their immediate trust change.

2.3.1 Lingering Trust Change

Colquitt et al. (2007, p. 911) suggested that “an unanswered question is whether trust propensity continues to impact trust once trustworthiness has been gauged.” In other words, it is inconclusive whether general trust could continue, after completion of the negotiation, to affect subsequent, long-term trust change. Some scholars suggest that general trust simply serves as the baseline for individuals to build initial trust in the trustee (Lewis and Weigert 1985), implying a limited effect of general trust over the long run. Consistent with this idea, empirical studies found that general trust loses its predicting power once social interactions generate concrete information regarding the counterparts (Gill et al. 2005).

However, other scholars suggest that trust propensity may continue to influence trust change over time. The central argument is that the formation of trust is a process of sense-making, in which general expectations of others affect how the information will be received and interpreted (Adobor 2005; Yang 2006). For example, Becker (1996, p. 47) argues that trust should always be connected to “good estimates of others’ trustworthiness.” Thus, even after the initial social interaction is completed, general trust may remain influential on how negotiators recall and make sense of the negotiation experience. Following this logic, we propose that,

H3

Negotiators’ general trust will be positively associated with their lingering trust change.

Social exchange theory suggests that economic exchange rests on a formal contract that specifies the terms of exchange, while social exchange entails unspecified obligation and norm of reciprocity (Blau 1964; Molm et al. 2000). It means that perception of a good relationship quality entails the nature of the social exchange, encouraging social actors to feel obligated to reciprocate. Will negotiators’ trust continue to increase, decrease, or maintain long after the negotiation is completed? A key difference between the two phases of trust change is that immediate trust change relies more on economic exchange quality due to the unacquaintance between negotiators and the competitive nature of negotiation, while lingering trust change relies more on social exchange quality due to established negotiation experience. Negotiators could therefore continue to ruminate the negotiation experience based on the information related to social exchange rather than economic exchange.

A circulating stream of research shows that the relational quality built in the negotiation affects subsequent interactions and agreement implementation (e.g., Mislin et al. 2011; Campagna et al. 2016; Hart and Schweitzer 2020). A key question is to know on which kind of information a negotiator relies to assess the level of trust in the counterpart since the negotiation has been concluded in the past. We argue that negotiators will mostly rely on relationship-related information and that, therefore, relationship satisfaction will be the best predictor of long-term trust change. Curhan et al. (2006) showed that relationship satisfaction is the only dimension predicting post-negotiation trust. It is worth noting that process satisfaction and relationship satisfaction has conceptual proximity to some extent, but we do not hypothesize the former one for the lingering trust change. This is because they fundamentally differ: process satisfaction still focuses on the focal task (e.g., the ease of reaching an agreement), while relationship satisfaction focuses on the people beyond one task (e.g., the positive impression of the counterpart). When negotiators recall their memories and draw inferences of trust after the negotiation, their perception of the people rather than the perception of the task becomes the most salient indicator. Thus, relationship satisfaction is the core facet that captures negotiators’ overall social exchange quality of the negotiation, and hence it will continue to shape their lingering trust change even after the negotiation. Therefore, we propose that,

H4

Negotiators’ relationship satisfaction will be positively associated with their lingering trust change.

3 Method

3.1 Participants and Procedure

We recruited 260 undergraduate students (147 women, Mage = 20.9) who were enrolled in the course introduction to negotiation at a French business school. This was the very first negotiation course for all participants in their undergraduate program. They voluntarily participated in this study as their first face-to-face role-playing negotiation exercise in this course. Before the exercise, participants had learned the basic distinction between distributive and integrative negotiations, but they have not yet learned any contents related to trust, which came later as part of the debriefing of this exercise. The general design of the study included three waves of time: one week before the negotiation (T0), the day of the negotiation (T1), and one week after the negotiation (T2).

First, at T0, all participants completed a pre-negotiation survey which measured their general trust. When participants reported their general trust, they did not know the identity of their counterparts for the upcoming negotiation in one week. We followed previous studies to adopt this approach to capture general trust with comparable items of post-negotiation trust measures at T1 and T2 (Yao et al. 2017).

Second, at T1, participants completed a face-to-face negotiation simulation New Recruit (Neale 1997), which has been widely used in research (e.g., Overbeck et al. 2010; Swaab et al. 2011). It was an eight-issue integrative negotiation involving a job recruiter and a job candidate who intended to reach an agreement regarding details of a job offer. This simulation has been widely used to study post-negotiation trust (Kurtzberg et al. 2009). We randomly assigned participants into the role of either a recruiter or a candidate. Participants read their role materials and independently prepared for 20 min, and then they negotiated for 25 min trying to reach an agreement. All 130 dyads reached agreements in our study. Once they did, they completed a post-negotiation survey which measured their satisfaction, trust in counterparts (T1), and previous relationship with counterparts.

Third, at T2, participants again reported their trust in counterparts (T2). During the one week between T1 and T2, participants did not have any assignments that induced them to engage in any social interactions with their classmates.

3.2 Measures

3.2.1 Trust

We adopted the three-item scale from Levine and Schweitzer (2015) to measure participants’ trust at T0, T1, and T2. The items were: (1) “I would trust her/him as my negotiation counterpart”, (2) “I would be willing to make myself vulnerable to her/him in negotiation”, and (3) “I am confident that she/he would send me accurate information in negotiation”. In the T0 survey, participants reported trust in general others without knowledge of the identity of their counterparts. In the T1 and T2 surveys, because the negotiation simulation had already been conducted, participants reported trust in the negotiation counterpart with whom they negotiated. In doing so, we measured general trust at T0 and trust in counterparts at T1 and T2 using essentially equivalent and comparable items. Participants answered on a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). Cronbach’s α were 0.62, 0.61, and 0.68 for T0, T1, and T2, respectively. It is worth noting that the measures at T1 and T2 were trust measures but not how we modeled immediate and lingering trust change, and we will explain the modeling method in the following section.

3.2.2 Satisfaction

We used the 16-item Subjective Value Inventory (Curhan et al. 2006) to measure negotiator satisfaction. In this inventory, four items measured each of the four dimensions of satisfaction: (1) outcome satisfaction (e.g., “I am satisfied with my own outcome”), (2) self-image satisfaction (e.g., “This negotiation made me feel more competent as a negotiator”), (3) process satisfaction (e.g., “I would characterize the negotiation process as fair”), and (4) relationship satisfaction (e.g., “I am satisfied with my relationship with my counterpart as a result of this negotiation”). Participants responded on a seven-point Likert scale. Cronbach’s α were 0.74, 0.54, 0.78 and 0.86, respectively.

3.2.3 Economic Gain

We measured participants’ individual economic gains obtained in the negotiation and controlled it in our analysis to be in line with previous studies on negotiator satisfaction (e.g., Curhan et al. 2010). The two roles received different payoff tables, but they had the equivalent potential to achieve the economic gain ranging from − 8400 to 12,400. As a result, participants achieved 4924.62 (SD = 1953.22) economic gain on average, and there was no significant difference between the two roles (t (129) = 0.08, p = 0.93).

3.2.4 Previous Relationship

We controlled for participants’ previous relationship with their counterpart because previous relationship closeness could affect trust building dynamics (Berscheid et al. 1989). We asked participants that “prior to today’s negotiation, what was the relationship between you and your counterpart?” We used the two-item scale from Fischer and Roseman (2007): (1) “How close were you with this person” and (2) “How well did you know this person”. Participants answered via a five-point Likert scale (1 = not close at all to 5 = very close and 1 = not well at all to 5 = very well). The two items were highly consistent and averaged per participant (Cronbach’s α = 0.83). To reflect the more “objective” facet of the mutual relationship (Cronbach’s α = 0.81 between two negotiators within one dyad) as well as to alleviate the multicollinearity problem, we averaged the two scores from the two negotiators within one dyad to represent their overall previous relationship at the dyad level.

3.3 Analytical Strategy

Our research question aimed at modeling trust change over time, and our design entailed repeated measures in a dyadic data structure, so we combined the growth curve modeling and Actor-Partner Interdependence Modeling (APIM) to reflect the dyadic, repeated nature of our data. All models were estimated with the R package lavaan (Rosseel 2012). Missing data were random, because participants with and without missing data did not significantly differ in their gender, economic gain, trust, outcome satisfaction, self-image satisfaction, process satisfaction, or relationship satisfaction (ps > 0.05). Thus, we relied on a full-information maximum likelihood estimation.

Growth curve modeling is a Structural Equation Modeling (SEM) technique that has been widely used to model changes over time with longitudinal data (Olsen and Kenny 2006). In growth curve model, the description of changes in the outcome variable over time is based on the interpretation of a latent intercept (i.e., the baseline level) and a latent slope (i.e., the amount of change in the outcome variable over time). Following the method, we modeled time (i.e., T1 versus T2) as the predictor variable and trust as the outcome variable. In our analysis, we centered trust at T1 and trust at T2 based on the grand mean of the sample’s trust level at T0. After this transformation, the latent intercept represents negotiators’ trust change immediately after the negotiation, and the latent slope represents negotiators’ trust change over the week between T1 and T2 after the negotiation. In sum, we use the latent intercept of our model to test our H1 and H2 which focused on the immediate trust change, and we use the latent slope of our model to test our H3 and H4 which focused on the lingering trust change over the week.

APIM is a multi-level analysis specifically for analyzing dyadic data (Kenny et al. 2006). When we test hypotheses related to immediate trust building (i.e., H1 and H2), the strong interdependence of the negotiation would theoretically witness both actor effect (i.e., recruiters’ satisfactions affect recruiters’ trust building) and partner effect (i.e., candidates’ satisfactions affect recruiters’ trust). APIM treats individual negotiators as nested within dyads to test both actor and partner effects simultaneously, so it is particularly useful in negotiation research to capture the interdependent nature (e.g., Curhan et al. 2010). When we tested hypotheses related to lingering trust change (i.e., H3 and H4), we did not incorporate partner effects anymore. This is because we expected that participants would have, to the best of our knowledge, no interactions in the week following the simulation and, consequently, that the lingering trust change would be theorectially influenced only by the actor’s characteristics. Specifically, we specified a dependent error covariance structure to account for the non-independence of observations coming from the same dyad (Olsen and Kenny 2006). To do so, we estimated covariances between recruiters’ unique residuals and candidates’ unique residuals at each time point.

We chose the indistinguishable model to treat the two roles (i.e., recruiter and candidate) as essentially equivalent. First, these two roles had equivalent payoff structures which allowed them to achieve essentially equivalent economic gains. The fact that individual economic gains did not differ between recruiters and candidates in our sample supported this assumption. Second, the results based on the minimal Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) procedure (Vrieze 2012) showed that indistinguishable model assuming role equivalence (AIC = 1107.8, BIC = 1147.8) had a better relative fit than the distinguishable model assuming role difference (AIC = 1104.0, BIC = 1138.3). Consequently, we constrained effects to be equal across roles in all subsequent analyses. Regarding absolute model fit, we followed the recommendations of Schumacker and Lomax (2004) and used three statistical indices: (1) chi2/df ratio: a good model should be less than 3, (2) Comparative Fit Index (CFI): a good model should be more than 0.90, and (3) Root Mean Square Error of Approximation (RMSEA): a good model should be less than 0.08. Note that the fit indices were computed using the appropriate null and saturated (I-SAT) models as the adjustment for indistinguishable dyads (Peugh et al. 2013). We represent the model in which intercepts and slopes are constrained to be equal for recruiters and candidates in Fig. 1.

Fig. 1
figure 1

Growth curve model with APIM on trust change in negotiations

Full size image

4 Results

We reported means, standard deviations, and correlation analysis results of all variables in Table 1.

Table 1 Descriptive statistics and correlation analysis results
Full size table

4.1 General Findings on Trust Building

We found that the growth curve model without any predictors had a good absolute fit (chi2/df = 0.09, CFI = 0.92, RMSEA = 0.00). This model offers important insights into the overall trust building dynamics, without yet considering predictor variables. There were several important findings. First, the estimated mean of the latent intercept was positive (B = 0.39, p < 0.01). As a reminder, the intercept represents the difference between trust levels reported immediately after the negotiation and before the negotiation. Thus, the positive intercept suggested that engaging in a negotiation increased negotiators’ trust in general. This finding is consistent with previous studies (e.g., Naquin and Paulson 2003; Yao et al. 2017).

Second, the estimated mean of the latent slope was not significantly different from zero (B = 0.05, p = 0.50). As a reminder, the slope represents the change of trust over the week after the negotiation. Thus, the non-significant slope meant that negotiators’ trust in their counterparts did not change significantly over the week. This suggested that when we see all negotiators in the sample as a whole, their lingering trust change during the one week after the negotiation neither increased nor decreased, but rather remained unchanged.

Third, the actor’s immediate trust change (actor’s latent intercept) and the partner’s immediate trust change (partner’s latent intercept) were positively correlated (r = 0.14, p < 0.05). This suggested that two negotiators’ immediate trust change are highly correlated due to the interdependent negotiation process—when one negotiator trusts her/his counterpart, the counterpart usually trusts back.

Finally, all participants’ immediate trust change (actor’s latent intercept) and their lingering trust change (actor’s latent slope) were positively correlated (r = 0.13, p < 0.01). This suggested that negotiators’ immediate trust change affects their lingering trust change over the following one week—when one negotiator increases her/his trust in one negotiation, it is more likely she/he will continue to increase trust afterward at least for one week.

4.2 Specific Findings on Hypotheses Testing

We included all predictors and control variables in the growth curve model to test our full model involving H1 to H4. The absolute fit of the model including all variables was good (chi2/df = 1.13, CFI = 0.96, RMSEA = 0.04). We reported the estimates of regression coefficients in Table 2.

Table 2 Results of growth curve model with APIM on trust change in negotiations
Full size table

H1 and H2 predicted the effects of negotiators’ general trust and outcome satisfaction on immediate trust change. To test the effects on immediate trust change, we focused on the latent intercept of the model. We found that negotiators’ general trust level was a positive predictor of immediate trust change (B = 0.27, p < 0.01). To gain a better understanding of this effect, we estimated the coefficients of the latent intercept at different levels of general trust. We found that at relatively high levels of general trust (+ 1 SD), the intercept was positive (B = 0.62, p < 0.01). At relatively low levels of general trust (− 1 SD), the intercept was positive but only marginally significant (B = 0.20, p = 0.07). This finding supported our H1.

In addition, we also found a significant intraindividual effect of outcome satisfaction (B = 0.29, p < 0.01). We found that at relatively high levels of outcome satisfaction (+ 1 SD) the intercept was positive (B = 0.80, p < 0.01), whereas at relatively low levels of outcome satisfaction (− 1 SD), it was not significantly different from zero (B = 0.03, p = 0.85). This finding supported our H2.

H3 and H4 predicted the effects of negotiators’ general trust and relationship satisfaction on lingering trust change. To test the effects on lingering trust change, we focused on the latent slope of the model. To begin with, we did not find a significant effect of general trust on the lingering trust change (B = −0.08, p = 0.34). Thus, H3 was not supported.

In addition, we found a positive intraindividual effect of relationship satisfaction on lingering trust change (B = 0.29, p < 0.05). The simple slope analysis showed that among individuals who were more satisfied with the relationship with their counterparts (+ 1 SD), there was a significant increase of trust over the week from T1 to T2 (B = 0.35, p < 0.05). Among individuals who were less satisfied with the relationship with their counterparts (− 1 SD), there was no significant evolution in trust in the week following the negotiation (B = −0.23, p = 0.16). In sum, these findings supported our H4, and we plotted the effect of time and relationship satisfaction on lingering trust change in Fig. 2.

Fig. 2
figure 2

Lingering trust change over T1 and T2 for negotiators with high versus low relationship satisfaction

Full size image

5 Discussion

What contributes to building trust in negotiations? Our study suggests that negotiators’ general trust before the negotiation and their satisfaction with the negotiation both play a role. Negotiators’ dispositional general trust affects the level of trust immediately after the negotiation, but it does not affect further trust change over the following week. In addition, negotiators’ satisfaction with the outcome affects their trust immediately after the negotiation, while their satisfaction with the relationship further affects their lingering trust change over the following week. For negotiators with high relationship satisfaction, their trust will continue to increase even after the negotiation for at least one week; for negotiators with low relationship satisfaction, their trust will maintain unchanged afterward. These findings offer a comprehensive understanding of trust building by considering different phases of trust change and identifying their respective predictors.

We did not find empirical support for our third hypothesis, meaning that negotiators’ general trust may not be able to predict how their trust in counterparts evolves after the agreement is reached. Therefore, the findings with respect to general trust are consistent with some previous empirical studies, which show that general trust is only influential in the beginning stage of social interactions and not anymore after people have achieved concrete trustworthiness information of the other party (Gill et al. 2005).

5.1 Theoretical Contributions

Our study makes major theoretical contributions to the literature of negotiation, trust, and social exchange theory. First, our findings enrich the negotiation literature by documenting trust change dynamics after a negotiation. Although negotiators have numerous important tasks to perform after reaching an agreement, such as contract implementation and relationship maintenance, the existing negotiation literature has paid surprisingly little attention to post-negotiation stage (Jang et al. 2018; Hart and Schweitzer 2020). Trust is one of the key factors to maintaining a positive post-negotiation cooperation (Mislin et al. 2011; Campagna et al. 2019), but the process of trust change as a result of negotiation has not been thoroughly investigated. In this study, we used a longitudinal design to build and test an integrated model that considers trust levels at three moments to offer a comprehensive description of the trust change dynamics. In addition, our findings can contribute to the emerging stream of research that focuses on negotiator satisfaction, beyond economic outcomes (Becker and Curhan 2018; Olekalns and Smith 2018; Lewis et al. 2018). We contribute to and further extend this perspective by showing the different effects of outcome satisfaction versus relationship satisfaction on trust change in the short-term versus long-term.

Second, our findings add insights into the trust literature by documenting both immediate and lingering trust change and their respective predictors. The theoretical model of trust development (Lewicki et al. 1997, 2006) proposes that trust increases after a first interaction and then maintains. During the maintenance period, trust may continue to increase or remain unchanged. Our findings support this theoretical model and echo some previous empirical studies (e.g., van der Werff and Buckley 2017) by showing that the trust change pattern indeed contains the immediate trust increase phase and the subsequent trust maintenance phase. More importantly, we extend this model by showing how relationship satisfaction explains the bifurcation of trust in the maintenance phase: trust continues to increase among people who are satisfied with the relationship, while it remains the same among people who are less satisfied with the relationship. This finding shows that individuals’ subjective feelings of one focal interaction will continue to influence their trust dynamics in the post-interaction phase and opens a new research avenue to investigating the critical turning point of trust change trajectory in the post-interaction trust maintenance phase.

Third, our findings extend social exchange theory by showing how trust relies on both economic and social exchange in negotiations—a context in which the two types of exchange are typically involved. Social exchange theory has been overwhelmingly adopted to study trust in presumably collaborative contexts, such as intrateam trust (Cheng et al. 2016), overlooking negotiation—an inherently competitive context. Previous studies also mainly focus on the immediate, calculative phase of trust building (see review by Costa et al. 2018), overlooking the long-term trust building dynamics. People often conceptualize negotiation as an economic transaction context, but we suggest that it can still contain both economic exchange and social exchange due to its very nature of mixed motives (Thompson et al. 2010). Therefore, by extending the theory to examine a comprehensive trust building both in the short-term and long-term in negotiation context, we show that the salience of the two types of change may be different in different phases.

5.2 Practical Implications

Negotiators should be aware that trust evolves not only during the negotiation but also afterward. We acknowledge that it is a cliché to restate the importance of relationship building in negotiations, but we want to remind practitioners of two additional points. First, we should set a goal that values relationship building in negotiations, but we also should ensure that counterparts share this goal. Second, negotiators should not treat post-negotiation relationship maintenance as a separate and subsequent task due to the lingering effect of relationship satisfaction on trust change. Therefore, negotiators should plan out in advance whether trust building is a strategic goal for them so that they can have an overview of their strategies which further influence each other’s ultimate satisfaction.

Moreover, we know that trustful people build trust more swiftly, but it is not realistic to expect trustful counterparts in all situations. Negotiators may encounter distrustful counterparts due to various reasons, such as personality preferences, cultural norms, and corporate practices. That is to say, it may be difficult to build trust immediately through open communication as suggested by previous research (Yao et al. 2017). But we show that relationship satisfaction would influence negotiators’ trust perception as time goes by. Thus, it is possible to wait to see trust change over time, and hence negotiators should be patient regarding trust building as it will be a natural, reciprocal response from the counterpart over time.

5.3 Limitations and Future Directions

Our study has several limitations. First, the reliabilities of our trust measure were consistently between 0.6 and 0.7. Even though some scholars are lenient by arguing that “a general accepted rule is that α of 0.6–0.7 indicates an acceptable level of reliability” (Ursachi et al. 2015, p. 681), we still acknowledge that this was a limitation of our research. We adopted this measure from Levine and Schweitzer (2015) who used it to measure trusting attitude in trust game with a decent reliability (α = 0.84). We infer that the nuance between the originally used trust game context and our face-to-face negotiation context might explain the reduction in reliability. In addition, a small number of items is known to penalize Cronbach’s α (Eisinga et al. 2013). However, we argue that the merits of adopting this trust measure outweigh the possible limitations because this adoption allowed us to have comparable items of general trust and trust in counterparts in negotiation. Also, the T1 trust levels reported by two participants in one negotiation dyad were highly correlated (r = 0.29, p < 0.01), indicating the high interdependence of negotiation and the convergent validity of this measure.

Second, we examined the dynamic trust change between T1 and T2, but the choice of one week as the time lag may not reflect the full scope of trust change trajectory in real life. We chose one week to examine lingering trust change mainly based on our heuristics and practical feasibility, but lingering trust change may last for a shorter or a longer period depending on multiple factors. For example, what will happen to negotiators’ trust in two weeks, one month, or one year? How long does this lingering effect will last if the stake of the negotiation is bigger versus smaller? Is that possible that some repeated interactions can reinforce the lingering effect on trust? Those questions call for more scholarly attention in the future. But at least there is one thing that we can safely conclude from the current study: the immediate effect of outcome satisfaction on trust change will not contribute further to trust change one week later.

Third, although to the best of our knowledge, participants should not have interactions during the one week after the negotiation, we had no control over what participants actually did during this period of time. Even though we intentionally remove all assignments and tasks that could induce social interactions between participants during this one week, they still might have done so voluntarily. However, we argue that this limitation did not compromise our findings. If the social interaction between participants during the one week is totally random, this random noise would not affect the results. If this is not random, it is likely that high relationship satisfaction would induce more post-negotiation social interaction and finally increase the lingering trust change. This conjecture suggests that the post-negotiation social interaction might be one potential mechanism that explains the relationship between relationship satisfaction and lingering trust change. This potential mechanism does not at all contradict our findings, but rather offers a future direction to further investigate the bridge between one focal negotiation and future perception in the context of post-negotiation trust building.

We call for more future research that pays attention to what will happen after the agreement has been reached. The dominant paradigm in the experimental negotiation research often recruits participants to negotiate once and they leave the lab without seeing each other in the future. However, this does not reflect accurately many real-life negotiations that involve post-negotiation follow-ups or services. Some emerging studies start to examine the link between a focal negotiation and post-negotiation interactions (e.g., Mislin et al. 2011; Hart and Schweitzer 2020), reminding researchers and practitioners not to consider signing the agreement as the terminal point of negotiations. With the present study, we share their point of view and call for more studies to further understand how and when post-negotiation interactions matter. Studies on this issue would help us achieve a better understanding of negotiation beyond the one-shot negotiation process.

6 Conclusion

Rome was not built in a day, neither was trust. Negotiators’ trust in counterparts could fluctuate due to the instrumental outcome, but it is the relationship satisfaction that shapes their trust in the long run. Ideally, we would love to gain our counterparts’ trust on the spot as well as over time gradually, but maybe sometimes it is necessary to make a wise trade-off between the one-shot economic outcome and the long-term relationship building, depending on the main purpose of the negotiators.