• trust;
  • relationship duration;
  • interorganizational relations;
  • meta-analysis;
  • trustworthiness


  1. Top of page
  2. Abstract

A common premise in prior research is that trust increases over time in relationships. Through a meta-analysis of 39 studies, we find that the bivariate correlation between trust and relationship duration (1) is on average positive and small, and (2) varies significantly across studies indicating the presence of unobserved moderators. We therefore build a theoretical framework to specify four different mechanisms—initial bias correction, change in relationship value, identification, and trust-based selection—that may affect the development of trust. We then argue that the relative strength of these mechanisms should influence whether trust increases, remains constant, or decreases over time. © 2013 The Authors. Strategic Management Journal published by John Wiley & Sons Ltd.


  1. Top of page
  2. Abstract

Trust is a key factor in understanding performance variation among relationships between organizations (Granovetter, 1985; Lado, Dant, and Tekleab, 2008; Li, Poppo, and Zhou, 2010; Luo, 2008; Poppo and Zenger, 2002; Uzzi, 1997) as well as within them (e.g., Davis et al., 2000; Fang et al., 2008; Wang, He, and Mahoney, 2009). Beyond its performance effects, trust (or the lack thereof) is also used to explain the structure of exchange relationships (Bradach and Eccles, 1989; Granovetter, 1985; Gulati, 1995; Gulati and Nickerson, 2008; Puranam and Vanneste, 2009; Reuer and Ariño, 2007; Ring and Van de Ven, 1992; Shah and Swaminathan, 2008; Uzzi, 1997; Williamson, 1975).

Scholars have often assumed that interactions over time affect trust in a relationship (Ring and Van de Ven, 1994). Many empirical studies measuring trust predict it to increase with the duration of a relationship (e.g., Anderson and Weitz, 1989; Chua, Ingram, and Morris, 2008; Doney and Cannon, 1997; Ganesan, 1994; Gulati and Sytch, 2008; Wang, Tomlinson, and Noe, 2010). Others not measuring trust may use relationship length as a proxy (or assume that trust increases over time) when examining the effects of relationship duration (e.g., Corts and Singh, 2004; Greve et al., 2010; Gulati, 1995; Parkhe, 1993; Vanneste and Puranam, 2010). Yet, different studies report a positive regression coefficient between trust and duration (e.g., Anderson and Weitz, 1989), a negative one (e.g., Poppo, Zhou, and Ryu, 2008), or no relation (e.g., Doney and Cannon, 1997).

We study the evidence on trust and relationship duration through a meta-analysis of 39 studies. We extend prior meta-analyses in several ways. Colquitt, Scott, and LePine (2007), Geyskens, Steenkamp, and Kumar (1998), and Swan, Bowers, and Richardson (1999) do not explore relationship duration as an antecedent of trust. Dirks and Ferrin (2002) meta-analyze the link between trust and duration in the context of leadership using five studies. Our more comprehensive aggregation of results shows that the correlation between trust and relationship duration is positive but small (Cohen, 1992), and we find that the correlation varies more across studies than expected by sampling error alone—suggesting that unobserved moderators matter.

We also build a set of theoretical arguments on why and how the relation between trust and relationship duration may vary. Using Coleman's (1990) formal description of trust, which captures both rational and heuristic-based trust (McEvily, 2011), we show that common trust conceptualizations imply at least four mechanisms influencing the development of trust over time: initial bias correction, change in relationship value, identification, and trust-based selection. Our framework links trust and relationship duration via the relative strength of these potentially countervailing mechanisms.


  1. Top of page
  2. Abstract

Identification of studies

To identify studies for our meta-analysis, we first searched the Business Source Complete and PsycINFO databases for the period 1887–2010. We used the keywords trust* (to capture studies on trust or trustworthiness) together with at least one of history, duration, length, time, and frequency as well as at least one of data, empirical, test, statistical, finding*, result*, and evidence. This filtering resulted in 1,435 and 2,351 articles (respectively) culled from the two databases. Second, we considered all articles published since 1980 that mentioned “trust” (or a variant thereof) in the following journals (based on Geyskens, Steenkamp, and Kumar, 2006): Academy of Management Journal, Administrative Science Quarterly, International Journal of Research in Marketing, Journal of Applied Psychology, Journal of Business Research, Journal of International Business Studies, Journal of Management, Journal of Marketing, Journal of Marketing Research, Journal of Personality and Social Psychology, Management Science, Marketing Science, Organizational Behavior and Human Decision Processes, Organization Science, Personality & Social Psychology Bulletin, Psychological Science, and Strategic Management Journal. Third, we examined the references listed in review articles on trust (Colquitt et al., 2007; Dirks and Ferrin, 2002; Ebert, 2007; Geyskens et al., 1998; Swan et al., 1999). Fourth, we sent a request for unpublished studies to approximately 350 trust researchers. Finally, we went through the reference section of each relevant paper identified previously.

We included a study if it met three conditions. First, it had to employ a construct of trust or trustworthiness broadly consistent with the definitions of Mayer, Davis, and Schoorman (1995) and Rousseau et al. (1998), see the Theory section. Second, it had to report the sample size and pairwise correlation between trust and relationship duration; these values are required for conducting a meta-analysis. Third, the study had to be at the interpersonal or interorganizational level (we shall report any differences in findings for these two levels).

Data set

We used three procedures to achieve independence among the correlation coefficients in our meta-analysis. First, for studies with multiple independent samples, we separately included the correlation from each sample. Second, for samples with more than one correlation per unit of analysis (i.e., interpersonal or interorganizational) due to multiple operationalizations of a trust construct, we calculated a single composite correlation (for the formulas used to make this calculation, see Hunter and Schmidt, 2004). Third, for studies based on the same or overlapping data sets, we included only one correlation in our meta-analysis. We chose the correlation from the study with the largest sample or, if the sample sizes were equal, from the oldest article.

We used standard meta-analytical procedures to establish that the most influential studies did not clearly increase or decrease the sample weighted average correlation (Huffcutt and Arthur, 1995). These procedures—along with those already discussed under “Identification of studies”—yielded a final set of 39 studies reporting k = 47 effect sizes with a total of N = 9,632 observations. Studies included in the meta-analysis are preceded by a * in the References.

Meta-analytic calculations

We follow the Hunter and Schmidt (2004) approach, which allows for the correction of statistical artifacts in an attempt to estimate the population correlation. We correct the observed correlation coefficients (r) for biases arising from three statistical artifacts: (1) dichotomization of the relationship duration variable, (2) measurement error in the trust variable, and (3) sampling error.

For each relationship duration variable, we know whether it is dichotomized and, if so, at what percentage the cutoff occurs (artifact 1). We have information on the reliability of nearly all trust variables (artifact 2). We lack information on four effect sizes; for these, we take the average reliability for the trust variable. For artifacts 1 and 2, we correct each correlation coefficient following Hunter and Schmidt (2004). We then meta-analyze the resulting corrected correlations to obtain a sample-weighted average correlation and variance. The variance is also corrected for sampling error (artifact 3).

Finally, we investigate publication bias, which exists if published studies systematically report effect sizes that differ from those reported in unpublished studies. Because the number of unpublished studies in our data set is small (with k = 2), we instead compare studies that use trust and relationship duration as independent variables with studies with one construct (i.e., trust) as the dependent variable. It is unlikely that a single correlation between two independent variables systematically affects the probability of publication. If this is true, then comparing the two sets of studies gives an indication whether publication bias is present. The 29 effect sizes reported when trust is the dependent variable are, on average, 0.02 higher than the 18 effect sizes reported when both trust and relationship duration are independent variables. A t-test reveals that this difference is not significant (p = 0.53), so we find no evidence of publication bias.


  1. Top of page
  2. Abstract

Table 1 presents the meta-analytic findings. The sample size–weighted average observed correlation between trust and relationship duration is 0.11. The estimate of the average corrected population correlation is 0.12 (the 95% confidence interval ranges from 0.09 to 0.16). The correlation between trust and relationship duration can therefore be classified as small (Cohen, 1992). Furthermore, statistical artifacts (sampling and measurement error) explain only 22.5 percent of the variance in observed correlations. The chi-square test for homogeneity is significantly different from 0 (p < 0.001), which indicates that the observed variation is greater than what would be expected to arise from chance alone. Taken together, these findings suggest the presence of unobserved moderator variables—as does the so-called 95 percent credibility interval (within which 95% of the population correlations lie (Hunter and Schmidt, 2004). It ranges from -0.08 to 0.32, which implies significant variation in population correlations (and not merely variation in estimates of the mean). Because this credibility interval is large and (unlike the confidence interval) includes 0, it suggests that the calculated average correlation is more likely to be the average of subpopulations (some of which exhibit a negative correlation) than the average of a single population (Whitener, 1990).

Table 1. Meta-analytic results for correlation between relationship duration and trust
kNrinline image95% confidence interval (inline image)95% credibility interval (ρ)% variance accounted forQ
  1. k = number of effect sizes; N = total number of observations; r = sample size–weighted average observed correlation; inline image = estimate of corrected average population correlation; % variance accounted for = percentage of observed variance accounted for by statistical artifacts; Q = chi-square test for homogeneity.

  2. ***p < 0.001.


In short, our meta-analysis reveals that (1) the bivariate correlation between trust and relationship duration is, on average, positive but small; and (2) there are indications that unobserved factors moderate this relationship. In two robustness tests, our results are unaffected by the unit of analysis (interpersonal versus interorganizational) or the measurement of “trust” (trust versus trustworthiness).


  1. Top of page
  2. Abstract

Prior literature suggests that trust and relationship duration are positively correlated; our meta-analysis shows that the correlation is positive but weak. Moreover, this correlation is likely moderated by unobserved factors, so that it could be 0 or even negative under some conditions. We theorize about these conditions by focusing on the mechanisms affecting the association between trust and relationship duration.

What is trust?

Scholars working within different literatures have converged on a common meaning of trust. Rousseau et al. (1998: 395) capture this consensus in their definition of trust as “a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behavior of another.” Similar definitions appear in Mayer et al. (1995: 712), Bhattacharya et al. (1998: 462), and Coleman (1990: 99).

To systematically analyze how trust changes over time, we use Coleman's (1990) formal description of trust that captures these key features (see also Nooteboom, 2002, who uses the same description). Consider a trustor who must decide whether or not to establish a relation with a trustee of unknown trustworthiness. To begin with, we assume that trustworthiness does not change during such a relation (we relax this restriction later). Because trustworthiness is not directly observable, the trustor can at best assign some subjective probability that the trustee will act in a trustworthy manner. This perceived trustworthiness (at the beginning of period t) is denoted pt. Suppose that the trustor's interest in entering such a relationship is increasing in its expected payoffs. We use At to denote the trustor's gains in period t from its relation with a trustworthy trustee (At > 0); conversely, Bt denotes the trustor's losses in period t from its relation with an untrustworthy trustee (Bt < 0).

Following Coleman (1990) and Nooteboom (2002), we write the trustor's expected payoff from relationship R in period t as

  • display math(1)

A trustor's willingness to engage in a relationship relative to not engaging in a relationship increases in the expected payoffs as given in Equation (1). Hence, trust increases with Et[R], because trust is the willingness to be vulnerable to the actions of another party, i.e., the willingness to engage in a relationship with the other party.1

This description embodies four key ingredients of trust common to existing definitions. First, trust is the willingness to engage in a relationship with the other party (i.e., an intention to act), not the actual decision to engage in the relationship (i.e., accepting vulnerability) (Mayer et al., 1995; Rousseau et al., 1998). Second, trust as the trustor's willingness to engage in a relationship, which increases with Et[R], is different from the perception of the trustee's trustworthiness (pt) (Colquitt et al., 2007; Hardin, 2002; Mayer et al., 1995). Third, trust is formed under uncertainty because the trustor can only guess the other's trustworthiness (as reflected in 0 < pt < 1). Because At and Bt are uncertain to occur, there is vulnerability, without which the role of trustworthiness (and trust) is limited (Yamagishi and Yamagishi, 1994). Hence, trust is different from the “shadow of the future” effect that arises in repeated games (e.g., Vanneste and Frank, 2013), where a long-time horizon or a low discount rate may guarantee cooperation even among untrustworthy partners because it is in their best interest (Friedman, 1971). Fourth, trust as an intention to act depends on the context-dependent payoffs At and Bt. We may trust a partner in one situation but not in another (Bhattacharya et al., 1998; Coleman, 1990; Deutsch, 1958).

How trust develops over time

Given this conceptualization of trust as an increasing function of the left-hand side of Equation (1), changes in trust must follow when the right-hand side changes. This can occur through at least four different mechanisms: initial bias correction, change in relationship value, identification, and trust-based selection. In the real world these mechanisms may co-occur, but we discuss each mechanism separately to enhance clarity. Consequently, our propositions have the usual ceteris paribus interpretation. Table 2 gives an overview of the mechanisms and their effect on the association between trust and relationship duration.

Table 2. When does trust increase over time?
MechanismElement: Et[R] = pt At + (1 − pt)BtOver time, on average, trust
1. Initial bias correctionp1Optimistic: trustor overestimates the level of trustworthinessPessimistic: trustor underestimates the level of trustworthiness
2. Change in relationship valueAt, BtAt : decreases upsideAt : increases upside
Bt : increases downsideBt : decreases downside
3. IdentificationTrustor: higher At and lower Bt Trustor: obtains higher utility from trusting
Trustee: pt increases
Trustee: becomes more trustworthy
4. Trust-based selectionEt[R] > Et[NR] Trustor terminates relationships if perceived trustworthiness is low
Mechanism 1: initial bias correction

Before entering any relationship, a trustor may be pessimistic, unbiased, or optimistic about the partner's trustworthiness (in the formal description, this is denoted by p1). A pessimistic (optimistic) trustor is one who underestimates (overestimates) that trustworthiness; an unbiased trustor makes the correct estimation. The trustor's initial bias has an effect not only on the willingness to begin a relationship but also on the evolution of trust within a relationship once it is established, as we explain next.

Engaging in a relationship provides the trustor with firsthand evidence, influencing the trustor's estimate of its partner's trustworthiness. Thus, the trustor gradually revises its assessment of the trustee to approach the true level of trustworthiness. For a pessimistic initial assessment, this may lead to increasing trust (Anderson and Weitz, 1989; Lewicki and Bunker, 1995; Shapiro et al., 1992). Conversely, an initially optimistic trustor is more likely to revise its beliefs downward, decreasing trust over time. For unbiased trustors, trust will remain constant over time on average. Because trust increases in the perceived trustworthiness of the trustee, initial bias correction will affect whether trust increases or decreases over time.

Proposition 1a: With an optimistic (pessimistic, unbiased) initial belief about the other's trustworthiness, on average trust will decline (increase, not change) over time.

As an illustration, suppose initial beliefs about trustworthiness are more optimistic when an acquirer and its target are from the same industry than when they are from different industries. Such optimism could be explained by an in-group bias by which people ascribe better qualities to others from the same group (Tajfel, 1982). One such quality is trustworthiness (Brewer, 1996; Platow, McClintock, and Liebrand, 1990). If post-merger integration carries informative signals about firms' trustworthiness, the optimistic initial bias should result in trust forming more rapidly (or decreasing less rapidly) after between-industry than after within-industry acquisitions.2

The research of Pillutla et al. (2003) and Weber et al. (2005) suggests a moderator of Proposition 1a.3 Trusting acts (by the trustor) may affect trustworthiness (of the trustee). For example, a decision by the trustor to engage in a relationship with the trustee may make the trustee more trustworthy. Hence, a more optimistic trustor may endogenously lead to a more trustworthy trustee. As long as the endogenous increase in trustworthiness does not lead to a reversal in the bias (i.e., optimistic becomes pessimistic, or vice versa), Proposition 1a holds. The effect of Proposition 1a will be weaker in the presence of the moderator and stronger in its absence.

Proposition 1b: If the trustor's trusting acts affect the trustee's trustworthiness, then the effect of initial bias correction on trust is weakened: an optimistic bias will lead to a smaller reduction in trust, whereas a pessimistic bias will lead to a smaller increase in trust.

If the trustor can expand or shrink the scope of the relationship, then a change in perceived trustworthiness has an additional effect on trust. Consider a relationship in which the trustor can either choose project 1 with a large scope or project 2 with a small scope. The interesting case is when both the upside and downside of project 1 are greater (i.e., A1t > A2t and |B1t| > |B2t|). If a project has a greater upside and smaller downside, then that project is preferred regardless of perceived trustworthiness, and project choice will not change over time. In contrast, the choice between the projects 1 and 2 above depends on perceived trustworthiness: project 1 is preferred if perceived trustworthiness is high, else project 2 (as long as project 2 provides more value than not being in the relationship).4 A pessimistic trustor may initially prefer project 2, but after awhile the trustor may prefer project 1 as the beliefs about trustworthiness are revised upward. If so, then trust not only increases because perceived trustworthiness increases (Proposition 1a), but also because for the new level of perceived trustworthiness, the expected value from project 1 is greater than that of 2 (else, the trustor would not have switched). The trustor's willingness to engage in the relationship (i.e., trust) goes up even more. Likewise, an optimistic trustor may over time reduce the perception of trustworthiness, and consequently prefer project 2 over 1, leading to a lesser decline in trust (relative to the situation where project 1 could not have been abandoned).5

Proposition 1c: The possibility to adjust the scope of the relationship positively moderates the effect of an initial bias on trust: an optimistic bias will lead to a smaller reduction in trust, whereas a pessimistic bias will lead to a greater increase in trust.

Mechanism 2: exogenous change in relationship value

Trust is context specific (Bhattacharya et al., 1998; Coleman, 1990; Deutsch, 1958); hence, a party might be trusted in one context but not in another. This intuition is captured by Equation (1), where the level of trust is a function of the relationship's value (i.e., payoffs At and Bt). So, if a trustor and trustee interact in different contexts over time, then trust can change even when perceived trustworthiness in the relationship remains constant.

Suppose that, in each period, a trustor and a trustee engage in a task with given payoffs. Then, all else (including perceived trustworthiness) being equal, trust is increasing in At and decreasing in Bt. Changes in relationship value thus imply the following statements:

Proposition 2a: If the gains from a partner's trustworthy behavior increase (decrease) over time, then trust will likewise increase (decrease) over time.

Proposition 2b: If the losses from a partner's untrustworthy behavior increase (decrease) over time, then trust will decrease (increase) over time.

It is important to note that these propositions refer to trust—the trustor's willingness to engage in a relationship with the trustee (i.e., an intention to act)—as distinct from a trusting act—which occurs when the trustor actually engages with the trustee (i.e., the actual accepting of vulnerability). A situation with high vulnerability (i.e., low At and/or high Bt) will lead to a low willingness to engage in a relationship (i.e., low trust), see Propositions 2a and 2b. However, a trusting act in a situation with high vulnerability must imply high perceived trustworthiness.

Propositions 2a and 2b describe exogenous changes in relationship value. Proposition 1c is about endogenous changes in relationship value. Their implications differ in the following way. If we consider the earlier projects 1 and 2 where 1 has a greater upside and downside (i.e., A1t > A2t and |B1t| > |B2t|), then exogenously forcing the trustor to switch projects may lead to lower or higher trust depending on perceived trustworthiness. In contrast, if the trustor is free to choose, then she will only switch projects if that makes her more willing to engage in the relationship; trust will necessarily go up.

Relationship-specific investments affect gains and losses in an exchange relation (Bidwell and Fernandez-Mateo, 2010; Williamson, 1975, 1985). For instance, Bt (the trustor's potential loss in period t) would increase in response to an increase in the possibility of being “held up” by an opportunistic partner. Consider a company that has invested in a relationship with supplier S1 but could transfer its investment to supplier S2 at some cost, and at higher cost to supplier S3 (but not elsewhere). Now assume that S2 is no longer an option because of some exogenous event (e.g., bankruptcy). We can then expect that, since Bt has increased (because S1 can now inflict larger damages by acting opportunistically), the firm's trust in the relation with S1 may decrease. Suppose now that a change in regulations or in consumer demand exogenously increases the supplier's potential profits At from the relation (and its investment) but not the losses from the buyer acting opportunistically (Zajac and Olsen, 1993); trust should now increase.

Mechanism 3: identification

Identification can work in both directions: a trustee may identify with a trustor and/or vice versa. Shapiro et al. (1992) and Lewicki and Bunker (1995) argue that interactions lead people to identify with each other over time as they internalize each other's preferences. Hence, as X and Y interact with each other repeatedly, X begins to care about good and bad outcomes for Y and considers them as also being good and bad for himself (and vice versa). This dynamic leads to increased trustworthiness (Kramer, 1999; Lewicki and Bunker, 1995; Poppo et al., 2008; Shapiro et al., 1992). As X begins to care about Y receiving good outcomes, X becomes more trustworthy from Y's perspective. If perceived trustworthiness (eventually) adjusts towards actual trustworthiness, then over time Y finds X more trustworthy and will have a higher level of trust in him. This process essentially increases the benevolence and possible integrity of the trustee (Mayer et al., 1995). The faster the trustee identifies with the trustor and the more the trustor recognizes this identification, the more rapidly trust increases over time.

Interacting repeatedly can also induce a trustor to identify with the trustee—the trustor cares also about the trustee's payoffs. Identification then increases the trustor's payoffs as they become an increasing function of the trustee's (positive) payoffs. In Equation (1), this is equivalent to increasing trust by increasing At or by decreasing Bt, where either change is endogenous to the relationship (unlike our previous mechanism involving exogenous changes in relationship value). Note that identification can lead to an increase but never decrease in trust over time.

Proposition 3: If partners identify more with each other over time, then trust will increase over time.

One implication of this claim is that trust develops differently across contexts in which the rate of identification differs. Consider the difference between services being outsourced offshore versus onshore (Srikanth and Puranam, 2011). For contracts of identical duration between the same client and vendor, the frequency of interaction is probably higher with onshore than with offshore outsourcing because of the geographical distances involved. If identification increases with the number of face-to-face interactions (Lewicki and Bunker, 1995; Shapiro et al., 1992) and if there are more interactions in onshore than offshore outsourcing, then identification between client and vendor personnel will develop more rapidly in onshore settings. Similarly, trust between the personnel involved should develop more quickly for onshore than for offshore outsourcing.

Mechanism 4: trust-based selection

Dealing with unsuccessful relationships plays a crucial role in the evolution of trust over time. Given the possibility of exit, a trustor will continue to interact only with partners that are trusted (Gambetta, 1988). Hence, such partners are more likely to continue the relationship (Anderson and Weitz, 1989; Doney and Cannon, 1997; Ganesan, 1994). A trustor's long-lived relationships will then be primarily with trustworthy partners because untrustworthy ones are deselected over time. Therefore, trust should be high in long-lived relationships.

The shorter relationships in which a trustor is involved will feature a mix of trustworthy and untrustworthy partners because, in such relations, the trustor has not yet gathered enough information on whether or not the trustee is trustworthy. Deselection implies that trustors will, on average, have more trust in long-lived than in short-lived relationships. Note that this mechanism affects the average level of trust in a portfolio of relationships of a single trustor because it results in that portfolio including mainly trustworthy partners. However, trust-based deselection does not affect the level of trust in a given relationship between a trustor and trustee.

Proposition 4: With deselection of untrustworthy partners, the average trust in a portfolio of relationships will increase over time.

One implication of this statement is that the rate of trust development in a portfolio of relationships depends on the presence of exit barriers. The easier it is to cancel contracts and discard untrustworthy partners, the faster is the increase in partners' average trustworthiness. So, if contract workers are more easily laid off than employees, then a firm's average trust may increase more rapidly in contract workers than in regular employees. Similarly, you may increase trust in friends more rapidly than in your family members because only the former can be deselected (“you can choose your friends but you can't choose your relatives”).


  1. Top of page
  2. Abstract

Our analysis indicates that numerous possible moderators of the relationship between trust and relationship duration exist because multiple (and sometimes opposing) mechanisms link the two. Mechanisms 3 and 4 lead to the prediction that trust increases with relationship length, while mechanisms 1 and 2 do not generate unambiguous predictions. Unless we know the mechanisms' direction and relative strength, we cannot predict whether trust increases or decreases over time.

This paper makes both theoretical and empirical contributions. Empirically, we assess evidence on the link between relationship duration and trust more comprehensively than prior work has done. Theoretically, we build on established trust definitions to clarify known and to suggest new mechanisms that influence how trust develops over time. Thus, we can distinguish among several related constructs: trust, trustworthiness, context, uncertainty, and propensity to trust.

Two established mechanisms (identification and trust-based selection) are readily represented in our formalization. It also advances our understanding of two novel mechanisms via which trust develops over time. First, trust between parties may increase as they learn about each other's trustworthiness (Anderson and Weitz, 1989; Lewicki and Bunker, 1995; Shapiro et al., 1992); however, learning leads to more trust only in the case of pessimistic initial beliefs about such trustworthiness. Second, exogenous changes in relationship value also affect trust as a situation-specific construct. Situations can differ in the potential gains (or losses) arising from a relation with a trustworthy (or untrustworthy) partner. For instance, exogenous changes in the value of relationship-specific investments or in the availability of alternative partners may lead to either an increase or a decrease in trust. If such changes occur systematically over time, then relationship duration and trust may be correlated.

Perhaps the most important implication of our results for future research is that merely correlating trust and relationship duration yields limited insight because a number of mechanisms could have either negative or positive (or no) effects. The net effect is indeterminate a priori. Further, the available evidence is seldom sufficient to determine which mechanisms are at play. Our meta-analytic results are consistent with multiple combinations of different mechanisms.

This difficulty in extracting the underlying mechanisms from available data suggests that examining moderators of the association (between trust and relationship duration) is more promising than looking at the main effect. While others may exist, we identify four classes of constructs that could be measured: initial beliefs, situational payoffs, mutual identification, and deselection. Isolating these would provide researchers the opportunity to study them in combination. For example, if trustworthiness is domain specific, then how does a changing situation affect the value of the relationship and the level of trustworthiness, and ultimately how trust develops? Finally, it would be illuminating to study how trustors revise their opinion as well as how (if at all) they generalize experiences with partners to their initial beliefs about new partners from the same population.

We acknowledge certain limitations. First, our meta-analysis relies on nonexperimental data. Using pairwise correlations between trust and relationship duration implies that uncontrolled factors (of which we identify some in our theory) may influence these correlations. Second, the studies we analyze use cross-sectional data. Although our theoretical mechanisms should be reflected in such data, inferring their existence or strength is difficult. Third, we conceptually distinguish different mechanisms affecting the development of trust over time, but the data at hand do not let us distinguish them empirically. Despite these limitations, a meta-analysis can better separate true effects from noise (i.e., sampling error) than individual studies.

How trust develops in the course of a relationship is key to many studies on the performance and structure of exchange relationships, including alliances, organization design, contracting, and social embeddedness. We propose ways in which a more sophisticated understanding of this relationship can be built. Because mechanisms likely have an unequal impact across domains, identifying when which mechanisms matter the most may enrich the theory specific to each.


  1. Top of page
  2. Abstract

The authors are grateful for the comments of Eva Ascarza, Guoli Chen, Felipe Csaszar, Kurt Dirks, Isabel Fernandez-Mateo, Don Ferrin, Inge Geyskens, Javier Gimeno, Madan Pillutla, Andrew Shipilov, seminar participants at Eindhoven University of Technology, INSEAD, and Tilburg University as well as for the research assistance of Cécile Girardon. The authors also appreciate the insightful and constructive comments of the editor Will Mitchell and the reviewers. Vanneste gratefully acknowledges financial support from the INSEAD Alumni Fund.


  1. Top of page
  2. Abstract
  • *Ahearne M, Gruen T, Jarvis C. 1999. If looks could sell: moderation and mediation of the attractiveness effect on salesperson performance. International Journal of Research in Marketing 16(4): 269284.
  • *Anderson E, Weitz B. 1989. Determinants of continuity in conventional industrial channel dyads. Marketing Science 8(4): 310323.
  • Bhattacharya R, Devinney TM, Pillutla MM. 1998. A formal model of trust based on outcomes. Academy of Management Review 23(3): 459472.
  • Bidwell M, Fernandez-Mateo I. 2010. Relationship duration and returns to brokerage in the staffing sector. Organization Science 21(6): 11411158.
  • Bradach JL, Eccles RG. 1989. Price, authority, and trust – from ideal types to plural forms. Annual Review of Sociology 15: 97118.
  • Brewer MB. 1996. In-group favoritism: the subtle side of intergroup discrimination. In Codes of Conduct: Behavioral Research and Business Ethics, Messick DM, Tenbrunsel AE (eds). Russell Sage Foundation: New York; 160171.
  • *Cardona P, Elola A. 2003. Trust in management: the effect of managerial trustworthy behavior and reciprocity. IESE Working paper D/496. University of Navarra, IESE Business School, Barcelona, Spain. Available at SSRN:
  • Chua RYJ, Ingram P, Morris MW. 2008. From the head and the heart: locating cognition-and affect-based trust in managers' professional networks. Academy of Management Journal 51(3): 436452.
  • *Chua RYJ, Morris MW, Ingram P. 2009. Guanxi vs networking: distinctive configurations of affect-and cognition-based trust in the networks of Chinese vs. American managers. Journal of International Business Studies 40(3): 490508.
  • *Chua RYJ, Morris MW, Ingram P. 2010. Embeddedness and new idea discussion in professional networks: the mediating role of affect-based trust. Journal of Creative Behavior 44(2): 85104.
  • Cohen J. 1992. A power primer. Psychological Bulletin 112(1): 155159.
  • Coleman J. 1990. Foundations of Social Theory. Harvard University Press: Cambridge, MA.
  • Colquitt JA, Scott BA, LePine JA. 2007. Trust, trustworthiness, and trust propensity: a meta-analytic test of their unique relationships with risk taking and job performance. Journal of Applied Psychology 92(4): 909926.
  • Corts KS, Singh J. 2004. The effect of repeated interaction on contract choice: evidence from offshore drilling. Journal of Law, Economics & Organization 20(1): 230260.
  • *Coulter KS, Coulter RA. 2002. Determinants of trust in a service provider: the moderating role of length of relationship. Journal of Services Marketing 16(1): 3550.
  • *Currall SC. 1992. Group representatives in educational institutions: an empirical study of superintendents and teacher union presidents. Journal of Applied Behavioral Science 28(2): 296317.
  • Davis JH, Schoorman FD, Mayer RC, Tan HH. 2000. The trusted general manager and business unit performance: empirical evidence of a competitive advantage. Strategic Management Journal 21(5): 563576.
  • Deutsch M. 1958. Trust and suspicion. Journal of Conflict Resolution 2(4): 265279.
  • Dirks KT, Ferrin DL. 2002. Trust in leadership: meta-analytic findings and implications for research and practice. Journal of Applied Psychology 87(4): 611627.
  • *Doney PM, Cannon JP. 1997. An examination of the nature of trust in buyer-seller relationships. Journal of Marketing 2(2): 3551.
  • *Dyer JH, Chu W. 2000. The determinants of trust in supplier-automaker relationships in the U.S., Japan, and Korea. Journal of International Business Studies 31(2): 259285.
  • Ebert TAE. 2007. Interdisciplinary trust meta-analysis. Working paper 2007–18, Ludwig-Maximilians-Universität München, Munich School of Management, Munich, Germany. 1134.
  • *Fang E, Palmatier RW, Scheer LK, Li N. 2008. Trust at different organizational levels. Journal of Marketing 72(2): 8098.
  • *Fried Y, Tiegs RB, Bellamy AR. 1992. Personal and interpersonal predictors of supervisors' avoidance of evaluating subordinates. Journal of Applied Psychology 77(4): 462468.
  • Friedman JW. 1971. A non-cooperative equilibrium for supergames. Review of Economic Studies 38(1): 112.
  • *Fryxell GE, Dooley RS, Vryza M. 2002. After the ink dries: the interaction of trust and control in US-based international joint ventures. Journal of Management Studies 39(6): 865886.
  • Gambetta D. 1988. Can we trust trust?. In Trust: Making and Breaking Cooperative Relations, Gambetta D (ed). Basil Blackwell: Cambridge, MA; 213237.
  • *Ganesan S. 1994. Determinants of long-term orientation in buyer-seller relationships. Journal of Marketing 58(2): 119.
  • Geyskens I, Steenkamp JBEM, Kumar N. 1998. Generalizations about trust in marketing channel relationships using meta-analysis1. International Journal of Research in Marketing 15(3): 223248.
  • Geyskens I, Steenkamp JBEM, Kumar N. 2006. Make, buy, or ally: a transaction cost theory meta-analysis. Academy of Management Journal 49(3): 519543.
  • Granovetter M. 1985. Economic action and social structure: the problem of embeddedness. American Journal of Sociology 91(3): 481510.
  • Greve HR, Baum JAC, Mitsuhashi H, Rowley TJ. 2010. Built to last but falling apart: cohesion, friction, and withdrawal from interfirm alliances. Academy of Management Journal 53(2): 302322.
  • Gulati R. 1995. Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal 38(1): 85112.
  • Gulati R, Nickerson JA. 2008. Interorganizational trust, governance choice, and exchange performance. Organization Science 19(5): 688708.
  • *Gulati R, Sytch M. 2008. Does familiarity breed trust? Revisiting the antecedents of trust. Managerial and Decision Economics 29(2-3): 165190.
  • Hardin R. 2002. Trust and Trustworthiness. Russell Sage Foundation Publications: New York.
  • Huffcutt AI, Arthur W. 1995. Development of a new outlier statistic for meta-analytic data. Journal of Applied Psychology 80(2): 327334.
  • Hunter JE, Schmidt FL. 2004. Methods of Meta-Analysis – Correcting Error and Bias in Research Findings. Sage: Thousand Oaks, CA.
  • *Klein R, Rai A. 2009. Interfirm strategic information flows in logistics supply chain relationships. MIS Quarterly 33(4): 735762.
  • *Korsgaard MA, Brodt SE, Whitener EM. 2002. Trust in the face of conflict: the role of managerial trustworthy behavior and organizational context. Journal of Applied Psychology 87(2): 312319.
  • Kramer RM. 1999. Trust and distrust in organizations: emerging perspectives, enduring questions. Annual Review of Psychology 50(1): 569598.
  • *Krishnan R, Martin X, Noorderhaven NG. 2006. When does trust matter to alliance performance? Academy of Management Journal 49(5): 894917.
  • *Kumar N, Scheer LK, Steenkamp JBEM. 1995. The effects of supplier fairness on vulnerable resellers. Journal of Marketing Research 32(1): 5465.
  • Lado AA, Dant RR, Tekleab AG. 2008. Trust-opportunism paradox, relationalism, and performance in interfirm relationships: evidence from the retail industry. Strategic Management Journal 29(4): 401423.
  • *Lagace RR, Dahlstrom R, Gassenheimer JB. 1991. The relevance of ethical salesperson behavior on relationship quality: the pharmaceutical industry. Journal of Personal Selling and Sales Management 11(4): 3947.
  • *Levin DZ, Walter J, Murnighan JK. 2011. Dormant ties: the value of reconnecting. Organization Science 22(4): 923939.
  • *Levin DZ, Whitener EM, Cross R. 2006. Perceived trustworthiness of knowledge sources: the moderating impact of relationship length. Journal of Applied Psychology 91(5): 11631171.
  • Lewicki RJ, Bunker BB. 1995. Trust in relationships: a model of development and decline. In Conflict, Cooperation and Justice, Bunker BB, Rubin JZ (eds). Jossey-Bass: San Francisco, CA; 133173.
  • Li JJ, Poppo L, Zhou KZ. 2010. Relational mechanisms, formal contracts, and local knowledge acquisition by international subsidiaries. Strategic Management Journal 31(4): 349370.
  • *Lui SS, Ngo H. 2004. The role of trust and contractual safeguards on cooperation in non-equity alliances. Journal of Management 30(4): 471485.
  • *Luo Y. 2002. Building trust in cross-cultural collaborations: toward a contingency perspective. Journal of Management 28(5): 669694.
  • *Luo Y. 2008. Structuring interorganizational cooperation: the role of economic integration in strategic alliances. Strategic Management Journal 29(6): 617637.
  • Mayer RC, Davis JH, Schoorman FD. 1995. An integrative model of organizational trust. Academy of Management Review 20(3): 709734.
  • McEvily B. 2011. Reorganizing the boundaries of trust: from discrete alternatives to hybrid forms. Organization Science 22(5): 12661276.
  • *Mohr A. 2010. Does success breed trust? Revisiting the trust-performance link in International Strategic Alliances. Working paper, Kent Business School, University of Kent, Canterbury, UK. 1-36.
  • *Möllering G. 2002. Perceived trustworthiness and inter-firm governance: empirical evidence from the UK printing industry. Cambridge Journal of Economics 26(2): 139160.
  • *Nguyen TV, Rose J. 2009. Building trust--evidence from Vietnamese entrepreneurs. Journal of Business Venturing 24(2): 165182.
  • Nooteboom B. 2002. Trust: Forms, Foundations, Functions, Failures and Figures. Edward Elgar Publishing: Cheltenham, UK.
  • *Norman PM. 2004. Knowledge acquisition, knowledge loss, and satisfaction in high technology alliances. Journal of Business Research 57(6): 610619.
  • Ortmann A, Fitzgerald J, Boeing C. 2000. Trust, reciprocity, and social history: a re-examination. Experimental Economics 3(1): 81100.
  • Parkhe A. 1993. Strategic alliance structuring: a game theoretic and transaction cost examination of interfirm. Academy of Management Journal 36(4): 794829.
  • *Perrone V, Zaheer A, McEvily B. 2003. Trust as an organizing principle. Organization Science 14(1): 91103.
  • Pillutla MM, Malhotra D, Murnighan JK. 2003. Attributions of trust and the calculus of reciprocity. Journal of Experimental Social Psychology 39(5): 448455.
  • Platow MJ, McClintock CG, Liebrand WBG. 1990. Predicting intergroup fairness and ingroup bias in the minimal group paradigm. European Journal of Social Psychology 20(3): 221239.
  • Poppo L, Zenger T. 2002. Do formal contracts and relational governance function as substitutes or complements? Strategic Management Journal 23(8): 707725.
  • *Poppo L, Zhou KZ, Ryu S. 2008. Alternative origins to interorganizational trust: an interdependence perspective on the shadow of the past and the shadow of the future. Organization Science 19(1): 3955.
  • Puranam P, Vanneste BS. 2009. Trust and governance: untangling a tangled web. Academy of Management Review 34(1): 1131.
  • Reuer JJ, Ariño A. 2007. Strategic alliance contracts: dimensions and determinants of contractual complexity. Strategic Management Journal 28(3): 313330.
  • *Rindfleisch A. 2000. Organizational trust and interfirm cooperation: an examination of horizontal versus vertical alliances. Marketing Letters 11(1): 8195.
  • Ring PS, Van de Ven AH. 1992. Structuring cooperative relationships between organizations. Strategic Management Journal 13(7): 483498.
  • Ring PS, Van de Ven AH. 1994. Developmental processes of cooperative interorganizational relationships. Academy of Management Review 19(1): 90118.
  • Rousseau DM, Sitkin SB, Burt RS, Camerer C. 1998. Not so different after all: a cross-discipline view of trust. Academy of Management Review 23(3): 393404.
  • Shah RH, Swaminathan V. 2008. Factors influencing partner selection in strategic alliances: the moderating role of alliance context. Strategic Management Journal 29(5): 471494.
  • *Shamir B, Zakay E, Breinin E, Popper M. 1998. Correlates of charismatic leader behavior in military units: subordinates' attitudes, unit characteristics, and superiors' appraisals of leader performance. Academy of Management Journal 41(4): 387409.
  • Shapiro DL, Sheppard BH, Cheraskin L. 1992. Business on a handshake. Negotiation Journal 8(4): 365377.
  • *Squire B, Cousins PD, Brown S. 2009. Cooperation and knowledge transfer within buyer–supplier relationships: the moderating properties of trust, relationship duration and supplier performance. British Journal of Management 20(4): 461477.
  • Srikanth K, Puranam P. 2011. Integrating distributed work: comparing task design, communication, and tacit coordination mechanisms. Strategic Management Journal 32(8): 849875.
  • Swan JE, Bowers MR, Richardson LD. 1999. Customer trust in the salesperson: an integrative review and meta-analysis of the empirical literature. Journal of Business Research 44(2): 93107.
  • Tajfel H. 1982. Social psychology of intergroup relations. Annual Review of Psychology 33(1): 139.
  • Uzzi B. 1997. Social structure and competition in interfirm networks: the paradox of embeddedness. Administrative Science Quarterly 42(1): 3769.
  • Vanneste BS, Frank DH. 2013. Forgiveness in vertical relationships: incentive and termination effects. Organization Science (Forthcoming).
  • Vanneste BS, Puranam P. 2010. Repeated interactions and contractual detail: identifying the learning effect. Organization Science 21(1): 186201.
  • *Wang Q, Bradford K, Xu J, Weitz B. 2008. Creativity in buyer-seller relationships: the role of governance. International Journal of Research in Marketing 25(2): 109118.
  • Wang HC, He J, Mahoney JT. 2009. Firm-specific knowledge resources and competitive advantage: the roles of economic-and relationship-based employee governance mechanisms. Strategic Management Journal 30(12): 12651285.
  • *Wang S, Tomlinson EC, Noe RA. 2010. The role of mentor trust and protégé internal locus of control in formal mentoring relationships. Journal of Applied Psychology 95(2): 358367.
  • *Wasti SN, Wasti SA. 2008. Trust in buyer–supplier relations: the case of the Turkish automotive industry. Journal of International Business Studies 39(1): 118131.
  • Weber JM, Malhotra D, Murnighan JK. 2005. Normal acts of irrational trust: motivated attributions and the trust development process. In Research in Organizational Behavior, Kramer RM, Staw B (eds). JAI Press: New York; 75101.
  • Whitener EM. 1990. Confusion of confidence intervals and credibility intervals in meta-analysis. Journal of Applied Psychology 75(3): 315321.
  • Williamson OE. 1975. Markets and Hierarchies – Analysis and Antitrust Implications. The Free Press: New York.
  • Williamson OE. 1985. The Economic Institutions of Capitalism. Free Press: New York.
  • Yamagishi T, Yamagishi M. 1994. Trust and commitment in the United States and Japan. Motivation and Emotion 18(2): 129166.
  • *Yang J, Mossholder KW. 2010. Examining the effects of trust in leaders: a bases-and-foci approach. Leadership Quarterly 21(1): 5063.
  • *Yilmaz C, Kabadayi ET. 2006. The role of monitoring in interfirm exchange: effects on partner unilateral cooperation. Journal of Business Research 59(12): 12311238.
  • *Young-Ybarra C, Wiersema M. 1999. Strategic flexibility in information technology alliances: the influence of transaction cost economics and social exchange theory. Organization Science 10(4): 439459.
  • Zajac EJ, Olsen CP. 1993. From transaction cost to transactional value analysis: implications for the study of interorganizational strategies. Journal of Management Studies 30(1): 131145.
  1. 1

    In other words, trust can be described as a monotonically increasing function f of Et[R] such that ∂f(Et[R])/∂Et[R] > 0. Variations in the propensity to trust (see Colquitt, Scott, and LePine, 2007; Ortmann, Fitzgerald, and Boeing, 2000), can be captured through a less or more steep (but always increasing in Et[R]) function f.

  2. 2

    A reviewer pointed out that, if people entering relationships tend to be positively biased in their expectations regarding trustworthiness or gains from the exchange, then subsequent corrections should lead to a decline in trust. We believe it is an empirical question whether such a bias is widespread.

  3. 3

    We thank a reviewer for encouraging us to explore this moderator.

  4. 4

    If the trustor prefers the highest expected value, then project 1 is chosen for high p, project 2 for intermediate p, and no relationship for low p. Formally, she will choose project 1 over 2 if pA1t + (1 − p)B1t > pA2t + (1 − p)B2t [LEFT RIGHT DOUBLE ARROW] p > (B2 − B1)/(A1 − A2 + B2 − B1). Project 2 is preferred over no relationship if pA2t + (1 − p)B2t > 0 [LEFT RIGHT DOUBLE ARROW] p > B2/(−A2 + B2). To ensure that project 2 is preferred for some intermediate p, it must be that the threshold p for project 1 is greater than the threshold for project 2: (B2 − B1)/(A1 − A2 + B2 − B1) > B2/(−A2 + B2), which is the case as long as A1/B1 ≠ A2/B2.

  5. 5

    We thank a reviewer for suggesting this mechanism.