Zmf 4.pdf

Trust and honesty 4 Trust There are different views on trust... - “It is impossible to go through life without trust; that is to be imprisoned in the worst cell of all, oneself.” Graham Greene -> If I dont trust people, there is no way to live -> I trust people all the time (for example i trust the...

Trust and honesty 4 Trust There are different views on trust... - “It is impossible to go through life without trust; that is to be imprisoned in the worst cell of all, oneself.” Graham Greene -> If I dont trust people, there is no way to live -> I trust people all the time (for example i trust the architect, when he builds my house, i leave my stuff in the train, when I need to go to the toilet etc.) - “There is no trust, no faith, no honesty in men; all perjured, all forsworn, all naught, all dissemblers.” William Shakespeare ➔ We see a very different view of people! Trusting People is normally a good idea. Not trusting comes in organization with a cost A view on trust in the 1970s: Important but not well investigated in management research Trust is to fuzzy to study about What is trust? A working definition -> Prüfungsfrage - - Trust is making yourself vulnerable to another person’s exploitation with the prospect of receiving some benefit if that other person honors the trust given to them (Rosseau et al., 1998; Fetchenhauer & Dunning, 2010) Examples: o Sharing another’s secret risks public embarrassment, but may receive social support o Handing money over to a bank risks never seeing it again, but may yield a safe storage place, etc. Trust, like cooperation can be a puzzle... Hobbes: „Der Mensch ist des menschens Wolf“ -> I’m scared of punishment and this reculates my behavior (I would shoot with a gun right now, if i wouldn’t be scared to go to prison Trust in close relationships: Positive reinforcement - - Trust in the shadow of uncertain exploitation can be built up over several interactions by gradual and repeated demonstrations of trust and cooperation o Two distrusting parties can come to a more mutually cooperative stance after signaling that they can be trusted ▪ » Tit-for-tat with slightly higher costs of cooperation each time they interact ▪ » Graduated reciprocation in tension reduction (GRIT, Osgood, 1962; Lindskold, 1978). o One party makes a small unilateral gesture of trust and the other party communicates that he or she is willing to match it. Over time, we will observe more trusting and more trustworthy behaviors as the relationship intensifies Problem: It works best in close relationships, but in the workplace we often work with people who are “strangers”. Trust among strangers - - - Many workplace and social interactions are short-termed, on-off types of interactions o Sourcing of materials with unknown quality o Trusting a doctor trying to tell you that you need an expensive treatment o Trusting a mechanic needing to “fix” your car o Trusting a remote colleague about a detail in a presentation o Sitting on the train and asking a stranger to “watch the bags” while going to the restroom Trust among strangers is the heart of “social capital” within a certain social group o Country o Organization Trust (social capital) is regarded as a pre-requisite for the functioning of modern organizations and countries Trust is the lubricant of efficient interaction -> Things work smoothly Social capital measurement in the World Value Survey - The WVS contains a one-item measure of trust: o “Generally speaking, would you say that most people in your country can be trusted or than one cannot be too careful?” - Substantial differences between countries: Trust can mean various things... (Thompson, 2012) - - - Deterrence-based trust o Based on consistency of behavior (threats and going trough with threats) ▪ Punishment ▪ Reward o Problems: Expensive to maintain (monitoring), Risk of backfire (reactance theory) Knowledge-based trust: o Grounded in behavioral predictability o Information is sufficient to predict behavior o Trust is a consequence of uncertainty Identification-based trust o Grounded in complete empathy with another person’s desires and intentions. o Typical in close relationships o Problem: Wrong signals about the relationship (e.g., prenuptial agreements) Correlates of World Value Survey Measure - Financial honesty Economic growth Corruption Etc. ➔ Good things happen if you have a lot of trust Trust research has increased since the advent of economic games - Dictator game Ultimatum game Prisoner’s dilemma Public Good Game Third-party punishment game Trust game Economic games: The dictator game - Two people interact in a one-shot interaction Player A is endowed with 100 tokens and has to decide how much (if any) to allocate to Player B The interaction ends -> Measure of pro-sociality (How much you give, is the measurement, because I could have kept it all for myself, in this case being social cost this many tokens) Economic games: The ultimatum game - Two people interact in a one-shot interaction Player A is endowed with 100 tokens and has to decide how much (if any) to allocate to Player B. This is called the proposal. Player B decides to accept or to reject the proposal. If Player B accepts, the money is distributed according to the proposal, if Player B rejects, both players leave without any money. -> Measure of second-party punishment (is Player B punishing unfair offers at his/her own expense) -> Measure of pro-sociality and expected acceptance (what does Player 1 offer, knowing that it may be rejected) Prisoner’s dilemma See lecture “Cooperation” Public good game - Groups of four interact Each member is endowed with 20 tokens and decides how much (if any) to contribute to a common pool. Common pool is multiplied by efficiency factor (e.g., 1.6) Common pool is equally distributed among all four players Dilemma: o It is individually best to free ride (payment is endowment plus 1/n of common pool) o It is socially best to contribute (more money gets multiplied by efficiency factor) o Efficiency factor has to be both sufficiently high and sufficiently low ▪ If too low: Inefficient (contribution of one needs to increase) ▪ If too high: Individually rational (MPCR needs to be smaller than 1) ➔ Make Cooperation socially but individuelly costly Public good with punishment - Stage 1: o Like before (PGG) Stage 2: o After the decision each person can see what the others have contributed and can use money to deduct from others (i.e., punish) o Efficiency factor in punishment (punishing 1 token has a consequence of >1 token) Economic games: Gift-exchange game - Two people interact in a one-shot game o Principal o Agent - - Principal cannot observe the worker’s effort, but has to set wage before agent chooses her effort level o Earnings can be, for example: 2 times the workers effort Agent chooses the effort level and has a positive cost of effort o Earnings: Endowment minus cost of effort Equilibrium prediction: o Principal anticipates the workers shirking behavior and, therefore, sets the minimal wage level o Agent minimizes effort o Inefficient outcome Typical results: Positive wages and positive effort (reciprocity) Measuring trust: The trust game as a sequential game Above Number refers to payoff of Player 1 If Player 1 decides to trust, Player 2 has the option to reciprocate (also trust) or to exploit ( not trust). If player 2 has exploidet he would have the higher payoff with 20 Trust game versus prisoner’s dilemma -> Prüfung!! - The trust game is a sequential game where players move sequentially o Player 2 gets to decide after player 1 has made her decision o Player 2 can condition her behavior to the behavior of player 1 - The prisoner’s dilemma is a simultaneous game where both players move under uncertainty about the action of the other o Player 2 cannot “respond”, but needs to act on her beliefs about player 1’s move - From the general structure, the games share a property o Rational behavior (fitness maximization) favors “no trust” and “no reciprocity” o Irrational behavior can yield higher payoff - The trust game can be formalized as a “sequential prisoner’s dilemma” - Solution concept: Backward induction o Player 1 tries to guess what Player 2 may do and calculates this into his/her decision - Empirical results are at odds with rational concept would suggest o Why? Trust behavior versus risk behavior? - The trust game may just boil down to a simple risk decision: o I can invest 5CHF and may get 10CHF back, depending on some positive probability ▪ Shakespeare’s thought: People are not trustworthy, I better keep my money” ▪ Greene’s thought: People are trustworthy, if I send the money, I will get 10 CHF back - Is trust just risk? o Imagine the following game: You can send money into a machine, and the machine knows the probability with which people typically reciprocate. It uses a random mechanism to decide to reciprocate the money or not. ▪ Do people send the money at higher frequency to the machine or to another human being? How well do measures of risk predict trust behavior? - Generally, trust behavior is not very well explained by risk-behavior o - People who found risky gambles to be attractive were no more likely to take their chances with another person in a trust game than those who found the gamble less attractive (Kanageratnam et al., 2009; Ashraf et al., 2006) o Measurements of risks in surveys and gambles do not predict behavior in the trust game (Ben-Nur & Halldorsson, 2010). o Measuring risk tolerance in various ways (survey, gambles, thrill-seeking, desire for novelty, disinhibition) does not show robust correlations with trust behavior. Trust – apparently – is much more than risk, What? -> Trust is not very well explained and it is more than risk! Can social expectations explain trust behavior? - Expectations-based theories on trust suggest that trust decisions are based on the estimated likelihood that trust is being reciprocated o Conditional cooperation.. Etc. - Classical economic approach with rational agents: o Do not trust because Player 2 will not reciprocate! - In how far do people belief that others will reciprocate trust? Berg et al.’s (1995) investment game - - The trust game was initially introduced as an investment game by Berg et al. (1995) The investment game is a continuous trust game o Rather than having to send the whole endowment as player 1, s/he gets to send any share of the initial endowment o Player 2 can reciprocate any amount s/he pleases Meta-study results (Johnson & Mislin, 2011) show that both trust and reciprocity is non-zero and frequent: People trust too much and too little (Fetchenhauer & Dunning, 2009, J. Econ. Psychol.) - Expectations-based models do not explain trust behavior well, due to two observations People trust too little: o People tend to be skeptical about the trustworthiness of others ▪ Expected return: 59.1% versus real return: 90.4% (Cornell U) ▪ Why so cynical? - People trust too much: o People trust much more (45%) than their expectation (30.7%) would suggest ▪ Why is that? - Their beliefs are off (they underestimate the actual trustworthiness), hence they should trust much more than they do - Their trust behavior does not match their belief, hence they should trust much less than they actually do - -> Expectations cannot alone explain trust behavior ➔ People trust more than they think they should (rationally) Why are people so cynical about other’s trustworthiness? - There is an informational asymmetry in the feedback depending on trusting and not-trusting o Trust: People reciprocate the trust (i.e., they are trustworthy) or they exploit - - - o No Trust: People will never learn if the interaction partner would have been trustworthy Study: People saw videos of people (no audio, 10-sec sequence of self- introduction) and had to estimate their trustworthiness o Trustworthiness is underestimated (cynicism), regardless of financial incentives Experience-sampling hypothesis: o Experience informs people in an asymmetric way o Feedback can decrease people’s cynicism Result: Unconditional feedback decreases people’s cynicism In a low-trust environment (e.g., high control), people never learn if people can be trusted... And? Can social expectations explain trust? - People’s expectations are – on average – off. They underestimate the trustworthiness of others Feedback asymmetry leads to low levels of updating expectations in an accurate way. Social expectations cannot fully explain trust behavior Are people just seeking to maximize the total payoff (i.e., efficiency concerns)? - An alternative explanation of trust in the trust game is people’s potential concern for the common good. o No trust: Total payoff is merely 5CHF o Trust: Total payoff is 20CHF - If people care about the maximum amount of money (irrespective who gets it), these efficiency concerns may underlie decisions to trust. - Trust game versus extended coin flip (Mensching et al., 2010) o Trust game: send 5 EUR, trustee decides to keep 20 EUR or send 10 EUR back, probability to meet a trustworthy trustee held constant at 50% o Extended coin flip: bet 5 EUR on a side of a coin, if correct receive 10 EUR and another random stranger is given 10 EUR ( fun to do in university: Approach random people and give them money...), if false, random person is given 20 EUR. o Result: Trust rate is larger than betting rate. ➔ People trust more, that they would bet in a similiar situation - Efficiency concerns do not explain behavior in the trust game to the full extent Relational concerns explain trust behavior: Emotional correlates - - - Results suggest that trust may not be an instrumental choice, but rather reflecting a relational concern o It is not about the prospect of receiving money, but about the expressive act of trusting itself (social norm, preference to trust irrespective of outcomes) Research into emotional correlates provides support for this hypothesis: o There are two layers of emotion relevant in the trust game ▪ How do I feel about the decision itself? (-> immediate emotions) ▪ How do I feel about the potential consequences? (-> anticipated emotions) o Anticipated emotions did not predict trust behavior, immediate emotions did o People feel strongly about the act of trusting itself Immediate and anticipated emotions were largely unrelated Relational dynamics of trust - What prompts people to send money even in situations in which they are pessimistic about receiving any money back? There is not crystal clear answer, but research is moving... o Relationship status of interaction partners matter - o 73% found it acceptable to pay a new employee $7 if the current one left o 83% found it unfair to decrease the wage to $7 of an existing employee o Similarities when asked about landlord/tenant relationships The relationship status between interaction partners may explain these differences Lawler et al. (2001, 2008) show how instrumental relationships become relational after some time Minimal relationship effect - Mere exposure to a relationship (“You are matched with another person...”) is enough to trigger expressive concerns Similarity to “Minimal group effect” (see Lecture on “Group processes”) People hold a norm to “be nice” Related phenomenon: Identifiable victim effect o People are more altruistic to “identified” victims, rather than an anonymous one o Famous Stalin quote: “One death is a tragedy. A million deaths is just a statistic” o Common strategy in donation campaigns Crossing the street to avoid the beggar” - People are sometimes motivated to “avoid” a relationship as if they know that they would give to a beggar when confronted, but then choose to cross the street to not see him In Dictator games: o People can either play a 10$ dictator game or o Exit the game for a payment of 9$ o More people exit than take the entire endowment or take 9$ and leave 1$ to the receiver Expressive trust as proximate mechanism of cooperation - As societies grow, cooperation is harder to maintain (i.e., cognitive complexity, reputation networks, etc.) Social norms can help and the norm to be nice can be a “social glue” Henrich et al. (2010, Science) found that dictator giving differed largely by degree of specialization of labor or market integration: o Among Americans, who buy 100% of their calories from strangers, giving is higher than among... o Hazda in Tanzania, who buy 0% of their calories from strangers. o Related finding: People from complex (i.e. larger, more divided in labor) societies are more willing to incur punishment when fairness is violated. - Evidence for expressive trust (Fetchenhauer & Dunning, 2009): o In hypothetical trust game, the trust rate is lower than in actual trust games o Being “already assigned” or “not yet assigned” to an interaction partner provides different results in trust rates, suggesting minimal relationship relevance. o If the other person does not know about being in a relationship, trust rates still grow ➔ In large complex societys we should trust and cooperate more The hidden cost of control - - Many companies rely on excessive control instruments o Time monitoring (do employees work long enough?) o Video taping (do employees steal goods/services?) o Performance measurements (do employees work sufficiently productive?) What are the effects of doing so? o Increased business performance (by means of deterrence, people are actually nasty and they need to be monitored) o Equal business performance (monitoring systems are costly, and the benefits just carry the cost) o o Decreased business performance (monitoring is too costly, given workers are actually not as nasty as companies believe they are) Even worse (the mere existence of monitoring actually decreases the performance) ▪ Distrust as a violation of social norms ▪ Reciprocal punishment for this distrust -> Employees would lower their effort Experiment on hidden cost of control (Falk & Kosfeld, 2006, Am. Econ. R.) - Two-stage principal agent game Agent chooses a productive activity x, which is costly to him but beneficial to the principal The cost of the agent is c(x)=x, while the benefit for the principal is 2x. The agent has an initial endowment of 120 tokens, the principal’s endowment is zero. Payoff functions: o πP = 2x o πA = 120-x - Before the agents exerts effort, the principal can restrict the action set by setting a minimal level (varies by treatment). Alternatively, the principal can leave the action set unrestricted - Principal can guarantee his minimal payment (2*minimum level) or leave it open (0) - Supplementary questionnaire to assess motivation of principals and agents ➔ It is better to let people go and let them do whatever they want, because if they got controlled, they would do just the bare minimum. ➔ Singnalling distrust has ist consequences Control leads to a system of low trust and low work motivation - Falk & Kosfeld (2006) show that control is perceived as low trust Fetchenhauer & Dunning (2010) can explain that this may be perceived as a relational violation, as trust is not instrumental, but rather an expressive act Not being trusted is a harsh social signal, it does not just confer risk attitudes Lack of trust can harm the integrity of people o Why am I not trusted? What is wrong with my firm? I am an honest person!!! Habits of virtue? (Peysakhovich & Rand, 2015, Management Science) - What may explain the variability in norms of trust and cooperation in organizations? Why is there a culture of cooperation in some firms, while there is a culture of skepticism in other firms? It may depend on how well norms are carried to novel situations Experiment: Participants are randomly allocated into one of two worlds o C culture: Cooperation pays off ( ->there is a higher shadow of the future) o D culture: Cooperation does not pay (-> there is a lower shadow of the future) o - Manipulation comes from the probability of another round (see lecture “Cooperation), and the benefits of defection After the experiment: Several games (DG; UG; TG; PGG) o How are behaviors different depending whether you come from a C or D culture? Honesty Honesty in classical economic theory (Becker, 1968): Rational crime - - - Cost and benefit analysis o If the cost of lying (getting caught and facing punishment) do not exceed the benefit of lying (cheating for a bigger reward) people cheat “rationally” Implications: o Threat of punishment (i.e., severity) should lower crime rate -> deterrence o Benefits of crime should increase crime rate -> it just pays o Risk of being detected (i.e., number of police officers) affects crime behavior -> too risky Important: o In the classical economic analysis, crime is not a moral, cultural, etc. problem, it is a mere instrumental (and financial) problem. Honesty in behavioral experiments - Die-under-the-cup paradigm (Fischbacher & Föllmi-Heusi, 2013) o Participants secretly roll a die under a cup and report their number by typing it into the computer screen ▪ 1=1Euro ▪ 2=2Euro ▪ 3=3Euro ▪ 4=4Euro ▪ 5=5Euro ▪ 6=0Euro o Participants are told to report their first number, but then could continue rolling to verify that the die is fair (i.e., not loaded) -> So the participants decide if they should be honest or not, because the instructor can not control their answer Die-under-the-cup: Results (Dis-)Honesty as a matter of business culture (Cohn et al., 2014, Nature) ➔ Field experiment with bankers Identity as a banker was experimentally increased or not Instead of die-roll, the experimenters used a coin-toss task Bankers lie more when they are reminded, that they are Bankers