1 of 44

Motivational Effects �of Feeling Trusted

Diya Abraham¹, Ondřej Krčál²

¹Masaryk University

²University of Reading

SEAM 2024

12^th Sempteber 2024

2 of 44

Motivating Example

2

You:

A

3 of 44

Motivating Example

3

M

HIGH

LOW

You:

A

4 of 44

Motivating Example

4

M

B

LOW

HIGH

You:

A

But now the manager says that he would like your colleague, employee B, to handle the high importance task and he assign the less important task to you. But now you’re wondering WHY the manager made this decision… Suppose you know this decision can’t have been based on competence or skill because the task itself does not require much skill. Still, you know the manager knows you both and he clearly preferred to trust employee B with the more important task. The question is – would this affect your motivation to work for the manager? i.e., Beyond the fact that the task itself is less important, would just the fact that the manager chose to trust your colleague over you affect your motivation to work for the manager?

I hypothesize that despite the fact that this is for you a completely payoff irrelevant decision it would still affect your motivation to work for the manager. Because as an employee, you care about how you are perceived by management and you’ve just received a signal that you are perceived as less trustworthy than your peers.

This paper is about quantifying the effect of finding out this information on the employee’s motivation to work for the manager.

5 of 44

Research question

How does revealing which of two employees a manager prefers to trust affect their motivation to work for the manager?

5

6 of 44

Method

6

A laboratory experiment.

Responsibility allocated unequally between two employees by a manager/computer. Payoff irrelevant for employees

7 of 44

Method

7

A laboratory experiment.

Responsibility allocated unequally between two employees by a manager/computer. Payoff irrelevant for employees 🡪 rules out material reciprocity concerns playing a role.

8 of 44

Method

8

Employees then engage in a stated effort task for the manager

A laboratory experiment.

Responsibility allocated unequally between two employees by a manager/computer. Payoff irrelevant for employees 🡪 rules out material reciprocity concerns playing a role.

9 of 44

Method

9

Employees then engage in a stated effort task for the manager 🡪 rules out intrinsic motivation and beliefs about perceived skill playing a role.

A laboratory experiment.

Responsibility allocated unequally between two employees by a manager/computer. Payoff irrelevant for employees 🡪 rules out material reciprocity concerns playing a role.

10 of 44

Contribution to the literature

Social comparisons and effort provision: Bartling and Von Siemens (2011, JEP), Greiner et al. (2011, EL), Charness et al. (2016, EE), Clark et al. (2010, ILR Review), Gächter and Thöni (2010, JEBO), Nosenzo (2013, EI), Angelova et al. (2021, LE), Breza et al. (2018, QJE), Cohn et al. (2014, JEEA), Ockenfels et al. (2015, MS).

🡪 We study the effect of being more or less trusted while holding wages constant.

Distrustful intentions: Falk and Kosfeld (2006, AER); Belot and Schröder (2016, MS).

🡪We examine distrust stemming from comparisons with how much one’s peers are trusted.

Discrimination and labor supply: Gagnon et al. (2020, WP), Glover et al. (2017, QJE)

🡪 We quantify the reciprocal response of workers to being distrusted by an employer.

Subjective evaluations and reciprocity: Brandts et al. (2006, LE).

We quantify the effect of not being selected.

10

11 of 44

Experimental design

11

Part B:

Part A:

Part C:

Part D:

12 of 44

Q1. Have you often lied for personal benefit in the heat of the moment?

1 = yes, 0 = no

Q2. If you had to choose, which do you think is more important in a discussion?

1 = being honest, 0 = making people feel comfortable

Q3. How do you like to travel?

1 = plan ahead, 0 = spontaneously

Q4. Do you mind being the center of attention?

1 = yes, 0 = no

12

Part A:

Personality questions

13 of 44

13

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

sends x₁effort points

sends x₂effort points

E1

E2

M

Manager = 3x₁ + 3x₂

E_i = 150 + 0.5(40 – x_i)

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

14 of 44

14

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

15 of 44

15

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

16 of 44

16

E1

E2

M

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

17 of 44

17

sends x₁effort points

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

E1

E2

M

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

18 of 44

18

sends x₁effort points

sends x₂effort points

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

E1

E2

M

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

19 of 44

19

Manager = 5x₁ + 1x₂

E_i = 150 + 0.5(40 – x_i)

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

sends x₁effort points

sends x₂effort points

E1

E2

M

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

20 of 44

20

Random treatment: Computer randomly determines employees’ importance

Deliberate treatment: Manager decides employees’ importance

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

Task 2: Treatment manipulation

Part C:

sends x₁effort points

sends x₂effort points

E1

E2

M

Manager = 5x₁ + 1x₂

E_i = 150 + 0.5(40 – x_i)

Employees have a fixed wage of ECU 150 (EUR 6) and are endowed with 40 effort points.

21 of 44

21

Task 2: Treatment manipulation

Part C:

Post Experimental Questionnaire

Part D:

Part A:

Personality questions

Task 1: Baseline measure of effort provision

Part B:

beliefs, self-esteem, emotions, demographic info.

22 of 44

Procedures

452 participants recruited in Nov-Dec 2020 via hroot (Bock et al. 2014) from the subject pool of MUEEL.

Experiment was implemented online in z-Tree unleashed (Duch, Grossman and Lauer 2020).

Average payment of CZK 151 (~EUR 6) for 35 minutes.

Study design, hypotheses and analyses pre-registered on AsPredicted.org.

22

23 of 44

Summary of set-up & hypotheses

23

Random

(N=148)

Low Importance Employees (74)

High Importance Employees (77)

Deliberate

(N=153)

High Importance Employees (74)

Low Importance Employees (76)

24 of 44

Summary of set-up & hypotheses

24

Random

(N=148)

Hypothesis 1 (main):

Low importance Deliberate < Low importance Random

Hypothesis 2 (exploratory):

High importance Deliberate > High importance Random

Low Importance Employees (74)

High Importance Employees (77)

Deliberate

(N=153)

High Importance Employees (74)

Low Importance Employees (76)

25 of 44

Summary of set-up & hypotheses

25

Random

(N=148)

Hypothesis 1 (main):

Low importance Deliberate < Low importance Random

Low Importance Employees (74)

High Importance Employees (77)

Deliberate

(N=153)

High Importance Employees (74)

Low Importance Employees (76)

Hypothesis 2 (exploratory):

High importance Deliberate > High importance Random

26 of 44

Main experimental results

26

27 of 44

Balance

27

28 of 44

Personality q’s uncorrelated with Task 1 effort

28

OLS models of effort sent in Task 1 on responses to personality questions

29 of 44

29

OLS regression for low importance employees

30 of 44

30

OLS regression for low importance employees

Main treatment effect

31 of 44

Guilt aversion

Employees’ beliefs about manager’s expectation in Task 2 by experimental condition

Low Importance

High Importance

d = -6.652***

d = 3.165*

Low Importance employees: beliefs significantly lower in Deliberate compared to Random condition.

We have correlation with work-hours sent to manager : Spearman rank correlation r_s = 418, p < 0.001.

Random

Deliberate

Random

Deliberate

32 of 44

Employees take managers’ decisions personally

32

Proportion who stated a given factor influenced managers’ decisions in Task 2

effort expected

At the end of the experiment managers in the deliberate treatment were asked what drove their decisions and employees in this treatment were asked what they believed drove the manager’s decisions – and the options were how similar the employee was to the manager, how much the manager liked each of them and how many hours they expected to send. So on the left side are what managers reported actually drove their decisions and on the right side are the employees’ beliefs about what influenced these decisions. So here you can already see there is a discrepancy in what managers report vs. what employees believe and the biggest difference is in the light blue bar. So over 60% of managers’ report that their decision was just based on how to maximize the effort points they receive from the employees. Now perhaps they use the questions to that end since that is, after all, the only information they have. But employees seem to think well this is impossible, there is no way for them to objectively assess how much effort they would provide and instead managers must have based their decision on more subjective criteria like how much they liked one of them or how similar they thought one of them was to themselves. And I think this is quite a common thing, so despite managers caring mainly about objective performance when they make these sorts of decisions, employees perceive their decision to have a more subjective flavor (so to be driven by some kind of favoritism).

33 of 44

Emotions experienced by low importance employees

33

Word	Frequency
Disappointed	20
Surprised	14
Good	8
Sad	5
Bad	5
Angry	3
Shocked	3
Betrayed	2
Underestimated	2
Appreciate	2
Upset	2
Honest	2
Mean	2
Hurt	2
Fault	2
TOTAL	73

Word	Frequency
Okay	7
Bad	5
Responsible	5
Confused	4
Worry	3
Good	3
Surprised	3
Nice	2
I-dont-know	2
Satisfied	2
Selfish	2
Weird	2
Valuable	2
Normal	2
Fair	2
TOTAL	50

Random

Deliberate

34 of 44

Emotions experienced by high importance employees

34

Word	Frequency
Pleased	16
Surprised	14
Trust	12
Good	10
Happy	5
Appreciated	3
Responsible	3
Glad	2
Bad	2
Motivated	2
Interesting	2
Ok	2
Proud	2
Liked	2
Satisfied	2
TOTAL	79

Word	Frequency
Good	7
Responsible	4
Pleased	4
Fair	3
Hope	3
Ok	3
Surprised	4
Bad	4
Disappointed	3
Enjoy	2
Guilty	2
Happy	2
Fine	2
Care	2
Pleasant	2
Hard	2
Slavery	2
Uncomfortable	2
Worthy	2
Inferior	2
Special	2
TOTAL	58

Deliberate

Random

35 of 44

Emotional experience of distrust

35

RAs’ emotion ratings by experimental condition

Low Importance

High Importance

Random

Deliberate

Random

Deliberate

36 of 44

Emotional experience of distrust

36

RAs’ emotion ratings by experimental condition

d = -0.825 ***

Low Importance

High Importance

Random

Deliberate

Random

Deliberate

37 of 44

Emotional experience of distrust

37

RAs’ emotion ratings by experimental condition

d = -0.825 ***

Low Importance

High Importance

Random

Deliberate

Random

Deliberate

d = 1.058***

38 of 44

Emotional experience of distrust

38

39 of 44

Emotional experience of distrust

39

40 of 44

40

Suggestive evidence that emotions mediate the main treatment effect.

DV: effort sent in Task 2

41 of 44

Conclusion

New evidence that workers care about how they are perceived:

Learning they are less trusted has a negative effect on effort provision.
Learning they are more trusted does not impact effort provision.
Suggestive evidence of lower guilt & negative emotions driving main treatment effect.

Managerial decisions do not need to be payoff relevant to have an effect on motivation:
importance of implementing fair task allocation procedures.

42 of 44

Thank you�ondrej.krcal@econ.muni.cz

42

43 of 44

A model of intention-based reciprocity

w: employee’s payoff

m: manager’s payoff

α: shape parameter

θ_t: reciprocity parameter

θ_Rand,high , θ_Rand,low , θ_Delib,high , θ_Delib,low

Cox, Friedman and Gjerstad, 2007: “A Tractable Model of Reciprocity”

other regarding utility is analogous to utility over commodities but the marginal rate of substitution depends on the emotional state of the decision maker – could be affected by relative payoff opportunities, reciprocity, positive or negative, relative status

The Cox and Sadiraj (2005) two-player model includes nonlinear indifference curves for other-regarding preferences. The utility function has the form X with the parameter restriction θ ∈ [0, 1). As in the Fehr-Schmidt model, I like own income and my marginal rate of substitution depends on whose income is higher but, in contrast to Fehr-Schmidt, I always like or am at least neutral with regard to your income and always have a decreasing marginal rate of substitution between my income and your income as my income increases.

The main alternatives so far to these distributional preference models are equilibrium models that try to capture the reciprocity motive in terms of beliefs regarding intentions.

Whether utility

increases in other’s payoff y depends on the sign of θ. Sometimes it is useful to decompose θ(r, s) as θ(r, s) = θ0 + ˜θ(r, s), where ˜θ(0, 0) is zero, and θ0 is the residual benevolence

or malevolence. Thus the typical individual will have standard selfish preferences when

θ(r, s) = 0, i.e., ˜θ(r, s) = −θ0.

44 of 44

A model of intention-based reciprocity

Step 1: Parameterizing the model for the Random treatment.

Step 2: Show the parameterized predicts effort in Task 1.

Step 3: Use the α from the parameterized model and estimate the θ_Delib,high , θ_Delib,low

Cox, Friedman and Gjerstad, 2007: “A Tractable Model of Reciprocity”

θ_Delib,low	=	0.32
θ_Delib,high	=	0.73

α	=	0.32
θ_Rand	=	0.62

other regarding utility is analogous to utility over commodities but the marginal rate of substitution depends on the emotional state of the decision maker – could be affected by relative payoff opportunities, reciprocity, positive or negative, relative status

The Cox and Sadiraj (2005) two-player model includes nonlinear indifference curves for other-regarding preferences. The utility function has the form X with the parameter restriction θ ∈ [0, 1). As in the Fehr-Schmidt model, I like own income and my marginal rate of substitution depends on whose income is higher but, in contrast to Fehr-Schmidt, I always like or am at least neutral with regard to your income and always have a decreasing marginal rate of substitution between my income and your income as my income increases.

The main alternatives so far to these distributional preference models are equilibrium models that try to capture the reciprocity motive in terms of beliefs regarding intentions.

Whether utility

increases in other’s payoff y depends on the sign of θ. Sometimes it is useful to decompose θ(r, s) as θ(r, s) = θ0 + ˜θ(r, s), where ˜θ(0, 0) is zero, and θ0 is the residual benevolence

or malevolence. Thus the typical individual will have standard selfish preferences when

θ(r, s) = 0, i.e., ˜θ(r, s) = −θ0.