An Experimental Study of Quality Signaling through Seed Money and Matching Gifts

Last registered on September 20, 2021

Pre-Trial

Trial Information

General Information

Title
An Experimental Study of Quality Signaling through Seed Money and Matching Gifts
RCT ID
AEARCTR-0008256
Initial registration date
September 19, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
September 20, 2021, 6:06 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Primary Investigator

Affiliation
University of Houston

Other Primary Investigator(s)

PI Affiliation
Texas A&M University
PI Affiliation
Texas A&M University

Additional Trial Information

Status
On going
Start date
2021-04-05
End date
2022-05-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
The nonprofit sector is known for significant presence of highly inefficient organizations. Apart from the well-publicized fraudulent nonprofits, such as the Cancer Fund of America, there are many organizations that may not blatantly scam donors, but are nevertheless doing a poor job of providing public benefits. Donors' lack of information is likely a contributing factor. While the presence of rating agencies can help donors, the staggering number of charitable organizations makes the available information imperfect and costly to obtain. Consequently, a big challenge for well-run nonprofits is finding ways to credibly inform donors of their quality and distinguish themselves from their poorly performing counterparts.

We aim to investigate experimentally the role that leadership giving plays in conveying information to donors. Leadership gifts can be in the form of an unconditional lump sum called “seed money” or a promise of matching small donations by a fixed ratio called “matching gift”. Our goal is to understand how the type and size of the leadership gift impact donors' beliefs about the quality of the public good and their willingness to donate.

This is based on our novel theoretical model of fundraising, which reveals that the fundraiser's choice of leadership scheme depends crucially on the lead donor's information. If the lead donor is fully informed about the quality, either exogenously or endogenously via costly information acquisition, the charity relies on the lead donor to reveal the quality to subsequent donors through the size of her donation. This eliminates the signaling concern of the high quality charity when choosing her optimal fundraising scheme. Consequently, consistent with the existing theoretical literature, we find that in absence of signaling considerations, the fundraisers should optimally choose the matching scheme, as it alleviates the free rider problem present in public goods provision. However, if the leader is not fully informed about the quality, the lead donor's gift becomes a less reliable source of information and the fundraiser may have to employ the fundraising scheme itself to inform donors. In this case, the model gives rise to a strong prediction regarding the use of the two schemes to signal quality information to donors. While, as any signaling game, multiplicity of equilibria may arise, we find that in any equilibrium with partial information acquisition by the lead donor, the use of seed money in equilibrium is associated with higher expected quality compared to matching gift. Intuitively, since seed money is less effective in alleviating the free-rider problem compared to matching, it becomes a costly signaling tool for the high quality fundraiser that aims to differentiate itself from the low quality fundraiser. As a result, in equilibrium the high quality fundraiser solicits for seed money more often and for matching gift less often than the low quality fundraiser. This, in turn, gives rise to an equilibrium belief among donors that seed money is indicative of higher quality.

We investigate the theoretical predictions regarding the use of seed money as a signal of high quality in controlled lab experiments. The lab is an ideal environment for this as it allows us to control the amount of information available to donors, which is hard to achieve in the field. Moreover, we are the first to endogenize the choice of leadership scheme in the lab to understand the information that each scheme carries, how donors form beliefs, and how they respond to different schemes. In particular, our experimental design consists of a piece-wise linear public good game, in which the marginal per-capita return (MPCR) varies with the total amount raised and the quality of the public good. This ensures sufficient variation in the equilibrium contribution amounts to make both channels of signaling, namely the scheme choice and the lead donor's gift, possible. Since the ability of matching to alleviate the free-rider problem, and thus raise more contributions relative to seed money, is a key driving factor for seed money to emerge as a costly signal of quality in the incomplete information setting, our investigation consists of four treatments: a control setting, focusing on a complete information, and three main treatments with incomplete information. The control treatment compares the two schemes and studies the endogenous choice of scheme when the quality is known to all players. The three main treatments extend the framework to incomplete information and vary the probability of verification by the lead donor in order to establish how the choice of scheme and the lead donor's contribution choice interact to convey information to the downstream donor. This will allow us to measure the signaling value of each scheme.
External Link(s)

Registration Citation

Citation
Krasteva, Silvana, Marco Palma and Piruz Saboury. 2021. "An Experimental Study of Quality Signaling through Seed Money and Matching Gifts." AEA RCT Registry. September 20. https://doi.org/10.1257/rct.8256
Experimental Details

Interventions

Intervention(s)
Intervention Start Date
2021-04-05
Intervention End Date
2022-05-31

Primary Outcomes

Primary Outcomes (end points)
(1) Scheme choice (lump sum or matching) by Player 1
(2) Contributions decisions by Player 2 and Player 3
Primary Outcomes (explanation)
Please refer to the experiment design

Secondary Outcomes

Secondary Outcomes (end points)
(1) Beliefs by PLAYER 1 and PLAYER 2 regarding the decisions made by downstream players
(2) Beliefs by PLAYER 2 and PLAYER 3 regarding the quality of the public project
(3) Eye tracking and facial expression data during the experiment
(4) 5 page survey about demographics, CRT and math skills, family background, charity activity, comments about the experiment
Secondary Outcomes (explanation)
Please refer to the experiment design

Experimental Design

Experimental Design
In this experiment, we recruit subjects to participate in a lab experiment conducted in Human Behavior Lab at Texas A&M University. We implement a 4x1 between subjects design, in which subjects are randomly assigned to one of 4 treatments: Control, Full Information, No Information, and Partial Information, corresponding to no quality variation, fully informed lead donor, uninformed lead donor, and partially informed lead donor, respectively.

Within each treatment, donors will be randomly assigned to 3 possible roles of PLAYER 1, PLAYER 2, and PLAYER 3, corresponding to the fundraiser, the lead donor, and the follower donor, respectively. In all treatments, the fundraiser (PLAYER 1) has the first move. She is given the choice between lump sum (seed money) and matching gift in order to establish the fundraisers' preference between the two fundraising schemes. Then, the lead donor (PLAYER 2) and the follower donor (Player 3) sequentially contribute to an induced-value public project. Under lump sum, each donor (PLAYER 2 and PLAYER 3) makes lump sum contributions. Under matching, PLAYER 3's choice set remains the same, but PLAYER 2 commits to matching PLAYER 3's contribution by a fixed percentage.

Under Control, the quality of the public project is known to all players. In each of the main treatments, Player 1 is privately informed about the quality of the public project, which is equally likely to be high or low (random realization). PLAYER 2 is informed with probability 1 in the Full Information treatment, 0 in the No Information treatment, and 0.5 in the Partial Information treatment. PLAYER 3 is uninformed in all main treatments.

Using this experimental design we test the following 4 main hypotheses:
1. Under all treatments, average contributions under seed money, are lower than average contributions under matching, when PLAYER 2 observes the quality of the public project.
2. PLAYER 1 will choose matching more frequently than seed money, under Control, Full Information and No Information.
3. Under Partial Information, average contributions under seed money, are higher than average contributions under matching, when PLAYER 2 does not observe the quality of the public project.
4. PLAYER 1 will choose seed money under Partial Information, more frequently than under Full Information and No Information.
Experimental Design Details
The game between lead donor (PLAYER 2) and follower donor (PLAYER 3) proceeds in a similar fashion across the 4 treatments. PLAYER 2 and PLAYER 3 are each given an initial endowment of 40 tokens, which they allocate sequentially between private consumption and a public project. In order to simplify donors' contribution decision, under lump sum, each of PLAYER 2 and PLAYER 3 is allowed to a make lump sum contribution in multiples of ten up to their endowment of 40 tokens, resulting in 5 possible strategies for each player in {0,10,20,30,40}. Under matching, PLAYER 3's choice set remains the same as lump sum, but PLAYER 2 makes a commitment to a matching percentage in {0%,25%,50%,75%,100%} of PLAYER 3's contribution. In order to keep the set of possible contribution amounts the same across the two schemes, the resulting matching contribution by PLAYER 2 will be rounded down to the nearest multiple of 10.

Under the Control, the return form the public project is commonly known to have an MPCR of 0.7 for contributions up to 40 and 0.1 for higher contributions. In the main treatment, there will be two types of public project, low and high quality, that are publicly known to be equally likely. The MPCR for the low quality project is the same as in the Control, and that of the high quality project is 1.2 for contributions up to 60 and 0.6 for higher contributions. We provide payoff tables to the subjects during the experiment, since computing payoffs in the matching scheme entails an additional step of calculating the contribution amount of PLAYER 2 from the match ratio and we prefer to avoid this complexity impacting the choice of scheme and the contribution decisions.

The above mentioned game is played for 10 incentivized rounds with random-matching. Additionally, before the 10 rounds, players will play 10 other incentivized rounds with random matching, where the game is exactly the same as the following 10 rounds, except for the fact that the fundraising scheme is exogenously set by the experimenter to lump sum or matching gift (5 rounds of each with random order). During these first 10 rounds, PLAYER 1 is a non-strategic player and simply collects a payoff from the contributions of the other two players. This ensures that PLAYER 1 understands how the game is played by downstream players, before making her own strategic decisions. We will also be sure to collect enough balanced data on both fundraising schemes.

In each of the 20 incentivized rounds, as well as 3 practice rounds, subjects are randomly and anonymously re-matched into groups of 3 with one of each role. However, a subject’s role remains the same throughout the session. One of the incentivized rounds will be randomly chosen for monetary payment based on the choices made in that round.

In each round, we collect data on PLAYER 1's scheme choice, and PLAYER 2 and PLAYER 3's contribution decisions. We also elicit beliefs by PLAYER 1 and PLAYER 2 regarding the decisions made by downstream players, and beliefs held by PLAYER 2 and PLAYER 3 regarding the quality of the public project. We incentivize them to make a correct guess by awarding 4 bonus tokens for correct guesses. We also collect behavioral responses to complement the experimental economic data. Eye tracking data will be recorded using Tobii Spectrum eye-tracking devices to reveal visual attention to information presented and how the format and complexity of the experiment affect the respondent's choices. The information collected includes the time to first fixation or how long it took participants to look at an area of interest for the first time; fixation duration or how long they looked at each area; and fixation count or how many times they look at an area. The eye tracking data will help us understand the process and strategic considerations used by participants to make their choices. We will also monitor participant's emotions using facial expression analysis to detect overall positive, negative, or neutral emotional responses. All the data is synchronized and recorded simultaneously to obtain a complete behavioral picture of participants.
Randomization Method
Subjects will be assigned to treatments based on session signup (they have no prior knowledge of the treatment that is pre-assigned to each session). Role in the game and group matching is done in the lab randomly by computer program coded in z-Tree.
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
There is no clustering.
Sample size: planned number of observations
(1) Scheme choice by Player 1: 2000 rounds of game played by 600 subjects randomly re-matched into groups of 3 for each round (2) Contributions decisions by Player 2 and Player 3: 4000 rounds of game played by 600 subjects randomly re-matched into groups of 3 for each round
Sample size (or number of clusters) by treatment arms
600 subjects recruited from Texas A&M University student population
This number may vary based on available funding.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
300 subjects
IRB

Institutional Review Boards (IRBs)

IRB Name
Texas A&M Institutional Review Board
IRB Approval Date
2020-03-09
IRB Approval Number
IRB2018-1438M

Post-Trial

Post Trial Information

Study Withdrawal

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials