What Makes Pyramid Schemes Work? A Followup

Last registered on February 15, 2019

Pre-Trial

Trial Information

General Information

Title
What Makes Pyramid Schemes Work? A Followup
RCT ID
AEARCTR-0003880
Initial registration date
February 15, 2019

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
February 15, 2019, 4:14 PM EST

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

Locations

Region

Primary Investigator

Affiliation
University of Cologne

Other Primary Investigator(s)

PI Affiliation
University of Queensland
PI Affiliation
University of Queensland

Additional Trial Information

Status
In development
Start date
2019-02-15
End date
2020-02-25
Secondary IDs
Abstract
Via an online experiment, we investigate why people invest in pyramid schemes. We focus on cognitive biases that could explain entry decisions.
External Link(s)

Registration Citation

Citation
Dogan, Gonul, Kenan Kalaycı and Priscilla Man. 2019. "What Makes Pyramid Schemes Work? A Followup." AEA RCT Registry. February 15. https://doi.org/10.1257/rct.3880-1.0
Former Citation
Dogan, Gonul, Kenan Kalaycı and Priscilla Man. 2019. "What Makes Pyramid Schemes Work? A Followup." AEA RCT Registry. February 15. https://www.socialscienceregistry.org/trials/3880/history/41622
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
Our study consists of three treatments. We will first run a baseline treatment to get enough subjects to use in the matching procude of the complexity treatment. This baseline treatment will have a small number of subjects. In the first complexity treatment, we will reduce the complexity of decisionmaking in a large group by making subjects consider how they would behave in a smaller group. Further, we will study witihin subject consistency of investment behavior in a pyramid scheme to an analogous lottery. In the second complexity treatment, we will reduce complexity by making subjects realise the payoffs in a large group in which everyone invests. We will do this by asking subjects to consider what their earnings would be in different levels of the tree.
Intervention (Hidden)
In our earlier experiment on the pyramid scheme, AEARCTR-0003057, we found that subjects who expect more investors are significantly more likely to invest. This relationship suggests that subjects mistakenly expect to earn more money if more subjects invest. In the pyramid scheme, if everyone invests, half of the subjects lose all their money, and at most a quarter can have positive investment returns. We conjecture that, subjects mistakenly expect higher earnings with a larger percentage of investors, and this is due to the complexity of the decision. To test the complexity mechanism, we introduce two treatments that reduce complexity.

First, we will run a baseline treatment with 20 subjects as in our initial baseline. We will use 7 investors in the baseline for later treatments. Two complexity treatments test whether subjects' bias arises because of the group size (payoffs in large groups are hard to intuit than in small groups), or because of the lack of realisation that most subjects do nor earn money in the experiment.

In the first complexity treatment, there are two investment decisions, and one of those will be randomly chosen for payment. in the first part, using the strategy method, we will ask subjects their investment decision in a tree with 7 other investors where we will vary their position in the tree from 1 to 8. One of these decisions will be radomly implemented for part 1. In the second part, we will run the standard investment decision with 200 subjects. of the that tests whether subjects understand the payoffs if everyone invests in a small group of subjects. to get enough subjects to use in the matching procude of the complexity treatment.

In the second complexity treatment, after the instructions, we will show the subjects a tree in which 200 people invest. Subjects would then answer questions regarding what they earn if they are in the bottom branch of the tree, second bottom, and the third bottom. These branches will be highlighted to make subjects realise how many other investors are in the same position. After they answer these questions correctly as well as our standard comprehension questions, they will proceed to the main experiment in which 200 subjects decide.

In both treatments, after the investment decision, they will estimate the number of investors, will state what they think the experiment is about, and then there will be a dictator decision, and two lotteries. We will randomly choose one of these lotteries, and pay for one randomly chosen decision. The first lottery is a Holt Laury lottery that we have used in the earlier set of experiments. The second lottery is the equivalent to the lottery induced by the investment decisions with 20, 40, 60,...,200 people investing plus their estimated number of investors investing. To generate each lottery, we generated 10000 draws of trees, calculated the payoffs of subjects, and then determined the probability of each payoff outcome. Since the outcome space is very large, we reduced it to probabilities associated with earnings 0, 2, 4, 6, 8, 10, 12, 16, 20, 24, 28 $ versus a sure amount of $4. This reduction is done for each possible amount,probability pair in such a way that the nearest possible amounts have added probabilities and the expected payoffs are the same.

Subjects then solve the race game five times in order to check their klevel reasoning. In the final questionnaire, we ask two financial literacy questions that are standard in surveys (OECD, 2018). We also ask them their gender, income, education level, number of years of education, whether they buy lotteries, whether they buy warranties, and whether they trust others.
Intervention Start Date
2019-02-15
Intervention End Date
2019-02-26

Primary Outcomes

Primary Outcomes (end points)
Percentage of investors
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
investment decisions in the 8 person game, the number of attempts for correct answers in the second complexity treatment, their beliefs about the number of investors, the amount of giving in the dictator decision, the switching point in HoltLaury lottery, the switching point in the investment induced lottery, the lottery decision based on one's beliefs on the number of investors, whether a subject won the race game , the number of times the subject has won the race game, financial literacy, gender, years of schooling, trust and warranty decisions.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Subjects are endowed with $4 and have to decide whether to invest or keep their endowment. 200 participants make this decision, and this is common knowledge. Subjects' decisions are implemented based on a randomly drawn tree, and the payoffs realise.
Experimental Design Details
For each participant, we will draw a position in the randomly drawn tree. At the top of the tree, there are two persons, and each person is matched with two further people (followers). If a person decided to invest, and is part of an active branch of the tree, her earnings would be 2 times the number of first degree followers, 1 time the number of second degree followers, 0.5 times the number of third degree followers and so on. Participants get detailed instructions about this earnings structure and have to answer comprehension questions with example trees.
Randomization Method
MTurk randomisation device. On our end, we will randomly draw the tree via a computer program.
Randomization Unit
individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
420
Sample size: planned number of observations
420
Sample size (or number of clusters) by treatment arms
20 for the baseline, 200 for each complexity treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Based on our earlier experiment, about 58 percent invest in baseline. Based on Gpower software, with 80 percent power at 5 percent alpha, with 400 subjects in total (200 baseline, 200 complexity), we can detect an effect that gives at most 43.6 percent investors.
IRB

Institutional Review Boards (IRBs)

IRB Name
University of Queensland
IRB Approval Date
2016-02-18
IRB Approval Number
201600074

Post-Trial

Post Trial Information

Study Withdrawal

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

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