The wisdom of crowds in equity crowdfunding
Last registered on May 18, 2017

Pre-Trial

Trial Information
General Information
Title
The wisdom of crowds in equity crowdfunding
RCT ID
AEARCTR-0001817
Initial registration date
May 18, 2017
Last updated
May 18, 2017 10:45 AM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
Saïd Business School, University of Oxford
Other Primary Investigator(s)
PI Affiliation
HEC Paris, France
Additional Trial Information
Status
On going
Start date
2015-03-30
End date
2017-06-30
Secondary IDs
Abstract
A large and growing literature demonstrates the impact of early stage investments on start-up success, and the factors that affect the terms of financing. But, what factors drive the investment process of early-stage investors, that is, how do they choose which start-up to fund?
The ideal setting to establish causality would compare an investor's reaction to two identical firms that differ only in the characteristic of interest. Such a setting is not feasible using observational data, but here we approximate it with a randomized field experiment using the correspondence testing methodology pioneered in labour economics. The experiment takes place on Seedrs, a European online equity investment platform that matches start-ups and angel investors. We observe these start-up companies at the stage at which they approach investors to raise capital through Seedrs.
Seedrs present "campaigns" to investors on their web site featuring start-ups that are raising capital. On the primary window for each campaign, and on the top banner, the web site contains information on how much money the firm aims to raise, equity offered, current valuation, how much it has raised to date, and number of days of the campaign remaining. In addition, Seedrs provides several additional pieces of information below the top banner: the founding team, lead investors and amounts, company legal information, and the idea. Investors may choose to remain anonymous, or they may show their name and provide investment experience. Through clickable tabs further pages provide information about the start-up team (position, time commitment and ownership share), the market, updates, Q&A and all investors.
In the experiment, we randomly choose which of the categories of information which is first presented to a potential investor, and exploit the variation across angels' reaction within each start-up. We randomly display on the first page an investor with a leading amount as lead investor (or not), and either as anonymous or with their name displayed, to determine whether their identity matters, both for entrepreneurs and for unrelated investors. We infer which factors drive investors' decisions by measuring each investor's interest in the company by the "click-through" on any of the tabs to subsequent information. On the subsequent investor page all the above information is displayed for all subjects, but on this investor page we still randomly either permanently hide or display the investment pedigree of the investor to determine if this is an important feature for follow-on investments. We also record whether the order of information, and whether knowing the investment pedigree, matters for subsequent investments.
External Link(s)
Registration Citation
Citation
Vulkan, Nir and Thomas Åstebro. 2017. "The wisdom of crowds in equity crowdfunding." AEA RCT Registry. May 18. https://www.socialscienceregistry.org/trials/1817/history/17776
Experimental Details
Interventions
Intervention(s)
Psychological studies have shown that what people see first may have a large effect on subsequent actions people take. We also think there are subtle effects at work, where people "herd" based on very small signals, for example information displayed about other investors. For instance, in the current landing page, investors can quickly know the amount invested by the top 5 investors. Even though this information is only a partial signal about the value of a project, it is likely to have a strong impact on potential investors' behaviours.

To test this hypothesis, we propose to design a simple experiment where we will manipulate only one factor: the presence of the "Top Investors" panel on the landing page. We will run the experiment on all new campaigns launched during a period of 30 days. The manipulation will last the first 30 days of each campaign. Therefore the experiment will last a maximum of 60 days. We expect to get around 20,000 observations during this period.
Intervention Start Date
2016-04-29
Intervention End Date
2016-08-31
Outcomes
Outcomes (end points)
In order to check the impact of the manipulation on members' behaviours, we will measure several parameters: - the number of tabs the member clicks on - the time spent by the member on a campaign page - if the member clicks on the company's website link - if the member shares the project through any social media - if the member adds the campaign to his/her favourite bookmark - the amount invested by the member (if any) on the project
Outcomes (explanation)
None of the outcomes will be constructed from the main variables in this experiment.
Experimental Design
Experimental Design
For each campaign, members will be randomly assigned to either the treatment or the control group.

- Members allocated to the treatment group will see a landing page without the "Top Investors" panel.
- Members allocated to the control group will see a landing page with the "Top Investors" panel on the left.

All the other elements of the landing page will be exactly identical for members of the treatment group and of the control group. In order to strengthen the effect of the manipulation, we suggest making more salient the information about "Top Investors". To do so, we suggest changing, for the control group, the order between the "Team" and the "Top Investors" panels. This would allow the members to see the "To Investors" panel without having to scroll down the page. The "Top Investors" panel will be completely removed for the treatment group.
Experimental Design Details
Not available
Randomization Method
Randomization is achieved using Optimiser software.
Randomization Unit
Each registered Seedrs user is a unit.
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
40,904 members
Sample size: planned number of observations
40,904 members
Sample size (or number of clusters) by treatment arms
20,522 members control, 20,382 members treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
IRB Approval Date
IRB Approval Number