Fraudulent Behavior by Entrepreneurs and Borrowers
Christa Hainz
5.1 Introduction
The crowdfunding market is a child of the digital revolution and, although
still in its infancy, it is growing rapidly. Prosper.com, one of the first
crowdfunding platforms to engage in peer-to-peer lending, was founded
in 2006. Just like new products new markets have to demonstrate that
they satisfy needs that would otherwise be unmet. The need addressed by
crowdfunding platforms is to bring supply and demand of capital
together.1 As funding decisions involve significant risks, the platforms
need to build up the reputation that transactions take place in a fair and
trustworthy manner. Otherwise investors are not willing to invest.
To build up this reputation it is important to limit fraudulent behav-
ior. From other financial markets we know that fraud has severely nega-
tive repercussions on the market. There is evidence from the United
States that households in states that are also home to firms involved in
corporate fraud cases reduce their stock market participation (Giannetti
and Wang 2016). In Germany the so-called Neuer Markt (a stock market
for small- and medium-size innovative growth firms) was dissolved in
2003 only a few years after its launch in 1997. One of the main reasons
was that some major corporate scandals, such as misstatement of turn-
over and insider trade, eroded its reputation (Burghof and Hunger 2004).
A similar effect could occur in the crowdfunding market in case of fraud.
As the market is still very young, the negative effects of fraud cases might
be very strong and potentially unfold a destructive power. Fraud by plat-
forms will exert similar negative externalities.
In this chapter we investigate fraud by borrowers and entrepreneurs.
Fraud has many different faces. We use the definition provided by
Cummings et al. (2016, 4) for reward-based crowdfunding and formu-
late it for crowdfunding in general. The investor must verify that the fol-
lowing five different elements are present in order to prove fraud on the
part of a firm: (1) the firm must have made a false statement related to a
material fact, (2) the firm must have known that the statement was
untrue, (3) it must have been the firm’s intention to deceive the investor,
(4) the investor must have reasonably relied on the statements of the firm,
and (5) the investor must have been injured, which is most likely the case
if funds are lost.2
We will begin this chapter by taking an economic perspective on
fraudulent behavior. We use the sketch of a model with asymmetric
information to highlight the role of uncertainty and discuss mechanisms
to reduce the underlying incentives problem. We then review the existing
evidence on potentially fraudulent behavior in the three different crowd-
funding markets and highlight their limitations. We subsequently discuss
those factors that influence the detection of fraud and conclude by offer-
ing some policy implications.
5.2 Asymmetric Information and Fraud
The behavior of agents and the relationship between agents and their
principals is studied in contract theory. The idea underlying the models
in contract theory is that the agents, or in the case of financial services the
firms represented by their managers, have better information than their
principals, the financiers; and that the former use the information asym-
metry for their own benefit. The contract theoretical models deal with
fraudulent behavior without calling it fraud. As the definition of fraud
has shown, the challenge is to demonstrate that an agent’s behavior is
fraudulent and that the agent took his actions by intent. Contracts can
specify variables that can be observed and verified. However, the agent’s
behavior cannot be stipulated in a contract because it cannot be observed
and verified. At the point in time at which the contract is written there is
uncertainty about the outcome of a project. In the context of finance this
means that the capital that is invested in a project does not generate a
return with certainty, but that there is a distribution of returns. The ulti-
mate return is, in contrast to the agent’s behavior, observable and verifi-
able and therefore can be the subject matter of a contract. We will discuss
two problems of fraud depending on the point in time when it takes
place; the agent can deceive the principal before or after the contract is
concluded.
5.2.1 Adverse Selection
At the point in time before the principal and the agent enter into a con-
tractual relationship the principal cannot observe the agent’s type, that is
whether the agent has a high- or low-risk production technology. The
agent’s type will influence the distribution of the returns and ultimately
returns are observable.
An example of adverse selection from crowdfunding is Kobe beef jerky.
In a Kickstarter campaign Magnus Fun Inc. offered Kobe beef jerky
shortly after the import of Kobe beef to the United States was allowed,
but still heavily regulated. The original goal was to raise USD 2,374. In
fact more than 3,000 backers offered over USD 120,000. A team of filmmakers detected inconsistencies in the figures of Magnus Fun Inc.
and the campaign stopped briefly before it would have been completed
and the money of the principals could have been lost.3
5.2.2 Moral Hazard
The second problem of asymmetric information arises after the contract
is concluded because the agent cannot commit to a certain behavior, such
as investing the money as promised or exerting effort in managing the
project. This problem can be referred to as moral hazard.4 By exerting
effort the agent increases the probability that the project generates a high
return, enabling the agent to make payments to its principal. Similarly
the agent can divert the funds instead of investing them appropriately,
meaning that the project stands a relatively low chance of proving
successful.
There are two fraud cases from crowdfunding that can serve as exam-
ples for moral hazard. Jen Hintz raised USD 26,000 on Kickstarter for
FibroFibers, an indie yarn-dyeing business. In reality she did not spend
the money on her business, but instead used it to finance her move from
North Carolina to Massachusetts. Another example comes from
GoFundMe. A mother raised money for paying the cancer treatments for
her daughter. The daughter, however, was healthy and the money was
spent otherwise (Fredman 2015).
We want to use the following simple model to illustrate the moral
hazard model for the crowdlending market. Therefore the contracting
parties are called borrower and lender. We study credit contracts in which
borrowers first receive credit and then decide on where to invest the
money. If the borrower invests the money in the proposed project the
probability of success, that is of being able to repay the loan, is pH. If he
does not invest the money as proposed, but uses it for his own purposes,
he will get a private benefit b with certainty, but the project will never
succeed. The borrower has a return of X in the case of success and zero in
the case of failure; returns are assumed to be verifiable. Furthermore, we
assume that investment I is efficient from a social welfare perspective only
if the borrower decides to invest the money instead of taking the private
benefit, that is pH X − I > b. However, the choice of the borrower is not
observable and causes a moral hazard problem. We assume that the bor-
rower possesses assets totaling the amount of A that can be liquidated by
the lender in the case of failure. Thus, the borrower’s liability is limited to
A (<I). The payoffs are depicted in Fig. 5.1. It is worth noting that in the
case of investing as proposed, the payoff might be 0 whereas it is certainly
0 in the case of fraud. Thus, for the investor it is impossible to distinguish
between fraudulent and non-fraudulent behavior in this case because the
agent’s investment decision is not observable. But the lender gets an
imperfect signal as to the borrower’s behavior. Therefore, the contract
terms are the means of solving the moral hazard problem; they must be
set such that they give the borrower an incentive to behave
non-fraudulently.
The principals offer a contract {R; A} to the borrower, in which R is the
repayment in the successful state and A is the liability in case of default.
Although crowdfunding contracts do not specify collateral, borrowers are
liable with all their assets in case of default and A measures the borrower’s
liability.5 In order to solve the moral hazard problem, the credit contract
must satisfy an incentive compatibility constraint (1), which states that
the net payoff for the borrower must be higher when investing in the
project than when taking the money and spending it on for its own pri-
vate benefit. When investing the money the borrower will be successful
with probability pH, generating a return of X and repaying R to the lender.
If the project fails, the borrower will lose all of his assets totaling the
amount of A. When the borrower spends the money for its own benefit,
he gains a private benefit of b, but will certainly lose its assets A
This equation helps us to understand the problems that may arise
because a project is credit financed. As we assumed that pH X − I > b,
nobody would undertake a fraudulent project with its own means.
However, if it is possible to find a lender that provides a loan, the bor-
rower does not have to bear all the costs of his (non-)investment and
therefore may have an incentive to take the money from the lender and
spend it on its own purposes, getting a private benefit of b. Equation (2)
states the condition a credit contract has to fulfill so that the borrower
will opt for the investment. Comparative statistics provide interesting
insights. The higher the private benefits from diverting the funds, the
higher the incentive to opt for diverting the funds. On the other hand,
the more profitable the investment project, that is the higher the proba-
bility of success pH and the return in case of success X, the lower the
incentive to divert the funds. Most importantly, the terms of the credit
contract influence the borrower’s incentives. The higher the repayment R
and the lower the liability A, the more attractive it is for the borrower to
divert the funds. The lower the liability of the borrower, the more diffi-
cult it will be to write an incentive-compatible contract. Here it is impor-
tant to bear in mind that the lower the difference between (R−A), the
higher the incentive not to divert the funds and invest them as
proposed.
Ultimately there are two ways to address problems of asymmetric
information. The first way is to reduce the information asymmetry. The
second way is to write a contract that gives the borrower an incentive not
to exploit its information advantage. The simple model above has shown
that in the case of moral hazard the difference between the repayment in
the case of success and failure, that is (R−A), should be low.
In the banking context, reducing information asymmetry after the
contract is signed is reached by monitoring the borrower. To this end the
borrower has to document the development of his business regularly by
showing balance sheet and other data to the loan officer. However, moni-
toring imposes a fixed cost on the bank, making it unattractive for small
loan sizes. For microcredit new contractual forms have emerged as a
result. The first microfinance bank, the Grameen bank in Bangladesh,
initially only granted microloans to groups of borrowers with joint liabil-
ity. The idea was to exploit the knowledge that individual borrowers have
about their peers. Thus borrowers would exert pressure on their peers to
repay the loan because otherwise the well-performing borrowers would
have to repay for their defaulting peers.
The microfinance loans have another important feature to improve
incentives. Borrowers can build up a credit history. A good credit history
gives them access to future loans and the size of those loans increases over
time. This means that default leads to a loss of reputation. In our simple
model above this could be captured as a higher liability whereby borrow-
ers do not lose physical assets, but their reputation. The same mecha-
nisms exist when a borrower and a bank have a longer-term relationship
and when information-sharing devices exist in a credit market.
We have just discussed the mechanisms that could solve the moral
hazard problem. Similar mechanisms exist for adverse selection. For
crowdfunding to be successful it must develop ways to solve the problems
created by asymmetric information, as otherwise it will attract fraudulent
projects that are not financed by financial intermediaries that have mech-
anisms in place that solve these problems. An adverse selection problem
therefore exists between different lenders, that is, between platforms and
more generally between the more traditional financial market and the
crowdfunding platforms. In the end there are several adverse selection
problems, one between the lender and the borrower and another one
between different lenders.
5.3 Empirical Evidence on Fraud
No systematic evidence on cases of fraud in crowdfunding has been col-
lected to date. We will provide some evidence on (what we will call)
performance problems in the three different areas of crowdfunding, such
as non-deliveries and defaults.
5.3.1 Reward-Based Crowdfunding
Reward-based crowdfunding differs in several aspects from crowdlending
and crowdinvesting. Firstly, it does not necessarily give a monetary payoff
to the backers, but does provide them some other reward, such as the
product or a giveaway which, for example, may be a documentary of how
a product is made or a project t-shirt. Perhaps as a result it is often not
perceived as an investment by the backers. Legal scholars argue that the
backers’ motivation to provide money is not to finance the development
of the product, but rather to buy rewards or goods. This argument is
illustrated by the Pebble Smartwatch project in which most backers (96
percent of the 68,929) pledged at least USD 99 which was the threshold
above which one obtained the product. If backers wanted to see the
Pebble Smartwatch to be developed, the fraction of contributions below
the threshold should have been (much) higher. From a legal point of view
the parties enter a contract for the design and manufacture of a specific
good. But the important difference to other contracts for purchasing
products on- or off-line is that the goods in reward-based crowdfunding
have not been produced at the point in time the contract is concluded
(Moores 2015).6 This means that there is more uncertainty involved
when purchasing a good via reward-based crowdfunding, which might
often not be fully acknowledged by the parties of the contract, and par-
ticularly by the buyer.
Mollick (2014) studies data on performance problems on Kickstarter.
He uses data on Kickstarter projects from its start in 2009 until July
2012. During that period over 23,000 projects were successfully funded
on the platform (which equals a 48.1 percent share of all proposals sub-
mitted). To see how the projects perform over time the author analyzes
the final outcome of the 471 projects in the categories of Design and
Technology, which had specified delivery dates before July 2012. Among
these 471 projects 381 had outcomes that were clearly identifiable.
Within this group there were 14 projects that failed (or 3.6 percent)
either issuing a refund (3 projects) or stopping to respond to backers
(11). However, among the better-performing projects delivery on time is
not the rule, as only 24.9 percent of the projects were not delayed.
Another 33 percent did not deliver as promised until the end of the sam-
ple period. The projects with a delay (126 projects or 33 percent) deliv-
ered on average 2.4 months later.7
These figures provide some evidence on the performance problems in
reward-based crowdfunding. However, the reasons underlying these
problems can be manifold and range from intentionally deceiving inves-
tors to slipping into such deception, or even a mixture of both.8 Fraud is
only one possible explanation. If a project grants a refund, technical
problems are more likely to explain non-delivery than fraud. If an indi-
vidual stops responding, it could well be that “he ran away with the
money.” This happens in 2.9 percent of the cases in the sample. It is
important to bear in mind that these projects operate under greater
uncertainty than traditional sales, as products have not been produced at
the point in time when they are sold. As a result, Mollick (2014) finds
that delays are more likely if products are promised as compared to giveaways.
The other factors increasing the risk of a delay are the size of
the project and the degree of overfunding. These findings may provide
some indication that performance problems increase with the complexity
of the project as the latter results in uncertain outcomes.
Alternatively, one could look at fraud directly. The challenge here is
that fraud—in contrast to delivery—is not readily observable. Cumming
et al. (2016) search for fraud cases for projects on the two most popular
platforms (Kickstarter and Indiegogo) in nine countries during the period
2010–2015. They not only collect data from the websites of the two plat-
forms but complement it by searching for fraud cases themselves. They
find only 207 fraud cases (which corresponds to a rate of 0.01 percent).
The figures on fraud cases (0.01 percent) and non-deliveries (about 3
percent when deducting the refunds from the non-deliveries) could act as
lower and upper bounds for fraudulent behavior in reward-based crowd-
funding. As fraud is not readily observable, the fraud cases that this figure
is based on are only the tip of the iceberg (we will discuss the detection of
fraud cases below). By contrast, non-deliveries will exaggerate fraudulent
behavior because in an uncertain world non-fraudulent projects also fail.
