Gilles Chemla on Prediction Markets, Crowdfunding, & The Wisdom Of Crowds

This is the first of a two part interview with Professor Chemla.

Prediction Markets and Information Efficiency

Aiden Singh: In finance, we discuss and debate whether the stock market is efficient - meaning do they accurate price in available information. Do you believe that concept applies to prediction markets? And if so, do you believe that prediction markets are efficient?

Gilles Chemla: Prediction markets are indeed fascinating and they differ from traditional financial markets in several respects.

Their appeal lies in their simplicity and transparency. The core idea behind prediction markets is that they provide a relatively straightforward mechanism to extract information from prices. Unlike complex derivatives or structured financial products, prediction markets typically consist of simple securities that pay one unit if a specific event occurs, and zero otherwise. This simplicity is what makes them powerful tools for aggregating information and forecasting outcomes such as election results.

Regarding efficiency, the performance of prediction markets depends on several factors.

In theory, if participants are not financially or informationally constrained, these markets should reflect true probabilities accurately. For example, if a candidate or other actor attempts to manipulate the market, participants who are indifferent to the outcome but motivated by profit can trade against such manipulation. Provided there are enough participants with the ability to arbitrage, the market has a self-correcting mechanism.

In practice, however, there are limitations. Prediction markets are often small, raising questions about whether sufficient arbitrage activity exists to ensure accurate pricing. Unlike large financial markets, where institutional investors provide liquidity and oversight, smaller prediction markets may struggle to maintain the same level of informational efficiency. Moreover, like any forecasting tool, prediction markets are not perfect. Polls have well-known limitations, and while prediction markets are useful complements, they too are subject to noise, biases, and structural constraints.

- - - - - - -

Crowdfunding, Information Cascades, and the Wisdom of Crowds

Aiden Singh: Related to prediction markets, and the question of whether they efficiently aggregate the wisdom of crowds to make accurate predictions about the future, is the topic of crowdfunding. You’ve conducted research into the crowdfunding of new upstart business ventures and whether that process successfully harnesses the wisdom of crowds. What did you find?

Gilles Chemla: Crowdfunding is a multi-faceted phenomenon, and its effectiveness depends on the type of platform and the underlying incentives.

In the United States, for example, the sale of securities online was prohibited until relatively recently.

The first widely adopted crowdfunding platforms were reward-based, allowing consumers to pre-order products such as prototypes of watches before they were commercially available.

Reward-based crowdfunding serves a critical function: it enables entrepreneurs to test market demand. Entrepreneurs often exhibit overconfidence, assuming that their products will appeal broadly. Crowdfunding provides an objective measure of consumer interest. By analyzing pre-orders or contributions, entrepreneurs can estimate potential demand and assess whether production is financially viable. In this sense, reward-based platforms efficiently harness the wisdom of crowds, funding projects that demonstrate genuine market appeal while filtering out those that do not.

Equity-based or lending-based crowdfunding operates differently. Here, investment decisions are influenced not only by individual preferences but also by perceptions of how others evaluate the opportunity. If a project attracts few early investors, it may struggle to secure further funding, resulting in undercapitalization. Conversely, initial success can trigger an informational cascade, attracting additional investment. These dynamics can be strategically manipulated, for instance when entrepreneurs seed campaigns through friends or family.

Aiden Singh: So, crowdfunding mimics traditional financial markets in some respects, including that investors are trying to anticipate the future behavior of other investors. In what ways do they differ?

Gilles Chemla: Traditional financial markets were not easy to establish. Their origins can be traced to early enterprises such as the East India Company and to the financing needs generated by the Industrial Revolution. In the early stages of market development, limited liability was often viewed as a deterrent to investment. At the same time, banks were too small to finance large-scale projects, and many individuals kept their wealth outside the financial system.

The development of stock markets made it possible to pool capital on a much larger scale. Over time, financial markets evolved into the structured exchanges we recognize today. Major exchanges such as the New York Stock Exchange, Nasdaq, and the London Stock Exchange are now subject to extensive regulation, particularly with respect to mandatory disclosure. Firms with long operating histories are required to provide detailed information to investors, while professional analysts and traders process that information and help ensure that prices reflect available knowledge as accurately as possible.

Crowdfunding, by contrast, is far less structured. Projects typically present only a prototype, and entrepreneurs often have no established track record. The success of a project depends largely on the crowdfunding campaign itself and on the reactions of potential backers.

Aiden Singh: How good of a job to these crowdfunding initatives do in allocating capital? What’s the success rate?

Gilles Chemla: Despite their apparent lack of information and professional oversight, crowdfunding can be remarkably effective. For example, on Kickstarter, 97 percent of successfully funded projects deliver the promised rewards. That outcome is striking, given that contributors are often non-professional investors evaluating a single prototype.

These results suggest that even in relatively unstructured markets, collective decision-making can function effectively under the right conditions.

Aiden Singh:  So 97% of the time, when the crowd funds a project on Kickstarter, the prototype is ultimately delivered as a finished product?

Gilles Chemla: That is correct. There may occasionally be delays, but the overall delivery rate remains very high. Several factors help explain this outcome.

First, technological risk is often limited. The projects featured on Kickstarter are typically innovative consumer products, for which production feasibility is generally not highly uncertain. In most cases, the main question is not whether the product can be built, but whether there is sufficient demand for it.

Second, the platform is designed to limit opportunistic behavior. A potential concern in any funding environment is that an entrepreneur might raise money without delivering the promised product. Kickstarter addresses this through an all-or-nothing funding model: entrepreneurs specify a target, and if that target is not met, backers are fully refunded. This structure gives entrepreneurs a strong incentive to complete the project, since the payoff from successful delivery exceeds any gain from walking away with the funds.

Lending-based platforms operate somewhat differently. Although non-professional investors typically have limited information, the interest rates they set often track default risk better than traditional credit scores such as FICO. While they may underperform professional investors and formal econometric models, they still aggregate useful information.

Equity-based crowdfunding introduces another layer of signaling. These platforms often require meaningful self-funding, together with investment from friends and family, typically around 30 percent of the total amount sought. That early commitment signals confidence in the project and can encourage additional investors to participate. In this sense, both lending-based and equity-based crowdfunding embody elements of crowd wisdom, though they do so through different mechanisms and with varying degrees of efficiency.

Aiden Singh: The data on Kickstarter outcomes raise an interesting question about venture capital funds. Estimates suggest that roughly 75% of venture capital-backed companies fail to return money to their investors.

By contrast, crowdfunding platforms such as Kickstarter also finance very young firms, yet the apparent success rate is much higher, with around 97 percent of funded projects ultimately delivering the promised product. 

Why do we observe such different outcomes?

Gilles Chemla: That is an excellent question, and it provides a useful entry point into the broader discussion of venture capital and private equity. The distinction between venture capital and crowdfunding is quite instructive.

Reward-based crowdfunding is especially interesting because it is often used not only by small start-ups but also by large firms such as ZTE - the Chinese mobile phone company - which use these platforms to test prototypes and obtain proof of concept even when external financing is not the primary objective. In that sense, crowdfunding is often a mechanism for validation as much as funding. Many projects that succeed on these platforms do so after failing to secure venture capital financing elsewhere.

Venture capital, by contrast, operates under very different incentives. Successful venture capital funds can afford to lose money on 90 to 95 percent of their investments as long as a few deliver extremely high returns. This creates a “lottery ticket” logic: entrepreneurs seeking venture capital must present their projects as potentially transformative and capable of producing very large exits. Modestly profitable or incremental projects are often unattractive, not because they are poor businesses, but because their potential upside is too limited for the venture capital model. Venture capital therefore focuses on identifying the next unicorns, rather than on financing projects that are simply viable or useful.

Reward-based crowdfunding follows a different logic. Projects are funded when enough individuals express interest in the product itself. Contributors often commit small amounts and their decisions are driven, at least in part, by consumption value rather than purely financial return. As a result, the underlying risk-return trade-off is fundamentally different from that of venture capital.

Venture capital also requires specialized expertise, access to networks, and the ability to identify promising opportunities within emerging technological waves such as artificial intelligence. Early investors in successful firms can earn exceptional returns, but doing so requires both scale and sophistication. Smaller or less connected funds are often unable to participate effectively in these opportunities. Moreover, successful venture capital funds can build reputation and attract more talent and capital over time, which helps explain why performance is often highly concentrated.

In short, the difference between venture capital and crowdfunding reflects a difference in objectives, incentives, and selection mechanisms. Venture capital is designed to capture rare, outsized successes through intensive screening and concentrated risk-taking, whereas reward-based crowdfunding relies on broad participation and proof of concept through collective demand.

——————

Editing by Harpreet Chohan.