The full version of our latest paper on successful crowdfunding campaigns is published by Journal of Business Research. In collaboration with Israel Felipe, Cristiana Leal, and Danilo Braun Santos, I started from the following idea: one of the most emblematic FinTech techniques is that of crowdfunding allowing entrepreneurs of new ventures to fund their efforts by drawing on relatively small contributions from a relatively large number of individuals (pledgers) through internet platforms without the use of standard financial intermediaries. Crowdfunding financing has grown exponentially in recent years, especially reward crowdfunding. A large number of projects that find financing in this market would not have access to any traditional source of financing during the early stage.     

Understanding the determinants of crowdfunding campaign success over its duration is critical for entrepreneurs and campaign supporters. As campaigns are usually defined as all-or-nothing (a campaign is deemed successful and the pledged money is collected only if the targeted amount is reached in a given limited duration of time), entrepreneurs are required to present campaigns in an attractive fashion as pledgers tend to contribute to campaigns that are perceived as worthy and are expected to succeed. Both entrepreneurs as well as pledgers are interested in a given campaign not only to achieve its target goal, but also to achieve it as quickly as possible. As fast as the target goal is reached, the project can start and the rewards can be received sooner. Notwithstanding this aspect, a number of campaigns have met with failure. It is the waste of time for the pledger and the waste of time, effort, and, eventually, money invested for campaign preparation for the entrepreneur, with there being no type of return, not even a non-financial return.

In this paper, we provide a replication and extension of previous studies focused on the role of success of crowdfunding campaign determinants by using survival analysis models. Specifically, we aim to identify the factors that impact the time-to-success of reward crowdfunding campaigns and introduce the new factor variables and new methods in the literature. In this sense, besides the factors already analyzed in the literature, we study the new drivers in the new institutional context and provide new results based on the statistical procedures not yet used in crowdfunding literature.

We argue that the success of reward crowdfunding depends on its perceived viability, social merit, and impact. The pledgers may not only be motivated by self-benefit (the number of rewards received), but also the campaigns’ expected social benefits, that is, the types of projects and their ability to reduce inequality and promote social cohesion. We claim that, in cities with higher inequality, crowdfunding can connect the people in need of funding with the people attempting to contribute to inequality reduction through their pledges, and thus, can positively influence the time-to-success of crowdfunding campaigns. People are satisfied by contributing to projects that make an impact, have social merit, and generate macro-economic implications in terms of income redistribution. In this sense, we argue that the pledgers disregard self-benefit and prefer social benefits, contributing to higher rates for social projects, rather than other project types such as art, and they do so even more intensively in cities with higher income inequalities. We also address the unobserved heterogeneity of the campaigns’ quality to ensure the robustness of our results.

From the point of view of entrepreneurs, this knowledge makes it possible to improve the delineation of the fundraising campaign profiles to increase the probability of success and accelerate the rhythm of fundraising. In terms of the current literature and the preeminence of the subject, we have noted that the research on the appropriate models to analyze the time-to-success of crowdfunding campaigns is a subject of equal relevance and little investigation on it, which put together, characterizes the present work as pioneering. In this article, we are particularly interested in gathering new empirical evidence regarding the success of campaigns by assuming an extended definition of success for reward crowdfunding campaigns. This is given that we contemplate not only the fact that the target amount has been achieved, but more importantly, the aspect of time-to-success.

We use parametric and semi-parametric models of survival analysis (an umbrella term covering data analysis that describes the expected duration of time until a well-defined event occurs) to examine the determinants of the time-to-success of campaigns. The application of this technique is original in this field, and it generates a better estimation of the model parameters as the survival models incorporate information about the censored and uncensored observations (successful and unsuccessful campaigns) and the duration modeling in the estimation. We use a unique database of 4,262 reward crowdfunding campaigns hosted between 2011 and 2016 on the largest crowdfunding platform in Brazil, one of the largest economies of the world. Considering Brazil is a large country with significant socioeconomic asymmetries, reward crowdfunding could prove to be a relevant instrument for reducing inequality. The platform we studied raised more than R$ 38 million (1 USD ≈ 3.2 R$ in December 2016 and 5.50 R$ in September 2020) in the period of analysis in an all-or-nothing system, which highlights reward crowdfunding as a relevant form of funding for a variety of entrepreneurial projects.

The dependent variable in the survival analysis consists of two parts: the moment of the event (the campaign’s time-to-success) and the status of the event (whether the campaign was a success or failure). The use of censoring in reward crowdfunding campaign survival analysis is a proper strategy to jointly investigate the success and the time-to-success of campaigns. We also carryout controls for the unobserved heterogeneity of the fundraising campaigns and censoring on the 59th and 60th day of the campaigns.

Estimates for Time-to-Success and the Smoothed Hazard Estimate (for Campaigns of 60 and 59 days)

Source: Calculation by the authors. Note: This figure presents a graphical representation of the survival function with probabilities estimated by the Kaplan-Meier method including 95% confidence bands, and on the right, curves that describe the instantaneous rate of occurrence of the event over time, namely the hazard function estimated for the time-to-success of reward crowdfunding projects with fundraising campaigns up to 60 days (Panel A), and up to 59 days (Panel B).

Our results are equally new. They reveal and suggest that the specific attributes of campaigns, such as their location, can influence the time-to-success of reward crowdfunding campaigns. In general, the results show that the campaigns that achieve success more rapidly are characterized by a lower target amount, a larger number of pledges, a smaller number of rewards promised in exchange for pledges, and are predominantly non-art projects located in cities with greater income inequality. In addition, we found out that the covariates adopted in the empirical model influence time-to-success in a non-constant manner during the fundraising period. This has motivated the estimation of parametric models, which ratify the results obtained by the popular Cox proportional hazard (PH) models.

Our results are robust to unobserved heterogeneity. This study contributes to crowdfunding literature in three ways. First, to the best of our knowledge, there is still no conclusive evidence regarding the influence of campaign attributes on time-to-success. We assume the campaigns classified as failures could be successful if the active management of the fundraising process was viable during the fundraising period in such a way that it would increase the chances of attaining the target amount. Survival analysis models have advantages over the other binary response models, especially because they rarely allow the time variable to present missing values. In conventional binary analysis, unlike survival analysis, if some of the observations disappear before the end of the observation period, it implies the loss of relevant information about the analyzed event.

Second, we believe that the influence of campaign location characteristics is an open question that concerns the concentration of crowdfunding campaigns in certain cities with greater income inequality. We also consider the geographic attributes of the location of reward crowdfunding campaigns. In this respect, we should point out that our results are supported by the data collected of an emerging economy in which social inequalities are explicit and the cost of capital is a limiting factor for the new ventures. Very few research studies on reward crowdfunding have discussed the attributes of the locations in which the campaigns are centered, which may bring out the information regarding the social and economic development of the given region. An investigation of the local attributes regarding these campaigns can generate knowledge to promote the effectiveness of crowdfunding campaigns. Third, while the literature is essentially characterized by OLS and logit models, we use survival analysis. Given that we also use robust models that violate the main assumption of the most disseminated model of survival analysis, we produce new evidence pertaining to the non-constant impact of the determining factors during the fundraising period. These results may be useful for entrepreneurs, pledgers, platforms, and regulatory agents.

If the crowdfunding literature has grown rapidly to the point where the success drivers of campaigns have been reasonably well documented, there is room to consider the possibility of new successful campaign drivers, in addition to the extensions and generalizations based on new evidences arising from new methods and new relevant institutional environments. Our study, even though it uses variables already documented in the crowdfunding literature, contributes to the theoretical and empirical development of this field, especially by promoting the generalization and extension of previous empirical findings. We bring out new and revealing evidence that has not yet been explored from an institutional context and explore unique data, new variables, and statistical procedures that are not yet used in the crowdfunding literature. Therefore, we provide additional evidence to help build a cumulative body of knowledge in crowdfunding literature.