Do project quality and founder information signals always matter? Evidence from equity and reward crowdfunding

This study provides a comparative analysis of the success factors of equity and reward-based crowdfunding. We find that information signals related to venture/project quality are important determinants of both equity-based and reward-based crowdfunding success. However, signals on founder information matter only for equity-based but not for reward-based crowdfunding. Our study exemplifies the varying role of success factors that depend on the type of crowdfunding projects and provides a basis for further research to compare different forms of crowdfunding.


| INTRODUCTION
Entrepreneurs seek innovative ways to attract funding for new ventures. Crowdfunding has now emerged as a popular form of alternative finance where entrepreneurs call for contributions, usually in small amounts, from the public or the "crowd", instead of a limited group of professional investors to fund their ventures. It is a unique way of financing noticeably different from other forms of entrepreneurial finance like angel investment, venture capital, and initial coin offerings (Block et al., 2018b;Block et al., 2021;Rossi et al., 2021). Crowdfunding is typically arranged via internet-based websites (the "platforms) where project owners (the "founders") propose their ideas, and individuals (the "backers" or "funders") contribute money to support the projects and/or receive some kind of benefits, rights, privileges or monetary rewards in return. In this way, entrepreneurs who had no or little access to the capital market can reach out to a large number of retail investors, spreading over many geographic areas and countries. 1 An important avenue of crowdfunding research is to understand the factors that contribute to the success of crowdfunding projects. These factors are mostly related to (i) venture/project-specific characteristics such as project description, the use of media, and provision of updates, and (ii) entrepreneur/founder-specific characteristics such as experience, capabilities and skills (Hsieh et al., 2019). The main purpose of our study is to examine which information signals regarding venture/project and entrepreneur/founder characteristics are significantly related to crowdfunding success.
We specifically focus on reward-based and equitybased crowdfunding projects that exhibit several differences in project and founder characteristics. First, reward-based projects are usually launched by individuals or firms, whereas equity-based crowdfunding is primarily launched by private firms (Vismara, 2018a). Second, the motivations of the funders are different. For equity crowdfunding, the funders invest money with the expectation to receive a share of future profits, whereas for reward-based projects non-financial motives like the delivery of a product or service dominates (Cholakova & Clarysse, 2015). Third, the objectives intended to be achieved with the funds raised from crowdfunding are different. Founders in reward-based crowdfunding want to test their products/services in the market, while founders in equity-based crowdfunding often want to finance the expansion of the business. This difference determines different timing when entrepreneurs use reward-based crowdfunding at an early stage and equity-based crowdfunding at a later stage of a firm's life cycle (Lukkarinen et al., 2016). Fourth, the target amount of the project is different across these two types. The capital requirement of reward-based projects is often small and aims at a large pool of funders while equity-based projects require a high contribution amount and a relatively more sophisticated funder (Belleflamme et al., 2014;Vulkan et al., 2016). All these differences have implications on the effort founders exert to signal information and reduce information asymmetry vis-à-vis project and founder.
The distinct features of reward-based and equitybased crowdfunding provide interesting contexts to test signal flexibility. Literature has shown that the effectiveness of signals differs due to the cost and specificity of signals, and the needs and knowledge of signal receivers (Ertug et al., 2016;Scheaf et al., 2018;Vanacker & Forbes, 2016;Vismara, 2018b). In our setting, we have multiple signals like project demo and video which are relatively more costly to produce than signals like the availability of founder's personal website. 2 In addition, we have different signal receivers like reward-based crowdfunding investors with a small average contribution and equitybased crowdfunding investors with a relatively large amount of contribution. Using the taxonomy developed in Bafera and Kleinert (2022), this paper studies the signal effectiveness by looking at the quantity and type of signals sent to different receivers in a noisy environment like crowdfunding platforms.
In addition to analysing the two crowdfunding types, our study is also interesting for an additional reason. The data we analyse come from multiple online platforms of varying sizes and allow us to consider whether oftenregarded success factors work well across various platforms. This is different from many other studies that typically collect data from one platform only. 3 Crowdfunding platforms differ from each other with respect to the provision, ease and security of information, the amount of funds raised (all-or-nothing approach), and the levels of competition for funding. An analysis of multiple platforms enables researchers to identify common success factors that are robust to platform-specific characteristics. Giudici et al. (2018) analyse thirteen platforms from Italy but one reward-based crowdfunding only. Ralcheva and Roosenboom (2020) examine two equity crowdfunding platforms from the UK. Block et al. (2018a) study two while Hornuf and Schwienbacher (2018b) study multiple equity crowdfunding platforms from Germany. In this study, we focus on five platforms that include four reward-based platforms and one equity-based platform.
Another way the paper differentiates from the rest in the literature is that we pay attention to an emerging economy Vietnama less-developed capital market where the increased availability of the internet infrastructure and the growth of private enterprises have made crowdfunding a viable financing alternative for small businesses. To date, almost all crowdfunding studies focus on platforms operating in developed capital markets in North America, Western Europe, and Australia, which share somewhat similar institutional and financial systems. Our research complements two recent studies by Bi et al. (2017) and Hsieh et al. (2019) who examine rewardbased crowdfunding projects from China and Taiwan, respectively. We use the context of Vietnama country that has moved from a centrally planned economy to a more industrial and market-based economy by undertaking financial system liberalizations and reforms (Le et al., 2019). Vietnam's open economy with substantial international trade has been flourishing year over year, with an annual growth of more than 5% since 2012. Higher economic growth coupled with private sector development brought stock market investment into public highlight. Furthermore, individual investors are the main traders on the stock market accounting for 85% of trading activities in Vietnamese stock exchanges (Tran et al., 2018). The level of real interest rateslending rates adjusted for inflationhas been relatively high at around 7%, which made crowdfunding an attractive option for small businesses in Vietnam. A rapid increase in internet services and online payments coupled with the growth of private enterprises created new opportunities to develop crowdfunding. Since the first successful crowdfunding campaign in 2014, the total amount of capital raised via crowdfunding has been modest in Vietnam. Nevertheless, according to Statista, it is expected to grow at 3.2% for the period of 2023-2026 compared to 0.8% in the US for the same period. The data indicate a great potential for crowdfunding to develop in Vietnam in the coming years.
In sum, we make three contributions to the literature on crowdfunding success. First, we examine equity-based and reward-based projects with their distinct differences to show the effect of signals in different contexts. Second, we include multiple platforms to seek common indicators that affect success. Last, we add a piece of evidence on successful crowdfunding from an understudied economy where there are great growth opportunities but also high hurdles for entrepreneurship.
We hand-collect information on all projects undertaken during 2014-2016 on the five largest crowdfunding platforms in Vietnam. The sample includes 38 (32%) reward-based and 80 (68%) equity-based projects. We use logistic and ordinary least squares regression techniques, adopt robust measures of funding success, and consider the endogeneity issues around signals and target amount. Our results show that signals related to project quality and founder information positively influence crowdfunding success. This holds entirely for equity-based crowdfunding. But for reward-based crowdfunding projects, we find only project quality signals matter. The finding highlights the varying role of success factors depending on the type of crowdfunding.
The rest of this paper is structured as follows. In Section 2, we briefly review the literature that examines crowdfunding success factors, and develop hypotheses regarding the impact of signals on funding success. Section 3 is devoted to the description of the research methodology and data. Section 4 presents the empirical results. Section 5 concludes and provides directions for future research.

| Crowdfunding and its success factors
Crowdfunding takes place in different forms, depending on how the return or reward is designed. We focus on equity-based and reward-based crowdfunding. In equity crowdfunding, the backers provide funding in exchange for shares in the start-up company. Reward-based crowdfunding is characterised by offering rewards to backers in the form of a product or service. In this case, the decision to fund a project is based on the expectation to receive some form of utility from the product or service.
An important avenue of crowdfunding research is to understand the factors that contribute to the success of crowdfunding projects. Agrawal et al. (2015), Giudici et al. (2018) and Josefy et al. (2017) examine the impact of geography and the local environment on funding success. Colombo et al. (2014), Mollick (2014) and Vismara (2018a) study the impact of endorsement by early backers. Hornuf and Schwienbacher (2018a) study the funding dynamics and find that larger investment and posting updates by founders increase subsequent investments.
Several studies specifically focus on the success of reward-based crowdfunding. Mollick (2014) finds that projects with a higher level of preparedness and a larger social network of founders are associated with an increased chance of success. Courtney et al. (2017) find that the use of media and the past success of founders increase the likelihood of crowdfunding success. Bi et al. (2017) and Shahab et al. (2021) find that providing online information and online feedback has a significant positive effect on funding decisions. Lagazio and Querci (2018) confirm that organizational and marketing factors influence the success of reward-based crowdfunding campaigns. Hsieh et al. (2019) observe that social-movementrelated crowdfunding projects exhibit significantly greater success than general projects.
Decision-making with respect to equity-based crowdfunding is relatively more complex as investors, motivated to receive an ownership stake and a financial incentive, require much more information regarding the risk involved. Ahlers et al. (2015) document that information about project quality and risk increases the success of equity crowdfunding. Block et al. (2018a) document that information updates provided during the campaigns significantly increase both the number of backers and the dollar amount of investments. Vismara (2018a) studies the positive effect of social capital on funding success. Shafi (2021) finds that equity fundraising success is most strongly related to the characteristics of the business such as product or service attributes. Barbi and Mattioli (2019) show that the education and professional experience of the project team help in raising more funds. Ralcheva and Roosenboom (2020) and Rossi et al. (2021) find that crowdfunding success is positively related to higher equity retention by founders. The finding indicates that investors perceive an entrepreneur's financial commitment as a good signal.

| Theory and hypotheses
Asymmetric information is a fundamental problem that entrepreneurs face in their search for external financing because they have more information regarding a firm's quality in comparison to outside investors (Cosh et al., 2009). The risks to outside investors, associated with uncertainty, make external finance more expensive or even unavailable. Several idiosyncratic features of crowdfunding exacerbate information asymmetry between entrepreneurs and investors. First, unlike conventional investors such as venture capitalists or business angels, crowdfunding investors are usually retail investors. They lack sufficient knowledge about the industry, technology, management, experience and capabilities of the entrepreneurs as well as other types of information to make a decent assessment (Schwienbacher & Larralde, 2012). Second, the level of uncertainty in crowdfunding is higher than in other forms of investment, because most crowdfunding projects are "work-inprogress" that can constantly change and incorporate feedback from the crowd (Colombo et al., 2014). Third, the fear of unwanted exposure may make entrepreneurs even more reluctant to publicise sensitive information (Schwienbacher & Larralde, 2012). Crowdfunding thus provides a context where the core idea of signalling theory can be applied to develop testable hypotheses (Block et al., 2018a;Connelly et al., 2011). Ahlers et al. (2015) argue that the average crowdfunding investor is unlikely to have the time, capacity, and incentive to investigate activities of firms and their business model in detail. To mitigate information asymmetry and successfully attract more funds, a high-quality entrepreneur needs to signal the superior quality of the project to convince potential investors (Belleflamme et al., 2014;Courtney et al., 2017). Researchers try to identify various signals that are effective in drawing funds from investors and increasing the likelihood of crowdfunding success.
Nevertheless, Scheaf et al. (2018) show that signals can vary in their effectiveness due to signal flexibility and receiver characteristics. Flexible signals are effective across different exchange contexts and signal receivers. A costly signal like media coverage can enhance its flexibility to work in different contexts like reward-based and equity-based exchange. Receivers differing in their needs, knowledge, and human capital attend to different information and interpret it differently. Similar propositions are made by Vanacker & Forbes (2016), who explain how signals are received differently by different resource providers based on differences in their information environment. For example, human resource providers are more influenced by a new venture's media prominence (a relatively general, more readily accessible and easierto-interpret reputation signal), whereas financial resource providers, who are already surrounded by a richer information environment, look for more specific signals such as the venture's industry-specific experience. Ertug et al. (2016) find that reputation signals specific to certain receivers can make a difference in the outcome. Investors of reward-based projects value signals about the ability of timely delivery of rewards/products, while investors of equity-based projects value signals of future financial success (Cholakova & Clarysse, 2015).
Signals in the context of crowdfunding can be classified as related to the quality of the project or the credibility of the founder (Courtney et al., 2017;Huang et al., 2022). While starting a crowdfunding campaign, the founder needs to write an introductory text to describe the purpose of the project, how it works and what the expected deliverables are. A comprehensive project description thus increases the chances of funding success. Signals reflecting project quality (e.g., videos, updates, spelling mistakes, and length of text) also lead to higher funding success. Most crowdfunding platforms advise to give an introductory video in their pitch, which is "a clear signal of at least minimum preparation" and therefore leads to higher success probabilities (Bi et al., 2017;Colombo et al., 2014;Mollick, 2014). Similarly, platforms recommend providing updates soon after the project launch because speedy updates imply a well-prepared entrepreneur (Mollick, 2014). Another signal of project quality is spelling mistakes. More errors indicate a lack of basic proofreading and therefore signal insufficient preparedness and quality, reducing funding success (Mollick, 2014).
Completeness is another indicator of project quality (Bi et al., 2017). Word count, apart from showing the degree of preparedness, can proxy the amount of information delivered to readers. For example, in the description of a film project, the creator may give details about the main plot, the crew, the expected time of preview, and the use of funds. All this information is especially important to investors to assess the potential of the project. The more detailed it is, the less information asymmetry is attained and the more likely will be the realization of investment.
The positive relationship between fundraising success and project quality signals has been confirmed for both equity-based (Ahlers et al., 2015;Block, Colombo, et al., 2018) and reward-based crowdfunding projects (Bi et al., 2017;Courtney et al., 2017). In the context of Vietnam where individual investors dominate the stock market, and both entrepreneurs and investors tend to be failure-averse, the value of such signals is expected to be even higher. Therefore, we put forward our first hypothesis as: Hypothesis 1. Project quality signals increase the likelihood of crowdfunding success in both equity-based and reward-based projects.
Investors also seek information regarding the founder to alleviate their concern about the ultimate success or failure of the project. Founders are interested to increase the likelihood of the project's success and therefore provide information concerning their capabilities and skills. Zott and Huy (2007) argue that founders who are able to show their capability (educational degree, relevant research) and commitment to the venture (financial sacrifice, delayed personal gratification) are likely to get more funding. Ahlers et al. (2015) show that human capital is a determinant of project success because it is better at "identifying and exploiting business opportunities", "defining and realizing a venture's strategy" and "building a positive basis for future learning". In the context of Vietnam, where most traders in the capital market are individual investors (Tran et al., 2018), signals conveying information on founders are expected to be more valuable. Especially for equity-based projects, more signals on founders can lead to more funding success.
The effectiveness of different signals varies with the context and the characteristics of receivers (Scheaf et al., 2018). In the case of reward-based projects, backers usually expect a return in the form of a product or a service rather than the future success of the business. If there is a good prospect of product /service delivery such as a prototype available, backers may not look for signals about the founder. In addition, the funding amount for rewardbased projects is usually small. Backers are likely to spend a limited amount of time and effort searching for further information beyond the project, such as the founder's educational background, industry experience, or management skills. Therefore, we hypothesise the following: Hypothesis 2. Signalling information on founders increases the likelihood of crowdfunding success in equity-based projects, but not necessarily in reward-based projects.

| METHODOLOGY AND DATA
We use the following model to examine the factors leading to crowdfunding success: The dependent variable funding success is usually measured as a binary indicator variable (see, for example, Ahlers et al., 2015;Courtney et al., 2017;Mollick, 2014). We, therefore, use a logistic regression analysis. As a robustness check, we use numerical measures of funding success and perform an Ordinary Least Squares (OLS) regression analysis. We also undertake tests for endogeneity and perform an analysis by splitting the sample into subsamples based on the two crowdfunding types.

| Dependent variable: funding success
We initially measure funding success as a dummy variable which takes the value of 1 if the actual funding amount is equal to or higher than the target amount of funding, and 0 otherwise. Two alternative measures used are the dollar amount of funds actually raised and the number of backers contributing to the crowdfunding project. Because of the large variation of these two variables, we use natural logarithms in the regression analysis.

| Independent variable: project quality
Project quality is estimated on the basis of online information that founders provide on crowdfunding platforms. We collect five different types of information provided on each project: video, product demo, updates, spelling mistakes, and project website/page; and then use an aggregate measure of project quality by constructing a composite index as the sum of these five indicators.
Video is a dummy variable to indicate whether founders provide a video in the project description. Most crowdfunding platforms strongly advise creators to include a video as an illustrative introduction of the project and project team. Videos can say more than words and show adequate preparation of founders. These can serve as a signal of high-quality projects, and as such should be associated with a higher success rate.
Product demo is a dummy variable to indicate whether a demo is available (in the introductory video, pictures, or provided links). A demo is understood as a version of the finished product, showing how it looks and works (if possible). In this way, it should be important to demonstrate the serious attempts and passion of the founders for the project as well as to communicate signals about its feasibility. A product demo helps increase the certainty of a project by providing additional information and thus is useful in attracting funders.
Update is a dummy variable to indicate whether the founder of a project posts updates about the progress of the project. Quick and frequent updates show an effort of the founders to provide the latest information to the public, indicating that the founders are potentially responsible and serious, and that the project is making progress, which should increase the chance of successful fundraising.
Spelling mistake is a dummy variable to indicate whether a project pitch is free of spelling mistakes. Mistakes resulting from insufficient preparedness negatively affect the chance of success. In contrast, a mistake-free project description may indicate higher quality and signal greater success. In this analysis, we use Vspell online spell-checking tool and manual scanning to identify spelling mistakes in the description of projects.
Project website/page is a dummy variable to indicate whether there is a separate website or fan page dedicated to the project. The existence of such pages demonstrates the high levels of effort put in by project founders and provides additional information to investors about the project. It can be a hint of quality and is expected to imply greater success.

| Independent variable: founder information
Founders of crowdfunding projects aim to reduce investors' concern about information asymmetry in order to achieve a high likelihood of successful funding. We identify a variety of information sources such as the presence of a personal website, blog, Facebook/Twitter page, or detailed introduction of the founders in project pitches. The reason is that, through these means, potential investors can get in touch and explore more about the founder's qualifications, achievements, and characteristics and get a better idea of his/her trustworthiness and credibility. Founder information is a dummy variable to indicate whether a personal website, social media page, blog or additional information (who they are, what they have done, how they come up with the idea) on the founders is provided.

| Control variables
We include the funding target amount, minimum pledge amount and the number of pledge levels as control variables. 4 Funding target amount is the desired amount of funding set by founders. Since the platforms included in this analysis operate on an "all or nothing" basis in which founders receive nothing if the funding goal is not reached, they should be careful in choosing a realistic and attainable target in order to avoid waste of time and effort. Due to the wide dispersion of funding goal amount and minimum pledge amount, we use the natural logarithm of funding goal and of minimum pledge amount in regressions. A minimum pledge refers to the minimum amount of money that can be pledged, which is usually set by project founders. Conceptually, the smaller the minimum pledge is the more funders the project can attract, because it is easier to convince people to spend small amounts of money than larger ones. The number of pledge levels is the number of funding options available. Different pledge levels are associated with different privileges, with larger contributions being rewarded with more valuable and exclusive benefits. A project with many pledge options allows for more flexibility and therefore is more accessible to potential backers. We also control for the specific effects of platforms when possible. Rossi et al. (2021) report that differences across platforms matter. Table 1 gives an overview of all variables and their measurement.

| Data
We hand-collect information regarding all projects on the five largest crowdfunding platforms in Vietnam. 5 Four platforms deal with reward-based projects, and one 4 | RESULTS Table 2 presents the descriptive statistics of all variables for reward-based and equity-based crowdfunding projects in the sample. We observe that the average success rate among reward-based projects is 63% whereas the rate is 43% for equity crowdfunding projects. These two types of crowdfunding projects also differ in many other aspects. The funding amount is not capped at the funding target and a campaign can actually raise more than what is requested. On average, a typical equity-based project is able to attract more than double (227%) of the target amount with an average funding amount of USD 25,154. Again, this number is substantially lower for the rewardbased crowdfunding initiatives (83% and USD 4044). The average equity crowdfunding project seeks to raise USD 33,436, which is nearly six times more than the average reward-based project (USD 6035). This large difference between equity-based projects and reward-based projects in terms of funding amount and target capital is also documented in Vismara (2016). There is a considerable variation across the sample as can be seen from the medians which are substantially below the means. A typical reward-based project seems to be able to attract more backers than an equity-based one (250 versus 30). On the other hand, reward-based projects require a lower minimum investment (5 versus 220) and provide more pledge options than equity-based projects (6 versus 3). The overall picture depicts that reward-based crowdfunding is targeted at the general public with small amounts and diverse needs while equity-based crowdfunding is attractive to more "professional" investors with larger investments. Descriptive statistics of different signals are presented in Panel B of Table 2. We observe that on average 66% of the equity-based projects provide an introductory video, slightly more than that used by reward-based projects (55%). A product demo reduces uncertainties about the project. As much as 84% of the reward-based projects include a preview of the (nearly) finished product in their project pitches, compared to 66% for equity crowdfunding. It is quite interesting to see that only 1% of equity funding projects provide updates about project progress to keep current investors informed or to attract new investors. Yet more than half of the reward-based projects (53%) provide updates. Compared with the average number of updates (3.97) reported by Mollick (2014), the average update for all projects in this study is only 0.36 (less than one update due to many zero updates, from an untabulated result). It appears that in Vietnam founders seem to use less communication via the platforms. The description text's spelling accuracy demonstrates the degree of preparedness and, to some extent, the education level of the founder. Both reward-based and equity-based firms have almost the same percentage of spelling mistakes (25%). This percentage is much higher than the 2% reported by Mollick (2014). One possible explanation is the lack of proofing tools in Vietnamese language. Given that project pitches are limited in terms of length and presentation, the provision of a separate project web page is an important additional source of information, which enables investors to evaluate the quality and prospects of the project better. About three-fourths of equity crowdfunding projects in our sample have a website or a Facebook page, compared to 53% for reward-based projects. Finally, project quality, calculated as the sum of five dummy project quality variables, shows that, on average, a reward-based project scores 2.71, compared to 2.31 for an equity-funded project.

| Descriptive results
A founder is deemed to be of higher credibility when (s)he reveals additional information about qualifications and experience in the project description or his/her personal website, blog, LinkedIn or social network pages. As such, potential funders can get a better idea of the project team's capabilities and are more inclined to invest. We find that 82% of reward-based projects in our sample provide this kind of information, mostly by linking to a personal Facebook account. This percentage is lower (63%) for equity-based projects. A comparison between the two categories of crowdfunding projects shows that rewardbased projects tend to signal quality via product demo, updates and founder information, while equity-based projects tend to use video and project websites.
The last two columns of Table 2 report the statistical significance of the mean and median difference between reward-based projects and equity-based projects along various variables. The high significance of the differences in the independent variables suggests that it is not appropriate to include the crowdfunding type with the independent variables in the same regression. Therefore, we show the differences between the two crowdfunding types by using the split-sample analysis. Table 3 reports the Spearman correlations between variables. Project quality and founder information are positively correlated with success in the full sample (Panel A) and the subsample of equity crowdfunding projects (Panel C). For the subsample of reward-based projects (Panel B), founder information has little correlation with project success. Among the independent variables, pairwise correlation coefficients are relatively low except for the correlation between the logarithm of minimum pledge amount and the number of pledge levels in the full sample and the correlation between project quality and the logarithm of minimum pledge amount in the reward-based projects subsample. We use different model specifications to deal with any potential concern for multicollinearity. Table 4 reports the result of logistic regression of the likelihood of crowdfunding success on project and founder variables for the full sample in Panel A, and split samples in Panel B. In Panel A, across various model specifications, project quality positively affects the chance of success. To be precise, an exponential coefficient 8 of 2.86 in Model (4) means that if the project quality increases by one point (having one more signal), the odds of crowdfunding success will increase by 2.86 times. The finding is in line with Bi et al. (2017) and Mollick (2014). Project quality signals are an effective way for founders to communicate and provide useful information about the project's underlying quality to potential investors. This mitigates information asymmetry, increases backers' trust and makes them more confident to invest in the crowdfunding project, increasing its likelihood of success.

| Regression results of success factors
As for the founder information, we find that the estimated coefficient is statistically higher than 1 [exp(B) = 3.60 in Model (4)], indicating a positive impact on the likelihood of success. Crowdfunding projects that include additional information about their founders are approximately three times more likely to succeed than those, which do not. The finding confirms the value of founder information signals studied earlier in the form of the founder's network size (Mollick, 2014;Vismara, 2016) or qualifications (Ahlers et al., 2015).
Regarding the control variables, the funding target has a statistically significant and negative impact on funding success. Its exponential coefficient is always below 1, suggesting that setting a high target reduces the probability of successful crowdfunding. The minimum pledge amount and the pledge levels do not show a statistically significant impact on funding success.
Panel B reports the results for reward-based and equity-based projects respectively. In reward-based crowdfunding projects, backers receive a reward for backing a project. This reward can include receiving a product at a reduced price before it is marketed to the public. On the other hand, in equity-based crowdfunding, backers are more like investors, and are expected to receive a share of the profit from a project. We find that the impact of project quality is statistically significant for both reward-based and equity-based projects. Models (3) and (6) show that having one more project quality signal increases the odds of success by 3.68 times and 2.84 times, respectively. The results support Hypothesis 1 that project quality signal is an important determinant of success for both types of crowdfunding projects Regarding the impact of founder information, we find that it is statistically significant only for equity-based projects. Model (6) shows that the provision of founder information increases the odds of success of equity-based projects by 3.69 times. For reward-based projects, founder information does not show a statistically significant impact. This insignificant impact of founder information signal may arise from the fact that an average rewardbased project attracts 250 backers with a minimum pledge amount of only US $5, in contrast to only 30 backers with a minimum pledge amount of US $220 in the case of equity-based projects. With a small amount of investment at stake and often a product as a reward, backers of reward-based projects are not so keen on learning about the founder in their funding decision. 9 Since we find statistically significant effect for equity- based crowdfunding projects, but not for rewards-based projects, we conclude that Hypothesis 2 is only partially supported.

| Robustness checks
We undertake several tests to ensure the robustness of our results. In addition to measuring funding success with a binary indicator, we use the number of backers and the actual amount of funding as alternate metrics. We perform OLS regressions. Because these two measures have large variations, we use logarithmic transformation. The results are presented in Table 5. Panel A reports results on all projects. We find that project quality has a significant positive impact on funding success. More project quality signals are associated with the ability to attract more backers and more funding. These  Table 1. Coefficients are reported in exponential form, with Wald statistics in parentheses. The Wald statistic tests the hypothesis that the exponential coefficient is 1. The asterisks *, **, *** refer to significance level at 5%, 1% and 0.1%, respectively.
results are consistent with what we found earlier in Table 4. Regarding the control variables, a higher funding target (in Models 1 and 2), a lower minimum pledge requirement (in Model 1) and more pledge levels (in Model 2) attract more backers to fund the projects. However, these variables do not matter for the actual funding amount, as shown in Models (3) and (4). The regression results of the split samples of reward-based and equity-based projects are shown in Panel B. The results provide support to the positive effect of project quality in both crowdfunding types. However, the founder information variable does not show a statistically significant effect on funding success. This implies that founder information does not matter for the extent of success when measured by the number of backers and the actual amount raised. An endogeneity issue can arise when the target amount is chosen to maximise the chance of success. We deal with endogeneity using the instrumental variable approach. Following Cumming et al. (2019), we instrument the target amount with the industry average amount, which is measured for each firm as the average of all target amounts in the same industry, platform and year. The results are reported in Panel A of Table 6. Another concern for endogeneity arises from the level of information disclosure that can largely depend on the requirement of the platform. To deal with this, we replace the signal measures by taking the deviation from the average amount of signal of that platform. We present these results in Panel B. Both results are similar to the results presented earlier in Table 4. We also examine the individual impact of each project quality signal on funding success. Statistically significant results are found for equity-based projects only. The Note: The table reports the logistic regression results of funding success for the two subsamples of rewardbased and equity-based crowdfunding projects. To relieve the endogeneity arising from the variable target amount, we replace it with the industry average target in Panel A. To relieve the endogeneity arising from the signals of each platform, we replace project quality and founder information with the deviation from the platform average of the year in Panel B (Dev_Project quality and Dev_Founder information). All variables are defined in Table 1. Coefficients are reported in exponential form, with Wald statistics in parentheses. The Wald statistic tests the hypothesis that the exponential coefficient is 1. The asterisks *, **, *** refer to significance level at 5%, 1% and 0.1%, respectively.
T A B L E 7 Individual signals as crowdfunding success factors. results are reported in Table 7. For reward-based projects, we find no evidence of influence from individual signals as seen in Panel A. Results in Panel B show that video, demo and founder information have a statistically significant coefficient of 6.69, 10.79, and 3.99, respectively. It means, for equity-based projects, the provision of a video, a demo and founder's information are effective signals to increase the chance of funding success. We also undertake additional analyzes on the interplay of signals as in Courtney et al. (2017) and Scheaf et al. (2018). We interact visual cues (like demo and video) with other signals (like spelling mistakes, website and founder information) to see whether signals enhance each other, and find no statistically significant results.

| DISCUSSIONS AND CONCLUSIONS
Crowdfunding has emerged as a novel and plausible financing method for many small businesses. A large number of investors can put modest sums into cashstarved businesses. In this study, we simultaneously examine reward-based and equity-based crowdfunding projects and examine the role of signals concerning venture/project quality and entrepreneurs/founders in predicting crowdfunding success. We hand-collect unique data from the five largest online platforms in Vietnama country that opened up the communist economy to private enterprises and experienced one of the fastest economic growths in Asia in the last two decades. We observe significant differences between rewardbased and equity-based crowdfunding projects. The equity-based project is, on average, larger and has fewer investors but higher average funding per investor, and a higher success rate than a reward-based project. Rewardbased crowdfunding projects provide more product demos, updates, websites and founder information than equity-based projects.
The empirical analysis of all projects together suggests that signals of high project quality and founder information are associated with a higher probability of success. In particular, crowdfunding projects are more like to be successfully funded if they signal high quality by providing a combination of a video, a product demo, a description text free of spelling mistakes, a project website, and a detailed description of the founders. These findings confirm the role of signals in explaining funding success. With more signals on project characteristics and founders, the information asymmetry problem becomes less severe and crowd investors have more trust to invest in a project.
However, we do not find a consistent impact of these signals when we separate reward-based projects from equity-based projects. Signalling entrepreneur/founder information is an important success factor for equity crowdfunding, but it does not play a key role in the success of reward-based crowdfunding. This is potentially due to the relatively smaller amount of investment involved in reward-based projects. For backers in rewardbased projects, their return is in a form of a product. The information specific about the product is sourced from product demo and video, while the access to founder information seems not directly relevant. This echoes the concept of audience-specific signals in Ertug et al. (2016) that signals specific to receivers can improve the outcome. From the perspective of decisions in these two types of crowdfunding, backers of reward-based projects make non-financial decisions that combine a consumption decision and a donation decision (Bollaert & Schwienbacher, 2021), while backers of equity-based projects make investment decisions. The later decision demands relatively more signals to reduce information uncertainty. Signori and Vismara (2018) observe that sophisticated investors possessing superior information are able to pick firms with more promising business models. From the perspective of motivation, in rewardbased crowdfunding, backers participate in the design of the product, or co-create the product, and benefit from pre-ordering the product below the market price. This cocreating process gives the backers a sense of community, and it is argued that they fund the project with a community logic (Wessel et al., 2022). The focus on the product is reflected by the higher number of updates when compared to equity-based crowdfunding. Thus, signals about project quality are more relevant than signals about founders in reward-based projects. In equity-based crowdfunding, backers expect to share the future profit of the project, they are argued to have a market logic with a focus on the economic return of the investment. Thus, all signals that are relevant in evaluating the project's ability to generate cash flows in the future matter.
This study has general implications for entrepreneurs, platform managers and policymakers. For those who seek finance with crowdfunding, the most important lesson is that quality signals play an important role in funding success. Investors actually value credible signals and use them to make funding decisions. Therefore, entrepreneurs should look for ways to convey more information on the new venture as well as their own quality to potential funders by showing preparedness and full information. Moreover, carefully choosing an appropriate goal also contributes to a higher chance of success. Platform managers can facilitate start-ups by providing more information on both the project and the founder, and thus help reduce the information asymmetry problem. Policymakers need to design regulations in such a way that online investment can be a viable mode of financing startups. Increasing individual investor confidence would be a useful step towards raising capital for productive purposes, and eventually crowdfunding success.
Our study has an important implication for emerging countries. Vietnam's annual GDP growth rates with an average of 6.6% over the period 2014-2018 are higher than that of neighbouring countries. A fast-growing economy will create a favourable environment for entrepreneurs to start a business in general and start a business via crowdfunding in particular. According to the report of Global Entrepreneurship Monitor and VCCI (2015), the levels of entrepreneurial opportunities, confidence and entrepreneurial spirit are not high among Vietnamese people. The fact that the low percentage of people with actual startup intentions in comparison with the level of identified opportunities suggests a lack of trust in the market as well as a personal fear of startup failures. Vietnamese people still do not have a positive view of failures, thinking that failing is a bad or embarrassing thing. As a result, many may hesitate to share their ideas and call for investment publicly due to the fear of being criticised, disparaged and losing face if the project does not succeed eventually (Trieu, 2015). From investors' perspectives, their biggest fear is the founder's misuse of funds (Gerber & Hui, 2014). In the context of Vietnam, face-to-face communication and personal relationships play an essential role in business transactions: mostly people only trust and conduct business with those that they know well, or at least have met and talked to in order to mitigate information asymmetry and delivery risk (Turan, 2015). Consequently, it is very important for investors to receive quality signals before making the funding decision. A further investigation of success factors with a large sample will be of great practical importance to Vietnamese entrepreneurs and investors, and to many South Eastern countries whose people share similar traits.
Our study is not free from some limitations, which also provide avenues for future research. First, although we made an attempt to perform a comparative analysis of reward-based and equity-based crowdfunding, our analysis involved a relatively small sample of crowdfunding projects from five platforms in Vietnam. Thus, the results may not be generalised to other countries and platforms. A larger sample will provide more solid results. We could not also compare donation-based and lending-based crowdfunding projects. The motivations of backers who act as philanthropists and lenders may be different from those of customers and equity holders, leading to the emergence of different success factors. Second, due to limited data availability, we are not able to investigate some of the factors that are found significantly related to the likelihood of crowdfunding success by prior researchers. These include project duration and featuring of the project on a platform's website. Third, founder information signal is measured by a dummy variable. When more data are available, alternate and more comprehensive measure can be constructed and tested in subsequent studies. 10 As pointed out by Bafera and Kleinert (2022), the interaction between the signals, whether amplification, complementarity or conflict, can further be explored.
homepage. Unfortunately, because the websites used in this study do not make available of such data, we cannot use them in the analysis. 5 The names (number of projects) of the crowdfunding platforms are Betado (18), Comicola (11), Firststep (7), Fundingvn (80), and Fundstart (6). 6 In split sample analysis, we use three platform dummies for reward-based projects. 7 One reward-based project did not disclose information on minimum pledge amount and number of pledge level, thus was excluded from the analysis. 8 An exponential coefficient higher than 1 (exp(B) > 1) indicates a positive impact on the likelihood of success (B > 0). 9 Performing the Wald chi-square test shows no statistically significant difference in the regression coefficients of the two types of crowdfunding projects. This result does not allow us to claim a statistically different impact. 10 Researchers can collect information on the number of founding team members, their educational qualifications (Ahlers et al., 2015), their age (Ralcheva & Roosenboom, 2020), the size of their social networks (Mollick, 2014;Shane & Cable, 2002), or the physical traits (Mohammadi & Shafi, 2018) and use these as signals.