Managerial Responses and Customer Engagement in Crowdfunding

Round 1

Reviewer 1 Report

Report on “Managerial Responses and Customer Engagement in Crowdfunding”

I appreciated this paper which brings an original look at the performance of crowdfunding projects. The hypotheses are clearly stated and the empirical study is generally well conducted. I have some reservations, however, in what follows. The question of endogeneity / causality is in my opinion the main point of improvement of the paper before publication.

Main remarks:

1. In equation (4) a significant collinearity problem may arise. It would be a good idea to present the correlation matrix between the different variables and to indicate the FIVs for this equation, as it may be biased.
2. Problem of endogeneity which should lead to greater caution in interpretation. Indeed, a project is at the outset intrinsically good or bad. If it is good, it will perform well and succeed. If it is good it will attract reviews. This does not make the review a factor of success, but a consequence of success. Causality analysis needs to be better presented. How can causality be tested econometrically? I am not totally convinced that your robustness analysis (eq5 and table 7) fully answers the question of causality. I would like a more in-depth discussion of this essential question.
3. Lines 287-288: “While Column (2) and Column (3) of Table 5 present the length of customer review has a negative and significant effect on the crowdfunding performance.” It would be useful to have an interpretation of this effect which is not coherent with the hypotheses.

Other remarks:

4. Line 28 : Bargain et al. (2018) shows the same for the wine sector. That reference should be added :Bargain, O., Cardebat, J. M., & Vignolles, A. (2018). Crowdfunding in the wine industry. Journal of Wine Economics, 13(1), 57-82.
5. Line 299: you mean eq (4) which becomes eq (5) ?
6. Line 308: you mean table (4) and not (6) ?

Author Response

Point 1: In equation (4) a significant collinearity problem may arise. It would be a good idea to present the correlation matrix between the different variables and to indicate the FIVs for this equation, as it may be biased.

Response 1:  Thanks for your suggestion. To avoid multicollinearity problem, we revise equation (4) and equation (5) in our revised manuscript. To investigate whether the collinearity problem exists, we report the variance inflation factors of these independent variables. All VIFs are below the conventional cutoff 10, the highest being 5.15 on Total_Backer.

Point 2: Problem of endogeneity which should lead to greater caution in interpretation. Indeed, a project is at the outset intrinsically good or bad. If it is good, it will perform well and succeed. If it is good it will attract reviews. This does not make the review a factor of success, but a consequence of success. Causality analysis needs to be better presented. How can causality be tested econometrically? I am not totally convinced that your robustness analysis (eq5 and table 7) fully answers the question of causality. I would like a more in-depth discussion of this essential question.

Response 2: Thanks for your suggestion. To avoid the endogeneity problem, we revise our equation by decomposing its error term, and control for time-invariant unobservable project heterogeneity with project fixed effects .

Point 3: Lines 287-288: “While Column (2) and Column (3) of Table 5 present the length of customer review has a negative and significant effect on the crowdfunding performance.” It would be useful to have an interpretation of this effect which is not coherent with the hypotheses.

Other remarks:

Response 3: Thanks for your suggestion and we have corrected it in our revised manuscript. To avoid multicollinearity problem, we have revised equation (4) in our revised manuscript. Table 5 shows the estimates of Equation (4): the length of responses, the speed of responses, and the volume of responses have a significantly positive effect on the final crowdfunding performance.

Point 4: Line 28 : Bargain et al. (2018) shows the same for the wine sector. That reference should be added :Bargain, O., Cardebat, J. M., & Vignolles, A. (2018). Crowdfunding in the wine industry. Journal of Wine Economics, 13(1), 57-82.

Response 4: Thanks for your suggestion and we have added it in our revised manuscript.

Point 5: Line 299: you mean eq (4) which becomes eq (5) ?

Response 5: Thanks for your suggestion. In equation (4), we have used the incremental amount pledged at time t as the dependent variable. In Equation (5), we use an alternative variable, the incremental backers.

Point 6: Line 308: you mean table (4) and not (6) ?

Response 6: Thanks for your suggestion and we have corrected it in our revised manuscript. “Compared with Table 5, the responsiveness has the same sign.”

Author Response File: Author Response.pdf

Reviewer 2 Report

In the paper „Managerial Responses and Customer Engagement in Crowdfunding“,  the authors address the role of user-generated content on Kickstarter. The paper has an interesting idea and contains several empirical analyses. However, there are some shortcomings that would not allow to publish the paper as it is. In the following, I detail on my thoughts about the paper.

-- Abstract:

I would like to read a definition/explanation about what kind user-generated content one can expect in crowdfunding. What exactly is meant by managerial response in this context? Later in the paper, it would be important to describe how such texts looks like.

-- Introduction:

The authors find a good start into the topic. However, I miss references and evidence for several statements. Why does crowdfunding realize “the optimal resource allocation among different individuals”? Why are the authors so sure that the allocation is optimal? References/proofs are missing. Moreover, the authors say “User-generated content is significantly increasing in crowdfunding”. Again, references are missing. Is it really increasing? How do the authors know? How much has it increased over what time period?

I would like to read more about which problem the authors address and try to resolve. What exactly is the contribution of the paper? (Theoretical and practical.) The fact that social media, user-generated content, managerial responses are important in these days is well known – but what is special about the given case/analysis. This need to be explained.

The language of the paper needs to be revised, e. g., what is meant by “Understanding the effect … is significant”? See also below some examples of mistakes (section: Minor issues) – but these are not all language-related mistakes.

-- Literature Review:

What is meant by “we search one of the main researches”?

The first sentences of the crowdfunding literature review are no good start into this section. Here, general information is given that I would expect to read earlier.

The paper is about user-generated content (UGC). But why do the authors only review crowdfunding papers without UGC? There are quite some crowdfunding papers that deal with the UGC. This is completely left out – but this is an important basis for the paper and the later analysis. Without a proper literature analysis, the reader cannot judge about the contributions of the paper.

“It is useful to study how the relationship between user-generated content and crowdfunding performance and how this effect varies across different features of projects.” Is this a complete sentence? And why is it useful? Valuations always need supporting arguments. Why do the authors argue that social media and crowdfunding research is “still insufficient”?

If I count right – only one paper (in the UGC section) is referred to which deals with social media and crowdfunding together. However, there are much more papers about this combination. The authors need to refine their literature review. Maybe the authors judge the research streams as “insufficient” because they have not invested much time about the literature review.

The review is quite superficial and contains only very few findings. This should definitely be optimized.

-- Theory and Hypothesis Development:

Until here, still no information is given what kind of UGC is prevalent on kickstarter. How does reviews and responses look like? I only now comments and updates on kickstarter. But “reviews” (product reviews?) as in Amazon (about product quality) are not known to me. Where are they available on Kickstarter?

“Audience driven is enhanced”? Please reformulate.

I miss the theory behind authors conclusion that the interaction between the creator and the backers might lead to writing more reviews? Still, I don’t know where the reviews are – but if comments are meant, I do not know why backers should post more questions/requests when these have already been posted/discussed. The comments are normally used to discuss questions or problems. Why should they discuss more then? Another explanation is that there is more discussion because they like the project. Then this is just a side effect but no explanation of funding success. The authors need to deal with such ideas to better guide the reader.

The hypothesis part lacks theoretical backing. I would like to read more about why these dependencies are expected.

-- Research Design:

Why is it interesting to know the function used for the sentiment – if nothing else is said about how/where/in which program everything is done?

The authors need to detail more on what they include as time-varying control variables: Z is not described at this place. How does date and day look like? And what is the last variable of the equations?

I doubt that the data set is free of selection bias. The authors state that 60% of the projects in their data set are financed. However, there should be a rate of 30-35% of financing on Kickstarter. I would prefer seeing the analyses with a data set that is more comparable to the real situation on Kickstarter. Have the authors compared their data set to the real numbers on Kickstarter? If not, they should do so.

-- Robustness Checks:

The authors introduce “interaction” terms. However, they miss to argue why they decide for this way of “robustness” check. How can the results be interpreted? The more money a project wants to collect, the worse is responsiveness for performance? This is not intuitive. The authors should explain it.

-- Implication and Conclusion

The authors argue that “the dataset is not big enough” … big enough to do what? To have reliant results? Moreover, the authors think that “other control variables of project characteristics are not exhaustive”. What is missing? Why are they not exhaustive? Why don’t the authors solve this problem?

-- Minor issues:

Is scared or scarce?

P3/L125: “are focus on” right grammar?

P3/L131: “precious studies” or previous?

P4/L147: “The quality (…) is uncertainty”?

P7/L243: “resview”

etc.

All in all, I like the idea/topic and the analyses. However, I am not convinced about how the paper is written. More time needs to be invested in the elaboration of the paper. I hope the hints above help to improve this paper. If the paper can be improved, it should have a chance. In this revision, more theoretical foundation is needed. I wish the authors luck to improve the paper!

Author Response

-- Abstract:

I would like to read a definition/explanation about what kind user-generated content one can expect in crowdfunding. What exactly is meant by managerial response in this context? Later in the paper, it would be important to describe how such texts looks like.

Response 1: Thanks for your suggestion and we have added it in our revised manuscript. User-generated content means “the volume of comments, linguistic features of comment text, and the length of comments.” Managerial response in this context is “responding rate, length, and speed.”

-- Introduction:

The authors find a good start into the topic. However, I miss references and evidence for several statements. Why does crowdfunding realize “the optimal resource allocation among different individuals”? Why are the authors so sure that the allocation is optimal? References/proofs are missing. Moreover, the authors say “User-generated content is significantly increasing in crowdfunding”. Again, references are missing. Is it really increasing? How do the authors know? How much has it increased over what time period?

Response 2: Thanks for your suggestion and we have corrected it in our revised manuscript. We have revised “the optimal resource allocation among different individuals” as “crowdfunding realizes dynamic resource allocation among different individuals”. And we have removed “User-generated content is significantly increasing in crowdfunding”.

I would like to read more about which problem the authors address and try to resolve. What exactly is the contribution of the paper? (Theoretical and practical.) The fact that social media, user-generated content, managerial responses are important in these days is well known – but what is special about the given case/analysis. This need to be explained.

Response 3: Thanks for your suggestion and we have added it. This paper studies how do managerial responses influence the following crowdfunding performance, and how a creator’s response style (e.g. responding rate, length, and speed) influences the following backers’ comment style.

The language of the paper needs to be revised, e. g., what is meant by “Understanding the effect … is significant”? See also below some examples of mistakes (section: Minor issues) – but these are not all language-related mistakes.

Response 4: Thanks for your suggestion and we have corrected the language in our revised manuscript.

-- Literature Review: 

The first sentences of the crowdfunding literature review are no good start into this section. Here, general information is given that I would expect to read earlier.

Response 5: Thanks for your suggestion and we have revised it as “our research builds on earlier work on crowdfunding, managerial responses and user-generated content.”

The paper is about user-generated content (UGC). But why do the authors only review crowdfunding papers without UGC? There are quite some crowdfunding papers that deal with the UGC. This is completely left out – but this is an important basis for the paper and the later analysis. Without a proper literature analysis, the reader cannot judge about the contributions of the paper.

Response 6: Thanks for your suggestion and we have added crowdfunding papers with UGC in “Section 2.2 User-generated Content and Managerial Responses”.

“It is useful to study how the relationship between user-generated content and crowdfunding performance and how this effect varies across different features of projects.” Is this a complete sentence? And why is it useful? Valuations always need supporting arguments. Why do the authors argue that social media and crowdfunding research is “still insufficient”?

Response 7: Thanks for your suggestion and we have corrected it in our revised manuscript. We have revised “still insufficient” as “A growing number of studies have confirmed the user-generated content has a significant influence on firm performance. While little researches have studied the relationship between managerial responses and user-generated content in crowdfunding, especially to different textual features.”.

Due to the lack of theoretical backing, we have removed the research about “the relationship between user-generated content and crowdfunding performance and how this effect varies across different features of projects”.

If I count right – only one paper (in the UGC section) is referred to which deals with social media and crowdfunding together. However, there are much more papers about this combination. The authors need to refine their literature review. Maybe the authors judge the research streams as “insufficient” because they have not invested much time about the literature review. The review is quite superficial and contains only very few findings. This should definitely be optimized.

Response 8: Thanks for your suggestion and we have added some related papers in the UGC section.

-- Theory and Hypothesis Development:

Until here, still no information is given what kind of UGC is prevalent on kickstarter. How does reviews and responses look like? I only now comments and updates on kickstarter. But “reviews” (product reviews?) as in Amazon (about product quality) are not known to me. Where are they available on Kickstarter?

Response 9: Thanks for your suggestion and we have revised “reviews” as “comments”. Comments come from backers and replies come from project creators.

“Audience driven is enhanced”? Please reformulate.

Response 10: Thanks for your suggestion and we have revised “Audience driven is enhanced” as “Audience-driven motivations are enhanced”.

I miss the theory behind authors conclusion that the interaction between the creator and the backers might lead to writing more reviews? Still, I don’t know where the reviews are – but if comments are meant, I do not know why backers should post more questions/requests when these have already been posted/discussed. The comments are normally used to discuss questions or problems. Why should they discuss more then? Another explanation is that there is more discussion because they like the project. Then this is just a side effect but no explanation of funding success. The authors need to deal with such ideas to better guide the reader.

Response 12: Thanks for your suggestion. We think that comments include not only questions or requests but also their opinions or attitudes to this project. Psychological motivations can increase the volume of online reviews. As a customer relationship management channel, managerial responses can satisfy customers' desire for social interaction and feel valued by managers. Creators who are willing to listen and interact can enhance the business trustworthiness and inspire backers to write reviews. That will encourage customers to engage more and write more comments to express their opinions. Backers feel obliged to express their opinions differently than what previous reviewers have done. Additionally, backers would also feel the need to describe the opinion in more detail.

To avoid this endogeneity problem, we revise our equation by decomposing its error term, and control for time-invariant unobservable project heterogeneity with project fixed effects .

The hypothesis part lacks theoretical backing. I would like to read more about why these dependencies are expected.

Response 13: Thanks for your suggestion and we have added some related researches and theoretical backing in our revised manuscript.

-- Research Design:

Why is it interesting to know the function used for the sentiment – if nothing else is said about how/where/in which program everything is done?

Response 14: Thanks for your suggestion and we have revised it in our revised manuscript. “Using the functions convertToBinaryResponse and convertToDirection with R 3.5.1, we detect the sentiment direction of reviews (Feuerriegel and Proellochs, 2018).”

The authors need to detail more on what they include as time-varying control variables: Z is not described at this place. How does date and day look like? And what is the last variable of the equations?

Response 15: Thanks for your suggestion and we have added it.

“ is a vector of time-varying control variables, which includes , the category of project i; , the number of collaborators; , the goal of project i; , whether or not this project had a video. and are the date and the duration of project i on its tth day, which are used to control time effect. is the time-invariant unobserved project heterogeneity. represents time-variant unobservable project heterogeneity.”

I doubt that the data set is free of selection bias. The authors state that 60% of the projects in their data set are financed. However, there should be a rate of 30-35% of financing on Kickstarter. I would prefer seeing the analyses with a data set that is more comparable to the real situation on Kickstarter. Have the authors compared their data set to the real numbers on Kickstarter? If not, they should do so.

Response 16: Thanks for your suggestion. Table 1 shows that 32.5% of the projects in our data set are financed. Table 5 shows that 68% of the projects with customer comments are financed.

-- Robustness Checks:

The authors introduce “interaction” terms. However, they miss to argue why they decide for this way of “robustness” check. How can the results be interpreted? The more money a project wants to collect, the worse is responsiveness for performance? This is not intuitive. The authors should explain it.

Response 17: Thanks for your suggestion and we have corrected it. We have revised our robustness checks as using an alternative variable, the incremental backers, to examine the relationship between managerial responses and crowdfunding performance.

-- Implication and Conclusion

The authors argue that “the dataset is not big enough” … big enough to do what? To have reliant results? Moreover, the authors think that “other control variables of project characteristics are not exhaustive”. What is missing? Why are they not exhaustive? Why don’t the authors solve this problem?

Response 18: Thanks for your suggestion. We have revised it as “There are a few limitations to our research that could be opportunities for exploration in the future. First, the dataset comes from a single platform. Further researches should collect data from various information channels and different online platforms. Second, future research examining how the relationship between managerial responses and user-generated content varies across different project characteristics could explore this further.”.

Author Response File: Author Response.pdf