The Effect Of Service Providers' Reputation On The Revenue On The Crowdsourcing Platform

The prosperous development of the e-commerce platform has broke the traditional trading patterns,and significantly reduced the transaction cost,but also caused the information asymmetry between buyers and sellers and a bigger risk on online transactions.Reputation mechanism effectively reduces the uncertainty of online transactions,and establishes the trust between buyers and sellers.At present,there have been numerous essays studying the effect of online reputation on sales based on the traditional e-commerce platform,but due to problems such as data and research methods,the results are not consistent.Most studies only focus on the unreliable correlation of online reputation and sales,and can't prove the causal effect.Few studies estimate the causal effect of online reputation and sales,based on traditional retail markets.As the online crowdsourcing market has its unique"customization" production principle and the "reverse auction" trading mode,the causal effect of online reputation and sales in this field still worth studying.Therefore,from the perspective of service providers,this article studies the causal effect of online reputation and earnings based on crowdsourcing platform.A total of 7495 sellers'data is scraped from the Chinese largest service crowdsourcing platform(zbj.com),including reputation level,rating score and shop attributes on September 1 and October 1 as well as the corresponding monthly sales during September and October.Utilizing the discontinuity of the online reputation level,we conduct the sharp regression discontinuity design and estimate the causal effect of online reputation on online sales with nonparametric methods.All data is pooled together to estimate the average effect of the overall sample,and found that when online reputation scores cross the critical point,there will be a leap of 0.33 to 0.59 in the average logarithmic sales.Then,we clarify the efficacy of the RD estimates,by proving that the independent variable is not the manipulated and control variables are continuous.In addition,we also conduct a series of robustness tests,by setting different bandwidth,adding virtual time variables and control variables,checking the artifact thresholds as well as trying parametric methods.And all the research results prove that the estimation of the overall sample is robust.Furthermore,we divide the whole sample into several groups of subsamples,draw the RD estimates respectively,and find results are significantly different:(1)grouping by the reputation level,the overall causal effect of online reputation tend to increase and then decline as reputation scores rises,while there is significant causal effect on shops' sales with nearest level 4 to 5 and nearest level 6 to 7,the strongest causal effect on shops' sales with nearest level 6 to 7,and no significant causal effect on shops' sales with nearest level 8 to 10;(2)grouping by shop subjects,the online reputation has significant causal effect on individual shops,but no effect on company shops;(3)grouping by whether VIP or not,the online reputation has significant causal effect on non-VIP shops,but no effect on VIP shops;(4)grouping by whether joining employer guarantee or not,the online reputation has significant causal effect on shops without employer guarantee,but no effect on company shops with employer guarantee.