摘要
随着网络购物日益受到人们的欢迎,大型电商平台产品销售页面呈现的信息对消费者的购买决策具有越来越重要的影响。本文基于国内最大的B2C电商平台——天猫商城产品销售页面提供的丰富信息,重点关注了点评类网络口碑(Word-of-Mouth,WOM)与电商服务质量(e-commerce Service Quality,e-SQ)类在线观察学习(Observational Learning,OL)这两类销售页面信息。借助一个时间跨度为20周的笔记本样本数据,本文运用面板向量自回归模型研究了产品销量、点评类WOM与e-SQ类OL信息之间的动态交互影响,并运用脉冲响应函数与预测误差方差分解方法,进一步分析了上述三者之间的相互预测能力。研究发现:(1)产品周销量与点评类WOM信息(点评量变化与正面点评标签百分比)之间均具有动态交互影响;与e-SQ类OL信息——"描述相符"之间只具有动态单向影响,与e-SQ类OL信息——"退款速度"之间不存在动态交互影响。(2)除了自身冲击外,正面点评标签百分比冲击对产品周销量波动的解释能力较大(从长期来看,超过30%);产品周销量冲击对点评类WOM信息波动具有一定的解释能力;导致e-SQ类OL信息("描述相符"与"退款速度")波动的因素几乎全部源于其自身。这些研究发现为大型B2C电商平台上的网络卖家提供了一些有价值的管理启示。
With online purchase becoming increasingly popular, information shown on product sales pages from large e-business platform is playing a more and more important role in consumers' purchase decisions. This paper mainly concerns two typical categories of information, online word-of-mouth (WOM) and e-commerce service qual- ity (e-SQ) based on observational learning (OL), appearing on the product sales page of the largest B2C e-busi- ness platform in china-tmall, corn whose product sales pages provide rich information. This paper focuses on explo- ring the dynamic interactive effects among product sales, online WOM and e-SQ based on OL information, and fur- ther investigating the forecast capability between each other through impulse response function analysis and forecast error variance decomposition. To answer these questions, we collect a unique panal data with a span of 20 weeks only containing notebooks from July 29th to December 09th, 2013. After removing the invalid and abnormal samples, this data contains a total of 3820 valid samples, including 191 notebook products. This paper adopts panel vector auto-regression (PVAR) model to deal with the data. This model is used exclusively to study the dynamic relationships between multiple en- dogenous variables. It can not only reveal the dynamic characteristic of time series of single variable, but also the dynamic relationships between different variables, which is suited to the problems of this paper well. This paper use the volume changes of reviews and the percentage of positive reviews labels to measure WOM information, and the consistency of the description on page information to actual situation ( "description consisten- cy" ) and the refund speed of sellers ( "refund speed" ) to measure OL information about e-SQ. The findings show that: (1) Product week sales have dynamic interactive effects with online review WOM variables (that is, the vol- ume change of reviews and the percentage of positive reviews labels) , just have dynamic unidirectional effect with e-SQ based on OL information-- "description consistency", and have no dynamic interactive effect with e-SQ based on OL information-- "refund speed". (2) Besides its own impulse, the impulse of the percentage of positive reviews labels attributes the largest share to the fluctuation of product week sales (in the long term, more than 30% ). The impulse of product week sales has a certain explanatory capability to the fluctuation of online reviews WOM information. The main factor of the fluctuation of e-SQ based on OL information ( that is, "description consis- tency" and "refund speed") is almost entirely from its own. These findings provide several valuable managerial insights for online sellers on the large B2C e-commerce platform. Firstly, the exaggerated or false page information may bring about the increase of product sales in the short term, but not favor of the sellers' revenue in the long term. So the online sellers should improve the description confirmation of product information. Secondly, online sellers should pay attention to monitoring the trend of the changes about the number of positive and negative reviews, prevent the spread of these negative comments in more consumers and guide the follow-up consumers to write more positive reviews. This study has some limitations need to be further explored. Firstly, the empirical sample only collects one kind of product (that is, notebook). It means that the universality of the conclusions in this paper should be checked up further. Secondly, PVAR model used in this paper doesn' t control other exogenous variables. Adding some important exogenous variables will improve the effectiveness of the conclusions in this paper.
出处
《经济管理》
CSSCI
北大核心
2016年第2期91-101,共11页
Business and Management Journal ( BMJ )
关键词
网络口碑
在线观察学习
电商服务质量
面板向量自回归
online word-of-mouth
online observational learning
e-commerce service quality
panel vector auto-regression