基于网络搜索的销量与市场份额预测:来自中国汽车市场的证据  

Predicting Sales and Market Share with Online Search: Evidence from Chinese Automobile Market

作　　者：王炼[1,2] 宁一鉴[2] 贾建民[2,3] 

机构地区：[1]西南石油大学经济管理学院,四川成都610500 [2]西南交通大学经济管理学院,四川成都610031 [3]香港中文大学商学院,香港999077

出　　处：《管理工程学报》2015年第4期56-64,共9页Journal of Industrial Engineering and Engineering Management

基　　金：国家自然科学基金资助项目(71090402);西南石油大学青年教师"过学术关"计划资助项目

摘　　要：本文提出,网络搜索行为是一个衡量消费者购买意愿的指标,能够用于购买行为的预测。尽管信息搜索一直被看作消费者决策过程当中的重要一环,但是有关在线搜索与线下交易之间关系的实证研究有限。本文以消费者在搜集信息和购买决策过程中留下的真实数据为研究基础,探讨在线搜索与线下交易之间的关系。实证结果显示,网络搜索对汽车销量存在显著的正向影响。此外,本文基于fractional logit模型验证了消费者网络搜索对市场份额的影响。保留样本分析表明,在其他信息不可获取时,网络搜索数据能够对销量和市场份额预测起到重要作用。Consumers need information to make informed decisions. Information comes from various sources, including personal experiences, public sources, social networks, and online information retrieval tools. As the Web becomes more ubiquitous, consumers have free access to information at unprecedented breadth and depth, and more consumers are using the Internet to gather product-related information. The digital traces that consumers leave behind while doing online searches can be used to study consumer intentions.Literature review shows that although information search is regarded as an essential phase in the consumer decision process, empirical research on the relationship between online search and offline transaction is limited. Following previous studies on Web-based searches, the researchempirically investigates the issue using sales data and search volume data of 64 models of automobiles on sale in the Chinese automobile market to test the predictive power of consumer online search data. To account for heterogeneity of different automobile models and in conformity with traditional sales forecasts, automobile models＇ specific characteristics are included as covariates in the analysis. The research also considerstime-variant factors to account for the temporalvariance in panel data, such as purchase taxpolicies, oil prices, and seasonal effects. Firstly, theresults reveal that consumer online search has a significant effect on sales of automobiles. The incremental F-test shows that the difference between the proposed model and the control model is statistically significant, demonstrating that the inclusion of the search and lag of search measures can significantly increase the amount of explained variance compared with the traditional forecasting variables. It is shown that the inclusion of control variables does not change the direction and strength of the objective design coefficients. Secondly, the same results are confirmed by studying the market share of automobile brands based on a fractional logit model. Esti

关 键 词：网络搜索 销量预测 市场份额 汽车营销 

分 类 号：F272.1[经济管理—企业管理]