互联网大数据下消费搜索指数的构建与应用  

Construction and Application of Consumption Search Index Under Internet Big Data

作　　者：曾洁华 钟若愚[3] Zeng Jiehua;Zhong Ruoyu(School of Economics and Finance,Xi’an Jiaotong University,Xi’an 713599,China;Qianhai FOF(Shenzhen),Shenzhen Guangdong 518038,China;College of Economics,Shenzhen University,Shenzhen Guangdong 518061,China)

机构地区：[1]西安交通大学、经济与金融学院,西安713599 [2]前海方舟(深圳)资产管理有限公司,广东深圳518038 [3]深圳大学、经济学院,广东深圳518061

出　　处：《统计与决策》2023年第4期16-22,共7页Statistics & Decision

摘　　要：文章结合2013年1月1日至2020年6月30日的百度指数这一高频互联网大数据,利用主成分分析法构建高频居民消费搜索指数。在此基础上,将高频搜索指数拓展进ADL-MIDAS混频模型,对居民消费情况展开实证分析和预测。研究发现:所构建的消费搜索指数能较好捕捉居民在双十一、春节、新冠肺炎疫情期间的潜在消费行为变动情况,总体上,居民对消费相关事件的网络关注度变化会对居民消费水平产生显著影响,消费存在显著的耐久效应,收入增长对消费的促进作用明显。此外,引入高频消费搜索指数的混频模型,对比基准模型,在多步向前动态预测和实时预测上都呈现较高的预测精度和较好的稳定性。Combined with high-frequency Internet big data drawn from the Baidu Index for the period from January 1, 2013to June 30, 2020, this paper uses the principal component analysis method to construct a high-frequency consumer search index.On this basis, the paper extends the high frequency search index into ADL-MIDAS mixing model, and carries out the empirical analysis and forecast of residents’ consumption situation. The study finds that the constructed consumption search index performs well in capturing potential consumer behavior changes during Double Eleven(i.e., November 11, also known as Singles’ Day),Spring Festival, and the COVID-19 pandemic period.In general, changes in the degree of consumers’ attention to consumptionrelated events will have a significant impact on their consumption;consumption has a significant durable effect, and income growth plays an obvious role in promoting consumption.In addition, with the mixing model of high frequency consumption search index introduced, compared with the benchmark model, higher prediction accuracy and better stability are shown in multi-step forward dynamic prediction and real-time prediction.

关 键 词：互联网大数据 百度指数 消费指数 混频数据回归 居民消费 

分 类 号：F014.5[经济管理—政治经济学]