利用网络搜索大数据实现对CPI的短期预报及拐点预测——基于混频抽样数据模型的实证研究
|
摘要
消费者价格指数(CPI)对国家宏观经济运行监测及管理具有十分重要的作用,及时有效预测其未来走势有利于国家及时采取合理调控措施。为此,从价格决定理论出发,构建商品价格波动与个体网络搜索行为之间关系的逻辑框架,并建立利用网络搜索指数日频率数据实现对月度CPI预测的混频抽样数据模型(MIDAS)。模型模拟结果表明:(1)利用网络搜索高频数据的预测方法能提高CPI样本内拟合及样本外预测的精度;(2)利用网络搜索高频数据能够提高捕捉CPI趋势中"拐点"的成功率;(3)利用网络搜索高频数据能够在早于官方公布数据前大约半个月时间提供具有较高预测精度的CPI预测值。
The CPI index plays an important role in monitoring and managing the operation of a country's macro economy, and the timely and effective prediction of its future trend is conducive for the country to take reasonable regulatory measures in time. Therefore, starting from the price determination theory, this paper constructs a logical framework for the relationship between commodity price volatility and individual web search behavior, then it establishes a mixed-frequency data sampling model(MIDAS) for monthly CPI prediction by using the daily frequency data of network search index. The findings show that:(1) the prediction method using web searching high frequency data can improve the accuracy of both the inside matching and outside prediction of CPI samples;(2) the high frequency data using web search can improve the success rate of capturing the "inflection point" in the CPI trend;(3) the high frequency data using web searching can provide the value of CPI forecasting with higher prediction accuracy approximately half a month before the official release of CPI data.
The CPI index plays an important role in monitoring and managing the operation of a country's macro economy, and the timely and effective prediction of its future trend is conducive for the country to take reasonable regulatory measures in time. Therefore, starting from the price determination theory, this paper constructs a logical framework for the relationship between commodity price volatility and individual web search behavior, then it establishes a mixed-frequency data sampling model(MIDAS) for monthly CPI prediction by using the daily frequency data of network search index. The findings show that:(1) the prediction method using web searching high frequency data can improve the accuracy of both the inside matching and outside prediction of CPI samples;(2) the high frequency data using web search can improve the success rate of capturing the "inflection point" in the CPI trend;(3) the high frequency data using web searching can provide the value of CPI forecasting with higher prediction accuracy approximately half a month before the official release of CPI data.
引文
[1]崔百胜.基于动态模型平均的中国通货膨胀实时预测[J].数量经济技术经济研究, 2012,(7):76-91.
[2]程建华,任婷.我国通货膨胀及其转折点研究的理论与现实问题[J].经济问题探索, 2014,(9):10-16.
[3]汤丹.基干SVAR模型的中国核心通货膨胀估计及预测评价研究[J].宏观经济研究, 2015,(1):109-115.
[4]赵静,方兆本,朱俊鹏.公司债信用利差对产出和通胀的预测——基于BMA模型的研究[J].数理统计与管理,2018,(1):179-190.
[5]刘金全,张达平,张都.通货膨胀波动与货币政策调控机制研究——基于TVP-VAR模型的实证分析[J].当代财经, 2016,(4):51-60.
[6]许文立,程建华.我国通胀预期的动态特征及其政策启示,[J].当代财经, 2017,(5):14-23.
[7]张崇,吕本富,彭赓.网络搜索数据与CPI的相关性研究[J].管理科学学报, 2012,(7):50-59.
[8]张涛,刘宽斌.“大数据”在宏观经济预测分析中的应用[J].财经智库, 2018,(3):65-83.
[9]孙毅,吕本富,陈航.大数据视角的通胀预期测度与应用研究[J].管理世界, 2014,(4):171-172.
[10]卢洪涛,李纲.网络搜索关键词“稳定—突变”特征的识别——基于关键词集中度变化率的分析[J].图书情报知识, 2014,(3):80-86.
[11]袁铭.基于网购搜索量的CPI及时预测模型[J].统计与信息论坛, 2015,(4):20-27.
[12]董莉,彭凯越,唐晓彬.大数据背景下的CPI实时预测研究[J].调研世界, 2017,(8):51-54.
[13]Guzman G.. Internet Search Behavior as an Economic Forecasting Tool:The Case of Inflation Expectations[J].Journal of Economic&Social Measurement, 2011, 36(3):4187-4199.
[14]Li X., Shang W., Wang S.. A MIDAS Modelling Framework for Chinese Inflation Index Forecast Incorporating Google Search Data[J]. Electronic Commerce Research&Applications, 2015, 14(2):112-125.
[15]董倩.基于网络搜索数据的雾霾经济与CPI相关性研究[J].调研世界, 2016,(12):58-60.
[16]Ghysels E., Santa-Clara P., Valkanov R.. The MIDAS Touch:Mixed Data Sampling Regressions[R]. Cirano Working Papers, 2004, 5(1):512-517.
[17]Clements M. P., Galvao A. B.. Macroeconomic Forecasting With Mixed-Frequency Data[J]. Journal of Business&Economic Statistics, 2008, 26(4):546-554.
[18]刘金全,刘汉,印重.中国宏观经济混频数据模型应用——基于MIDAS模型的实证研究[J].经济科学, 2010,(5):23-34.
[19]Kuzin V., Marcellino M., Schumacher C.. MIDAS vs. Mixed-Frequency VAR:Nowcasting GDP in the Euro Area[J]. International Journal of Forecasting, 2011, 27(2):529-542.
[20]Ghysels E., Valkanov R.. Forecasting Volatility with MIDAS[M]//Bauwens L., Hafner C., Laurent S.. Handbook of Volatility Models and Their Applications. New Jersey:John Wiley&Sons, Inc, 2012.
[21]Ghysels E., Santa-Clara P., Valkanov R.. Predicting Volatility:Getting the Most out of Return Data Sampled at Different Frequencies[J]. Journal of Econometrics, 2006, 131(1-2):59-95.
[22]Forsberg L., Ghysels E.. Why Do Absolute Returns Predict Volatility So Well?[J]. Social Science Electronic Publishing, 2007, 5(1):31-67.
[23]Andreou E., Ghysels E., Kourtellos A.. Should Macroeconomic Forecasters Use Daily Financial Data and How?[J]. Journal of Business&Economic Statistics, 2013, 31(2):240-251.
[24]刘汉,刘金全.中国宏观经济总量的实时预报与短期预测——基于混频数据预测模型的实证研究[J].经济研究,2011,(3):4-17.
[25]王维国,于扬.基于混频回归类模型对中国季度GDP的预报方法研究[J].数量经济技术经济研究, 2016,(4):108-125.
(1)数据来源:Stat Counter,它是美国一家网站通讯流量监测机构,提供各种类型的统计报告以及网站流量统计服务。网址http://gs.statcounter.com/.
(1)数据来源:《第41次中国互联网络发展状况统计报告》,年份之间的月份数据,基于匀速增长假设插值补全。
(2)数据来源:数据来源:Stat Counter。
(1)本文CPI “拐点”表示CPI值在当期前后变化方向发生变化,例如t-1期到t期升高(降低),而t期到t+1降低(升高),则t期为一个“拐点”,除此之外均不为“拐点”。
[2]程建华,任婷.我国通货膨胀及其转折点研究的理论与现实问题[J].经济问题探索, 2014,(9):10-16.
[3]汤丹.基干SVAR模型的中国核心通货膨胀估计及预测评价研究[J].宏观经济研究, 2015,(1):109-115.
[4]赵静,方兆本,朱俊鹏.公司债信用利差对产出和通胀的预测——基于BMA模型的研究[J].数理统计与管理,2018,(1):179-190.
[5]刘金全,张达平,张都.通货膨胀波动与货币政策调控机制研究——基于TVP-VAR模型的实证分析[J].当代财经, 2016,(4):51-60.
[6]许文立,程建华.我国通胀预期的动态特征及其政策启示,[J].当代财经, 2017,(5):14-23.
[7]张崇,吕本富,彭赓.网络搜索数据与CPI的相关性研究[J].管理科学学报, 2012,(7):50-59.
[8]张涛,刘宽斌.“大数据”在宏观经济预测分析中的应用[J].财经智库, 2018,(3):65-83.
[9]孙毅,吕本富,陈航.大数据视角的通胀预期测度与应用研究[J].管理世界, 2014,(4):171-172.
[10]卢洪涛,李纲.网络搜索关键词“稳定—突变”特征的识别——基于关键词集中度变化率的分析[J].图书情报知识, 2014,(3):80-86.
[11]袁铭.基于网购搜索量的CPI及时预测模型[J].统计与信息论坛, 2015,(4):20-27.
[12]董莉,彭凯越,唐晓彬.大数据背景下的CPI实时预测研究[J].调研世界, 2017,(8):51-54.
[13]Guzman G.. Internet Search Behavior as an Economic Forecasting Tool:The Case of Inflation Expectations[J].Journal of Economic&Social Measurement, 2011, 36(3):4187-4199.
[14]Li X., Shang W., Wang S.. A MIDAS Modelling Framework for Chinese Inflation Index Forecast Incorporating Google Search Data[J]. Electronic Commerce Research&Applications, 2015, 14(2):112-125.
[15]董倩.基于网络搜索数据的雾霾经济与CPI相关性研究[J].调研世界, 2016,(12):58-60.
[16]Ghysels E., Santa-Clara P., Valkanov R.. The MIDAS Touch:Mixed Data Sampling Regressions[R]. Cirano Working Papers, 2004, 5(1):512-517.
[17]Clements M. P., Galvao A. B.. Macroeconomic Forecasting With Mixed-Frequency Data[J]. Journal of Business&Economic Statistics, 2008, 26(4):546-554.
[18]刘金全,刘汉,印重.中国宏观经济混频数据模型应用——基于MIDAS模型的实证研究[J].经济科学, 2010,(5):23-34.
[19]Kuzin V., Marcellino M., Schumacher C.. MIDAS vs. Mixed-Frequency VAR:Nowcasting GDP in the Euro Area[J]. International Journal of Forecasting, 2011, 27(2):529-542.
[20]Ghysels E., Valkanov R.. Forecasting Volatility with MIDAS[M]//Bauwens L., Hafner C., Laurent S.. Handbook of Volatility Models and Their Applications. New Jersey:John Wiley&Sons, Inc, 2012.
[21]Ghysels E., Santa-Clara P., Valkanov R.. Predicting Volatility:Getting the Most out of Return Data Sampled at Different Frequencies[J]. Journal of Econometrics, 2006, 131(1-2):59-95.
[22]Forsberg L., Ghysels E.. Why Do Absolute Returns Predict Volatility So Well?[J]. Social Science Electronic Publishing, 2007, 5(1):31-67.
[23]Andreou E., Ghysels E., Kourtellos A.. Should Macroeconomic Forecasters Use Daily Financial Data and How?[J]. Journal of Business&Economic Statistics, 2013, 31(2):240-251.
[24]刘汉,刘金全.中国宏观经济总量的实时预报与短期预测——基于混频数据预测模型的实证研究[J].经济研究,2011,(3):4-17.
[25]王维国,于扬.基于混频回归类模型对中国季度GDP的预报方法研究[J].数量经济技术经济研究, 2016,(4):108-125.
(1)数据来源:Stat Counter,它是美国一家网站通讯流量监测机构,提供各种类型的统计报告以及网站流量统计服务。网址http://gs.statcounter.com/.
(1)数据来源:《第41次中国互联网络发展状况统计报告》,年份之间的月份数据,基于匀速增长假设插值补全。
(2)数据来源:数据来源:Stat Counter。
(1)本文CPI “拐点”表示CPI值在当期前后变化方向发生变化,例如t-1期到t期升高(降低),而t期到t+1降低(升高),则t期为一个“拐点”,除此之外均不为“拐点”。