Evolution of Consumption Statistics Driven by Big Data

Transcription

1 Evoluion of Consumpion Saisics Driven by Big Daa - An Example from Minisry of Inernal Affairs and Communicaions, JAPAN - June 2018, Beijing Tomoaki OGAWA A fellow from he Governmen of Japan SDG Monioring Secion/ Developmen Daa and Oureach Branch Saisics Division/ Deparmen of Economic and Social Affairs/ UN

2 Srucure of Consumpion Trend Index (CTI) - Japanese governmen has conduced Family Income and Expendiure Survey (FIES) since Is purpose is o invesigae he srucure of household budge. On he oher hand, marke players and invesors have used he saisics produced by FIES o grasp he rend of he consumpion and predic he economic performance of Japan. (JPY) (Persons) 320, Consumpion expendiure per household (Lef) 300, Consumpion Trend Index (CTI) comprises: imely macro rend (Toal Consumpion Trend Index) mainly for marke players; coverall micro rend (Household Consumpion Trend index) mainly for researchers; and ohers. Consumpion Trend Index 1 Toal Consumpion Trend Index Esimaing he monhly rend of Household Final Consumpion Expendiure of SNA by applying he sae space model 2 Household Consumpion Trend Index Esimaing he monhly rend of Household Expendiure by enhancing he daa from FIES 3 Oher Consumpion Trend Esimaing he amoun of gifs from companies o cliens and cusomers (Japanese radiional business cusom) and consumpion of foreign visiors, ec. by using Big Daa 280,000 0 (million households) Average number of household members (righ) Toal number of households

3 Time Ouline of Toal Consumpion Trend Index - The consumpion rend of an enire sociey is esimaed by using ime series analysis (he sae space model). - Timely esimaion was achieved by adding Big Daa (uilized hrough he bias correcion) o he daa sources. - CTI Research Consorium launched on 28 July 2017 o promoe he use of Big Daa generaed by privae secors. Esimaion by Using he Sae Space Model - By applying he sae space model, one of he echniques of ime series analysis, he final consumpion expendiure of households, one par of GDP, is esimaed as Toal Consumpion Trend Index. - Toal Consumpion Trend Index is esimaed monhly, while GDP iself is esimaed quarerly. x = Fx 1 y = Hx + u + Gv log Eliminaing Bias and Unifying he Various Kind of Big Daa - Survey daa can be uilized for eliminaion of he bias of Big Daa and can unify some Big Daa from various resources. - Toal Consumpion Trend Index made of he unified daa can provide imely esimaion. Eliminaing Bias BIGDATA Unifying Daa Survey Daa Survey Daa Populaion Unified Daa 2

4 Ouline of Household Consumpion Trend Index - Expendiure Monior Survey for One-Person Households launched on Aug FIES was improved on Jan Household Consumpion Trend Index has been developed as a comprehensive saisics covering enire household consumpion by using FIES, FIES for Expensive Iems and Expendiure Monior Survey for One-Person Households. - Accuracy of Household Consumpion Trend Index will be improved when Big Daa is pracically available. Improvemen of FIES - Online diary, enhancemen of he measuremen of cashless ransacion and oher improvemen of FIES were adoped on January Expendiure Monior Survey for One-Person Households - The sample of former FIES consised of 8,000 households wih more han one person and 800 single-person household. Hence, saisics of single-person households could be released quarerly. - To produce an overarching saisics for every monh, Expendiure Monior Survey for One-Person Households launched on Augus Pracical analysis - Household Consumpion Trend Index is made of: FIES, FIES for Expensive Iems, Expendiure Monior Survey for One-Person Households, ec. Medium- o long- erm New Daa Source Pas Source Big Daa Furher Developmen Households wih more han one person FIES for Expensive Iems BIGDATA FIES POS daa Single-Person Households Expendiure Monior Survey for One-Person Households Elecronic Paymen, Credi Cards, ec. MEMBER S CARD Accuracy will be improved when Big Daa is pracically available. 3 SOUMU TARO

5 Esablishmen of CTI Research Consorium (from 28 July 2017) CTI Research Consorium MEMBER S CARD SOUMU TARO Privae Companies (23 Daa Holders) Observer As an advisor o he projec Saisics Bureau, Saisical Research and Training Insiue, and Naional Saisics Cener Academics * who mus keep secrecy Shor Term(2017FY) Medium Term(2018FY FY) Long Term The Firs Sep Grasping characerisic of each Big Daa Grasping characerisics of daa which is volunarily provided from each company Developing he compuaion mehod, hrough he research for bias correcion, impuaion, ec. The Second Sep Trial producion of new index by Big Daa Developing new index hrough addiional rials wih pas daa The Third Sep Regular release of new index based on Big Daa Organizing a eam for releasing index on a regular basis Daa Holders comprises: loyaly card companies, credi card companies, giganic reail companies, railway companies, makers of APP for elecronic housekeeping book and daa providers of poin-of-sale. 4

6