Comparative Study On CPI Compilation With Scanner Data

With the development of networking, cloud computing and other technologies,the importance of the role of big data is growing; researchers at home and abroad aremaking deepen research in the application of big data. The scanner data are structureddata and can be a new source of CPI. Comparing with the traditional manualcollecting method, scanner data have great advantages in the compilation of CPI.Foreign statistics departments attach great importance to this new method. In order toprovide reference for government statistics department using scanner data in CPI, theexperience of four countries using scanner data, such as the collection of price, thechoice of representative products and the compilation of the index is studied. Besides,this paper selects the cheese and beer groups as the research objects and compiles theelementary aggregates indexes using the Jevons and T rnqvist compilation equations，and the compilation results are compared with the RYGEKS index which are basedwith T rnqvist and Fisher price index, and are regarded as benchmark index. Theresults are as follows: T rnqvist-RYGEKS index has an upward bias comparing withFisher-RYGEKS index; monthly chained Jevons price index has a downward biascomparing with two types RYGEKS index, and the bias is large; but monthly chainedT rnqvist index has a relatively small deviations comparing with two types RYGEKS,but the direction is not clear. Whether the bias is downward or upward, not onlyrelates to the choice of compiling formula, but also the character of the product. Inaddition, according to the developed countries’ compilation experience and theresearch results, this paper made recommendations in six areas about the applicationof scanner data in China.