Item replacement and quality change in apparel price indexes

Monthly Labor Review

Volumn 129, Issue 12, 2006, Pages

  Brown, Craig ^a   Stockburger, Anya ^a  

^a BUREAU OF LABOR STATISTICS   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 33847073386     PISSN: 00981818     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (32)

1. See Chapter 17 of the BLS Handbook of Methods (Bureau of Labor Statistics, 2006) for a complete discussion of the construction of the CPI and the role of commodity analysts.
2. The Bureau took a significant step forward in improving the quality of microlevel data used to compute the CPI when it began using computer-assisted data collection in 2002. Prior to that changeover, the collection, transmission, and transcription of data were subject to a large degree of human error. Now, automatic checks and reviews help improve the completeness and accuracy of CPI data.
3. The Bureau uses the term "experimental" in contrast to "official" to denote statistics that it produces outside of its regular production systems and, consequently, with less than full production quality. For security reasons, BLS researchers cannot produce experimental statistics until after the publication of the corresponding official statistics. To obtain experimental series referred to in this article, contact either of the authors.
4. See Dennis Fixler, Charles Fortuna, John Greenlees, and Walter Lane, "The Use of Hedonic Regressions to Handle Quality Change: The Experience in the U.S. CPI," paper presented at the Fifth Meeting of the International Working Group of Price Indices, Reykjavik, Iceland, August 1999, pp. 6-9, for a review of the difficulties inherent in constructing accurate price indexes for apparel items.
5. In September 2005, the apparel sample of the CPI consisted of 6,316 priced quotes. At that time, 4,018 quotes represented unique items that were identified as being available for sale year round (nonseasonally). The remaining 2,298 seasonal quotes represented unique items that were identified as being available for sale only during certain months of the year.
6. Modern research on hedonic theories and methods generally starts with Zvi Griliches, "Hedonic Price Indexes for Automobiles: An Econometric Analysis of Quality Change," in The Price Statistics of the Federal Government, General Series No. 73 (New York, Columbia University Press, 1961).
7. For more information on early work with hedonic regression models for apparel items in the CPI, see Efthemia V. Georges and Paul R. Liegey, Jr., An Examination Using Hedonic Regression Techniques to Measure the Effects of Quality Adjustment on Apparel Indexes, internal report (Bureau of Labor Statistics, 1988);
8. Paul A. Armknecht and Donald Weyback, "Adjusting for Quality Change in the U.S. Consumer Price Index," Journal of Official Statistics, June 1989, pp. 107-23;
9. Paul R. Liegey, Jr., "Adjusting Apparel Indexes in the Consumer Price Index for Quality Differences," in Murray F. Foss, Marilyn E. Manser, and Allan H. Young, Price Measurements and Their Uses, National Bureau of Economic Research Studies in Income and Wealth, 57 (Chicago, University of Chicago Press, 1993), pp. 209-26;
10. Paul R. Liegey, Jr., "Apparel price indexes: effects of hedonic adjustment," Monthly Labor Review, May 1994, pp. 38-45;
11. and Nicole Shepler, Analysis of Hedonic Regression: Applied to Women's Apparel in the Consumer Price Index, internal manuscript (Bureau of Labor Statistics, 1994).
12. The number of apparel item replacement scenarios reported during the 3 years of research for this article (34,788) includes 948 scenarios (2.7 percent) that are not technically item replacements. These scenarios are referred to as reinitiations and are the result of resampling the price quotes and selecting replacement items through the initiation procedures determined by the Bureau. Reinitiations differ from item replacements in that data collectors are not directed to select the next-most-similar item to sample when reinitiating a price quote. Data collectors are directed to reinitiate quotes for a number of reasons, most notably when an item is ineligible in the specific category being sampled or when a category is selected for item rotation (a method of keeping the sample current by periodically reinitiating the quotes). Commodity analysts are required to review reinitiations in the same manner that they review item replacements. Reinitiations are included in this analysis because they share many characteristics with item replacements in terms of how they are reviewed.
13. Class-mean imputation is explained in greater detail in the BLS Handbook of Methods, Chapter 17, p. 23.
14. Cell-relative imputation for noncomparable item replacements is used primarily in the foods and services sectors of the CPI. The method also is used to impute a price relative for nonreplacement items for which prices are unavailable during a collection period. Cell-relative imputation is explained further in the BLS Handbook of Methods, Chapter 17, p. 23.
15. A "hedonic price index" is often defined as an index that uses prices imputed directly from a hedonic regression, either as a time dummy coefficient or by using the coefficients of the characteristics variables to impute a value for the dependent variable. The hedonic adjustment method used by the Bureau is what Mick Silver and Saeed Heravi call "patching" and simply means that adjustments for quality differences are made to noncomparable models and the adjusted, patched" price is used for price comparisons. (See Mick Silver and Saeed Heravi, "Scanner Data and the Measurement of Inflation," Economic Journal, June 2001, pp. 384-405.)
16. A complete academic description of quality adjustment and conventional price index methodologies can be found in Jack Triplett, Handbook on Hedonic Indexes and Quality Adjustments in Price Indexes: Special Application to Information Technology Products, OECD Science, Technology and Industry Working Papers, No. 9 (Geneva, OECD Publishing, 2004).
17. Triplett, Handbook on Hedonic Indexes, notes that the direct comparison method may be the appropriate way of handling changes in quality under the "most similar item" replacement rule employed by the Bureau. The rule instructs its data collectors to choose item replacements with characteristics similar to those of the item that "disappeared."
18. The experimental indexes were calculated with an in-house program that mimics the functions of the official CPI algorithm. This approach represents a noteworthy improvement over previous BLS experimental indexes used for research purposes, which were limited in their ability to replicate the official algorithm.
19. The distinction between upward and downward discrepancies is made in Jack Triplett, "Determining the effects of quality change on the CPI," Monthly Labor Review, May 1971, pp. 27-32 (see especially note to table 4, p. 30),
20. and in Liegey, "Apparel price indexes," p. 41. Borrowing from Liegey, an "upward discrepancy" means that the experimental index rises more than the official price index if prices are rising or falls less than the official index if prices are falling. Conversely, a "downward discrepancy" means that the experimental index rises less than the official price index if prices are rising or falls more than the official index if prices are falling.
21. Michael J. Boskin, Ellen R. Dulberger, Roberti. Gordon, Zvi Griliches, and Dale W. Jorgenson, Final Report of the Advisory Commission to Study the Consumer Price Index (Washington, DC, U.S. Government Printing Office, 1996).
22. David E. Lebow and Jeremy B. Rudd, "Measurement Error in the Consumer Price Index: Where Do We Stand?" Journal of Economic Literature, March 2003, pp. 159-201.
23. Charles R. Hulten, "Quality Change in the CPI," Federal Reserve Bank of St. Louis Review, May/June 1997, pp. 87-111;
24. and Bart Hobijn, "On Both Sides of the Quality Bias in Price Indexes," Federal Reserve Bank of New York Staff Reports, no. 157, December 2002, pp. 1-40. Neither of these articles offers an estimate of quality change bias specifically for the apparel item sector of the CPI.
25. Robert J. Gordon, "Apparel Prices 1914-93 and the Hulten/Brueghel Paradox," paper presented at the CRIW Conference on Price Index Concepts and Measurement, Vancouver, British Columbia, Canada, June 28-29, 2004.
26. Liegey, "Apparel price indexes."
27. Before the substitution relative procedure began in 1992, the apparel section used a linking method to ensure that apparel items did not exit the sample while on sale. Instead of using cell-relative imputation, analysts compared an overlap price estimated from the old item's last collected regular price with the new item's price.
28. B. Peter Pashigian and Brian Bowen, "Why Are Products Sold on Sale?: Explanations of Pricing Regularities," Quarterly Journal of Economics, November 1991, pp. 1015-38.
29. Liegey, "Apparel price indexes."
30. Liegey recalculated indexes for 7 item categories (3 men's apparel items and 4 women's apparel items), 2 aggregate indexes (men's and women's apparel), and an aggregate apparel index, for a total of 10 experimental, nonhedonic indexes. The analysis presented herein calculates experimental indexes for 10 item categories (3 men's apparel items, 3 women's apparel items, 1 boys' apparel item, 1 girls' apparel item, and 2 footwear items), 3 aggregate indexes (men's apparel, women's apparel, and footwear), and an aggregate apparel index, for a total of 14 experimental, nonhedonic indexes.
31. Liegey, "Apparel price indexes."
32. Triplett, "Determining the effects of quality change."