Modernizing the Consumer Price Index for the 21st Century (2022) / Chapter Skim
Currently Skimming:
3 Higher-Level Aggregation and Shifting Consumer Behavior
3 Higher-Level Aggregation and Shifting Consumer Behavior
Pages 63-84
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 63... ...
carries out this process using the Consumer Ex penditure Survey (CE) , which has for decades been the most comprehensive 1 As discussed below, Statistical Offices in many countries use national accounts data to estimate upper-level CPI weights, and data from household expenditure surveys to derive the more detailed weights (Eurostat, 2018)
|
From page 64... ...
Similarly, the new CPI weights in 2022 will be based on expenditure patterns from 2019 and 2020. As a result, the CPI weights tend to be outdated in representing consumer purchases taking place in a given period; for example, it is known that consumption behavior in 2020 changed sharply due to COVID-19 and so weights derived from 2019 to 2020 may not reflect what is going on in the market in subsequent periods.
|
From page 65... ...
, a modestly streamlined questionnaire, and an increased survey sample size -- the CE based method of establishing expenditure weights remains problematic. The survey generates high respondent burden and has experienced declining response rates.
|
From page 66... ...
SOURCE: Panel-generated, using BEA data. SOURCE: Panel-generated, using BEA data.
|
From page 67... ...
260) , while concluding that research at the time was too in complete to propose solid recommendations about CE sample size, offered analysis of the impact that "changing the survey sample size would have on the accuracy of expenditure weights and, in turn, the relationship between weight accuracy and index variance." A00858 -- Consumer Price Index REV.indd 67 8/15/22 11:54 AM
|
From page 68... ...
A preliminary version of the C-CPI-U uses a constant expenditure shares formula.9 The final chained CPI uses a Törnqvist formula, which accounts for consumer substitution and is the superior index for most applications where index revisions are acceptable. This Törnqvist formula is applied and the index is revised after monthly weights become available.
|
From page 69... ...
A00858 -- Consumer Price Index REV.indd 69 8/15/22 11:54 AM
|
From page 70... ...
use the national accounts approach for the higher-level weights, and consumer expenditure surveys for the lower level weights within each higher-level aggregate. Many other NSOs use a similar methodology.
|
From page 71... ...
They are recommending a variety of source data, including national accounts, to estimate expenditure shares.17 Although somewhat different from their consumer survey counter parts, PCE data underlying the national accounts also come with chal lenges in terms of their applicability to estimating CPI weights. PCE and CPI item categories are not all comparably defined.
|
From page 72... ...
A00858 -- Consumer Price Index REV.indd 72 8/15/22 11:54 AM
|
From page 73... ...
. 22 See "Consumer expenditures during COVID-19: An exploratory analysis of the effects of changing consumption patterns on consumer price indexes." www150.statcan.gc.ca/n1/ pub/62f0014m/62f0014m2020010-eng.htm; and "Adjusting the Consumer Price Index to the new spending realities during the pandemic," The Daily (Oct.
|
From page 74... ...
In the context of market disruptions during COVID-19, Reinsdorf (2020) recommended data blending procedures for updating CPI weights to ensure that they are less distorted by rapid changes in expenditure patterns exhibited by consumers brought on by extraordinary circumstances.26 As illustrated in the Statistics Canada example above, one such approach would use credit card and other payments data to produce a complementary index of short-term price change with weights reflecting spending patterns during the pandemic.
|
From page 75... ...
Updating the weights every year can be done using the CE data that BLS already collects.27 Under this setup, the rolling weights would still lag real 27 In the near term, one reason for continuing to use expenditure estimates based on 2 years of data is to maintain sample sizes needed to support metropolitan statistical areas level price indexes. A00858 -- Consumer Price Index REV.indd 75 8/15/22 11:54 AM
|
From page 76... ...
Ultimately -- in 10 to 20 years -- even more detailed price and quantity information will be available for use in the CPI from a range of nonsurvey sources that could allow much more frequent updating of item weights. In the intermediate run, BLS could develop a hybrid approach to estimating consumers' expenditure patterns that includes national accounts and credit card transactions data that could allow the reference period for weight updates to be compressed to just one year and the lag between the period to which they refer and the update to be shortened to a quarter or perhaps even just a month.
|
From page 77... ...
For example, credit card data identify only relatively broad expenditure categories, al though retailers often maintain and link information on detailed purchases to specific individuals/households. In contrast, transactions data from retail ers, such as scanner data or the data used for the Adobe Digital Economy Index,30 do include product detail.
|
From page 78... ...
National Accounts Data Because household surveys or other, alternative sources of information on household expenditures become available with a lag, agencies gener ally use the fixed basket approach that weights current price changes with expenditure share data from an earlier period (the Lowe formula) for their featured consumer price indexes.
|
From page 79... ...
ABS also provides a model for between-CE updates: Its Household Expenditure Survey provides the benchmark information on the expenditure weights of Australian households, although, as of 2018, ABS moved to annually re-weighting using HFCE data from the national accounts. 31 Fixler, de Francisco, and Kanal (2021)
|
From page 80... ...
The hybrid approach entails estimating recent expenditure patterns using the best data source for each level of aggregation and item considering the goals of timeliness and accuracy and the weaknesses of the CE. As described above, national accounts and payments (credit and debit card)
|
From page 81... ...
A00858 -- Consumer Price Index REV.indd 81 8/15/22 11:54 AM
|
From page 82... ...
As recommended in Chapter 7, this project would benefit from coordination with the Census Bureau, which would benefit from access to such price and quantity data from retail sales, and BEA, which would also benefit if CPI weights incorporated national account information. Sorting through these data infrastructure options, particularly the bold est ones, requires statistical agencies to envision the future data environment and what CPI users will need in 10 or 15 years.
|
From page 83... ...
recommend us ing a Fisher index, linking the first post-lockdown period to the last pre-lockdown period, for measuring price changes during this aberrant period. A00858 -- Consumer Price Index REV.indd 83 8/15/22 11:54 AM
|
From page 84... ...
84 MODERNIZING THE CONSUMER PRICE INDEX FOR THE 21ST CENTURY suggested in this chapter would provide useful comparison to the official CPI and in disruptive episodes like COVID-19 would provide one gauge of how far off weights were in the official CPI. A00858 -- Consumer Price Index REV.indd 84 8/15/22 11:54 AM
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.