An Introductory Look at the Chained Consumer Price Index
CCPIU Key Features
Price Index Number Formula
 The CCPIU will be a new supplemental index, and will not replace the CPIU and CPIW.
 The CCPIU will use a different price index number formula from that used in the current CPIU and CPIW.
 The CPIU and CPIW use a Laspeyres formula to average indexes together. The CCPIU will use a "superlative" Tornqvist formula.
Chaining Interval
The CPIU and CPIW are chained (that is, they use updated expenditure weights) every two years. The CCPIU will be chained monthly
CCPIU Publication Policy
Inaugural publication of CCPIU will occur in August 2002. The monthly index series will begin with January 2000 numbers; the reference base will be December 1999=100.
A limited set of urban population, U.S. city average, not seasonally adjusted indexes will be published. Indexes will be published for the US, for the categories listed below.
Published CCPIU Index
Series 
SA0 
All items 
SAM 
Medical care 
SAF 
Food and beverages 
SAM1 
Medical care commodities 
SAF1 
Food 
SAM2 
Medical care services 
SAF11 
Food at home 
SAR 
Recreation 
SEFV 
Food away from home 
SAE 
Education and communication 
SAF116 
Alcoholic beverages 
SAE1 
Education 
SAH 
Housing 
SAE2 
Communication 
SAH1 
Shelter 
SAG 
Other goods and services 
SAH2 
Fuels and utilities 
SAS 
Services 
SAH3 
Household furnishings 
SAC 
Commodities 
SAA 
Apparel 
SAD 
Durables 
SAT 
Transportation 
SAN 
Nondurables 
SAT1 
Private transportation 
SA0L1E 
All items less food and energy 
SETG 
Public transportation 
SA0E 
Energy 
CCPIU Revision Policy
There are limits on publishing the CCPIU in realtime, because expenditure data is only available with a two year lag. We will publish an Initial CCPIU index in real time that is subject to revision. We will publish a revised Interim index in February of the following year. We will publish a Final CCPIU in February of the second year. The initial and interim indexes will approximate a superlative index with the information available at the time of publication. The final CCPIU will use prices and expenditure weights from the measurement period.
CCPIU Indexes—calculation formulas
Initial Release
 indexes will be calculated using a geometric mean formula
 the geometric mean indexes will be adjusted using historical information on the ratio of the Tornqvist formula index to the geometric mean formula index.
 The expenditure weights will be those used by the CPIU.
Interim Release
 indexes will be calculated using a geometric mean formula
 the geometric mean indexes will be adjusted using historical information on the ratio of the Tornqvist formula index to the geometric mean formula index.
 The expenditure weights will be those used by the CPIU.
Final release
 indexes will be calculated using the Tornqvist formula
 both expenditure weights and prices will correspond to the measurement period.
Formulas are shown in the Technical Appendix available from http://www.bls.gov/cpi/super_cpi.pdf
The chart below shows the timing of CCPIU revisions.
TIMING OF CCPIU
REVISIONS 


2003 
2004 
2005 

2003 
INITIAL 

FINAL 
2004 

INITIAL 

2005 


INITIAL 
CostofLiving Indexes
As stated in the BLS Handbook of Methods, "The CPI provides an approximation to a CostofLiving Index as a measure of consumption costs." – Appropriate reflection of consumer substitution behavior can help provide an even closer approximation.
Appropriate reflection of consumer substitution behavior can help provide an even closer approximation.
Price Index formulas
Laspeyres formula
 uses spending behavior in previous periods
 assumes fixed quantity proportions
 does not reflect substitution
Geometric mean formula
 uses spending behavior in previous periods
 assumes fixed budget shares
 reflects substitution
Superlative formulas (such as Tornqvist formula)
 uses spending behavior in both the current and a previous period (consumer behavior observed, not assumed
 reflects observed substitution
Superlative indexes incorporate both current period and previousperiod spending behavior. Under certain assumptions, they will more closely approximate a Costof Living index more closely than either a Laspeyres or a
geometric mean index.
The following example shows simple examples of index construction of all three types:
Consumer Substitution
Consumer Substitution
2Good, 1Consumer Example 
Good 
Month 1 
Month 2 
Price 
Quantity 
Cost 
Price 
Quantity 
Cost 
Movie Ticket 
$5 
6 
$30 
$8 
4 
$32 
Video Rental 
$2 
10 
$20 
$2 
9 
$18 

$50 

$50 
Month 2
 Movie ticket price increases 60%
 Video rental price increases 0%
 Movie ticket quanity decreases 33%
 Video rental quantity decreases 10%
Laspeyres index increases 36.0% $68Superlative (Tornqvist) increases 33.8% $67Geometric Mean index increases 32.6% $66
For Lowerlevel index aggregation, CPIU, CPIW and CCPIU use a "Hybrid " approach:
 The Laspeyres formula is used for some items, such as electricity, rent, and hospital services. Zero substitution within an item category is implicitly assumed.
 The Geometric Mean formula is used for most items, such as apples, jewelry, and admissions. Unitary substitution within an item category is implicitly assumed.
For Upperlevel aggregation, the CPIU and CPIW use a Laspeyres formula, and assume zero substitution across items. The CCPIU will use a superlative formula, and will reflect substitution across items.
The "Chained" in the title refers to the frequency of expenditure baseperiod updates. Currently, the CPIU and CPIW are updated every two years; the CCPIU will be updated every month.
Year 
Simulated
Biennially
Updated
CPIU 
Simulated
CCPIU 
1990 
3.64 
3.52 
1991 
3.76 
3.69 
1992 
2.90 
2.58 
1993 
2.80 
2.67 
1994 
2.58 
2.49 
1995 
2.76 
2.46 
Average annual percent change 
3.07 
2.90 
Percentage difference per annum 

0.17 
CCPIU and CPIU Contrasted 

CPIU 
CCPIU 
Initial 
Interim 
Final 
Lowerlevel Index Formula 
Hybrid 
Hybrid 
Hybrid 
Hybrid 
Upperlevel Index Formula 
Laspeyres 
Adjusted
Geometric
Mean 
Adjusted
Geometric
Mean 
Tornqvist 
Expenditure baseperiod: 

Evenyear Indexes 
Biennial,
Lagged 23
years 
Biennial,
Lagged 23
years 
Biennial,
Lagged 12
years 
Previous
Month 
Oddyear Indexes 
Biennial,
Lagged 34
years 
Biennial,
Lagged 34
years 
Biennial,
Lagged 23
years 
Previous
Month 
Expenditure currentperiod: 
none 
none 
none 
Current
Month 
Weight Update Frequency 
Biennial 
Biennial 
Biennial 
Monthly 
Publication Schedule 
1 Month Lag 
1 Month Lag 
2 to 13
Month Lag 
14 to 25
Month Lag 
Last Modified Date: March 14, 2002