Calculation method of the Regional Difference Index

of Consumer Prices

I．Introduction ..........................................................................................................................................2
II．Method of calculating the RDI .............................................................................................................2
1. Calculation formula for the RDI .........................................................................................................2
(1) Quasi- Fisher formula ......................................................................................................................2
(2) Quasi- Laspeyres formula .................................................................................................................2
(3) Quasi- Paasche formula ....................................................................................................................3
2. Calculation of average prices .............................................................................................................3
(1) Price data ........................................................................................................................................3
(2) Procedure of calculating average prices .............................................................................................3
(3) Treatment of items that are not surveyed in the Structural Survey ........................................................5
3. Calculation of weights .......................................................................................................................5
(1) Preparation of expenditures for calculating weights ............................................................................5
(2) Procedure for calculating weights .....................................................................................................6
III．Related information ............................................................................................................................8
1. About the Structural Survey ...............................................................................................................8
2. List of Items .....................................................................................................................................8
3. List of Aggregation from Items to Groups ...........................................................................................8
4. Notes on the Statistical Tables ............................................................................................................8
5. Terminology .....................................................................................................................................8
6. National Survey of Prices ..................................................................................................................8

1
I．Introduction
The Regional Difference Index of Consumer Prices (RDI) is an index that indicates the regional
differences of the price level based on the average prices of Japan of goods and services
purchased by households nationwide. The “Total” and the “Ten major groups” indices in each
region (10 districts, prefectures, and capital cities and ordinance-designated cities 1 ) will be
published in the year following the survey year.
The RDI is calculated from the result of the Retail Price Survey (RPS) (the Trend Survey and
the Structural Survey2). The items to perform the calculation of the RDI (hereinafter “RDI items”)
are the items used in the calculation of the CPI, except for the “imputed rent” and the “items
surveyed only in Okinawa-ken”. (See “List of items” for details.)
Also the RDI for prefectures was published in the National Survey of Prices 3, which had been
conducted once every five years, was ended after the last survey in 2007. However, a simple
comparison is impossible owing to the differences of the sampling method and the survey period.

II．Method of calculating the RDI

1. Calculation formula for the RDI
(1) Quasi- Fisher formula
The RDI is calculated by the following Quasi- Fisher formula (the geometric mean of the
RDI obtained by the Quasi- Laspeyres formula and the RDI obtained by the Quasi- Paasche
formula),

I ag = I ag
L P
･I ag ,

I : RDI of Quasi - Fisher formula

I L : RDI of Quasi - Laspeyres formula

I P : RDI of Quasi - Paasche formula

a : Comparative Regions (10 districts, Prefectures, capital cities and ordinance - designated cities)
g : Item group

where, the Quasi- Laspeyres formula and the Quasi- Paasche formula are defined as
follows:

(2) Quasi- Laspeyres formula

The RDI of Quasi- Laspeyres formula for a group g at a region a is calculated as a
weighted arithmetic mean of item prices for a based on the average prices for Japan with

1
The ordinance-designated cities here are similar to the ordinance-designated cities of the 2010 based Consumer Price Index (CPI),
which have been delegated by August 3, 2009.
The RDI for this area classification was published as a reference index of the CPI until 2013’s result. From 2014’s result, the RDI
for this area classification will be published with a change of the formula as a result of the Structural Survey.
2
Refer to the website of the Statistics Bureau of Japan for the details of the Trend Survey. In regards to the Structural Survey, the
results of survey on price differences between regions are used. In that survey, 56 items・58 specifications are surveyed in 88 cities
excluding the municipalities covered by the Trend Survey. See “Outline of the Structural Survey” for details.
3
Refer to the website of the Statistics Bureau of Japan for the details of the National Survey of Prices.

2
 respect to weights W0i for Japan. The average prices used in this formula are the arithmetic
mean values.
0 :Japan
 P ai 
∑  
 P 0i W0i
i∈g 
i : Item

L
I ag = × 100 , P : Average price (weighted arithmetic mean) .
∑W0i
i∈g P ai
: Price index of item i
P 0i
W0i : Weight of item i for Japan

(3) Quasi- Paasche formula

The RDI of Quasi- Paasche formula for a group g at a region a is calculated as a
weighted harmonic mean of item prices for a based on the average prices for Japan with
respect to weights Wai for the region a . The average prices used in this formula are the
harmonic mean values.

Pˆ :Average price (weighted harmonic mean)

∑W
i∈g
ai
Pˆai
P
I ag = × 100 , :Price index of item i .
∑
Wai Pˆ
0i

i∈g  Pˆai  Wai : Weight of item i for region a

 
 Pˆ 
 0i 

2. Calculation of average prices

(1) Price data
A. The price data surveyed by the Trend Survey of the year.
B. The price data surveyed by the Survey on Price Differences between Regions (conducted in
an odd month) of the Structural Survey of the year.
Note that, since the Structural Survey is not conducted in even month whereas the Trend
Survey is conducted in every month, the prices of even month for each municipality covered
by the Structural Survey should be estimated, in order to fill the price differences between
odd month and even month. The estimation will be done as following procedure.
(a) Calculating the ratio of the average price of even month to the average price of previous
odd month for each 10 district by using the price data of the Trend survey (the data of
A).
(b) Estimating the price data of even month of the Structural Survey by multiplying the
price data of previous odd month of the Structural Survey by the ratio of corresponding
district.
(2) Procedure of calculating average prices
The average price of each item by municipality is calculated by the following procedure.
Note that, the treatment of items that are not surveyed by the Structural Survey is explained in
“2. (3) Treatment of items that are not surveyed in the Structural Survey”.

3
 A. Calculation of monthly average prices of each item by municipality
(a) Basic formula
The average monthly price of each item by municipality is calculated from the price data
of (1) by the simple arithmetic mean. Note that, for fresh food and cut flowers, whose daily
price changes are large, surveys are conducted three times a month (every 10 days in a
month) to determine the exact price for the month. The average price for those items per a
month is calculated by averaging prices for each 10-day period. Prices in the observation
period during the current month are calculated as the simple arithmetic mean of the average
prices for each 10-day period.
(b) Model formula (Items with complicated rate structures = Model item 5)
Some items such as electricity, water and sewerage charges are not appropriate to
calculate their average prices with the basic formula, since they have various rate structures,
with prices that vary according to the purchased conditions. Those items have been
specified by several specifications according to the purchased conditions, and then the
average prices for those items are calculated with special formulae, called model formulae,
which are designed by using a typical utilization case of each item as a model. It is in
principle based on the CPI method.
B. Calculation of annual average price of each item by municipality ( P ji , j : municipality)
The annual average price of each item (except for flesh foods) by municipality is
calculated as the simple arithmetic mean of the average prices for 12 months. Note that, the
average prices for fresh foods are calculated as the weighted arithmetic mean of the average
prices for 12 months with respect to the monthly weights 6.
If the specification is changed in the middle of the year, it is necessary to adjust the price
differences (quality differences) between new and old specifications caused by the revision
of specification, for the calculation of the annual average price. The adjustment is done by
multiplying the rink coefficient, which is created by month, to pertinent price data.
C. Calculation of annual average price of each item by region (Japan, districts, prefectures)
(a) Average prices by weighted arithmetic mean: Used for the Quasi- Laspeyres formula
The average price of each item i by region is obtained as a weighted arithmetic mean
of the average price for each municipality P ji with respect to weight of all items by
municipality j ( C j = ∑ W ji 　
).
i
∑C P
j∈k
j ji
k : Japan, districts, prefectures
P ki = 　 W : Weight for item i in municipality j
∑C
j∈k
j ji
C ：Weight of all items in municipality j
j

(b) Average prices by weighted harmonic mean: Used for the Quasi- Paasche formula

5
Appropriate items are the items that the symbol of the column of “Model items” in “List of Items” is “○”.
6
Since monthly purchase quantities greatly fluctuate by item, in the calculation of CPI, the annual average prices of fresh foods are
calculated by the weighted arithmetic mean of the monthly prices with respect to the monthly weights. The treatment of the annual
average prices of fresh foods in the CPI is adopted also for the calculation of the RDI.

4
 The average price of each item for each region (Japan, districts, prefectures) is obtained
as a weighted harmonic mean of the average price for each municipality P ji with
respect to weights for item i by municipality j ( W ji ).
∑ W ji  ∑ PjiQ ji 
 
j∈k  j∈k
P =
ˆ 　　= ，　P Q = W 
ki 1  ∑ Q ji ji ji 
∑ W  ji 
j ∈ k P ji
ji  j∈k 

Q : Quantity for item i in municipality j

ji

(3) Treatment of items that are not surveyed in the Structural Survey
The items that are not surveyed in the cities covered by the Structural Survey are treated in
accordance with the price substitution criteria for CPI as follows. See “List of Items” for the
price substitution criteria for each item. Note that, the items whose criteria symbol is blank
correspond to the items that are surveyed in all municipalities of the Trend Survey (D).
A. The items with uniform price throughout the country
The uniform price throughout the country surveyed in the Trend Survey is substituted for
the price of the Structural Survey.
B. The items with uniform price within each prefecture
The uniform price surveyed throughout the prefecture in the Trend Survey is substituted
for the price of the Structural Survey within each prefecture.
C. The items that are surveyed only in the municipalities of the Trend Survey with more than
a certain population size
Some items are surveyed only in the prefectural capitals or municipalities of the Trend
Survey with more than a certain population size. For these items, the average prices of a
neighboring municipality of the Trend Survey in the same prefecture are substituted for the
average prices of municipalities of the Structural Survey (in case of the items surveyed only
in the prefectural capital, the average price of the prefectural capital is used).
D. Items that are surveyed in all municipalities covered by the Trend Survey
The yearly average price by prefecture obtained by the weighted arithmetic mean of the
average prices by municipality of the Trend Survey with respect to the adjustment
coefficient (which is same as the adjustment coefficient for the weights) is substituted for
the price of each municipality of the Structural Survey in the same prefecture.

3. Calculation of weights
(1) Preparation of expenditures for calculating weights
A. Expenditure data
The data is the results of latest five years of Family Income and Expenditure Survey
(FIES) (including agricultural, forestry and fisheries two or more persons households).
The scope of the FIES items used for the calculation of weights is limited to the
household consumption expenditure.

5
 Note that, the following items of consumption expenditure are excluded from the
calculation of the weights: Remittances, money gifts, religious contributions, other
obligation fees (fees for neighborhood association, firefighting and street lamps) and
donation 7.
B. Calculation of the average expenditures by FIES item for FIES municipalities per a month
of one household
(a) Calculation of the monthly average expenditures (for 12 months) for each year
By using the data of A, the monthly average expenditures (for 12 months) by FIES
municipality is calculated.
(b) Calculation of the monthly average expenditures for five years
By using the data of (a), the monthly average expenditures for five years are calculated.
Note that, for the FIES items that are not suitable to use the results of five years of
FIES results caused by the revision of the budget item classification and the change of
various systems, the results of FIES under the latest classification and systems are used to
calculate the average expenditures.
(c) Correction of the expenditures by means of the district wise prorated ratio
By using the average expenditures of (b), calculate the district-wise prorated ratio for
each correction target FIES item, which prorate the expenditure of the subgroup
containing the target item to its component items. Next, correcting the expenditure of
each target FIES item is corrected with the multiplication of the above ratio to the
expenditure of subgroup containing the target item.
Note that, as a principle, this procedure is adopted for the subgroups having a large
amount of expenditure rate of its component items whose dispersion among FIES
municipalities and the years is large. For the items that are not estimated their expenditure
by proration, the average expenditure calculated in (b) is used for the weights.
(2) Procedure for calculating weights
A. Weights by FIES municipality
The weights by FIES municipality is calculated by allocating the FIES items into the RDI
items, in accordance with the latest allocation criteria (2010 based criteria) of the weights
for the CPI.
(a) When there is one to one correspondence between the FIES item and the RDI item, the
correspondence is maintained.
(b) When there are no RDI items corresponding to the FIES items, the amount of
expenditure of the FIES item is split or integrated with the data being allocated to the
corresponding RDI item, in accordance with the latest allocation criteria of the weights
for the CPI.

7 The reasons for excluding these items from the calculation of the weights are as follows: Markets do not generally exist for such
items; the relationship between payment and counter value is not clear or they involve income transfer to other households.

6
 B. Allocation of municipalities
(a) Allocate the FIES municipalities into the municipalities of the RPS (the municipalities
combined the Trend Survey and the Structural Survey; say “RDI municipality” for
short).
(i) Allocation to the municipalities of the Trend Survey
Since there is one-to-one correspondence between the FIES municipalities and the
municipalities of the Trend Survey according to the strata 8 of FIES, as the weights of
each municipality of the Trend Survey, is substituted for the weights of corresponding
FIES municipality.
(ii) Allocation to the municipalities of the Structural Survey
As the weights of each municipality of the Structural Survey, is substituted for the
weighted arithmetic mean of the weights for FIES municipalities with respect to the
latest number of household for tabulation after the adjustment for each stratum in FIES
by the corresponding district and the city group.
(b) Revision in accordance with the actual situation of RDI municipalities
For some RDI items such as Water & sewerage charges and Kindergarten fees, the
weights are revised in accordance with the actual situation of RDI municipalities.
(c) Adjustment of the expenditure in each RDI municipality with the consideration of the
number of households
The weights obtained from the above (b) do not consider the number of households in
each RDI municipality. In order to consider the number of households in each
municipality, the adjustment coefficient calculated from the number of households
consisting of two or more persons in each municipality obtained from the latest
Population Census is multiplied to the weight of (b), and then the final weights ( W ji ) are
obtained.
C. Aggregated items and regions
(a) Aggregated items
The weights for groups are aggregated from the weight for each RDI item. (See “List
of Aggregation from Items to Groups” for details.)
(b) Aggregated regions (Japan: W0i , regions: Wai )
The weights for Japan and regions are aggregated from the weight for each RDI
municipality obtained by the above (a).

8 Stratum is a classification of all municipalities in japan in order to extract the municipalities surveyed, in accordance with the
population size, the geographic location, and the industrial traits. The stratum for extracting the municipalities of the RPS is equal
to that of the FIES.

7
III．Related information
1. About the Structural Survey
http://www.stat.go.jp/english/data/kouri/kouzou/index.htm
2. List of Items
（Excel file）
http://www.stat.go.jp/english/data/kouri/kouzou/zuhyou/hinmoku_info_e.xls
（PDF file）
http://www.stat.go.jp/english/data/kouri/kouzou/pdf/hinmoku_info_e.pdf
3. List of Aggregation from Items to Groups
http://www.stat.go.jp/english/data/kouri/kouzou/pdf/aggregation_e.pdf
4. Notes on the Statistical Tables
http://www.stat.go.jp/english/data/kouri/kouzou/pdf/note_e.pdf
5. Terminology
http://www.stat.go.jp/english/data/kouri/kouzou/pdf/terminology_e.pdf
6. National Survey of Prices
http://www.stat.go.jp/english/data/zenbutu/index.htm