Estimators for IPCA core inflation
Francisco Marcos R. Figueiredo
Roberta Blass Staub
June 7th, 2001
Research Department
Central Bank of Brazil
1 - Motivation
 Relevant
inflation for Central Bank
 Core inflation as a common indicator
within Mercosur
Research Department – Central Bank of Brazil
2 – Core inflation measures used by Central Banks
Australia
Belgium
Canada
Chile
Colombia
Czech
Finland
France
Greece
Israel
Japan
Netherlands
New Zealand
Norway
Phillipines
Poland
Portugal
Singapore
Spain
Sweden
United Kingdom
United States
CPI less mortgage interest payments, government
controlled prices and energy prices
CPI less potatoes, fruit and vegetables
CPI less indirect taxes, food and energy items
CPI excluding perishable goods and energy
An average of four measures (CPI excluding food
and three limited influence estimator)
CPI less fees and controlled and regulated prices
CPI less housing capital costs, indirect taxes, and
government subsidies
CPI less change in taxes, energy prices, food prices
and regulated prices
CPI less food and fuel
CPI less goverment goods, housinh, fruit and
vegetables
CPI less fresh foods
CPI less vegetables, fruit, and energy
CPI less commodity prices, government controlled
prices, interest and credit charges
CPI less electricity prices and indirect taxes
A statistical trend line
A set of three measures (CPI less officialy controled
prcies, CPI less prices with highest volatilities and
a 15% trimmed-mean)
10% trimmed mean of the CPI
CPI less cost of private road transportation and
accommodation
CPI less mortgage interest payments
CPI less housing mortgage interest and effects of
taxes and subsidies
Retail price index less mortgage interest payment
CPI less food and energy items
Weighted median (FRB of Cleveland)
Research Department – Central Bank of Brazil
3 – Core inflation in Brazil



FGV - trimmed mean with smoothed
components (IPC-BR)
IPEA - common trend of prices and
smoothed trimmed mean
Pichetti & Toledo - dynamic factor index
and asymmetric trimmed mean
 Central
Bank - trimmed mean with
smoothed components
Research Department – Central Bank of Brazil
4 – Desired characteristics of a core measure






Timely computable
Forward looking nature
Good performance in description of the
evolution of the inflation in the past
Easy understanding
Definitive
Theoretical base
Research Department – Central Bank of Brazil
5 – Statistical features of core inflation



Less volatile than the headline inflation
Long run stable relation with inflation
Inflation attractor (time precedence)
Research Department – Central Bank of Brazil
6 - Core inflation estimates



Exclusion method
Symmetric trimmed mean with smoothed
series
Asymmetric trimmed mean
Research Department – Central Bank of Brazil
7 - Exclusion method
 IPCA less
food at home and
administered prices
 Excluded items amount 46% of IPCA
basket
Research Department – Central Bank of Brazil
8 – Exclusion method
Chart 1 - IPCA and exclusion core, Jan/96 through Apr/2001
2.00
IPCA
Exclusion Core
Monthly percent changes
1.50
1.00
0.50
-0.50
Research Department – Central Bank of Brazil
abr/01
jan/01
out/00
jul/00
abr/00
jan/00
out/99
jul/99
abr/99
jan/99
out/98
jul/98
abr/98
jan/98
out/97
jul/97
abr/97
jan/97
out/96
jul/96
abr/96
jan/96
0.00
9 – Symmetric Trimmed mean
x 
1
w x
 iI
1 2
100
i i
 
 
I  i |
 Wi  (1  )
100 
 100
Research Department – Central Bank of Brazil
%
(1-2)%
%
10 - 20% Trimmed mean
Items
Inferior
Superior
Total
More frequent trimmed
Potatoes
Vegetables
Fish
Cereals
Fruits
Sugar and sweets
Fats and oils
Household fuel
Poultry and eggs
Meat
Dairy products
Flour and prepared flour mixes
Television, sound equipment
Motor fuel
Bedding and bath clothes
Beverages
52.46
52.46
45.90
45.90
62.30
42.62
54.10
1.64
40.98
36.07
39.34
42.62
45.90
4.92
36.07
44.26
39.34
37.70
40.98
36.07
18.03
32.79
21.31
72.13
31.15
34.43
29.51
21.31
16.39
55.74
24.59
16.39
91.80
90.16
86.89
81.97
80.33
75.41
75.41
73.77
72.13
70.49
68.85
63.93
62.30
60.66
60.66
60.66
Less frequent trimmed
Maintanance and repair commodities
Decorator items
Tuition, other school fees and childcare
Reading materials
Vehicles
Personal care services
Hospital and other medical care services
Prepared food
Professional services
Domestic services
14.75
16.39
0.00
16.39
14.75
16.39
14.75
9.84
13.11
3.28
14.75
13.11
27.87
9.84
8.20
6.56
6.56
9.84
3.28
4.92
29.51
29.51
27.87
26.23
22.95
22.95
21.31
19.67
16.39
8.20
Obs. The smoothed items are in boldface.
Research Department – Central Bank of Brazil
11 – 20% Trimmed mean
Chart 2 - IPCA and 20% trimmed-mean core, Jan/96 through Apr/2001
2.00
IPCA
Trimmed 20
Monthly percent change
1.50
1.00
0.50
-0.50
Research Department – Central Bank of Brazil
abr/01
jan/01
out/00
jul/00
abr/00
jan/00
out/99
jul/99
abr/99
jan/99
out/98
jul/98
abr/98
jan/98
out/97
jul/97
abr/97
jan/97
out/96
jul/96
abr/96
jan/96
0.00
12 - Bryan & Cecchetti core

Bryan and Cecchetti (2001)



Asymmetric trimmed mean centered at 60th
percentile
24-month moving average
Optimal trims


14.4% - lower tail
9.6% - upper tail
Research Department – Central Bank of Brazil
13- Bryan & Cecchetti core
Chart 3 - IPCA e Bryan & Cecchetti core , Jan/96 through Apr/2001
2 .0 0
IP C A
1 .5 0
1 .0 0
-0 .5 0
Research Department – Central Bank of Brazil
abr/01
jan/01
out/00
jul/00
abr/00
jan/00
out/99
jul/99
abr/99
jan/99
out/98
jul/98
abr/98
jan/98
out/97
jul/97
abr/97
jan/97
out/96
jul/96
0 .0 0
abr/96
0 .5 0
jan/96
M onthly percent change
B rya n & C e cch e tti C o re
14 - Statistics of core inflation measures
Table 1 - Statistics of core inflation indicators - Jan/96 to Apr/2001
Statistics
IPCA
Exclusion
Trimmed 20
B&C
Acumul. Change
37.22
27.19
35.43
30.94
Standard-deviation
0.44
0.34
0.31
0.30
RMSE*
0.40
0.27
0.19
0.26
RMSE**
0.36
0.29
0.17
0.26
* The benchmark is a 13-month centered moving average of IPCA
** The benchmark is a 24-month centered moving average of IPCA
Research Department – Central Bank of Brazil
15 - Granger causality tests



Exclusion core
 does not cause IPCA
 is caused by IPCA
Trimmed 20 core
 causes IPCA ( 2 lags)
 is not caused by IPCA
Bryan & Cecchetti core
 causes IPCA
 is not caused by IPCA
Research Department – Central Bank of Brazil
16 - Impulse response from a bivariate VAR
IPCA and Trim 20 - Response to One S.D. Innovations ± 2 S.E.
Response of TRIM20 to TRIM20
Response of TRIM20 to IPCA
0.0015
0.0015
0.0010
0.0010
0.0005
0.0005
0.0000
0.0000
-0.0005
-0.0005
1
2
3
4
5
6
7
8
9
10
1
2
Response of IPCA to TRIM20
3
4
5
6
7
8
9
10
7
8
9
10
Response of IPCA to IPCA
0.004
0.004
0.003
0.003
0.002
0.002
0.001
0.001
0.000
0.000
-0.001
-0.001
1
2
3
4
5
6
7
8
9
10
Research Department – Central Bank of Brazil
1
2
3
4
5
6
17 - Concluding remarks


A essential tool to track the inflation trend
path
Use a set of core inflation indicators
Research Department – Central Bank of Brazil
18 - Further research


Use bootstrapping procedures to check
the optimal trims in symmetric and
asymmetric trimmed means
Combine different core inflation estimates
Research Department – Central Bank of Brazil