Page 172 - Special Topic Session (STS) - Volume 3

STS538 Pedro Luis do N. S. et al.
2. Overview of the Brazilian CPI
IBGE compiles the official inflation measure for Brazil (called IPCA), as part
of a larger set of consumer price indices for different reference periods and
target populations (defined according to income profile and geographical
coverage) [Miranda et al., 2019, IBGE, 2013]. The set of indices produced is part
3
of a framework called National System of Consumer Price Indices (NSCPI) . In
this system, INPC and IPCA are the most important indices. INPC measures
inflation for low income families (1 − 5 Brazilian national minimum wages).
IPCA has a broader scope and measures inflation for the Brazilian population
with income in the 1 − 40 Brazilian national minimum wages range. IPCA is
adopted by the Brazilian Central Bank to monitor tar-gets set for national
inflation and to define monetary policy.
Both indices are compiled and published monthly for 16 Brazilian states.
The state-level indices are then aggregated to compose the ‘national’ indices
[Miranda et al., 2019, IBGE, 2013]. Each of the 16 states have their own basket
of goods and services determined by local consumer habits observed in the
Household Budget Survey (HBS). Hence, the national basket contains every
element that appears in a given local basket with its respective national weight.
The basket is derived according to plutocratic criteria and the indices adopt a
domestic scope [Miranda et al., 2019]. In a country of continental dimensions
such as Brazil, derivation and use of local baskets is an important ingredient
to guarantee that the diversity of habits is represented in the NSCPI indicators.
Also, the calculation of indicators for different ranges of income is also an
important feature of the system, since the Brazilian society is also
characterized by large income inequality.

3. Challenges and opportunities for the use of big data in the NSCPI
a. Web data
The exploration of web data on prices for the compilation of CPIs was
triggered by the pioneering work of Cavallo and collaborators within the scope
of the MIT Billion Prices Project [Cavallo et al, 2016]. This approach is appealing
since apparently the prices and descriptions of products announced by web
retailers are of good quality, ‘free’ and can be downloaded massively in high
frequencies by means of automated web scrapers. Following this trend, many
NSOs are experimenting with the use of web data in the production of their
official CPIs [Van Loon et al, 2018, Breton et al., 2016, da Silva et al., 2019].
For a country with the geographic and socio-demographic peculiarities of
Brazil, the adoption of web data in replacement of traditional sources should
be considered with caution, since some deficiencies of web data can be
intensified here. One important methodological caveat is that so far, the HBS

3 SNIPC, in Portuguese.
161 |I S I W S C 2 0 1 9