Page 173 - Special Topic Session (STS) - Volume 3

STS538 Pedro Luis do N. S. et al.
only provides weights for products commercialized in brick-and-mortar stores
and those need to be used as proxies for products traded via web. For the
2017 - 2018 HBS being finalized, questions on the kind of store (online or
offline) where purchases were performed were introduced. With such
information more details on the web shopping process will be known and
derivation of weights for online shopping might be possible.
Another issue stems from the fact that ‘location’ of the web site hosts, the
location from where the consumer orders the goods, and the delivery address
for the goods may all be different, and therefore there would need to be
decisions regarding to which region the particular transaction should be
allocated, given our approach to build the national CPI from aggregation of
regional indices. In addition, though prices offered online often have no
distinction according to different regions of the country, different delivery fees
apply depending on delivery address, and such fees may also depend on the
amount of goods purchased, thus making it difficult to associate a proper price
for a single good. Level and evolution of prices at web and brick-and-mortar
stores are not necessarily the same, though in a recent study [Cavallo, 2017]
found no evidence of price discrimination between online and offline prices
for big retailer chains, with an observation of ≈ 5% difference in Brazil.
In Brazil, only about 75% of the households have some form of internet
access [IBGE, 2018]. The average velocity of internet connection in Brazil is
three times lower than the global aver-age and the access is much lower for
rural households. However, internet access is growing fast due to the use of
mobile devices and promoting an important increase of e-commerce in Brazil.
According to a recent report [ebit, 2019], 7 out of 10 Brazilians owned a mobile
phone in 2018, an increase of 7% over 2017. This increase is strongly correlated
with the enhancement of the e-commerce observed in 2018, 12% larger in
volume of sales than those observed in 2017 [ebit, 2019]. In 2018, 58 million
Brazilians (27% of the country’s population) performed at least one online
purchase. Of these, 10 million performed their first online purchase in 2018,
64% of which by means of a mobile phone. The growth in the volume of sales
of the m-commerce in 2018 over 2017 was 41%, amounting to ≈ 42% of the
volume of e-commerce sales in 2018 [ebit, 2019]. The number of online orders
was 123 million in 2018 (increase of 11% over 2017), however this led to a
modest average of ≈ 2 purchases a year for the online buyers in 2018, mostly
concentrated in sectors like make-up, and clothing and accessories which
respond to ≈ 40% of the orders.
Given the peculiarities of the e-commerce in Brazil discussed above, the
adoption of web data for the NSCPI should proceed with caution. IBGE is
experimenting a parsimonious introduction of web data to improve CPI
compilation in combination with traditional sources. The first approach that
IBGE is experimenting with involves web scraping for prices of products where

162 |I S I W S C 2 0 1 9