Page 171 - Special Topic Session (STS) - Volume 3

STS538 Pedro Luis do N. S. et al.
rise to m-commerce) have revolutionized consumer habits. An important by-
product of these changes is the large number of new data sources and vast
amounts of in-formation on transactions now potentially available to support
CPI measurement and estimation.
National Statistics Offices (NSOs) aim to provide timely and accurate
statistics to portray the societies that they serve. To remain relevant and
achieve their goals they need to keep track of change and to adapt to the
evolving data landscape. Typical NSOs also face challenges to produce more
while constrained by decreasing budgets, weak legal frameworks regarding
their right to access some of the new data sources (private or even
governmental), and facing pressures from new private statistics producers that
are taking advantage from emerging ‘big data’ sources [Cavallo et al, 2016].
CPIs are some of the most important economic statistics produced by
NSOs around the world [Stoevska, 2018]. The original goal of a CPI was to
provide an approximate measure of the cost of living for a given target
population. In recent times, CPIs have been used as deflators for Na-tional
Accounts, as well as core macroeconomic indicators that Central Banks
consider when defining monetary policy [ILO, 2004, Eurostat, 2018].
In this scenario, the adoption of alternative and potentially cheaper data
sources for CPI estimation by the NSOs seems attractive, if not mandatory.
However, this is far from straightforward. Such data sources are not designed
for statistical purposes, and hence their incorporation in CPI estimation might
require development and adoption of new methodologies. In some cases, they
may even be unusable given CPI quality measurement standards currently in
place. In addition, access to such new data sources might not be easy, because
NSOs lack adequate legal frame-works supporting rights of access.
The traditional approach for estimation of CPIs relies on the measurement
of the price variations for the goods and services in a ‘basket’ derived at a
given point in time [ILO, 2004]. With the advent of the digital economy, new
goods and services are created and others disappear at a fast pace. Therefore,
maintenance of a representative ‘basket’ of goods and services consumed, and
the collection of prices for any new items presents major demands for CPI
compilers. Aiming to circumvent such difficulties, NSOs all around the world
are experimenting with the use of alter-native data sources for CPI
compilation. The most relevant sources under study include web data, scanner
data and administrative registers [Van Loon et al, 2018, Hov et al, 2018,
Mendon-ça et al, 2018, Breton et al., 2016, da Silva et al., 2019].
Here we describe some of the main challenges and opportunities related
to the adoption of such new data sources for CPI compilation in Brazil. We
focus on the main relevant issues that need to be accounted for the use of
web data and ‘electronic fiscal records’ for CPI compilation in Brazil.

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