Page 118 - Invited Paper Session (IPS) - Volume 2

IPS 188 G. P. Samanta
1. Introduction
Inflation expectations constitute an important ingredient to monetary
policy formulation, particularly under the Inflation Targeting approach.
However, the forward-looking assessment or forecasting inflation has been an
extremely challenging task. The time lag in the release of official statistics on
inflation often aggravates the problem further. To address the issue, the
conventional literature suggests two broad approaches, viz., developing
forecasting models and conducting surveys for measuring inflation
expectations. The modelling exercise usually attempts to exploit inter-
relationship between inflation and relevant economic variable and indicators,
either under some economic theories or extracting data-driven patterns. The
empirical estimation of such models uses data released by official statistics,
which are traditionally compiled offline at a fixed interval, and usually released
by compiling agencies with a substantial lag. At times, it may also employ the
survey-based results as additional information.
Surveys have been an alternative tool to quantify market expectation on
an economic variable/parameter and also to fill-in potential data gap or
providing nowcasting indicator for a target variable, mainly data on which are
released with lag. As regards surveys for inflation sentiments, international best
practices have devoted to assessing future inflation based on either qualitative
or quantitative responses from welldesigned and representative target group.
While qualitative responses help in assessing the direction of change or
movement of the future inflation rate, quantitative results directly reflect the
respondents' perception about the level of future inflation numerically.
Answering qualitative questions on a variable, mostly asking an opinion on ‘no
change' or direction of change of the variable, are easier than providing a
quantitative response, and may improve the response rate in a survey. There
has been a quite rich literature on converting the qualitative responses on a
variable to the corresponding numeric value of the variable. However,
conducting surveys may be costly in terms of monetary expenditure, time
requirement and human resources. Further, for the time requirement, survey-
based results may fail to capture information real-time basis. To address these
issues, many researchers have explored if vast metadata and documents
available freely in online resources could be useful in tracking and predicting
economic variables. With the advent of the internet and digital platform fast
growing habits of people on the internet searching, expressing opinions and
sentiments in social media and digital platform, a few researchers have
examined if the internet resources available at more timely and more frequent
manner than traditional data can be useful to assess expectations of economic
agents. There have been wide varieties of online resources such as websites of
business houses, online retailers, social media like Twitter, search queries in
internet search engine like Google, digital or printed documents uploaded by

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