Nowcasting Indian GDP

Published date01 April 2018
DOIhttp://doi.org/10.1111/obes.12219
Date01 April 2018
259
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 80, 2 (2018) 0305–9049
doi: 10.1111/obes.12219
Nowcasting Indian GDP*
Daniela Bragoli† and Jack Fosten
Department of Mathematics, Financial Mathematics and Econometrics, Universit`a
Cattolica, via Necchi 9, 29100, Milano, Italy (e-mail: daniela.bragoli@unicatt.it)
Department of Economics, University of East Anglia, Norwich, NR4 7TJ, UK
(e-mail: j.fosten@uea.ac.uk.)
Abstract
Nowcasting has become a useful tool for making timely predictions of gross domestic
product (GDP) in a data-rich environment. However, in developing economies this is more
challenging due to substantial revisions in GDP data and the limited availabilityof predictor
variables.Taking India as a leading case, weuse a dynamic factor model nowcasting method
to analyse these two issues. Firstly, we propose to compare nowcasts of the first release
of GDP to those of the final release to assess differences in their predictability. Secondly,
we expand a standard set of predictors typically used for nowcasting GDP with nominal
and international series, in order to proxy the variation in missing employment and service
sector variables in India. We find that the factor model improves over several benchmarks,
including bridge equations, but only for the final GDP release and not for the first release.
Also, the nominal and international series improve predictions over and above real series.
This suggests that future studies of nowcasting in developingeconomies which have similar
issues of data revisions and availabilityas India should be careful in analysing first- vs. final-
release GDP data, and may find that predictions are improved when additional variables
from more timely international data sources are included.
I. Introduction
In recent years, nowcasting has emerged as an important tool for producing timely predic-
tions of economic activity variables such as gross domestic product (GDP). Since GDP
figures are only published on a quarterly basis, and typically with a publication lag of
more than a month, nowcasting provides policymakers and other market participants with
a timely snapshot of the current state of the economy with which to inform their policy or
investment decisions. In developed economies such as the United States and the eurozone,
where rich and timely economic datasets are available, papers such as Evans (2005), Gian-
none, Reichlin and Small (2008) and Ba´nbura et al. (2013) have established that suitable
nowcasting methods can provide predictions of GDP which are both more timely and at
least as accurate as surveys of professional forecasters or na¨ıve benchmark models.
JEL Classification numbers: C38, C53, E37, O11, O47.
*We thank participants of several research meetings held at Now-Casting Economics Ltd., London, for their
feedback and advice. Wealso directly thank Now-Casting Economics Ltd. for access to data. This research did not
receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
260 Bulletin
However, in the context of developing economies, the problem of nowcasting is much
less straightforward due to issues of data timeliness, availability and quality. This is con-
firmed by quantitative measures such as the ‘Data Quality Index’ (DQI) constructed by
World Economics, where the best rankings are dominated by developed countries in
Europe and North America.1In this paper we focus on the case of India, not only be-
cause it is one of the world’s largest and fastest-growing economies, but also because there
has been recent media interest in the waythe Indian Central Statistical Office (CSO) heavily
revises its figures for GDP and periodically changes its methodology for calculating these
figures. Furthermore, nowcasting Indian GDP is of particular relevance given its publica-
tion lag of almost two months after the end of the quarter, a delay which is far longer than
in other developing countries, such as China (three weeks) and Indonesia (five weeks), and
developed countries such as the UK and US (four weeks) and Japan (six weeks), albeit
similar to countries such as Brazil and Canada.
The first main contribution of this paper is to provide an analysis of the GDP data revi-
sions process in India. To achieve this, we propose to compare nowcasts of the first GDP
release to the final release, based on a nowcasting methodology which uses a dynamic
factor model which enables the accommodation of a large number of predictors. Since
GDP revisions data are not available for developing economies like India in databases
such as the OECD Real-Time Data and Revisions Database, we firstly construct a series
of first-release GDP figures using the available press releases from the CSO. As a starting
point, we use statistical measures based on the ‘news’vs. ‘noise’ data revisions hypothesis
developed by Mankiw, Runkle and Shapiro (1984) and Mankiw and Shapiro (1986). How-
ever, in developing countries such as India, where revisions data have only only available
for a around a decade, we note that these measures may not always give a robust conclu-
sion about the revisions process. We, therefore, argue that our nowcasting methodology
may be preferred for comparing predictions of first- and final-release GDP in developing
economies, as it uses the co-variation of GDP with a large number of predictor variables.
This acts like a nowcasting version of the out-of-sample test ofAr uoba (2008). Of the few
existing studies of nowcasting Indian GDP, such as Bhattacharya, Pandey and Veronese
(2011) and Dahlhaus, Gunette and Vasishtha (2014), none look at the effect GDP revisions
have on the nowcasting procedure.
The second contribution is that we provide discussion and analysis of the effect of
missing variables in the context of predicting Indian GDP. Specifically, we note that many
variableswhich are typical in empirical nowcasting studies, for example employment series,
are not available in a timely fashion from the Indian statistical authorities. Furthermore,
there are variables which wouldbe par ticularlyrelevant for use in a model for Indian GDP,
such as trade in services, for which timely information is also not available. Starting with
a small set of real variables used in studies such as Ba´nbura et al. (2013), we expand
our study in two ways. Firstly, we include financial series such as stock prices and the
exchange rate of the rupee with the US dollar. This helps us to establish a dataset with
a more complete representation of the Indian economy. Next, we introduce international
series such as US and eurozone industrial production, motivated by the fact that these
1For more information, see http://www.worldeconomics.com/pages/Data-Quality-Index.aspx (last accessed: 5
April 2016).
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT