Assessment of Statistical
Quality of Real Sector Data
Categories in India
Ramesh Kolli,
Central Statistical Organisation
Statistical system in India
• A decentralised system
• Ministry of Statistics and Programme
Implementation(MSPI) is the nodal Ministry for
Statistics in India. It consists of A. Statistics
wing
Central Statistical Organisation (CSO)
National Sample Survey Organisation (NSSO)
Computer Centre.
B. Programme Implementation wing
Coverage and quality of data
• The MSPI attaches considerable importance to coverage
and quality of data by adopting following measures:-
Designs and conducts Surveys in accordance with statistical
techniques
Data collected personally by trained investigators
Monitors the quality of statistics constantly
Organises technical meetings
Statistical System recently reviewed by a National Statistical
Commission
Subscriber to the International Monetary Fund’s (IMF) Special Data
Dissemination Standards (SDDS)
Recently subjected some data systems to IMF’s Report on Standards
and Codes (ROSC)
(1) RELEVANCE
• Determination of relevance of statistics
• Main users –
Government
Reserve Bank of India
Research Institutions
International bodies and organisations
individual researchers
• relevance for statistics is determined on the basis
of domestic and international requirements
User-consultation groups
– Governing Council (of NSSO)
– Advisory Committee (National Accounts)
– Technical Committee (Prices)
– Standing Committee (industrial statistics)
– Number of other Committees/Group
comprise
eminent academicians
other users
Government
non-Governmental agencies.
other channels
Conferences and Seminars organised by independent Associations
User responses
Links with Ministries of Finance and
Central Bank
– MSPI
– Ministry of Finance
– Reserve Bank of India
– main users and producers of real, fiscal, financial
and external sector statistics.
Real sector – MSPI
Fiscal - Ministry of Finance
Financial and external - Reserve Bank of India
Contd.…..
• The CSO and these organisations
share the data with each other
hold regular discussions
are members of most committees on statistics
• Allocation of Business Rules – fixes
responsibilities
Users requests for key statistics
• Past requests, which were met (examples)
Demand for QNA
Reduction of timeliness in production index
• Current requests, under trial
reducing the timeliness of GDP-expenditure
(presently 10 months)
Quarterly GDP by expenditures
Changing the base year of price indices and production index
• plan to meet the demands on GDP by March, 2007,
others earlier
Publication of experimental series of
data
not officially published
but discussed in Advisory Committees and
Seminars
through technical papers
major users have access
(QNA, expanding the coverage in new series of
national accounts, production index and wholesale
price index)
Presentation of conflict of costs versus
quality to users
• main users are aware of these conflicts and
limitations in flash estimates
• examples
Advance estimates of national income
Provisional production index
(2) ACCURACY
Surveys are efficiently designed
Data collected through dedicated field staff
well trained in concepts and definitions of survey; provision for
supervision & scrutiny
Checks also provided in administrative data
In the case of national accounts
organised sector data are considered accurate
unorganised sector - surveys only once in 5 years
For other years, benchmark estimates moved with indicators and
deflators/inflators.
• Some GDE and GDI aggregates derived either as residuals or
through commodity-flow approaches
Importance attached to accuracy and
steps to improve accuracy
utmost importance
revisions in source data are monitored
available databases are thoroughly examined
choice of sources and statistical techniques adopted
are solely on statistical considerations
validation studies are planned in advance
sources and methods discussed in the Advisory
Committees
flash estimates tend to suffer on accuracy
Level of accuracy attached to key
economic statistics
not possible to fix a level of accuracy
revisions give some indication of weaknesses in
method of estimation and choice of indicators
• Quality changes in goods are accounted in estimates at
current prices, and for the volume estimates, price
deflators are generally used
Statistics which are relatively less accurate
Advance estimates of national income
Provisional production index
Corporate sector statistics
• Statistics under development
QGDE estimates
sequence of accounts for institutional sectors
• Publication of statistics on revisions
• Along with latest data, revised data for previous period released
• Revisions targets
no such targets fixed
CSO scrutinises and validates estimates
• no effort is generally made to either revise the source data or
adjust the discrepancies
Accuracy of surveys
Accuracy of the estimates is assessed in the
following ways:
Sub-sample estimates are compared.
Estimates from earlier surveys and other
sources are compared
Estimates of standard errors of characteristics
under study are analysed.
Links between statistical office and
respondents to surveys
Investigators and supervisors are the links
Need for surveys publicised through media
Data collectors also explain the purposes
Results of surveys discussed and disseminated
by organising Seminars
Individual data is treated as strictly
confidential.
Assessment of the impact of sampling,
non-response, etc. and rates on accuracy
Impact of sampling on accuracy is assessed
through standard errors of the estimates
Non-response is generally avoided by
collecting data from substitutes. There are no
such specific minimum requirements of
accuracy
However, estimates are published generally
after ensuring a minimum number of sample
units
(3) TIMELINESS
first estimates of GDP - 2 months before the
close of the year
revised after 5 months and 12 months, 24
months and 36 months
production index – 1, 2 and 4 months
consumer price indexes – not revised
(4) PUNCTUALITY
Advance Release Calendar
ARC for data categories covered under the
SDDS is disseminated on the Ministry’s
internet site, and on DSBB of IMF
• Administrative data conventionally follow a
calendar of releases, which users are aware
Table: Timeliness and Periodicity of data categories
Periodicity Timeliness
SDDS Data
Category SDDS India SDDS India
Real Sector
National accounts Q Q 1Q 1Q
Production M M 6W 6W
index/indices
Employment Q A 1Q 12M
Unemployment Q A 1Q 12M
Wages/earnings Q five years 1Q 12M
CPI M M 1M 1M
WPI M W 1M 2W
Fiscal sector
General govt. A A 2Q 9M
operations
Central govt. M M 1M 1M
operations
Central Q Q 1Q 1Q
government debt
Financial sector
Analytical a/c’s of M 2W 1M 15D
banking sector
Analytical a/c’s of M W 2W 1W
central bank
2
Interest rates D D D
Stock market D ... ...
External Sector
BOP Q M Q 1Q
Intl. reserves M W 1W 1W
Reserves template M M 1M 1M
Merchandise trade M M 8W 5W
International inv. A A 2Q 2Q
position
Exchange rates D D D
population A A ... 12-18M
Contd.… PUNCTUALITY
Unavoidable delays in publication
Delays have not occurred in Real Sector
Trade-off between punctuality and accuracy
Punctuality is most important in release of data
Generally, it is ensured that the requisite data is
obtained well before the dates of release
(5) ACCESSIBILITY
Data (for example, National Accounts) simultaneously
released - press notes/ Internet
At the time of release, detailed information is generally
made available
Further breakdowns are made available only through
publications, diskettes and CD-ROMs
contact persons details are given in the Ministry’s web-site
Feedback received regarding internet site
• Users generally appreciated
Contd.. ACCESSIBILITY
Support services for users
List of publications available
List of contact persons, mail ids, fax and tel.
nos.
procedure of obtaining hard/ soft copies of
publications are disseminated on internet site
(6) INTERPRETABILITY/CLARITY
• Metadata of key statistics
metadata of 17 data categories covered under the SDDS are
available on the IMF’s DSBB
summary methodologies are also posted for national accounts,
production index, labour market and producers and consumer
price indices
The detailed sources and methods of national accounts,
production index is posted on the Ministry’s internet site
• Feedback received from users regarding metadata
Users generally found the details contained in the metadata as
informative about the practices of compilation and dissemination.
(7) CONSISTENCY/COHERENCE
Statistics released are internally consistent with the
components and source data
QGDP data add up to the annual GDP
estimates at current prices are consistent with the constant
price data and deflators
estimates of GDP and employment are consistent in terms of
concepts, definitions and coverage
GDP expenditures differ slightly (less than 5%) with those
compiled by economic activity
CSO does not make any effort to reconcile the estimates and
presents the discrepancies
Contd.... CONSISTENCY/COHERENCE
GDP by economic activity is considered more reliable, as
some aggregates on expenditure side are estimated through
commodity-flow method.
GDI data are consistent with those made by the economic
activity, as either compensation of employees and operating
surplus/mixed income, are derived as residuals.
Estimates of GDP are not compiled through supply-use
framework, although Input-Output Tables are compiled
once in every five years
India has plans to switch over to supply-use based national
accounts, by March, 2007
Classification systems
generally in line with international systems.
National Industrial Classification (NIC), 1998, is 5-
digited, but is identical at 4-digit level with the ISIC
(Rev.3)
product classification followed is the Common Product
Nomenclature (CPN), which is compatible with the
earlier NIC at 3-digit level
institutional sector classification is in accordance with
the 1993 SNA, although separate accounts are not
prepared for the sectors.
Contd.…. Classification systems
Public sector (comprising general government, public
financial corporations and public non-financial
corporations) is taken as a major institutional sector
Non-profit institutions serving households (NPISHs) is
not distinguished with the households.
• The functional classifications adopted refer to the 1968
SNA.
• Status of 1993 System of National Accounts in India
– Presently, India follows a mix of 1968 SNA and 1993 SNA
Continuity of time series
Whenever a new series is introduced, either a
linking factor (in the case of price and
production indices) or a comparable historical
series (in the case of national accounts) is
provided to the users.
Series from 1950-51 was published when new
series of national accounts with base year 1993-
94 was introduced in 1999
Integration of statistical software
systems and databases
There is no centralised information warehouse
at one place as at present, databases of
individual data sets are maintained by the
respective compiling Divisions
Plans to develop a data warehouse in the
immediate future.
(8) CREDIBILITYTRANSPARENCY/INTEGRITY
Legal Status
Government of India rules
traditionally functions as an independent
body
• statistics released at the level of DG, CSO
Demarcation between MSPI and Government
Ministers
MSPI has Government Ministers in charge
Minister in charge is the final authority for
all statistical policies
Under the terms of the Indian Constitution,
the Minister is responsible to the Parliament
Internal Govt. access at the time of release
only
Transparency in the production of statistics
Methodologies widely disseminated/published/ presented
in various forums accessible to users
metadata pages on are reviewed every quarter
reports of surveys and administrative data contain details
of sampling design, methodology of data collection and
estimation procedures.
Revisions in the data are explained
• Delays in meeting pre-announced publication schedules
• Delays in releases from the pre-announced calendar have
so far not taken place.
(9) COST-EFFICIENCY
Constraints of costs on statistics
Collection of statistics is costly as the requirement of sample
sizes is usually quite large, in order to generate reliable
estimates at industry level and Regional/sub-regional level.
surveys in India have to be canvassed personally by the
investigators, given other constraints
Attempts made to design efficient survey techniques
surveys are subjected to continuous review and improvement.
The Governing Council ensures that the surveys designed are
cost efficient. Budgets are allocated after thorough discussions
Emphasis placed on staff development
Indian Statistical Service officers (who form the
pivot of the system) undergo a 2-year training.
Statisticians also undergo in-service trainings
trainings are provided by international agencies
NSSO officials receive continuous training
Zonal Training Centres in the NSSO and a
Training Division in the CSO also exist
Availability of skilled staff
Available skilled staff in the Organisation is
considered adequate
hired through competitive examinations.
qualifications of officers is high with most
having Masters degree in Statistics and some
acquiring doctorate degrees.
The staff are aware of the quality aspects of
data, which is produced by the Organisation
Project/process management techniques
targeted outputs are initially planned in a five-year
time span
components of these outputs along with the time-
schedules are drawn up
For example, India has plans to implement the major
features of 1993 SNA by March 2007 and a broad
action plan has been prepared to complete this task.
For regular outputs, an annual action plan is prepared
well in advance with month-wise components
These are monitored at least once in a month
Thanks
