Transcript P7.3.4.

Mainstriming Data and
Composite Indicators From
Tendency Survey as Official
Statistics
Gian Paolo Oneto
ISTAT
Third International Seminar on Early Warning and
Business Cycle Indicators
Ottawa, 27-29 May 2009
Mainstreaming
Sentiment data as
Official Statistics
..from my presentation in Ottawa
A key aspect that emerges from the current practices in both the
leading indicators approach and the standard forecast modeling
is the central role of business and consumer sentiment data
… it is time to recognize that the quality aspects of the current
practices prevailing in the production of such indicators are
worth to be investigated from the point of view of the OS
criteria.
The Eu setting is already advanced in terms of comparability
across countries, but could be further improved through
coordination with the standard statistical production (for
instance, concerning industry classification, reference years
and variables to be utilized in the weighting structure).
There is the perception that outside Eu the statistical
underpinnings of survey data are even less comparable to the
OS standards.
Moscow, 17-19 November 2010
2
Mainstreaming
Sentiment data as
Official Statistics
1. Why the interest of official statistics for the
production process of Sentiment Indicators ?
The Indicators drawn from the Business and Consumer Sentiment
surveys (in the following BCS or Sentiment indicators) have played in
the recent crisis a very important role in conveying early warnings of
relevant movement of the business cycle.
In many cases BCS indicators provided the most readily available
signals of cyclical up-turn and down-turn (while confirming recently
that they are less reliable or easy to interpret as far as cyclical
accelerations and decelerations are concerned) .
This development has strengthened the “status” of BCS Indicators in
the view of users and producers, but has also revived interest (and
doubts?) about the institutional underpinnings of their production
process. Due mainly to a specific history and tradition, BCS surveys
have remained in many countries outside the realm of Official
Statistics.
Moscow, 17-19 November 2010
3
Mainstreaming
Sentiment data as
Official Statistics
2. The question: is the inclusion in the framework of
Official Statistics useful for BCS indicators?
As far as the definition of Official Statistics (OS) is concerned ,
we can consider a specific framework of codified practices.
Here the framework is the one defined by the European
Statistical System Code of Practice that fixes a set of principles
covering “the institutional environment, statistical processes and
outputs” of Eu statistics.
Taking this framework as the reference point, it seems worth
starting to examine the potential (beneficial) effects of applying
the principles of Official Statistics to the production of BCS
indicators.
Then, questions arise about the possible ways to bring BCS and
the derived indicators into the realm of Official Statistics (that
does not mean to transfer the production oof this indicators to
NSIs).
Moscow, 17-19 November 2010
4
Mainstreaming
Sentiment data as
Official Statistics
3. Features of Official Statistics that producers of
sentiment indicators should adhere to
A tentative list of the key principles of OS that seem at stake in
the production of BCS Indicators when the process is carried out
outside official statistics:
Applying adequate quality standards (typical of Quality
assurance frameworks routinely utilised in OS).
Working in an Institutional setting able to ensure the best
standards of independence (crucial for the relationship with
respondents and users).
Using best practices to disseminate the results (accessibility,
clarity, equal treatment of all users...)
Moscow, 17-19 November 2010
5
Mainstreaming
Sentiment data as
Official Statistics
4. General problems concerning quality standards
in the BCS domain
Indeed sound methodologies and commitment to quality are not
at all exclusive features of Official Statistics.
However, the issue is how to ensure that quality standards are
always prevailing with respect to any other consideration (for
instance deriving from budget constraints or, to be explicit, from
market forces).
Official Statistics has developed comprehensive frameworks that
are specifically devoted to monitor the quality of the collection,
processing and dissemination of statistics: it seems
straightforward to apply it to BCS but the problem is how to
make this practice a generalised standard .
The quality framework for BCS developed by Oecd and by EC
(in terms of principles and methodologies) is an important step
but fall short of the comprehensiveness of standards developed
for OS domains (for instance for the PEEIs).
Moscow, 17-19 November 2010
6
Mainstreaming
Sentiment data as
Official Statistics
5. Specific methodological issues for BCS
Two examples of highly relevant methodological issues that is
very difficult to tackle outside the framework of Official Statistics:
The quality of the surveys depends crucially on using a
regularly updated statistical business register (BR) in the
definition of the sampling design and the reference list of units; for
producers outside OS this can be a condition impossible to fulfill.
It is crucial for the quality of the indicators (namely the estimate
of the share of each response modality) that the definition of the
weighting system is based on robust information; the availability
of detailed structural statistics (about value added or turnover) can
be insufficient outside the OS system.
A host of other issues concerning the maintenance of the BR and
the weighting system could be listed.
Moscow, 17-19 November 2010
7
Mainstreaming
Sentiment data as
Official Statistics
6 The relevance of the Institutional setting
There are important provisions built in the institutional setting of
Official Statistics that can affect positively the quality of BCS
surveys (in particular data collection):
Data collection is favored by the existence of a legal mandate
to collect information for statistical purposes; in particular, it is
important that business operators are compelled by law to
deliver the requested information.
The guarantee about the privacy of data providers, the
confidentiality of the information they provide and its use only for
statistical purposes are crucial to maintain a satisfactory degree
of collaboration (and response rate).
 The response burden (also across different surveys) must be
taken into account in the design of the surveys.
Moscow, 17-19 November 2010
8
Mainstreaming
Sentiment data as
Official Statistics
7 Aligning to OS the practices concerning the
dissemination of sentiment indicators
An aspect of practices used in the domain of BCS indicators that
can differ from those recommended by Official Statistics is the
one concerning the dissemination of results. In this respect, the
reference to the Eu Code of Practice is straightforward:
“– Statistical release dates and times are pre-announced.
– All users have equal access to statistical releases at the same time
and any privileged prerelease access to any outside user is limited,
controlled and publicised. In the event that leaks occur, pre-release
arrangements should be revised so as to ensure impartiality.
– Statistical releases and statements made in press conferences are
objective and non-partisan.”
“– Metadata are documented according to standardised metadata
systems.
– Users are kept informed on the methodology of statistical processes
and the quality of statistical outputs with respect to the ESS quality
criteria.
Moreover it must be stressed that users should benefit of data
dissemination instruments integrating all the available indicators
Moscow, 17-19 November 2010
9
Mainstreaming
Sentiment data as
Official Statistics
Few remarks about how to tackle the issue of
applying OS principles to BCS Indicators
 The process of discussing if the domain of BCS Indicators needs
to fully adhere to the principles of Official Statistics must be brought
at the adequate levels of governance of Statistical Systems (may be
in steps).
If a positive consensus emerge, the Statistical System has to
devise the channels to open the discussion with the producers that
are outside the OS framework (because of an institutional status or
because the relevant Code of Practice is not binding).
 The role of EC (DG EcFin and Eurostat) and Oecd can be central
in developing initiatives to step up the process of armonisation,
heading in the direction of proposing standards very close to the one
defined in OS.
 The pressure created by supranational initiatives should compound
with an active stance of national authorities in opening a
“negotiation” with producers of BCS indicators.
Moscow, 17-19 November 2010
10