Comments on the Draft National Policy on Official Statistics

Edited by Swaraj Barooah. Download a PDF of the submission here

Preliminary

CIS appreciates the Government’s efforts in realising the importance of the need for high quality statistical information enshrined in the Fundamental Principles of Official Statistics as adopted by the UN General Assembly in January 2014. CIS is grateful for the opportunity to put forth its views on the draft policy. This submission was made on 31st May, 2018.

First, this submission highlights some general defects in the draft policy: there is lack of principles guiding data dissemination policies; there are virtually no positive mandates set for Government bodies for secure storage and transmission of data; and while privacy is mentioned as a concern, it has been overlooked in designing the principles of the implementation of surveys. Then, this submission puts forward specific comments suggesting improvements to various sections in the draft policy.

CIS would also like to point out the short timeline between the publication of the draft policy (18th May, 2018), and the deadline set for the stakeholders to submit their comments (31st May, 2018). Considering that the policy has widespread implications for all Ministries, citizens, and State legislation rights (proposed changes include a Constitutional Amendment), it is necessary that such call-for-comments are publicised widely, and enough time is given to the public so that the Government can receive well-researched comments.

General Comments

Data dissemination

For data dissemination, the draft policy does not stress upon a general principle or set of principles, and often disregards principles specified in the Fundamental Principles of Official Statistics, which are the very principles the Government intends to draw its policies on official statistics from. Rather it relies on context-specific provisions that fail to summarise and articulate a general philosophy for the dissemination of official statistics, and fails to practically embody some stated goals. The first principle on Official Statistics, as realised by the United Nations General Assembly, clearly states that: “[...] official  statistics  that  meet  the  test  of  practical utility  are  to  be  compiled  and  made  available  on  an  impartial  basis  by  official statistical agencies to honour citizens’ entitlement to public information.”

Let us compare this with Section 5.1.7 (9) of the draft policy, which refers to policies regarding core statistics: it mentions a data “warehouse” to be maintained by the NSO which should be accessible to private and public bodies. While this does point towards an open data policy, such a vision has not been articulated in any part thereof.

The draft policy, at the outset, should have general guiding principles of publishing data openly and freely (once it meets the utility test, and it has been ensured that individual privacy will not be violated by the publishing of such statistics). This should serve well to inform further regulations and related policies governing the use and publishing of statistics, like the Statistical Disclosure Control Report.

A general commitment to a well-articulated policy on data dissemination will ensure easy-to-follow principles for the various Ministries that will refer to the document. The additional principles that come with open data principles should also be described by the policy document: a commitment to publishing data in a machine-readable format, making it available in multiple data formats (.txt, .csv, etc.), and including its metadata.

Data storage and usage

In the absence of a regime for data protection, it is absolutely necessary that a national policy on statistics provide positive mandates for the encryption of all digitally-stored personal and sensitive information collected through surveys. Even though the current draft of the policy mentions the need to protect confidential information, it sets no mandatory requirements on the Government to ensure the security of such information, especially on digital platforms.

Additionally, all transmission of potentially sensitive information should be done with the digital signatures of the employee/Department/Ministry authorising said transmission. This will ensure the integrity and authenticity of the information, and provide with an auditable trail of the information flowing between entities in the various bodies.

Data privacy

It is appreciable that Section 5.7.9 of the draft policy notes, “[a]ll statistical surveys represent a degree of privacy invasion, which is justified by the need for an alternative public good, namely information.” However, all statistical surveys may not be proportionate in their invasiveness, even if they might serve a legitimate public goal in the future.

The draft policy does not address how privacy concerns can be taken into account while designing the survey itself. A necessary outcome of the realisation of the possible privacy violations that may arise due to surveys is that all data collection be “minimally intrusive”, the data be securely stored (see previous comment section, ‘Data storage and usage’), and the surveyed users have control over the data even after they have parted with their information.

Since the policy deals extensively with the implementation of surveys, the following should details should be clearly laid out in the policy:

• The extent to which an individual has control over the data they have provided to the surveying agency.
• The means of redressal available to an individual who feels that his/her privacy has been violated through the publication of certain statistical information

Specific Comments

Section 5.1: Dichotomising official statistics as core statistics and other official statistics

Comments

The reasons for dichotomising official statistics has not been appropriately substantiated with evidence, considering the wide implications of policy proposals that arise from the definition of “core statistics.”

Firstly, the descriptions of what constitutes “core statistics” casts too wide a net by only having a single vague qualitative criterion, i.e. “national importance.” All the other characteristics of the “core statistics” are either recommendations or requirements as to how the data will be handled and thus, pose no filter to what can constitute “core statistics.” The wide net is apparent in the fact that even the initially-proposed list of “core statistics”, given in Annex-II of the policy, has 120 categories of statistics.

Secondly, the policy does not provide reasons for why the characteristics of “core statistics”, highlighted in Section 5.1.5, should not apply to all official statistics at the various levels of Government. Therefore, the utility of the proposed dichotomy has also not been appropriately substantiated with illustrative examples of how “core statistics” should be considered qualitatively different from all official statistics.

This definition may lead to widespread disagreement between the States and the Centre, because Section 5.2 proposes that “core statistics” be added to the Union List of the Seventh Schedule of the Constitution. How the proposal may affect Centre-State responsibilities and relations pertaining to the collection and dissemination of statistics is elaborated in the next section.

Recommendations

The policy should not make a forced dichotomy between “core” and (ipso facto) non-core statistics. If a distinction is to be made for any reason(s) (such as for the purposes of delineating administrative roles) then such reason must be clearly defined, along with a clear explanation for why such a dichotomy would alleviate the described problem. The definitions should have tangible and unambiguous qualitative criteria.

Section 5.2: Constitutional amendment in respect of core statistics

Comments

The main proposal in the section is that the Seventh Schedule of the Constitution be amended to include “core statistics” in the Union List. This would give the Parliament the legislative competence to regulate the collection, storage, publication and sharing of such statistics, and the Central Government the power to enforce such legislation. Annex-II provides a tentative list of what would constitute “core statistics”; as is apparent, this list is wide-ranging and consists over 120 items which span the gamut of administrative responsibilities.

The list includes items such as “Landholdings Number, area, tenancy, land utilisation [...]” (S. No. 21), and “Statistics on land records” (S. No. 111) while most responsibilities of land regulation currently lie with the States. Similarly, items in Annex-II venture into statistics related to petroleum, water, agriculture, electricity, and industry; some of which are in the Concurrent or State List.

Statistics are metadata. There is no reason for why the administration of a particular subject lie with the State, and the regulation of data about such subject should lie with solely with the Central Government. It is important to recognise that adding the vaguely defined “core statistics” to the Union List, while enabling the Central Government to execute and plan such statistical exercises, will also prevent the States from enacting any legislation that regulates the management of statistics regarding its own administrative responsibilities.

The regulation of State Government records in general has been a contentious issue, and its place in our federal structure has been debated several times in the Parliament: the enactment of Public Records Act, 1993; the Right to Information Act, 2005; and the Collection of Statistics Act, 2008 are predicated on an assumption of such competence lying with the Parliament. However, it is equally important to recognise the role States have played in advancing transparency of Government records. For example, State-level Acts analogous to the Right to Information Act existed in Tamil Nadu and Karnataka before the Central Government enactment.

Recommendations

We strongly recommend that “statistics” be included in the Concurrent List, so that States are free to enact progressive legislation which advances transparency and accountability, and is not in derogation of Parliamentary legislation.

The Ministry should view this statistical policy document as a venue to set the minimum standards for the collection, handling and publication of statistics regarding its various functions. If the item is added to the Concurrent List, the States, through local legislation, will only have the power to improve on the Central standards since in a case of conflict, State-levels laws will be superseded by Parliamentary ones.

Section 5.3: Mechanism for regulating core statistics including auditing

Comments

The draft policy in Section 5.3.2 says, “[...] The Committee will be assisted by a Search Committee headed by the Vice-Chairperson of the NITI Aayog, in which a few technical experts could be included as Members.” The non-commital nature of the word ‘could’ in this statement detracts from the importance of having technical experts on this committee, by making their inclusion optional. The policy also does not specify who has the power to include technical experts as Members in the Search Committee. The statement should include either a minimum number of a  specific number or members, and not use the non-committal word “could”

The National Statistical Development Council, as mentioned in 5.3.9, is supposed to “handle Centre-State relations in the areas of official statistics, the Council should be represented by Chief Ministers of six States to be nominated by the Centre” (Section 5.3.10). The draft does not elaborate on the rationale behind including just six states in the Council. It does not recommend any mechanism on the basis of which Centre will nominate states in the council.

Recommendations

The policy should recommend a minimum number of technical experts who must be included in the search committee, along with a clear process for how such members are to be appointed.

Additionally, the policy appropriately recognises the great diversity in India and the unique challenges faced by each State. Thus, each State has its unique requirements. Since in Section 5.3.11, the policy recommends that council meet at a low frequency of at least once in a year, all States should be represented in the Council.

Section 5.4: Official Machinery to implement directions on core statistics

Comments

The functions of Statistics Wing in the MOSPI, laid out in Section 5.4.7, include advisory functions which overlap with functions of National Statistical Commission (NSC) mentioned in Section 5.3.5. Some regulatory functions of Statistics Wing, like “conducting quality checks and auditing of statistical surveys/data sets”, overlap with the regulatory functions of NSC mentioned in Section 5.3.7.

In section 5.3.1, the draft policy explicitly mentions that “what is feasible and desirable is that production of official statistics should continue with the Government, whereas the related regulatory and advisory functions could be kept outside the Government”. But Statistics Wing is a part of the government and it also has regulatory and advisory functions. It will adversely affect the power of NSC as an autonomous body.

There are inconsistencies in the draft-policy regarding the importance and need of a decentralized statistical system. In section 3 [Objectives], it has been emphasized that the Indian Statistical System shall function within decentralized structure of the system.  But, in section 5.4.15, the draft says that decentralized statistical system poses a variety of problems, and advocates for a unified statistical system. Again, in section 5.15, draft emphasizes the development of sub-national statistical systems. These views are inconsistent and create confusion regarding the nature of statistical system that policy wants to pursue.

Recommendations

The functions of the NSC should be kept in its exclusive domain. Any such overlapping functions should be allocated to one agency taking into consideration the Fundamental Principles on Official Statistics.

The inconsistencies regarding the decentralisation philosophy of the statistical system should be addressed.

Section 5.5: Identifying statistical products required through committees

Comments

While Section 5.5.2 recognises data confidentiality as a goal for statistical coordination, it does not take into account the violation of privacy that might occur due to the sharing of data. For example, a certain individual might agree to share personal information with a particular Ministry, but have apprehensions about it being shared with other Ministries or private parties.

Recommendations

We recommend that point 4 in Section 5.5.2 be read as, “enabling sharing of data without compromising the privacy of individuals and the confidentiality/security of data.”The value of of the individual privacy stems from both the recent Supreme Court judgment that affirmed privacy as a Fundamental Right, and also Principle 6 of the of the Fundamental Principles of Official Statistics. Realising privacy as a goal in this section will add a realm of individual control that is already articulated in Section 5.7.9.

Annex-VII: Guidelines on Outsourcing statistical activities

Comments

Section 6 defines “sensitive information” in an all-inclusive manner and does not leave space for further inclusion of any information that may be interpreted as sensitive. For example, biometric data has not been listed as “sensitive information”.

Section 9.1, draft says, “[t]he identity of the Government agency and the Contractor may be made available to informants at the time of collection of data”. It is imperative that informants have the right to verify the identity of the Government agency and the Contractor before parting with their personal information.

Recommendations

The definition of “sensitive information” should be broad-based with scope for further inclusion of any kind of data that may be deemed “sensitive.”

Section 9.1 must mandate that the identity of the Government agency and the Contractor be made available to informants at the time of collection of data.

Section 9.6 can be redrafted to state that each informant must be informed of the manner in which the informant could access the data collected from the informant in a statistical project, as also of the measures taken to deny access on that information to others, except in the cases specified by the policy.

Section 10.2 can be improved to state that if information exists in a physical form that makes the removal of the identity of informants impracticable (e.g. on paper), the information should be recorded in another medium and the original records must be destroyed.