Microdata in China

1
Micro-data in China

• Sarah Cook
• Institute of Development Studies
• and
• James Keeley
• International Institute for Environment and
  Development
• University of Sussex

2
The Study key questions

• Through extensive interviews and review of data
  sources
• What are the main national social science data
  sources available through government agencies at
  national and sub-national levels?
• What other social science data sources are
  available (e.g. sample surveys by academic
  institutions, commercial survey groups, etc.)?
• What are the current restrictions on access to
  and use of social science data by overseas
  researchers?
• What are the options for negotiating access and
  use of such data whether for public access or
  on a case by case basis for specific projects?

3
Outline of key issues

• Availability China collects good quality
  national survey data, administrative data and
  smaller sample surveys.
• However, there is limited availability of meta
  data or accompanying information.
• The quality of data is reasonably high, although
  problems can be identified.
• The main issue is one of accessibility macro /
  aggregate data is readily accessible but micro
  data is extremely difficult for Chinese or
  international scholars to use.
• Various constraints and barriers to access can be
  identified, but
• opportunities exist for greater sharing within
  China and internationally

4
Chinas data collection system

• National Bureau of Statistics (NBS) responsible
  for national statistics and national accounts
• NBS is also the regulatory and supervisory body
  for data collection and implementation of
  National Statistics Law
• Administrative data (collected by government
  agencies)
• Academic and research institutions (including
  government, university, private)
• Other (e.g. public opinion firms, market research)

5
Key categories of micro-data

• censuses (e.g. population, economic, and
  agricultural censuses) (NBS)
• national sample surveys including 1 and 10/00
  sample surveys of demographic change labour
  force rural and urban households etc. (NBS)
• administrative data reported by government line
  agencies
• surveys undertaken by government agencies, e.g.
  Ministry of Agriculture (RCRE) longitudinal fixed
  point survey Ministry of Health (often in
  conjunction with NBS)
• small scale sample surveys undertaken by
  academic, government and other institutions.
  Only a few have nationally representative
  samples many are of limited value due to quality
  of sampling or implementation.

6
Examples of National Data Population, Labour
and Health
7
Data are available but hard to access

• Most micro-data collected by government are
  extremely hard and costly to access.
• Easy access only to macro (aggregate) data
• creates problems even for general descriptive
  overview e.g. lack of gender disaggregation).
• Limited availability of good quality meta data.
• High transaction costs of negotiation of terms
  and cost of use (including for Chinese users).
• Limited interaction between users and producers
  of data.

8
Some data are more easily accessible

• Small sample surveys collected by leading
  research institutions.
• Data supported by international funders Terms
  of funding often require open access (subject to
  some conditions)
• Numerous small-scale surveys exist but only a
  limited number are of sufficiently good quality.
• Those by leading institutions and international
  funding tend to be high quality.
• Are often accessible for small fee or through
  personal relationships or through collaborative
  project.
• For example

9
Data collected and held by CASS Institute of
Population and Labour Economics
10
Internationally funded data sets
11
Summary availability and access

• Almost all micro-data in China are difficult to
  access statistical data, administrative data,
  and even data collected by researchers.
• The current environment is not conducive to
  direct access to data, especially nationally
  representative micro-data sets collected by NBS.
• The major exception is research survey data
  involving Chinese academics where there has been
  international funding and where open access to
  the data has been stipulated by funders.

12
What are the barriers to access?

• Restrictions of the legal environment, in
  particular, the content or interpretation of the
  Statistical Law.
• Specific conditions relating to providing social
  science data to foreigners special permissions
  needed from NBS if foreigners are to be involved
  in a survey / data collection, or to be provided
  with primary data.
• Risk of making data available - potential
  misuse of data for which data collectors may
  ultimately be held responsible.
• Attitudes towards data sharing and its public
  use.
• Data viewed as a marketable commodity not as a
  public good.

13
Barriers to access legal environment

• Laws and regulations are a critical issue both
  for those seeking to access to micro-data
  (users), and for those seeking to place data in
  the public domain (producers).
• Four main types of regulation exist which place
  limitations on access to original data
• rules on data protection and protecting the
  anonymity of data subjects
• rules on state secrets, endangering national
  security or economic and social stability
• rules on who can carry out surveys
• rules on releasing data to foreigners
• Lack of clarity means responsibility for misuse
  of data by others could be blamed on collectors
  of original data.

14
Risks to data producers

• Grey areas or lack of clarity in relation to
  what counts as sensitive information,
  inappropriate use of information, and who can be
  held responsible.
• Data viewed as politically sensitive, for
  instance, include school dropout rates and
  unemployment data.
• Result is that both government and academic
  research community are risk-averse in sharing
  information.
• Researchers can present analysis based on primary
  data in academic and government policy
  communities, but this might become more sensitive
  if picked up in the media and e.g. used to
  highlight social problems.
• Researchers aim to avoid negative repercussions
  if analysis by a secondary user results in
  media or other attention.

15
Other barriers to access

• Data are not regarded as a public good (even when
  publicly funded) data is often not shared even
  among government or NBS departments.
• Cultural attitudes place little value on sharing
  data.
• Competition among researchers or institutions for
  publications and outputs.
• Little recognition attached to production of good
  quality data (e.g. if used by other researchers).
• Data as a commodity use of data as an economic
  resource to generate income.
• Few incentives for cooperation e.g. in creating
  data banks for shared use.

16
Practical obstacles to making data public

• Lack of accessible information about what data
  exists in particular fields.
• Lack of good meta-data especially in English.
• Costs of preparing data for public use.
• Potential time for responding to questions by
  other users and managing public access data.
• High costs of translation of associated
  descriptive materials or meta data.
• No institutions or funding sources dedicated to
  managing data in public domain or funding above
  activities.

17
Problems facing users

• Chinese and international researchers both face
  problems of access.
• Domestically there is limited sharing across
  research institutions and limited access to
  government or NBS data.
• Especially for international researchers,
  difficulties of accessing information about
  existing data sets and related meta data.
• Data use generally needs to be negotiated (and
  paid for) on a case-by-case basis, with high
  financial and transaction costs (even for major
  actors such as World Bank).
• Access depends principally on building
  relationships and research collaboration with
  researchers or institutes in China.
• Even then, restrictions are often placed on use
  which can make analysis more time consuming.

18
Promising developments do exist

• CASS discussion of data / resource library.
• Beijing University (CCER) with support from
  Michigan archive of older data sets.
• ISDPP Beijing Normal University Social Policy
  Analysis Information Center.
• Other initiatives in areas of health and
  population (with international funding).

19
What more can be done?

• Create incentives for making data a public good
• Increase resources to document existing micro and
  meta data in a consistent format
• Strengthen the capacity and financing to clean,
  document and manage data sets for greater
  accessibility
• Invest in institutional infrastructure such as
  data archives for ease of access and sharing
• Reducing the risks and clarify regulations on
  data use and sharing
• Funders conditions for making data public

20
What can ESRC offer China?

• Possible financial or technical assistance.
• Respond to interest among Chinese institutes in
  data management, preparation and related
  services.
• Identify and work with institutions concerned
  with making data more open.
• Training courses and sharing of training
  materials.
• Create easily accessible information / data base
  of available data sets or contacts.
• Provide access to international data sets of
  interest to Chinese researchers.
• Reciprocal arrangements with UK or EU research
  institutions or data banks.