1. Emily Witt (INEXDA, ECB) 14 November 2018
International Network for Exchanging Experience on Statistical Handling of Granular Data (INEXDA)
Emily Witt
(INEXDA, ECB)
14 November 2018

2. INEXDA‘s General Mission
INEXDA is an open network
to exchange experiences on statistical handling of granular data
for central banks, national statistical institutes and international organisations
INEXDA aims at
investigating possibilities to harmonise access procedures and metadata structures
developing comparable structures of existing data and
further fostering efficiency of statistical work with granular data
Ultimately, in line with G20 Data Gaps Initiative,
INEXDA aims to facilitate use of granular data for analytical, research and comparative purposes by users within and outside the participating institutions, within the limits set by the applicable confidentiality regimes.

3. INEXDA: The Granular Data Network
founded by 5 central banks on 6 January 2017
others joined:
guests:
Central banks of AT, CH, MX, RU
BIS, Eurostat
NSIs of DE, UK

4. Work Programme for the first two Years
1. Inventory of data in all member institutions
2. Inventory of existing data access procedures
3. Dissemination of INEXDA results
Agreement on unified metadata schema
Setup of a platform to collect and exchange metadata
Start harmonising metadata across INEXDA members
ECB pilot collection of information on access for researchers
ADRF for INEXDA proposed by Julia Lane (NYU)
Workshop on data access in Q1 2019
Presentations at conferences
Website

5. How to join INEXDA
Participation in INEXDA is open to other central banks, statistical institutes and international organisations.
INEXDA is governed by an MoU, that every member has to sign.
Sharing of granular data between INEXDA members not part of this MoU.
Interested institutions can join as guests before becoming a member.

6. Thank you for your attention!
Contact:

7. References INEXDA is governed by a Memorandum of Understanding
Ninth IFC Conference "Are post-crisis statistical initiatives completed?“ – INEXDA papers
Presentation: INEXDA - The granular data network
Presentation: An introduction to INEXDA's metadata schema
Working Paper: INEXDA - the Granular Data Network
IFC reports on
data-sharing: issues and good practices
the sharing of micro data – a central bank perspective
Recommendation # II.20 of the G20 Data Gaps Initiative
(page 40)
(page 5, bullet point 15)
Proceedings of the G20 Workshop on Data Sharing (January 31-February 1, 2017) and related IAG report

8. INEXDA is gaining momentum…
1st INEXDA meeting in Lisbon
2nd INEXDA meeting in London
3rd INEXDA meeting in Paris
4th INEXDA meeting in Basel
INEXDA members
(+CL, ECB, ES, TR)
Guests: AT, CH, BIS, DE (NSI), Eurostat, MX, RU, UK (NSI)
INEXDA members
(DE, FR, IT, PT, UK)
Guests: BIS
INEXDA members
(DE, FR, IT, PT, UK)
Guests: BIS, ECB, ES
INEXDA members
(+ECB, ES)
Guests: AT, BIS, CL, MX, TR, UK (NSI)
Jan 2017
Jul 2017
Jan 2018
Aug 2018
Memorandum of Understanding
Signing and publication
INEXDA Metadata Tool
by GESIS
Working groups
Dissemination
Metadata
ADRF
Modes of accreditation
Contracts for research projects/bodies
Modes of data provision
Output control
Risk management for published results

9. INEXDA’s Metadata Schema1
Resource Type 2
Resource Identifier 3
Name of Dataset 4
Creator 5
DOI Proposal 6
URL 7
Language of Resource 8
Publication Date 9
Availability 10
Sampled Universe 11
Sampling 12
Temporal Coverage 13
Time Dimension 14
Collection Mode 15
Unit Descriptions 16
Descriptions 17
Geographical Coverage 18
Keywords 19
Alternative Identifiers 20
Relations 21
Publications
Purpose is to foster harmonisation between INEXDA members and broaden metadata sharing within INEXDA and possibly outside
Based on the GESIS DOI registration service da|ra (GESIS is cooperating with DataCite).
Name of metadata items closely follows da|ra conventions to enable seamless DOI registration, if desired later in the project.
Basis for INEXDA metadata database that was established to store and view metadata from INEXDA members.

10. Digital Object Identifier (DOI)
DOIs are permanent and persistent identifier which is unique and cannot be deleted.
DOIs are a simple character string which provides a link to a resource.
In Germany DOIs are provided by the GESIS DOI registration service da|ra (GESIS is cooperating with DataCite).

11. Part 1: Basic Information
Resource Type 2
Resource Identifier 3
Name of Dataset 4
Creator 5
DOI Proposal 6
URL 7
Language of Resource 8
Publication Date 9
Availability 10
Sampled Universe 11
Sampling 12
Temporal Coverage 13
Time Dimension 14
Collection Mode 15
Unit Descriptions 16
Descriptions 17
Geographical Coverage 18
Keywords 19
Alternative Identifiers 20
Relations 21
Publications
Creator is a mandatory item in da|ra. May be used to provide more granular information on the data compiler
URL refers to the webpage which displays information about the dataset
Availability (controlled) describes the procedure under which the data can be accessed (eg download or on-site)
DOI Proposal provides the suggested DOI name of the dataset. A Digital Object Identifier (DOI) is a permanent, persistent identifier used for citing and tracking datasets

12. Part 2: Methods 1
Resource Type 2
Resource Identifier 3
Name of Dataset 4
Creator 5
DOI Proposal 6
URL 7
Language of Resource 8
Publication Date 9
Availability 10
Sampled Universe 11
Sampling 12
Temporal Coverage 13
Time Dimension 14
Collection Mode 15
Unit Descriptions 16
Descriptions 17
Geographical Coverage 18
Keywords 19
Alternative Identifiers 20
Relations 21
Publications
Sampling displays the type of sample design used to select the observations to present the population
Time Dimension provides information on
frequency of observations.
whether dataset structure is panel, time-series or cross-sectional
Structural breaks are defined as major events and revisions that have impacted the dataset
Examples of structural breaks include:
changes to the time frequency with which data is collected
changes to the set of collected variables
changes in the population or sampling

13. Part 3: Descriptions 1
Resource Type 2
Resource Identifier 3
Name of Dataset 4
Creator 5
DOI Proposal 6
URL 7
Language of Resource 8
Publication Date 9
Availability 10
Sampled Universe 11
Sampling 12
Temporal Coverage 13
Time Dimension 14
Collection Mode 15
Unit Descriptions 16
Descriptions 17
Geographical Coverage 18
Keywords 19
Alternative Identifiers 20
Relations 21
Publications
Unit Description provides information on the entities that are being observed in the dataset
Datasets may contain more than one unit of observation. For example, in a credit register information on the following units are collected:
Banks
Companies
Governments
Loans
Descriptions also contains detailed information on structural breaks in the dataset

14. Part 4: Relations and Publications1
Resource Type 2
Resource Identifier 3
Name of Dataset 4
Creator 5
DOI Proposal 6
URL 7
Language of Resource 8
Publication Date 9
Availability 10
Sampled Universe 11
Sampling 12
Temporal Coverage 13
Time Dimension 14
Collection Mode 15
Unit Descriptions 16
Descriptions 17
Geographical Coverage 18
Keywords 19
Alternative Identifiers 20
Relations 21
Publications
Describes relations between datasets and databases in the INEXDA metadata database…
… in a given country
… across countries
Examples of use cases in INEXDA context
Relation between datasets containing similar units (in different countries).
Dataset feeds into a ECB dataset.
Publications provides information on scientific publications related to the dataset.

15. ADRF for INEXDA proposed by Julia Lane
(New York University) – The five modules (1|2) 1
ADRF Documentation Module
Provides metadata incl. data catalogue, ownership & access
Shows missing / distribution values
Links users, authors, research products, codes and tools and data producers
Allows researchers to annotate datasets and provide codes to read in and reuse data
ADRF Collaboration Module
Collaboration and resource sharing via shared project workspaces and social tools (e.g. interactive chat, questions and answers, code sharing)
Supports analysis workflows with self-documenting, sharable code files 2
ADRF Security Module
Security is implemented in three layers: Cloud infrastructure, operational security and application layer security (FedRAMP certified)
Data security and confidentiality is enforced at the level of all 5 Safes: 3
People
Projects
Data
Environment
Results

16. The Administrative Data Research Facility (ADRF) holistic user centric data approach
Security Module
FedRAMP security certified
Data in cloud
Alternative: local servers
Data producer
Metadata
Training Module
Data
Data analysis
Code
Collaboration
Documentation
Module
Explorer links metadata, codes, tools, publications
Collaboration Module
Interactive chat and code sharing
Workspace and tools
Stewardship Module
Approval workflow, monitoring, reporting
Usage
Feedback
Data steward
Access
Workflows
Monitoring
Reporting
Data user