2024-06-10-世界银行-全球劳动力数据库用户手册_全球劳动力数据库的理解_使用和交互指南（英）_143页_2mb

Global Labor Database (GLD) User Manual Summary

Core Content of the GLD

The Global Labor Database (GLD) is a World Bank initiative aimed at harmonizing labor force surveys (LFS) and household surveys with labor modules. It provides a platform for researchers and analysts to access, use, and expand harmonized labor data for cross-country and cross-time comparisons.

Main Objectives

1. Create a harmonized database of labor surveys with reliable and comprehensive data for analytical work.
2. Allow users to customize and expand the harmonization by providing full access to the codes and documentation, enabling deeper analysis.

Key Features

• Open and transparent approach, allowing users to trace and deviate from the standard harmonization.
• Complements I2D2 and GMD by focusing more on labor market details and including industry and occupation classifications using ISIC and ISCO codes.
• Sustainability through collaboration and community engagement on GitHub.

Main Audience

The intended users of GLD include:

• Researchers and data analysts
• Practitioners in international development
• Statistical offices
• Ministries of labor, economy, and planning
• Government agencies analyzing labor data

Users can leverage GLD in two ways:

1. As-is harmonization: Using the harmonized data files directly.
2. Amended or hacked harmonization: Editing the harmonization code or adding new variables to suit specific research needs.

Principles of GLD

1. Coverage and Expansion

• As of April 2024, GLD includes 345 surveys from 24 countries (1 high-income, 9 upper middle-income, 11 lower middle-income, and 9 low-income).
• The countries covered are: ARM, BGD, BRA, CHL, COL, EGY, ETH, GEO, IDN, IND, LKA, MEX, MNG, NPL, PAK, PHL, RWA, SLE, THA, TUR, TUN, TZA, ZAF, ZMB, ZWE.
• Surveys are updated to include the latest data (from the previous four years) to ensure timeliness.
• Coverage is balanced across regions and income groups, though limitations may exist due to data availability and NSO restrictions.

2. Transparency and Data Access

• All harmonization steps, including documentation, code, and survey details, are transparent and traceable.
• Outputs such as harmonization code and survey documentation are freely shared on GitHub.
• Raw and harmonized microdata are access-restricted based on data privacy rules and NSO policies.
• Access to raw data is limited to World Bank colleagues or via NSO portals in some cases (e.g., Egypt).

3. Data Quality and Validation

• GLD ensures high-quality data to support reliable cross-country comparisons.
• Three main validation tools are used:
  • Manual validation with NSO and country office colleagues.
  • Automated checks for data integrity and coherence with external sources (e.g., ILO, WDI).
  • Automated checks for consistency across survey series over time.
• Users can report issues through GitHub or direct communication with the GLD team, which will then update the harmonization as needed.

Data Types and Storage

1. Raw Microdata

• Individual-level data as received from NSOs, aggregators, or colleagues.
• May be in formats like CSV, TXT, SPSS, etc.
• Converted to Stata .dta format for consistency.
• Access rights depend on the data sharer and are categorized into:
  • Public domain: Free for all users.
  • World Bank official use: Accessible within the World Bank.
  • Limited release: Access requires permission from the data sharer.

2. Harmonized Microdata

• Output of the harmonization process.
• Contains only variables that have been successfully coded.
• Stored in Stata .dta files.
• Access rights are inherited from raw data.

3. Harmonization Code

• Instructions to convert raw data to harmonized data.
• Stored as Stata .do files.
• Shared under the MIT License.
• Includes comments explaining the rationale behind coding decisions.

4. Other Code

• Includes:
  • Raw data conversion code
  • Quality check code
  • Code templates
  • Ecosystem tools code (e.g., for ISIC/ISCO code conversion)
• All code is shared on GitHub and under the MIT License.

5. Survey and Documentation

• Includes:
  • Questionnaires
  • Enumerator manuals
  • Reports
• Stored in PDF or spreadsheet format.
• Country Survey Details document changes in survey definitions (e.g., employment classification) and are accessible to all.

Data Storage Structure

• All data is stored on the GLD Server, organized by country and survey vintage.
• The GLD WB Staff Server is a subset:
  • Only includes surveys with freely shareable raw data.
  • Only contains the latest harmonized version.
• Folder structure example for Armenia:
  • ARM_2014_LFS (country and year)
  • ARM_2014_LFS_V01_M (master data version)
  • ARM_2014_LFS_V01_M_V01_A_GLD.dta (harmonized output)
  • ARM_2014_LFS_V01_M/Doc (contains questionnaires, reports, and technical documents)
  • ARM_2014_LFS_V01_M/Programs (contains harmonization code)

Access and Collaboration

• Access to the GLD Server is restricted to the GLD team and World Bank staff under specific conditions.
• The GLD WB Staff Server is accessible to all World Bank staff and is used for easier access to the latest versions.
• GitHub is the primary platform for:
  • Sharing code and documentation
  • Collaborating on harmonization and data expansion
  • Receiving user feedback and suggestions for improvements
• Users can contribute to GLD by:
  • Sharing new raw data
  • Providing harmonization code for new surveys
  • Collaborating on new or expanded harmonizations
  • Correcting or expanding existing harmonizations and tools