Using National Labour Data in Cross-National Comparative Studies on Precarious Employment

IASSIST 36th annual conference, Ithaca, NY: 2010.06.03

Walter W. Giesbrecht
York University, Toronto

Abstract:

Attempting to compare labour data from different countries or regions can be fraught with danger, since definitions of concepts can vary tremendously, and gaps exist in the data collected. Apparently similar concepts such as part-time work, or permanent vs. temporary employment, are often not strictly comparable using published aggregate data. Surveys designed to gather data for national policy reasons do not automatically generate data that are comparable for use in cross-national studies. Harmonizing these concepts and definitions involves considering their deployment in national contexts in order to assure that cross-national comparisons are truly comparing likes with likes. The Comparative Perspectives Database (still under development) is a project that is attempting to generate comparative multi-dimensional data tables on aspects of precarious employment, using the microdata from a total of seven surveys spanning thirty countries (Canada, United States, European Union (EU-27) and Australia). I will discuss in detail some of the problems we encountered in producing a codebook that would allow us to produce useful cross-national data tables, as well as describe other similar projects conducted elsewhere.

Link to other presentations in the same "Stir it up: comparative data" session

Gender & Work Database

• project began 2001
• looked at different aspects of the relationship between gender and work, particularly precarious employment
• only Canadian data
• details at www.genderwork.ca

typical multi-dimensional table

definitions

• standard employment relationship
  the employee has one employer, works full-time, year-round on the employer's premises under his or her supervision, enjoys extensive statutory benefits and entitlements and expects to be employed indefinitely
  • normative model for (white) male employment in Canada since WW2
• precarious employment
  forms of employment normally involving atypical contracts, limited social benefits and statutory entitlements, job insecurity, low job tenure, low wages and high risks of ill health

problems

• most OECD countries have experienced substantial growth in one of more forms of PE (self-employment, homework, on-call work, part-time work or temporary work)
• standard employment relationship remains the model upon which labour laws, legislation and policies are based
• individual surveys not well-equipped to look at this phenomenon

GWD --> CPD

• GWD looks at Canada only
• CPD (Comparative Perspectives Database) will look at many more countries
  
  

CPD home page (beta)

has this been done before?

Luxembourg Income Study (LIS)

• begun in 1983, LIS harmonizes nationally representative micro-level survey data for over 30 countries
• primarily income and wealth data, but includes labour market and demographic data as well
• limitations:
  • indirect access to the underlying microdata
  • most researchers must accept the LIS standardization rules
  • cross-sectional data only
• details at www.lisproject.org

Cross-National Equivalent File (CNEF)

• CNEF 1970-2008 contains equivalently defined variables for
  • British Household Panel Study (BHPS)
  • Household Income and Labour Dynamics in Australia (HILDA)
  • Korea Labor and Income Panel Study (KLIPS)
  • Panel Study of Income Dynamics (PSID)
  • Swiss Household Panel (SHP)
  • Canadian Survey of Labour and Income Dynamics (SLID)
  • German Socio-Economic Panel (SOEP)
• longitudinal data only; also contains set of constructed variables
• details at http://www.human.cornell.edu/che/PAM/Research/Centers-Programs/German-Panel/cnef.cfm

Comparative Perspectives Database

• rose out of a workshop held in Toronto in Nov 2007
• will contain data from
  • EU Labour Force Survey (EU-LFS)
  • European Community Household Panel (ECHP) and Statistics on Income and Living Conditions (EU-SILC)
  • Household Income and Labor Dynamics of Australia (HILDA)
  • U.S. Current Population Survey (CPS), primarily the Contingent Work Supplement (CWS)
  • U.S. Panel Study of Income Dynamics (PSID)
  • Canada: Labour Force Survey (CA-LFS)
  • Canada: Survey of Labour and Income Dynamics (SLID)

why not use existing studies?

• insufficient detail in areas of interest (# countries or variable characteristics)
  (e.g., countries involved, industry and occupational coding, not enough intersections between gender and immigration)
• wanted to use harmonized data wherever possible
• our concept of precarious employment differs conceptually from non-standard and contingent work
  • We are interested profoundly in the intersection between regulation, law, policy and statistics, and in asking how and in what ways certain regulatory decisions 'make' some jobs insecure and extend protections to other jobs. Standard measures obsure these important processes. Even harmonization is tricky; we want to harmonize in a conceptual way -- e.g., compare forms of work with similar characteristics of insecurity but that may in fact be different. The main point is that our concept of precarious employment differs conceptually from non-standard and contingent work.

more definitions

• Harmonized variables are those that exist for each country, whose coding differs in principle across countries, but preserves as much of the original detail as possible
• Standardised variables are those that have been recoded in order to fit a fixed coding scheme

(from www.lisproject.org)

construction of harmonized codebook

• focus on variables that permitted examination of dimensions of labour market insecurity
  • degree of certainty of continuing employment;
  • regulatory protection;
  • control over the labour process;
  • income level

five principles of harmonization

• practicality, i.e, do the data exist? Can they be compared?
• comparability, but not at any price
• meaningful classifications
• maintaining the smallest level of granularity
• pointing to silences/invisibilities in the data

example of difficulties: part-time work

V10G1: Full Time/Part Time -- based on survey's definition

• ECHP, SILC, EU-LFS - self identify
• CA-LFS, SLID - based on 30+ hours being full time
• HILDA, CPS, PSID - based on 35+ hours being full time

example of solutions: part-time work

V10G2: Full Time/Part time (based on 35 hours)

V10G3: Full Time/Part Time (based on hours with two types of part time)

• 14 or less hours
• 15-34 hours
• 35+ hours

Thanks to
Leah Vosko and Heather Krause
for their guidance and moral support.