LAMAS Working Group 13-14 June 2019 Agenda Item 3.1 TF Labour Market Flows: main results Hannah.Kiiver@ec.europa.eu
TF Labour Market Flows One meeting (Luxembourg, February 27&28 2019) TF Members: BE, ES, FI, FR, IE, IT, NL, PT, UK, CH; ECB, DG EMPL Main purpose: agree on a methodology to allow the production of year-on-year (annual) flows at Eurostat Discuss experience with quarter-on-quarter (quarterly) flows, agree on some additions for quarterly flows
Why do we need annual flows? Quarterly flows describe seasonality and, if adjusted for seasonality, short term movements The average of quarterly flows just shows the average of these short term movements Annual flows describe structural changes High demand for this information: DG EMPL (DG ECFIN, DG REGIO) ECB, WHO/ILO General public
Why another Task Force? TF Flows recommended a method for quarterly flows, why not just use this? During thattask force, two issues were identified which are more relevant for annual flows than they are for quarterly flows: Movements in and out of scope of the longitudinal sample population Attrition of the sample
Full transition matrix
Histogram of population changes in % for all years (2010-2017) and all countries
Annual flows – movements in and out of scope Eurostat does not have the necessary data to fill the matrix A simple correction approach from the BLS was judged to be of no additional value Preference to keep consistency, since no good solution can be found to account for movements
Annual flows - attrition Eurostat cannot differentiate between individuals moving internationally, nationally, non-response, refusal, death Attrition estimates encompass thus all of these categories Figures on next slide are expressed as average of share of those lost to the sample based on rotational pattern and interview wave when comparing two quarters (either consecutively, or two consecutive years)
Annual flows – attrition (2016-2017)
Annual flows- options to correct for differential attrition Initial reweighting step to match target period margins for ILO status (by age group and sex) Add more variables into reweighting Some relevant variables are missing Risks associated with excessive reweighting Modelling not an option Missing information on reason for drop-out No studies on type of bias, no basis for models
Recommendation for annual flows Start publication of annual flows based on current method used for quarterly flows (but restriction of age group 15-74 only in target period) simple and easily replicable no good alternative options Eurostat: test the BLS method to check impact for countries with large population changes Eurostat: test initial reweighting including additional variables (region, degree of urbanisation, educational attainment)
Breakdowns for annual flows Same approach as for quarterly flows Requests: EU aggregates by duration of unemployment, 8 levels Transitions between ISCO groups Transitions between hours worked categories Transitions between contract types
Recommendations for quarter-on-quarter flows Restriction of age group 15-74 only in target period Produce seasonally adjusted quarterly flows without additional raking step Simplify regression: skip interaction terms New flows from microdata with longitudinal weights Detailed breakdowns for EU aggregates
Publication and communication strategy Annual flows will be published as regular Eurostat statistics Breakdowns will be published as "experimental" statistics First publication will be accompanied by press release Statistics Explained page will be extended to cover annual flows as well
EoV Apologies to Austria for misrepresenting their method in the report BLS = Bureau of Labour Statistics Different rotational patterns – comparability questionable?
Eurostat would like to thank all members of the TF for their contributions and their time. The members of the LAMAS WG are invited to comment on the recommendations of the TF.