Full-time and part-time work and the gender wage gap

Objectives: Using Canadian data, we separate workers into full-time (FT) and part-time (PT) and document the following striking observation: that the overall gender wage gap is larger than either the full-time pay gap or the part-time pay gap, even after controlling for detailed personal and job characteristics. We link this to a differential selection by gender into full-time and part-time work, with women of higher earnings potential being overrepresented relative to men in the pool of part-time workers.

Data:

For annual earnings we use data from the Canadian Census, years 2000, 2005, 2010. We plan to include 2015 once the data becomes available in June 2018.

For hourly wages we use data from the Labor Force Survey (LFS) from 2000 until 2017, focusing on the months of March and September. The LFS has a six-month outgoing rotation panel, and this way we only record each individual once.

Methods:

(i) Linear regression analysis for annual earnings and hourly wage regressions for overall sample of workers and separately for FT and PT.

(ii) Blinder-Oaxaca mean gender wage gap decomposition for all specifications in (i)

(iii) Probability model of selection into FT and into PT

(iv) Counterfactual quantiles (Firpo-Fortin-Lemieux): re-weighting the part-time distribution based on the distribution of covariates in the full-time worker sample, by gender. Recentered influence function (RIF) regressions and gender wage gap decomposition.

Results/Expected Results: In terms of annual earnings, the overall gender gap has decreased from about 50% in 1995 to 40% in 2010, while the gender earnings gap for full-time and part-time workers has been consistent at about 25% and respectively 20% over this period. In terms of hourly wages, the overall pay gap has been constant at around 15% to 20%, without decreasing over time. There is notably no gender wage gap for part-time workers. We link differences in the annual earnings gap to differences in the relative labor supply by gender. We further link the differences in gender gaps between FT and PT workers to the selection of women with high expected productivity characteristics into PT jobs. We expect that reweighting the PT distribution with productivity characteristics from the FT sample by gender will produce a gender wage gap similar to the FT one.

Discussion: Our recommendations in addressing the gender pay gap are linked to selection into work and work-family friendly policies.