Attrition and the estimation of employment and wage equations: the Canadian survey of labour and income dynamics

Abstract

This article provides an analysis of the effects of attrition on employment and wages using the Canadian survey of labour and income dynamics. We consider a structural model composed of three freely correlated equations for nonattrition, employment and wages. The model is estimated using microdata from 22 990 individuals who provided sufficient information in the first year of the 1996–2001 panel. The main findings of this article are that attrition is not random. Attritors likely are less attached to employment and come from low-income population. The correlation between nonattrition and employment is positive and statistically significant, though small. Also, wage estimates are biased upwards. Observed wages are on average higher than wages that would be observed if all the individuals initially selected in the panel remained in the sample.

Keywords:

• panel data
• attrition
• selection
• Canada

JEL Classification:

• C33
• C34
• J21
• J31

Notes

¹ See, for instance, the Special Issue ‘Attrition in Longitudinal Surveys,’ of The Journal of Human Resources, Spring, 1998, Vol. 33, No. 2.

² For further information on the SLID design and data processing, refer to the SLID Microdata User's Guide – Statistics Canada (available at: http://www.statcan.gc.ca/pub/75m0001g/75m0001g1997000-eng.pdf).

³ In the empirical estimation, we consider the composite hourly wage for all paid-worker jobs held by the respondent duringyear t.

⁴ See Gourieroux and Monfort (1996) for discussion and statistical background.

⁵ There is little change in the results when increasing this number.

⁶ The remaining estimates are available upon request by contacting the corresponding author.

⁷ We estimated nonattrition and employment equations separately, and we obtained estimates very close to those when the equations are estimated jointly.

⁸ The gap is calculated by multiplying minus the selection coefficient times the mean value of the correction term.