Research Papers

The “Sheepskin Effects” of Canadian Credentials
This paper re-examines the “sheepskin effects” of educational credentials in Canada using data from the 1996 Census and Survey of Labour and Income Dynamics. I found that the estimated credential effects are sensitive to specifications. Regressions analysis in the standard model may not be adequate to control for the workers’ productivity difference unrelated to the credentials. Particularly, the misspecification of the earnings equation and pooling sample might introduce biases into the estimates of credential effects. With carefully constructed comparison groups, the estimated sheepskin effects of a Bachelor’s degree are smaller than that reported in Ferrar and Riddell (2002).
[Download the paper in pdf format]

The US/Canada Difference in Postsecondary Educational Choice
This paper attempts to tackle the puzzle of why more Canadians choose community colleges over universities than their American counterparts, when previous research has suggested that the return to community college education is low in Canada. Using data from the Survey of Labour and Income Dynamics for Canada and the National Longitudinal Survey of Youth 79 for the US, I estimate returns to education with a trinomial selection correction using various instruments. I simulate the educational choices of Canadians who face American returns to education, and vice versa. I found that Canadians have a relatively strong incentive to choose community colleges if occupational choices are controlled for. The second finding is that Canadian universities and colleges specialize in different types of human capital. Also, my analysis confirms that the elasticity of educational attainment to tuition and fees is low. Finally, the self-selection processes in the two countries are different. More able Americans have higher educational attainment while more productive Canadians prefer going to universities but not community colleges.
[Download the paper in pdf format]

The Determinants of Participation in Adult Education and Training in Canada. Joint with Jeffrey Smith (University of Michigan)
This paper examines the determinants of participation in, and the amount of time spent on, public and private adult education and training in Canada.  Using the master file data from the 1998 Adult Education and Training Survey, we estimate probit models of adult education and training (hereafter just “training”) incidence and hurdle models of total time spent in training.  Consistent with the literature, we find that relatively advantaged workers, such as those who have completed high school, are working full time, and work at large firms, acquire more training, often with financial help from their employers.  Direct government-sponsored training represents a relative minor component of total training, and is not well targeted to the disadvantaged.  This is both surprising and problematic, as the primary justification for government-financed training is to overcome credit constraints among the low skilled and the secondary justification is redistribution.  We find large differences among provinces in the incidence of training; this variation appears to result from differences in provincial policies related to training.
[Download the paper in pdf format]

The Labor Market Impacts of Adult Education and Training in Canada Joint with Jeffrey Smith (University of Michigan)
In this report, we use the data from the Adult Education and Training Survey (AETS) 1998 to estimate the impact of participating in adult education and training on the employment and earnings of Canadians.  We apply methods that assume “selection on observables”, including both standard regression-based methods and propensity score matching methods.  We also apply methods based on instruments or exclusion restrictions, including standard instrumental variables estimation and the well-known Heckman bivariate normal selection estimator.  This methods aim to deal with “selection on unobservables”.  We find that none of the methods we examine produce plausible estimates of the impact of adult education and training, although the methods that assume selection on observables produce more reasonable estimates than those that assume an instrument or exclusion restriction.  Based on the results of our analysis, we suggest improvements in the AETS data that would make it a better tool for estimating the labor market impacts of adult education and training.
[Download the paper in pdf format]

Issues in the Design of the Adult Education and Training Survey Joint with Jeffrey Smith (University of Michigan)
This paper considers issues associated with using the Canadian Adult Education and Training Survey (AETS) to study participation in, and the impacts of, public and private education and training courses and programs.  It is based on our experience in using these data to conduct both types of analysis, the results of which we report in “The Determinants of Participation in Adult Education and Training in Canada” and “The Labor Market Impacts of Adult Education and Training in Canada”.
[Download the paper in pdf format]