An Econometric Analysis of the 'Backward-Bending' Labor Supply of Canadian Women
Data and Variables of Concern
The cross-sectional data from the Survey of Labor and Income Dynamics (SLID) Record Layout, 2009 is used for the empirical analysis. The survey includes extensive data concerning demographics, employment, unemployment, occupational history, migration, education, earnings, and parental background of all individuals in Canada, excluding residents of the Yukon, the Northwest Territories and Nunavut, residents of institutions and persons living on Indian reserves. Overall, these exclusions amount to less than 3 percent of the population12 [SLID].
For the purpose of this study, two datasets from the 2009 SLID data are used: the external cross-sectional person dataset (ecp2009pr) and the external cross-sectional economic family dataset (ec2009ef) were merged together in Stata format13. Initially ecp2009pr dataset had observations of 50,900 for both male and female individuals, and the ec2009ef dataset had observations of 26,650 households. In order to match the identity of the given female individual across the two datasets that allows us to match the data from different datasets to the right person, the two datasets were merged. After merger, data on 32,065 females who are in and out of the labor force remain. Out of the 32,065 females, the composite hourly wage of all paid jobs were observed for 16,371 females, the total hours paid all jobs observed for 24,009 females, age of 28,325 females, marital status of 28,264 females, 28,325 of females living with a spouse and not, the number of years of work experience of 24,864 females, support payments of 28,325 females, highest level of education of 28,204 females, 32,065 of females living with a child less than six years old and not, and the non-female income of 28,325 females in total were observed. See the summary statistics in Table 1.3 (Appendix) for a more detailed account of all the variables used in the analysis.Figure 1.3: Histogram of hours of labor
The hours variable consists of 24009 observations with a mean of 1129.835 and a standard deviation of 922.4242. Although the histogram of the hours of labor does not look “normal,” in order to ensure that the probability distribution of the sample average of the hours variable follows a normal curve the bootstrap technique was used. It is important for the data to be normally distributed, as without it our statistical tests will not hold. After using the bootstrap technique, the sample average of the variable hours is likely to be normally distributed from 50 simulations with a 95% confidence interval of [118.33, 1141.34] and a bootstrap standard error of 5.869992.
Figure 1.4: Histogram of wage
The wage variable consists of 16371 observations with a mean of 19.89017 and a standard deviation of 11.81493. Likewise, after using the bootstrap technique, the sample average of the variable wage is likely to be normally distributed from 50 simulations with a 95% confidence interval of [19.70432, 20.07602] with a bootstrap standard error of .094824.
Table 1.4 (Appendix) shows that the out of 28,264 observations on the marital status of females, 49% are married, 8.8% are in common-law relationship, 3.5% are separated, 6.7% are divorced, 9.82% are widowed, and 22.2% are singles. Additionally, out of 28,325 observations on females living with spouse and not, 59.3% of females are living with spouse and 40.7% are not (Table 1.5, Appendix). In Table 1.6 (Appendix), it is noted that out of 31,819 observations on females from the ten Canadian provinces, 4.4% are from Newfoundland and Labrador, 2.7% are from Prince Edward Island, 5.9% are from Nova Scotia, 5.8% are New Brunswick, 19.3% are from Quebec, 28.2% are from Ontario, 6.7% are from Manitoba, 7.2% are from Saskatchewan, 10% are from Alberta, and last but not the least 9.8% are from British Columbia. By continuing this way, it is further noted that out of 28,204 observations on the highest level of education of female, 0.4% have never attended school, 0.8% have 1-4 years of elementary school, 7.2% have 5-8 years of elementary school, 7.2% have 9-10 years of elementary and secondary school, 6.6% have 11-13 years of elementary and secondary school (but did not graduate), 15.8% have graduated high school, 7.2% have some non-university postsecondary (no certificate), 5.6% have some university (no certificate), 30.3% have non-university postsecondary certificate, 2.2% have university certificate below Bachelor’s, 12.2% have Bachelor’s degree, and 4.4% have university certificate above Bachelor’s , Master’s, First professional degree in law, Degree in medicine, dentistry, veterinary medicine or optometry, Doctorate (PhD) (Table 1.7, Appendix). And lastly, the kidslt6 variable that was created as an indicator variable by generating a variable less than six from the age of youngest person in economic family shows that out of 32,065 observations, only 9% of women in the sample have a child less than six years old and 91% do not (Table 1.8, Appendix).
In order to obtain the desired parabolic relation, i.e., the backward-bending effect between hours and wage, the square of wage is used as an independent variable. Due to the non-linear relation of age and experience on hours in the existing literature, squares of age and experience are included. The variable province is included to account for the regional differences in labor opportunities captured by regional dummies, and to account for the strong correlation between human capital and labor supply, dummies for the highest level of education attained by the individual female are also included. Other variables in the model are included for similar reasons as well as for purposes of expanding on previous studies concerning the labor supply schedule of Canadian women.
The indicators were coded and the codes were kept for identification purposes in raw data. The males’ earnings or non-female income was measured by subtracting the earnings of the female individual from the total income of a household. The total income of a single household was measured by summing up market income, old age security pension, other government transfers, other taxable income, private retirement pensions, social assistance, universal child care tax benefit, employment insurance benefits, wages and salaries before deductions, and workers’ compensation benefits.Continued on Next Page »