Labor Economics 6th Edition by George J. Borjas: Human Capital

Ability Bias

If even two people have got same level of schooling but they may earn different level of earnings, and it is due to their differences in respective abilities, which is known as Ability Bias in course of calculating effect of schooling on people as if workers comprising different level of abilities do not fall on same wage-schooling locus. And this inability of understanding about ability bias can lead policy makers to take some policies which are not needed now or may lead to give more importance to schooling all people which may not be proven profitable societally even at some circumstances.

Estimating the Rate of Return to Schooling

After adjusting the data for differences in other worker characteristics such as race, sex and age, the typical method for estimating the rate of return to schooling is calculated using the data on the earnings and additional year of schooling of different workers and then finds out percentage wage differential. Generally used regression model regarding this is,

 log w= bs + Other variables

Where, w= worker’s wage

s= numbers of years of schooling pursued by the employee

b= coefficient that indicates percentage wage differential between two staffs who differ by one year of schooling when other variables are let to be constant

But even though the data set is adjusted to sex, race and age but this model cannot evaluate the real situation of wage differences in a worthy way, due to one year more schooling, due to the differences in ability such that IQ level of workers.

Even if we take persons comprising same ability level, then also the study shows different results, as if some early studies reported that the rate of return to schooling in a sample of identical twins was 3 percent and most recent studies show that it is 15 percent (Borjas, pp 251, 2013). It means there some other factors which are controlling the Discount Factor of laborers when to evaluate the effect of schooling in measure of wage differences and without measuring the discount factor properly rate of return to schooling cannot be measured in the justified way.

Maximization of Workers Lifetime Earnings

To be out of the problem that is inherited in Schooling Model, if we be able to find the age-earnings profile of a particular employee in both situation, i.e., when he has chosen to go to college and when he has chosen to stop after high school, then by comparing the two categories we can test the main hypothesis of schooling model. Though it sounds simple to conduct, but we cannot because once a worker makes a particular choice, we can only observe the earnings stream associated with that choice. And the solution comes to this thing comes from Borjas (2013),

“Even though we will never observe how much a worker who quits after completing high school would have earned if he had attended college, we do observe the earnings of those workers who did attend college. We could then predict the high school graduate’s earnings had he attended college by using the observed data on what college graduates actually make. Similarly, even though we do not observe how much college graduates would have earned had they stopped after high school, we do observe the earnings of high school graduates. We could then predict the college graduate’s earnings (had he not attended college) from the salary data for high school graduates.”

But this is valid when only those pupils and workers lie on similar wage-schooling locus.

Selection Bias and Corrections to it

Let us start with a numerical example,

Worker	Earnings in Blue-Collar Job	Earnings in White-Collar Job
Hafiz	$20,000	$40,000
Latifee	$15,000	$41,000

Here, Hafiz is better in blue-collar job and Wendy is better with White-Collar job that is why they are earn better  than other one at their own particular  area.

Assume that both faces discount rate of 10 percent. So,

Hafiz’s present value if he does not go to school= 20,000+ (20,000/1.1) = $38,182

Hafiz’s present value if he goes to school= 0+ (40,000/1.1) = $36,364

These suggest, Hafiz will not go to school as his Present Value is higher in terms of $ earned when he does not go to school.

Again for Wendy,

Latifee’s present value if she does not go to school= 15,000+ (15,000/1.1) = $28,636

Latifee’s present value if she goes to school= 0+ (41,000/1.1) = $37,273

These above calculations indicate that Latifee will go for school and then will take white-collar job.

But if we do overall comparison between Hafiz and Latifee, then we would say Latifee did wrong by taking educations and should have gone to the blue-collar job. Hence, this kind of situation is known as Selection Bias cause, Latifee necessarily is not efficient for blue-collar job but additional level of education has made him to do white-collar job. However it is known as selection bias as the inference is contaminated by rejecting their particular proficiency on respective distinct fields.

Perhaps, there are so many methods that can be used to reduce or remove selection bias, and one from those is to take large number of workers who are pursuing higher education or quitting after higher school only, and this confirms that on average workers take the schooling option that maximizes the present value of lifetime earnings. And empirically it is found out that the workers who have comparative advantage in doing skilled job, they will do that and other will do less skilled jobs, who have comparative advantage in it.

Schooling as a Signal

 According to Spence (1996), “The schooling model is based on the idea that education increases a worker’s productivity and that this increase in productivity raises wages. An alternative argument is that education need not increase the worker’s productivity at all, but that “sheepskin” levels of educational attainment (such as a high school or college diploma) signal a worker’s qualifications to potential employers.”

This above annotation means that education necessarily does not increase productivity of anyone but it gives signal to the job providers that high educated persons are furnished enough to carry on smart works which accelerate their wages than to other categories of labors. And to the employers schooling is a better system of signal rather than other differences in abilities to reveal as those cannot be found out easily and in minimum cost.