3. A researcher runs the following regression on a cross section of adults using

Question

3. A researcher runs the following regression on a cross section of adults using Census data:

                ln w = b0 + b1 X + b2 ED + e

where w are annual earnings, ED is years of schooling, and X is a vector of controls (urban/rural residence, age, gender, race, married). His point estimate of b2 is .09 and it is statistically significant.

                a. What is the interpretation of his estimate of b2?

                b. In what ways might b2 be a biased estimate of the true return to schooling?

                c. In what ways have researchers attempted to deal with the bias in part b)?

Explanation / Answer

a) b2 is interpreted as the difference in the predicted value in w for each one-unit difference in ED, if X remains constant.

b) To check the bias we have to check whether other variables’ like X coefficients change significantly when
the variable ED is added to the equation, b2 might be a biased estimate due to specification error. Specification error is when correct independent variables are not chosen or when stochastic error-term is not is right form. Here in this case one of the reason can be that people might include their part time studies as full-time schooling and add it in the years

c) researchers can improve their instrument ask specific questions which will reduce the risk of bias or accordingly correction in model can also be made assuming the maximum possible chance of bias

Related Questions

Navigate

• Browse (All)
• Browse 3
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.