The Mincer equation includes experience as a quadratic (X and X²) because...
AAll economic relationships are quadratic
BEarnings rise with experience (positive first term) but at a decreasing rate (negative second term), consistent with declining human capital investment over the career
CIncluding X² eliminates omitted variable bias
DThe logarithm of wages requires polynomial terms
Human capital theory predicts that earnings rise with experience because workers accumulate skills on the job, but that the rate of increase declines as workers age (the investment horizon shortens, reducing the return to further investment, and skills may depreciate). The quadratic captures this concave relationship: the positive coefficient on X gives the initial rate of earnings growth, and the negative coefficient on X² captures the flattening. Empirically, this specification fits experience-earnings profiles very well.
Question 2 True / False
The OLS estimate of the return to education from the Mincer equation provides an unbiased estimate of the causal effect of education on earnings.
TTrue
FFalse
Answer: False
OLS is biased primarily due to ability bias — unmeasured ability is correlated with both schooling (more able people get more education) and earnings (more able people earn more regardless of education), leading to an upward bias in the OLS return estimate. Other sources of bias include measurement error in schooling (which attenuates the estimate) and selection into schooling based on expected returns. IV methods attempt to address these biases by finding instruments that affect schooling but not earnings directly. Interestingly, IV estimates often exceed OLS estimates, possibly because measurement error bias (downward) exceeds ability bias (upward), or because the local effect for compliers exceeds the average effect.
Question 3 Short Answer
Why are instrumental variable estimates of the return to education often interpreted as 'local average treatment effects' (LATE) rather than average treatment effects?
Think about your answer, then reveal below.
Model answer: IV estimates identify the causal effect for 'compliers' — individuals whose education was actually changed by the instrument (e.g., those who attended more school because of compulsory schooling laws but would not have otherwise). This is the LATE, and it may differ from the average treatment effect (ATE) for the full population if compliers have different returns than always-takers or never-takers. Compulsory schooling compliers may have higher marginal returns (they were at the margin of dropping out) than the average person.
This is a critical econometric point. Angrist and Krueger's quarter-of-birth instrument identifies the return for students who stayed in school longer only because compulsory schooling laws compelled them — a specific subpopulation that may not represent the typical student. If these marginal students (who would have dropped out without the law) benefit more or less from education than the average student, the LATE differs from the ATE. This does not invalidate the IV estimate — it correctly identifies a causal effect — but it limits generalizability.