Returns to education on wages (Mincer, OLS)

This illustrates EcoLab's 5-step workflow with the most classic model in labor economics: the Mincer wage equation estimated by OLS. The figures are format illustrations.

Summary: regress log wage on years of schooling and experience to estimate the "return to education" (the payoff to each additional year of schooling).

---

Step 1 — Ideation

• Question: by what percentage does one more year of schooling raise wages?

Step 2 — Literature Review

Human-capital theory (Becker, Mincer); standardize citations; clarify variables and the log-lin form.

Step 3 — Data Collection

Variable	Symbol	Measurement	Source
Log wage	lnwage	log(hourly/monthly wage)	VHLSS; labor survey
Years of schooling	educ	years	VHLSS
Experience	exper, exper2	years and squared	from age − schooling − 6
Controls	gender, region	binary	VHLSS

Step 4 — Modeling

Mincer form (log-lin, with squared experience to capture concavity):

$$\ln(wage_i) = \beta_0 + \beta_1\,educ_i + \beta_2\,exper_i + \beta_3\,exper_i^2 + X_i\gamma + \varepsilon_i$$

Choose the Classical linear regression family → OLS, with robust standard errors (heteroskedasticity is common in micro data).

Illustrative results (format — not real results):

Variable	Coefficient	SE (robust)	p-value
educ	0.082	0.005	0.000
exper	0.031	0.004	0.000
exper2	−0.0005	0.0001	0.000
$R^2$	0.38

Sample interpretation: $\hat{\beta}_1 = 0.082$ ⇒ each extra year of schooling is associated with about 8.2% higher wages (approximately, given log-lin); experience is concave (rising then flattening).

Stata

* ---- Mincer wage equation (OLS, robust SE) ----
use "vhlss_wage.dta", clear

gen exper2 = exper^2

regress lnwage educ exper exper2 gender region, vce(robust)

R

# ---- Mincer wage equation (OLS, robust SE) ----
library(lmtest)
library(sandwich)

df <- read.csv("vhlss_wage.csv")

model <- lm(lnwage ~ educ + exper + I(exper^2) + gender + region,
            data = df)

# Robust standard errors
coeftest(model, vcov = vcovHC(model))

Python

# ---- Mincer wage equation (OLS, robust SE) ----
import pandas as pd
import statsmodels.api as sm

df = pd.read_csv("vhlss_wage.csv")
df["exper2"] = df["exper"] ** 2

X = sm.add_constant(df[["educ", "exper", "exper2", "gender", "region"]])
model = sm.OLS(df["lnwage"], X).fit(cov_type="HC1")
print(model.summary())

Step 5 — Reporting

Export an APA/Harvard… report with replication code.

Endogeneity caveat

educ may be endogenous (unobserved ability affects both schooling and wages) ⇒ OLS may be biased. See the instrumental-variables fix in IV example: Returns to education.

Video tutorial

Video Tutorial: Running the Mincer wage equation (OLS) in EcoLab

See also

• OLS · IV example · Catalog