Econometrics - Chapter 3 Cheat Sheet
1. Multiple Regression Model (MRM)
Extends simple regression by including 2+ independent variables (X2, X3, ...). Allows ceteris
paribus analysis by holding other factors constant. General form: Yi = β1 + β2X2i + β3X3i + ... + ui.
Population vs. Sample Regression Function
PRF: Yi = β1 + β2X2i + β3X3i + ui (true, unknown). SRF: ■i = β■1 + β■2X2i + β■3X3i + ûi
(estimated from data).
OLS in Multiple Regression
OLS estimates βs by minimizing Σ(ûi²). Coefficient interpretation: β■k = change in Y for one-unit
change in Xk holding other variables constant.
Assumptions of MRM
1) Linear in parameters 2) Fixed X in repeated samples 3) E(ui|X)=0 4) Homoscedasticity 5) No
autocorrelation 6) n > #parameters 7) Variation in X values 8) Correct model specification 9) No
perfect multicollinearity.
Properties of OLS (Gauss–Markov)
OLS estimators are BLUE (Best Linear Unbiased Estimators) under assumptions. Residuals sum to
zero, uncorrelated with each regressor. Passes through means of Y and Xs.
Goodness of Fit
R² = ESS/TSS. Measures proportion of variance in Y explained by model. Adjusted R² adjusts for
number of predictors: Adj R² = 1 − [(RSS/(n−k)) / (TSS/(n−1))].
Partial Correlation Coefficients
Measures association between Y and one X, holding other Xs constant. r12,3 = correlation between
Y and X2 controlling for X3.
Functional Forms
Cobb–Douglas: Y = β1X2^β2 X3^β3 e^u → log form: ln Y = β0 + β2 ln X2 + β3 ln X3 + u. Polynomial
regression: Yi = β0 + β1Xi + β2Xi² + ... + βkXi^k + ui.
Key Formulas
β■2 = [S_yx2 - r23 S_yx3] / [S_x2x2 (1 - r23²)], se(β■j) = sqrt[σ■² / ((1−Rj²) Σ(xji²))], R² = 1 −
RSS/TSS, Adj R² = 1 − [(RSS/(n−k)) / (TSS/(n−1))].
Example Interpretations
β■educ = 2 → 1 year ↑ education → $2 ↑ in wage (holding other vars constant). In Cobb–Douglas,
β2=0.47 means 1% ↑ labor → 0.47% ↑ output, ceteris paribus.
