Econometrics - Chapter 2 Cheat Sheet
1. Definition & Overview
Regression analysis studies the dependence of one variable (dependent Y) on one or more
variables (independent X) to estimate/predict the mean of Y given X.
Population Regression Function (PRF)
E(Y|X) = β1 + β2X + u, where u is the stochastic error term. PRF shows the average relationship
between Y and X in the population.
Sample Regression Function (SRF)
Estimated from sample data: ■i = β■1 + β■2Xi. Residuals u■i = Yi − ■i estimate the error term.
Linearity
Linear in parameters means βs enter the model linearly. May be nonlinear in variables but still linear
in parameters.
Ordinary Least Squares (OLS)
Method to estimate βs by minimizing the sum of squared residuals. Formula: β■2 = Sxy / Sxx, β■1
= ■ − β■2X■.
Interpretation of Coefficients
β■2: Change in Y for a one-unit change in X. β■1: Expected value of Y when X=0.
Properties of OLS
Unbiased: E(β■) = β. BLUE (Best Linear Unbiased Estimator) under Gauss-Markov assumptions.
Residuals sum to zero; residuals uncorrelated with X.
Precision of Estimates
Standard errors measure reliability. Smaller SE → more precise estimates.
Assumptions of OLS
1) Linear in parameters 2) Fixed X values in repeated samples 3) E(u|X)=0 4) Homoscedasticity 5)
No autocorrelation 6) n > number of parameters 7) Variation in X 8) Correct model specification 9)
No perfect multicollinearity.
Goodness of Fit (R²)
R² = ESS/TSS. Measures proportion of variance in Y explained by X (0 ≤ R² ≤ 1). Perfect fit if R² =
1, no relationship if R² = 0.
Key Formulas
Sxy = Σ(Xi − X■)(Yi − ■), Sxx = Σ(Xi − X■)², β■2 = Sxy/Sxx, β■1 = ■ − β■2X■, R² = ESS/TSS,
TSS = ESS + RSS.
Example Interpretations
Working time & income: β■2 = 0.75 → 1h increase in work → $0.75 increase in income. Rice price
& demand: β■2 = −1.375 → Price ↑ 1 → demand ↓ 1.375 tons.
