Econometrics

• Establishing causality
  • does “policy/behavior/program → outcome”?
  • $X\to Y$?
• Challenges
  1. Is it random noise?
     • Removing outliers
  2. Are there exogenous factors?
     • • $X$ is endogenous if $X$ correlates with $ϵ$ (error)—non-causal, lurking variable
     • • $X$ is exogenous if $X$ doesn’t correlation with $ϵ$—likely causal
• Regression Analysis: $Y_i={\beta }_0+{\beta }_1{X}_i+{ϵ}_i$
  • ${\beta }_0$: intercept. Less interesting
  • ${\beta }_1$: slope. More Interesting
  • $X$: independent variable
  • $Y$: dependeng variable
• ! be careful…
  • Correlation $\ne$ Causation
  • A higher slope doesn’t mean higher correlation; a higher Correlation coefficient ($\rho =\frac{{\sigma }_{XY}}{{\sigma }_X{\sigma }_Y}$) does mean higher correlation.
  • Statistical Significance $\ne$ Real-life Effects