{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

ec20-lecture2-v5_1

# ec20-lecture2-v5_1 - Economics 20 Lecture#2 Bivariate...

This preview shows pages 1–14. Sign up to view the full content.

1 Economics 20 Lecture #2: Bivariate Regression

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
2 Today Brief Review of Last time Rubin Causal Model Simple Linear Regression Model Terminology Key assumption Derivation of Least Squares estimation Next Class: Multiple Regression
The fundamental problem of causal inference is that: 3 1 2 3 4 5% 0% 5% 91% 1. People self-select into treatment, so we can’t take differences in their outcomes as causal 2. We can never observe the counterfactual 3. Convincing natural experiments are difficult to find 4. Experiments are often not possible in economics

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
No assumptions are necessary for experiments to produce causal estimates 4 0% 100% 1 2 1. True 2. False
The assumptions necessary for experiments to produce causal estimates are plausible because of random assignment 5 86% 14% 1 2 1. True 2. False

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Causal inference always requires us to make untestable assumptions 6 76% 24% 1 2 1. True 2. False
7 Last Time… Rubin Model of Causality: Individuals (i) have 2 potential outcomes: Y i1 = outcome if individual i gets “treatment” (last time: unemployed person gets a temp job ) Y i0 = outcome if i does not get the “treatment” θ = Average Treatment Effect (ATE) = E[Y i1 – Y i0 ] = E[Y i1 ] – E[Y i0 ] Fundamentally unobservable (why?)

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
8 What we can actually observe… We observe T i = = E[Y i1 |T i = 1] - E[Y i0 |T i = 0] = θ + {E[Y i 0 |T i = 1] - E[Y i0 |T i = 0]} What we can estimate, , represents the causal effect only if E[Y i 0 |T i = 1] = E[Y i0 |T i = 0], the untestable [ ] θ ˆ E ATE = true causal impact The counterfactual difference in outcomes between treated and non-treated groups absent the treatment (not observable!) 1 if person i gets the “treatment” 0 if does not θ ˆ
9 Econometric Models Econometric models = set of assumptions, relevant to a particular research question, describing the relationship between the outcome, treatment, and other observable and unobservable variables **No matter the model, some version of the independence assumption is required for causal inference** Today: simple linear regression model

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
10 The Simple Regression Model y i = β 0 + β 1 x i + u i
11 Simple Linear Regression Model (SLRM) y i = β 0 + β 1 x i + u i For situations where we believe there is a linear (straight line) relationship between two variables ↔ how much y rises with x is constant ( β 1 ) β 1 represents the causal effect of a one- unit change in x on y. This is: Always unknown (fundamental problem of causal inference) May be zero!

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
12 Terminology y i = β 0 + β 1 x i + u i We typically refer to “y” as the Dependent Variable, or Left-Hand Side Variable, or Explained Variable, Outcome Variable Or Regressand
13 Terminology, cont.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### Page1 / 42

ec20-lecture2-v5_1 - Economics 20 Lecture#2 Bivariate...

This preview shows document pages 1 - 14. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online