-
Notifications
You must be signed in to change notification settings - Fork 4
/
week_10_recap.tex
417 lines (332 loc) · 13 KB
/
week_10_recap.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
\documentclass[pdftex]{beamer}
%\documentclass[notes=show]{beamer}
%\documentclass[xcolor=dvipsnames]{beamer}
\usepackage{amssymb}
\usepackage{latexsym}
\usepackage{amsfonts}
\usepackage{amsmath}
\usepackage[absolute,overlay]{textpos}
\usepackage[english]{babel}
\usepackage[latin1]{inputenc}
%\usepackage{times}
\usepackage[T1]{fontenc}
\usepackage{tabularx}
\newcolumntype{Y}{>{\small\raggedright\arraybackslash}X}
\usepackage{graphicx}
\usepackage{bigstrut}
\usepackage{bbm}
\usepackage{mathrsfs}
\usepackage{epsfig}
\usepackage{array}
%\usepackage{natbib}
\usepackage{ulem}
\mode<presentation> {
%\usetheme[left,width=1.7cm]{Berkeley}
%\usetheme{default}
\usetheme{Boadilla}
\usecolortheme[RGB={103,102,204}]{structure}
%\usecolortheme{dove}
\useoutertheme{infolines}
\setbeamercovered{transparent}
}
%\renewcommand{\familydefault}{cmss}
%\renewcommand{\mathrm}{\mathsf}
%\renewcommand{\textrm}{\textsf}
\usefonttheme{serif}
\newcommand{\X}{{\mathbf{X}}}
\newcommand{\x}{{\mathbf{x}}}
\newcommand{\E}{\mathsf{E}}
\newcommand{\V}{\mathsf{Var}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Neue Kommandos für fette Mathebuchstaben innerhalb von Formeln %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newcommand{\bom}{\boldmath}
\newcommand{\ubom}{\unboldmath}
\newcommand{\mb}{\mathbf}
\newcommand{\fmalpha}{\mbox{\bom${\alpha}$}} %Fettes alpha
\newcommand{\fmbeta}{\mbox{\bom${\beta}$}} %Fettes beta
\newcommand{\fmgamma}{\mbox{\bom${\gamma}$}} %Fettes gamma
\newcommand{\fmdelta}{\mbox{\bom${\delta}$}} %Fettes delta
\newcommand{\fmepsilon}{\mbox{\bom${\epsilon}$}} %Fettes epsilon
\newcommand{\fmvarepsilon}{\mbox{\bom${\varepsilon}$}} %Fettes varepsilon
\newcommand{\fmzeta}{\mbox{\bom${\zeta}$}} %Fettes zeta
\newcommand{\fmeta}{\mbox{\bom${\eta}$}} %Fettes eta
\newcommand{\fmta}{\mbox{\bom${\theta}$}} %Fettes theta (ta)
\newcommand{\fmvarta}{\mbox{\bom${\vartheta}$}} %Fettes vartheta (ta)
\newcommand{\fmiota}{\mbox{\bom${\iota}$}} %Fettes iota
\newcommand{\fmkappa}{\mbox{\bom${\kappa}$}} %Fettes kappa
\newcommand{\fmla}{\mbox{\bom${\la}$}} %Fettes lambda (la)
\newcommand{\fmmu}{\mbox{\bom${\mu}$}} %Fettes mu
\newcommand{\fmnu}{\mbox{\bom${\nu}$}} %Fettes nu
\newcommand{\fmxi}{\mbox{\bom${\xi}$}} %Fettes xi
\newcommand{\fmo}{\mbox{\bom${\o}$}} %Fettes o
\newcommand{\fmpi}{\mbox{\bom${\pi}$}} %Fettes pi
\newcommand{\fmvarpi}{\mbox{\bom${\varpi}$}} %Fettes varpi
\newcommand{\fmrho}{\mbox{\bom${\rho}$}} %Fettes rho
\newcommand{\fmvarrho}{\mbox{\bom${\varrho}$}} %Fettes varrho
\newcommand{\fmsigma}{\mbox{\bom${\sigma}$}} %Fettes sigma
\newcommand{\fmvarsigma}{\mbox{\bom${\varsigma}$}} %Fettes varsigma
\newcommand{\fmtau}{\mbox{\bom${\tau}$}} %Fettes tau
\newcommand{\fmupsilon}{\mbox{\bom${\upsilon}$}} %Fettes upsilon
\newcommand{\fmphi}{\mbox{\bom${\phi}$}} %Fettes phi
\newcommand{\fmvarphi}{\mbox{\bom${\varphi}$}} %Fettes varphi
\newcommand{\fmchi}{\mbox{\bom${\chi}$}} %Fettes chi
\newcommand{\fmpsi}{\mbox{\bom${\psi}$}} %Fettes psi
\newcommand{\fmomega}{\mbox{\bom${\omega}$}} %Fettes omega
\newcommand{\fmimath}{\mbox{\bom${\imath}$}} %Fettes imath
\setbeamercolor{bibliography entry title}{fg=black}
\setbeamercolor{bibliography entry author}{fg=black}
\setbeamercolor{subsection in toc}{fg=structure}
\setbeamercolor{palette primary}{bg=structure, fg=white}
%\setbeamercolor{palette secondary}{bg=structure, fg=black}
%\setbeamercolor{palette tertiary}{bg=structure, fg=black}
\setbeamercolor{caption name}{fg=black} \setbeamersize{text margin
left=.8cm} \setbeamersize{text margin right=1cm}
\hypersetup{linkbordercolor={1 0 0}} \setbeamertemplate{navigation
symbols}{} \setbeamertemplate{headline}[default]
\setbeamertemplate{enumerate items}[default]
\newcounter{transfct}
\newcounter{begbs}
\newcounter{endbs}
\title[Introduction]{Econometrics 2 (Part 2)}
\author[Lychagin \& Mu\c co]{Arieda Mu\c co}
\institute[CEU]{Central European University}
\date{Spring 2020}
\AtBeginSection[] {
\begin{frame}<handout:0>
\frametitle{TOC}
\tableofcontents[currentsection]
\end{frame}
}
%\AtBeginSubsection[] {
% \begin{frame}<beamer>
% \frametitle{Outline}
% \tableofcontents[currentsection,currentsubsection]
% \end{frame}
%}
%\beamerdefaultoverlayspecification{<+->}
\pgfdeclareimage[height=.7cm]{logo}{rgs2}
\logo{\pgfuseimage{logo}}
\begin{document}
\frame{\titlepage}
\begin{frame}
\frametitle{Contact}
\begin{itemize}
\item Sergey Lychagin: \href{mailto:[email protected]}{[email protected]}\\
\sout{Office: Budapest campus, Nador 13, 513}
\item Arieda Mu\c co: \href{mailto:[email protected]}{[email protected]}\\
\sout{Office: Budapest campus, Nador 13, 507}
\item Boldizs\'ar Juh\'asz (TA): \href{mailto:[email protected]}{juhasz\[email protected]}
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Textbooks}
\begin{itemize}
\item Introductory Econometrics: A Modern Approach by Wooldridge
\item Mostly Harmless Econometrics by Angrist and Pischke
\item \textcolor{blue}{Mastering Metrics by Angrist and Pischke}
\item \textcolor{blue}{Casual Inference: The Mixtape by Cunningham}
\item Reading list of applied papers
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Outline: Tentative Schedule}
\begin{enumerate}
\item Economic research questions: causality
\item The experimental ideal
\item Linear regression
\item Instrumental Variables
\item Panel Data, Fixed Effects, Differences-in-Differences
\item Matching
\item Program Evaluation: Nonparametric Methods
\item \textcolor{blue}{Regression Discontinuity}
\end{enumerate}
\end{frame}
\frame{ \frametitle{Della Vigna and Card}
\begin{center}
\begin{figure}
\includegraphics[width=1\linewidth]{graphs/what_economist_do.png}
\end{figure}
\end{center}
}
\frame{ \frametitle{}
\begin{center}
\begin{figure}
\includegraphics[width=1\linewidth]{graphs/what_economist_do1.png}
\end{figure}
\end{center}
}
\frame{ \frametitle{Applied Microeconometrics Papers}
\begin{center}
\begin{figure}
\includegraphics[width=1\linewidth]{graphs/what_economist_do2.png}
\end{figure}
\end{center}
}
\frame{ \frametitle{}
\begin{center}
\begin{figure}
\includegraphics[width=0.8\linewidth]{graphs/detective.png}
\end{figure}
\end{center}
}
\begin{frame}
\frametitle{Statistical Models of Shoe and Leather}
Freedman, David A. (1991) "Statistical Models and Shoe Leather", Sociological Methodology, 21, 291-313
\bigskip
John Snow studies of the cholera epidemics in Europe in the 19th century and proves that cholera is a waterborne infectious disease
\begin{itemize}
\item In the $19^{th}$ century no microbiology, limited microscopes
\item Theory: diseases result from "poison in the air" - miasma
\item Cholera Europe in epidemic waves
\item Snow studied spatial pattern of epidemics along tracks of human commerce
\item Influence of water supply on incidence of Cholera?
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Is cholera a waterborne or an airborne disease?}
London in the 1800's: different water companies serve different areas
\begin{itemize}
\item Some companies take water from the Thames polluted by sewage
\item 2 companies
\begin{itemize}
\item Southwark \& Vauxhall: downstream from sewage discharges
\item Lambeth: intake point upstream
\end{itemize}
\item Both companies served the same parts of London during the 1853-54 cholera epidemic
\item Sometimes houses next to each other in the same street were served by the 2 different companies
\begin{itemize}
\item Each company supplies rich and poor, large and small houses, no difference in condition or occupation
\end{itemize}
\item Idea: compare number of cholera victims
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Method of Shoe and Leather}
\begin{itemize}
\item Snow surveyed houses in large parts of London
\item Water company
\item Cholera victims
\item 300,000 households involved
\item Reward: clear result
\end{itemize}
\end{frame}
\frame{ \frametitle{}
\begin{center}
\begin{figure}
\includegraphics[width=1\linewidth]{graphs/SnowMap_Points.png}
\end{figure}
\end{center}
}
\frame{ \frametitle{}
\begin{center}
\begin{figure}[t]
\includegraphics[width=1.1\linewidth]{graphs/freedman_tab1.pdf}
\end{figure}
\end{center}
}
\begin{frame}
\frametitle{The Experimental Ideal}
Social experiment is the most influential research design. Why?
\bigskip
Solves Selection Problem
\end{frame}
\begin{frame}
\frametitle{Potential Outcome Model}
$D_i=\{0,1 \}$ treatment variable (hospital care)
For each population unit $i$ we consider two \emph{potential outcomes} (health status)
\[\begin{array}{cl}
Y_{1i} & \text{outcome with treatment} \\
Y_{0i} & \text{outcome without treatment}
\end{array}
\]
The gain from treatment or \emph{causal effect} for unit $i$ is
\[ Y_{i1}-Y_{i0} \]
Problem: For each $i$, only one of $ Y_{i1}$ or $Y_{i0}$ is observed.
%\emph{Missing data problem}
\end{frame}
\begin{frame}
\frametitle{Observed outcome}
We observe
\begin{eqnarray*}
Y_i &=& \left\{ \begin{array}{cc}
Y_{1i} & \text{if}\;\; D_i=1 \\
Y_{0i} & \text{if}\;\; D_i=0
\end{array}
\right. = Y_{0i}+ (Y_{1i}-Y_{0i}) D_i
\end{eqnarray*}
In the population distribution of $Y_{1i}$ and $Y_{0i}$, we can compare the average health of \emph{treated} and \emph{non-treated}
\begin{eqnarray*}
\underset{\text{observed difference}}{\underbrace{E[Y_i|D_i=1]-E[Y_i|D_i=0]}} &=& \\
\underset{\text{average treatment effect}}{\underbrace{E[Y_{1i}|D_i=1]-E[Y_{0i}|D_i=1]}} &+& \underset{\text{selection bias}}{\underbrace{E[Y_{0i}|D_i=1]-E[Y_{0i}|D_i=0] }}
\end{eqnarray*}
\end{frame}
\begin{frame}\tiny
\frametitle{Observed outcome}
Remind that the observed difference is $E[Y_{i}|D_i=1]-E[Y_{i}|D_i=0]$\\
Which can be rewritten as:
\[
E[\underbrace{Y_{0i}+ (Y_{1i}-Y_{0i}) D_i}_{\text{$Y_i$ }}|D_i=1]-E[\underbrace{Y_{0i}+ (Y_{1i}-Y_{0i}) D_i}_{\text{$Y_i$ }}|D_i=0]
\]
From the properties of the conditional expectation we can rearrange the above equation as:
\begin{equation*}
E[Y_{0i}|D_i=1]+E [(Y_{1i}-Y_{0i}) D_i|D_i=1]-E[Y_{0i}|D_i=0]-E[(Y_{1i}-Y_{0i}) D_i|D_i=0]
\end{equation*}
Which can be rewritten as:
\[
E[Y_{0i}|D_i=1]+E [(Y_{1i}-Y_{0i})|D_i=1]-E[Y_{0i}|D_i=0]
\]
And is equivalent to:
\[
E[Y_{0i}|D_i=1]+E [Y_{1i}| D_i=1]-E[Y_{0i}|D_i=1]-E[Y_{0i}|D_i=0]
\]
Rearranging we get:
\begin{eqnarray*}
\underset{\text{average treatment effect}}{\underbrace{E[Y_{1i}|D_i=1]-E[Y_{0i}|D_i=1]}} &+& \underset{\text{selection bias}}{\underbrace{E[Y_{0i}|D_i=1]-E[Y_{0i}|D_i=0] }}
\end{eqnarray*}
\end{frame}
\begin{frame}
\frametitle{Random assignment as a solution}
Random assignment makes $D_i$ independent of the potential outcome. If $D_i$ is independent of $Y_i$ then $E[Y_i|D_i]=E[Y_i|D_i=1]=E[Y_i|D_i=0]=E[Y_i]$
\begin{eqnarray*}
E[Y_i|D_i=1]-E[Y_i|D_i=0] &=& E[Y_{1i}|D_i=1]-E[Y_{0i}|D_i=0] \\
&=& E[Y_{1i}|D_i=1]-E[Y_{0i}|D_i=1] \\
&=& E[Y_{1i}-Y_{0i}|D_i=1]=E[Y_{1i}-Y_{0i}] \\
\end{eqnarray*}
The observed difference in mean outcomes equals the \emph{average treatment effect}.
Examples:
\begin{itemize}
\item health treatments
\item government sponsored training programs
\item education production: effect of class size, teacher quality, etc. on student achievement
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Regression Analysis of Experiments}
Assume $Y_{i1}-Y_{i0} = \rho$ \emph{constant} treatment effect
\begin{eqnarray*}
Y_i &=& \alpha + \rho D_i + \eta_i
\end{eqnarray*}
\begin{eqnarray*}
E[Y_i|D_i=1] &=& \alpha + \rho + E[\eta_i|D_i=1] \\
E[Y_i|D_i=0] &=& \alpha + E[\eta_i|D_i=0]\\
E[Y_i|D_i=1]-E[Y_i|D_i=0]&=& \rho + \underset{\text{selection bias}}{\underbrace{E[\eta_i|D_i=1]-E[\eta_i|D_i=0]}}
\end{eqnarray*}
Selection bias amounts to correlation between regression error $\eta_i$ and $D_i$.
\end{frame}
\begin{frame}
\frametitle{Regression Analysis of Experiments}
We know about the selection bias
\begin{eqnarray*}
E[\eta_i|D_i=1]-E[\eta_i|D_i=0]= E[Y_{0i}|D_i=1]-E[Y_{0i}|D_i=0]
\end{eqnarray*}
If $D_i$ is randomly assigned, the selection bias is equal to zero.
Thus estimating the regression model results in the \emph{causal effect} $\rho$.
\bigskip
Regression model with covariates
\begin{eqnarray*}
Y_i &=& \alpha + \rho D_i + \beta X_i+ \eta_i
\end{eqnarray*}
If $X_i$ uncorrelated with $D_i$, including them will not affect estimate of $\rho$, but increase precision.
\end{frame}
\end{document}