What is E o etri s ?
• We will study key concepts and techniques for statisti al orks ith so ial data.
• Econometrics： a method for processing data and learn about general patterns in the population of interest.
• Econometrics is a powerful tool for estimating social relationship from a data.
• Key ingredient: Data – typically in the form of large observations.
• Data: the set of information.
• Statistical works play a central role in modern social science.
2
Ｏ Employment outlook 2013
Economic and policy quetion
• In generally, we are required to answer economic and policy questions:
– Can education grant program improve the quality of life of poor peoples? – Can environmental taxes reduce the emission level?
– Ca higher ages i du e higher o tri utio orkers? – To i rease i o e of poor, hat’s fa tor is i porta t?
• In many times, we should provide the quantitative answers about the questions
4
We can provide the answer using techniques of econometric. Ho ?
What you can do after this class?
–
⇒Descriptive statistics –
e.g., representative income of Chinese, representative numer of chidrent in Bangradesh
–
e.g., income gap between urban and rural area, education inequality between male and female
–
e.g., effect of microfinance on children health status, effects of higer education on income
Formal theory and statistical works
• (Economics and other field) theories provide rich hypothesis.
⇒should e tested usi g i terest ou tr ’s data.
• Typically, economics theories provide only qualitative policy implications (e.g., the government should set positive environmental taxes on CO2).
⇒ To get the quantitative implication (e.g., the best tax rate is 10$ per CO2), you should estimate key parameters in the model by using statistical methods.
6
Descriptive Research and Ecnometrics
• Descriptive research (e.g., interview for politician, CEO, and workers) can provide detail information about your interest society.
⇒important to find interest research question.
• But it is ot e ough to o tai poli i pli atio s e ause ge eralit is questionable.
e.g., You a i ter ie o l i di iduals ho ha e e tre e opi io or feeli gs for their society.
⇒ Statistical works can lead to more general policy implications.
To good statistical works: Mathematics and Statistics
• Mathematics and statistics are very important to study econometrics.
• In this class, students are not required to have studied statistics and advanced mathmatics.
⇒Only require basic knowledge of four arithmetic operations.
• We start to study basically statistics.
8
To good statistical works: Formal Theory
• Economics and other field theories provide good hypothesis and research question. ⇒Very important!!!
• Empirical analysis should always be linked to social reasoning.
• Exactly how mathematical a d ’tight’ this li k is aries a ross studies.
• Researchers setting out to test an economic theory usually begin by developing a formal(mathematical) economic model to make clear various relationships between variables.
• However it is also common (and acceptable) to use theory less formally - or even to rely entirely on intuition.
e.g.) Hospital effectiveness
Econometrics 10
• Research interests: The effect of hospital on health status.
• Data: National Health Interview Survey
We can observe the answer of following questions:
1. During the past 12 months, was the respondent a patient in a hospital overnight? ⇒ can observe who are recent hospital visitors.
2. Would you say your health in general is excellent(5), very good(4), good(3), fair(2), poor(1)? ⇒can observe the health status.
e.g.) Hospital effectiveness
• We can make two sub-samples (Visited hospital, or not), and can calculate mean health status of each samples as
(From Angrist and Pischke 2009)
• Represe tati e health status a o g Hospital group is lo er tha No - Hospital group.
⇒ Can we conclude that the hospital has a negative effect on the health status?
Group Number of respondants Representative Health Status
Hospital 7,774 3.21
Non-Hospital 90,049 3.93
e.g.) Hospital effectiveness
Econometrics 12
• There some possibility why representative health status among Hospital group is lower than Non-Hospital group.
1. 2. 3.
4.
Example: Hospital effectiveness
Hospital Health status
• A key challenge in modern econometrics is to ide tif i terest effe ts from non-interest effects ⇒ There are many approach:
– Control variables
– Instrumental variables methods
– Differences-in-differences strategies and Fixed effects model – Matching methods
14
Challenge in modern econometrics
Plan of the class
• We focus on the study of the basically concepts of Econometrics and practically issue to write empirical papers. ⇒ We cannot detail discussion of each issues. Espe iall , I do ’t sho the for al athe ati al proof
• Main material is slides.
• You can get slides in previous class from my
homepage(https://sites.google.com/site/keisukekawata7/).
Plan of the class
• We conduct STATA Exercises using the date from the Bangladesh Household Survey ← You can freely get the date(file name: Data files) from the web-site
http://go.worldbank.org/FE8098BI60 .
STATA: Commonly used software to statistical analysis. ← The student of IDP can take over from your supervisor. If you are not from IDP, please send me E-mail after toda ’s lass.
16
Plan of the class
1. Data and descriptive statistics 2. Probability
3. Statistics
4. Randomized experimental design 5. Exact matching
6. Linear regression
7. Double Difference & Fixed effect 8. Instrumental Variable
9. Regression in discontinuity design 10. Propensity score matching
Reference Books
• Stock and Watson (2011) Introduction to Econometrics
• Wooldridge (2008) Introductory Econometrics: A Modern Approach
• Hayashi (2000) Econometrics
• Hansen(2015) Econometrics
(web-site http://www.ssc.wisc.edu/~bhansen/econometrics/Econometrics.pdf)
18
Main textbook
• Khandker, Koolwal, Samad (2010) Handbook on Impact Evaluation
(web-site http://www.stata.com/bookstore/handbook-impact-evaluation/)
Grading Method
Grade will be evaluated by the following criteria:
• 20% --- Report
• 20% --- Homework assignments
• 30% --- Mini exams
• 30% --- Final exam
Your first task
• You get the te t ook Ha d ook o I pa t E aluatio from web-site http:// .stata. o / ookstore/ha d ook-impact-evaluation/ .
• You get the date from the Bangladesh Household Survey (the file name is Data files) from the website http://go. orld a k.org/FE 0 BI60 .
• STATA is installed in your PC.
20
