Lecture 1

advertisement
Econ 140
Empirical Relationships
Lecture 1
Lecture 1
1
Today’s Plan
Econ 140
• Syllabus & housekeeping issues
• Course overview
– What is econometrics?
– Two econometric examples/ Examples from the
textbook.
Lecture 1
2
Teaching Team
Econ 140
Professor: Andrew K. G. Hildreth
593 Evans Hall (510) 642 0446
hildreth@econ.berkeley.edu
Office Hours: Monday 2-3 pm & Wednesday 10-11am
Assistant: Judi Chan, (510) 643-1625 chan@econ.berkeley.edu
GSIs:
Tanguy Brachet: tbrachet@econ.berkeley.edu
Office Hours: location and time to be advised. Sections 104 & 106.
Heather Royer: hroyer@econ.berkeley.edu
Office Hours: location and time to be advised. Sections 101 & 107.
Kristy Piccinini: kpiccini@econ.berkeley.edu
Office Hours: location and time to be advised. Sections 103 & 105.
Lecture 1
3
Syllabus
Econ 140
• Textbook: Stock, J. and Watson, M., Introduction to
Econometrics, Addison-Wesley, 2002.
• Grading & ‘Harsh but Fair rules’.
• Final Exam:
• Lecture Etiquette
• Econometrics is a ‘doing’ or active learning subject.
– Use of EXCEL. Available in all labs: times in the TMF
– STATA: Anyone taking Econ195A.
• GSI’s - have homes to go to.
• When the going gets tough….
Lecture 1
4
Course Website
Econ 140
• emlab.berkeley.edu/users/hildreth/e140_f02/e140.html
• What you’ll find at the website:
– My picture (not a good one!)
– Excel files
– Lecture notes
– Problem Sets (& Solutions)
– Midterms (after the tests) & Solutions
– Supplemental handouts
• Also - faculty page has previous course plus other stuff.
Just alter: “e140_f02/” to “e140_sp02/” and so on.
Lecture 1
5
What is Econometrics?
Econ 140
• Broadly defined: the study of economics using statistical
methods
• Founding members of the econometric society described it:
“..as the quantitative analysis of actual economic phenomena based on the
concurrent development of theory and observation, related by appropriate
methods of inference.” --Samuelson, P., Koopmans, T. & Stone, R. Report
of the Evaluative Committee for Econometrica, Econometrica, 1954, p.
142
Lecture 1
6
Why Econometrics?
Econ 140
• When we read the newspaper or see announcements of
economic statistics or predictions, how are the statistics
and predictions derived?
• Some uses:
–
–
–
–
Lecture 1
Returns from investing in 1 more year of school
2000 Florida election
Macroeconomic indicators (Phillips Curve)
Production function estimates
7
Takeaways
Econ 140
• Econometrics is a doing subject!
• It is an art that must be learned through practice - working
out problems algebraically, using economic data, building
models using computer software
• No one exact way to present a statistical argument
• Course objective: providing you with knowledge of
econometrics in theory and application
• Vocational uses
– consultancy
– business planning
– politics or public policy
– lawyers, circuit court judge, Supreme Court judge
Lecture 1
8
Returns to Education
Econ 140
• Examining relationship between years of education and
earnings using Gary S. Becker’s 1964 theory on human
capital
• Comparing the cost and future returns of an additional year of
schooling
– Future earnings are function of schooling given by:
W=f (s) where s = given # years of schooling
– But there’s a simultaneity problem: do you earn more
because you have more schooling or do you pursue more
schooling to earn higher wages?
Lecture 1
9
Returns to Education (2)
Econ 140
• Test the relationship using cross-section data from Current
Population Surveys (CPS) for CA males in 1979 and 1995
• You can use the 1995 data to graph gross weekly earnings
vs. years of schooling, but it’s impossible to see any
relationships between earnings and years of schooling
• The same goes for the 1979 data - it’s a mess!
• To highlight an array in EXCEL, hold CTRL+SHIFT and
press the down arrow
Lecture 1
10
Returns to Education (3)
Econ 140
• Use conditional means to get a better approximation of the
earnings and education relationship
• Conditional mean: the mean value of a variable Y given
the value of another variable X
n
– General formula:  Yi
i 1
– In our case:
n
X  some value
n
 Wi
i 1
n
Lecture 1
S  some value
Wi= gross weekly earnings
S = years of schooling
11
Returns to Education (4)
Econ 140
• Using conditional means, you can compare the mean gross
weekly earnings associated with different years of
schooling - the graph is less messy
• There may be problems with our analysis !
– definitions of schooling changed
– boundary set for top coding changed: in 1979, it was
$999. In 1995 it was $1923
– Macro and microeconomic factors
Lecture 1
12
Chasing Butterflies
Econ 140
• What happened in Palm Beach, Florida during the 2000
election?
• Can we test the assertion that the butterfly ballot confused
voters and caused them to accidentally vote for Buchanan
rather than Gore?
• If Palm Beach County hadn’t used the butterfly ballot, can
we show that Gore would have won Florida?
• The course website has Excel datasets of voting outcomes
in Broward County, Palm Beach County, and Florida.
Lecture 1
13
Chasing Butterflies (2)
Econ 140
• Broward County is similar to Palm Beach in size and
demographics, but the butterfly ballot was unique to Palm
Beach
• Graphing the number of votes for Buchanan against those
for Gore in Broward County, we see that he received less
than 10 votes in any of the voting precincts
• Looking at the same graph for Palm Beach, we see that
Buchanan received many more votes there than he did in
Broward County.
Lecture 1
14
Chasing Butterflies (3)
Econ 140
• We can also look at the number of votes for a party vs. the
number of registered voters for that party
• We see a similar upward trend for Democrats and
Republicans
• However, for the Reform voters Palm Beach is an extreme
outlier - for the other 66 counties, there were less than
1,000 Reform votes cast. Palm Beach County had 3,407
Reform votes cast!
Lecture 1
15
Chasing Butterflies (4)
Econ 140
• You can use a confidence interval to test whether the Palm
Beach observation is statistically different from the others
– Regress the number of Reform votes on the number of
registered Reform voters by county, not including Palm
Beach
– We find the coefficients are highly statistically
significant
• 95% confidence interval means that there is a 5% chance
that an observation will lay outside that interval by error.
Notice that Palm Beach doesn’t lie in that interval.
• What degree of confidence do we need to include Palm
Beach in the confidence interval?
Lecture 1
16
Wrap up
Econ 140
• An overview of what’s to come
• An introduction to economic data and the idea of empirical
relationships between two measured variables.
– Example: years of education and gross earnings
– Votes cast in Florida and ‘Butterfly Ballot’.
• Problems inherent in using economic data to test empirical
relationships
• Conditional mean function
Lecture 1
17
Download