Economics 322 - Introduction to Econometrics - Professor Norman R. Swanson
Note: I have not taught this course in many years.
This is a course in basic econometrics. The aim of the course is to familiarize
students with applied econometrics, with partial emphasis on empirical analysis using computers.
The tools learned in the course are widely used by practicing economists
in disiplines including (among others): consulting, government, banking, teaching,
marketing, accounting, advertising, and sales.
SPECIAL NOTES!!!!!!!!! ...
Office Hours
Norman R. Swanson: MTH 11:10-12:00 NJH 301d (nswanson@econ.rutgers.edu)
Geetesh Bhardwaj: TF 3:00-4:00 NJH 306 (bhardwaj@econ.rutgers.edu)
Kingkit Cheung - Kreeger Learning Resource Center (kingkitrow@yahoo.com - caclc.rutgers.edu) - OH T 4:00-6:00 & Th 12:00-2:00
Other Notes
1. Eviews help- check out the Microcomputing Center office hours at New Jersey Hall - ask grad students on duty for help!
2. Basic Statistics Review Notes
3. Eviews Introduction and Instruction
4. Extra office hours - tba
CLASSES WHICH I WILL MISS AND WHICH SHOULD BE MADE UP
Make up dates and or guest lecturers on the appropriate dates, etc., will be discussed in class.
NOTES
SOFTWARE and PROGRAMS DISCUSSED TO DATE
COMPUTATIONAL DISCUSSIONS FOR ECONOMICS 322
GRADES FOR ECON322
DATASETS FOR POSSIBLE USE IN ASSIGNMENTS AND PROJECT
the details below are all tentative and subject to change until this statement is removed!
COMPUTER EXERCISES to ACCOMPANY ASSIGNMENTS ABOVE:
Assignment 3:
Computer problem: Using EVIEWS, continue your analysis begun in Assignment 2.
In particular, I would like you to run three new regressions in addition to the two run in the
prior assignment. Use the same dependent variable from before for all three regressions.
For each regression choose a different set
of "explanatory" or "independent" variables.
Fully report and explain your findings. For example, discuss whether or not your
"economic" intuition is "satisfied" by your statistical findings.
Additionally, include discussion of at least one goodness of fit measure, the magnitudes and signs of the
coefficients in your models (and whether of not these signs and magnitudes ``make sense"),
and an analysis of the significance of the variables in the model based on application of t-tests, an F-test of overall significance,
and at least one other F-test of some coefficient restriction.
As in the previous computer assignment, you may gather your
variables from any data source (including the datasets which appear in this webpage). Although not a requirement,
you may find it useful to use data for this assignment which corresponds to
the data which you will use in the final project. This part of assignment 3 is to
be worked on in the assigned groups. However, each student is required to hand in a
separate assignment 3, including both computer and algebraic problem sections.
More Details: There are no restrictions on which regressions to run. Simply run three
different regressions. After doing this, interpret the t-statistics, overall F-statistics,
and R squared values. Also, discuss whether the signs of the coefficients
which you obtained make sense, given your prior beliefs concerning how the economic
variables which you have modeled should interact, etc.
Please do not hand in computer output. Rather, summarize you regression results and
findings in written fashion (in pencil, if you want).
Assignment 4 (practice computer assignment):
This is an assignment which is meant to familiarize you with
data transformation, using lagged economic variables as regressors,
and model selection.
Pick 2 variables, say X and Y.
Construct two new variables, say DLX and DLY, which are the log
difference of these variables. The objective is to use DLY as
dependent variable in a regression with DLX as regressor.
Run a regression of DLY on DLX. Now create a first lag of DLX,
say DLXlag1. Regress DLY on DLX and DLXlag1. Now create a lag of
DLXlag1 (i.e. say you had variable Z(t), now you'll also have
variables Z(t-1) and Z(t-2)), say DLXlag2. Regress DLY on DLX,
DLXlag1, and DLXlag2. Continue this process until you've tried
up to 12 lags. Report your results. Select which of your 13
regressions that you feel is "best" based on some
model selection criteria of your choice.
Justify and explain your choice for "best" model.
EVIEWS MANUALS ARE AVAILABLE FOR SHORT TERM LOAN:
FINAL PROJECT: due in class
Only one unique project should be
handed in by group. The project should be typed, and at least 7
pages in length. Please do not include computer printouts, programs, log files, etc.
Rather, summarize your findings in prose, and with simple equations.
The final project is due in class on the day of our last lecture of the semester.
I would like each "group" to complete a econometric and statistical
analysis of an economic dataset which you have constructed, or which you have
chosen from among the various data series that I have given to you.
In order to complete the project, each group is expected to draw
on (1) computational tools learned in this class;
(2) basic econometrics and statistics learned in this class,
and in earlier statistics class;
(3) your knowledge of basic economic theory and models.
I would like you to start by posing one or more economic questions which are
relevant to your particular dataset. Mention why the issue(s) you've
raised is relevant, in the context of, for example, government policy,
social welfare, testing neoclassical versus Keynesian views, assessing
the rationality of economic agents, constructing "good"
forecasting models, etc. If it's not obvious what you'd like
to do, you may want to pick up a basic macro text, say, and look
for simple theories which link two or more economic variables together.
However, for the project it suffices to pick a group of variables and simply
give me an intuitive explanation of why you might expect the
variables to be related in a regression context.
Alternatively, you may want to consider constructing competing forecasting models and
comparing them. On the other hand, if you do choose to simply
posit the economic relationship between some group of variables,
remember to pay close attention to explaining your expectations concerning the signs of
the coefficients in your regression model, etc.
I would like you to discuss
your approach to answering these questions given your particular dataset.
In particular, discuss you expectations concerning estimated economic relationships,
signs of coefficients, slopes of graphs, evolution of you series over time (or across
individual), etc.
Finally, carry out your analysis in a systematic fashion. When you construct
regression models, say, then try lots of different models with different lags, etc.,
and report to me your "best" findings, being careful to discuss why you think
those findings are "best". Be careful to address potential nonstationarity
among your data if you're using time series - and GOOD LUCK!!!
Please don't hesitate to contact myself or your TA if you have any
questions.
WEB SITES WORTH LOOKING AT
HANDOUTS