-----------------------------------------------------------------------
               Econ 120B, Spring 1998  --  Homework #3 (5%)
-----------------------------------------------------------------------
This homework is due at 12:45 pm on the last day of class.  NO LATE
PAPERS WILL BE ACCEPTED REGARDLESS OF WHAT THE COURSE OUTLINE SAYS.  As
before, team up with others (maximum 3 persons per team) and submit a
single paper with all names.

First study the "walk-through" applications in Sections 4.6 and 4.7 
(fourth edition only).  

DATA4-10 described in Appendix D, Page D-10 (fourth edition only), has
cross section data on private school enrollment and its determinants
for the 50 States and the District of Columbia (n = 51).  You are to
carry out an empirical analysis of this data set using the techniques I
described in class.


FROM YOUR HOME COMPUTER
-----------------------
If you have ESLWIN at home and you prefer that, run it, select the data
file called data4-10 and carry out the steps indicated below in 1. 
When output appears through NOTEPAD, choose File, Page Setup, reduce
margins to 0.5i or 0.25i, and then Print.

If you want to work within DOS entirely, first click the MSDOS icon
from Main (Windows 3.1) or from Start, Programs (Windows 95) and exit
temporarily to DOS.  Then

    cd  \esl  
    esl  data4-10  

to get the ? prompt.  Then type the commands given below in 1.  For
printing, type the following commands from the C:\ESL prompt.

    copy  inpfile  inp     (say yes to overwrite previous inp if asked)
    esl  data4-10  -b  <  inp  >  out
    lpr  out  >  prn


FROM ECON 100 LAB
-----------------
If ESLWIN works and you prefer that, run it, select the data file
called data4-10 and type the commands listed below in 1.  When
output appears through NOTEPAD, choose File, Page Setup, and reduce
margins to 0.5i or 0.25i, and then Print.

If you want to work within DOS entirely, first click the MSDOS icon
from Main to get to DOS.  Then

    cd  \pgms\esl  
    esl  data4-10  

to get the ? prompt.  Then type the commands given in 1 below.
For printing, type the following commands from the C:\ESL prompt.

    copy  inpfile  inp     (overwrite previous inp if asked)
    esl  data4-10  -b  <  inp  >  out
    lpr  out  >  prn

1.  First estimate the most general "kitchen sink" model using the OLS
    procedure (ols 1 0 2 ....  9 ;).  Next use the "data-based model
    reduction technique" discussed in class (also in Chapter 4) to omit
    variables one at a time (do not omit the constant term under any
    circumstances) until all variables (ignoring the constant term)
    have regression coefficients significant at the 10 percent level of
    significance.  When you are done, type quit to exit ESL.  Follow
    the directions given above to obtain a printed output which you
    should submit.  ONLY ORIGINAL PRINTOUTS ARE ACCEPTABLE, PHOTO
    COPIES ARE NOT.

    In the printout explain why you omitted each of the variables.

2.  Using the significance of coefficients, omitted variable bias, and
    the model selection statistics choose the "best" model.  Explain
    why you choose that as the best model.

3.  In the best model you chose, state whether the signs of the 
    regression coefficients (again ignore the constant term) agree
    with your intuition or any of them "wrong"? Explain what you
    expected and why.

4.  Choose the original kitchen sink model as the unrestricted model 
    and your best model as the restricted model and conduct an
    appropriate test for the JOINT SIGNIFICANCE of the coefficients for
    the omitted variables, using a 10 percent level.  Be sure to write
    down the null and alternative hypotheses, the test statistic and
    its distribution including the d.f., the critical value (or range),
    the decision rule, and the decision.  Show all your steps and your
    conclusion in words as to whether the coefficients are jointly
    significant or not.

5.  In the final model you chose, interpret the results by stating
    what is significant and what is not, what the marginal effects
    are and whether they are sensible.  For answering this, use the 
    interpretation of results in Section 4.7 as a guide.