Assignment 8 Economics 31 Fall 1999
Complex Hypothesis - F tests
Reading: Mirer Chpt. 14
1. Using stata file moneysupply (data from www.economagic.com)
consider
the following money supply equation for the US economy for the
period
1958-1997, where M=M2 billions chained to 1992 dollars,
C=Currency
component of Money Stock, and D=total time deposits at all
Depository
Institutions.
M=C+D
a) interpret the equation
b) regress currency, and deposits on money stock and test each of
the
coefficients for individual significance
c) Create 3 new dummy variables for year, one for 1959 to 1973,
one for
1973 to 1979, and one for 1979 and beyond (see assignment 2 for
instructions) Why might someone choose these years? Argue for a
different
selection of years.
d) regress currency, deposits, and two of the three new dummy
variables, on
money stock and test each of the coefficients for individual
significance.
Should any of the two dummy variables you tested be dropped from the
model?Why?
2.Refer back to the data in problem 1.
a) Did any of the variables improve the explanatory power of
the
regression? (f-test)Test the hypothesis that all of the coefficients
from
part b are equal to O.
b) (f-test)Test the hypothesis that all of the coefficients in 1b
are equal
to O.
c) (f-test) Test the hypothesis that the equation from part 1d
represented
no improvement over the equation from part 1b.
3. Using the stata file monbasemodel, consider the following
monetary base
determination equation for the period 1959-1997 (data from
economagic.com)
where M=C+ B, M=monetary base, billions of dollars, C=Currency
component of
Money Stock, and B=Reserve bank credit.
a) Give an intuitive explanation for the proceeding equation.
b) Regress currency and deposits on money stock, and test the
significance
of each of the individual coefficients.
c) Test the significance of the overall regression in part (f-test)
4. Use the data in problem 3.
a) Create 2 new dummy variables for year, one for 1959 to 1973,
and one for
1992-97.
b)Why did we introduce these new variables?
c) Regress currency, deposits, and the dummy variables on money
stock.
Test each of the coefficients for significance. Have any of the
coefficients changed from b to d?
d) Test the significance of the overall regression (f-test).
e)How would you decide between the two models?
5. Utilyzing the statafile invexpend, consider the following
investment
expenditure model for the period 1959-1997 (economagic.com)
Y=C + I + G + NX, where Y=real gross domestic product, chained
1992
dollars, C=real personal consumption figures chained to 1992 dollars,
I =
gross private domestic investment, Billions of chained 1992
dollars,
G=government consumption expenditures and gross investment, billions
of
chained 1992 dollars, NX=real net exports, billions of chained 1992
dollars.
a) Why might you introduce another variable, T, for time? Do
economic time
series tend to move together? What sign do you expect for each of
the
coefficients?
b) Regress consumption, investment, government consumption/invest,
and
real net exports on real gross domestic product, and test each of
the
coefficients for significance.
c) Are these coefficients individually statistically different from unity?
6. Using the data in problem 5,
a) Create a new variable: percgov= G (government consumption
expenditures
and gross investment)/Y (real domestic product). Create two dummy
variables, pgovh=percgov>.25, and pgovl=percgov<.25.
b) Test for a significant difference in structure between
percgov>.25 and percgov<.25. (After specifying the
regressed variables, same as problem 7, part b, for the first
regression
add the qualification, if pgovh==1 and for the second regression add
the
qualification if pgovh==0).
(Chow test)
c) Theoretically, why might percgov be important, amungst other
economic
indicators? Is it actually according to your data?