BOSTON COLLEGE

Department of Economics

EC 760B

Spring 1997

Prof. Baum

Mr. Franke

PROBLEM SET FIVE

due Thursday, 24 April 1997, at classtime

1. Greene 8-5.

2. Greene 8-6.

3. Greene 9-1.

4. Do the following empirical exercise using UNIX RATS. You should place your RATS commands in a file (e.g. ps5.rat), using the editor, and then invoke RATS via the command:

$ rats ps5.rat > ps5.out

This will place the output in file ps5.out, which you may then examine with the editor. Note: you may break RATS commands into several lines; if you do, place a dollar sign ($) at the end of the line to be continued.

Your RATS job should start with the commands:

cal 1953 1 4

all 1975:4

open data /u/baum/classes/taylor.rdb

data(for=rats)

tab

which will read in the data and generate descriptive statistics. The dataset contains quarterly data from 1953 through 1975 on four economic variables: POTGNP (potential GNP), M1 (narrow money supply), GNPDEF (implicit deflator for real GNP), and GNP (real GNP).

The data relate to three political regimes: the Eisenhower (Republican) years (1953:1-1959:4), the Kennedy-Johnson (Democratic) years (1960:1-1968:4), and the Nixon-Ford (Republican) years (1969:1-1975:4). Compute growth rates for each variable (hint: 400 delta log x == xdot at an annual rate).

a. Calculate and report the growth rates of GNP during each of these three regimes. Test the hypotheses: (1) all three are identical; (2) Democratic administrations bring about higher GNP growth than Republican administrations.

b. Do the same comparison for the rate of price inflation, as measured by the rate of change in GNPDEF.

c. Fit an equation relating the growth rate of GNP to the growth rate of the money supply, lags 1 through 8, with a constant term. Test whether a form of this equation containing only 4 lags would be adequate.

d. Using the original (8-lag) version of the equation, consider whether the once-lagged growth rate of potential GNP adds meaningfully to the explanatory power of the equation.

e. Fit the equation of part (d) over the Eisenhower-Kennedy-Johnson years (1955-1968, allowing for lags), and generate ex ante forecasts of the Nixon-Ford years (using prj). Calculate the mean forecast error, the mean absolute error (MAE), and the root mean squared error (RMSE). Interpret each of these statistics, considering how the estimated equation performs out-of-sample. How would the RMSE change if you left 1975:1 out of the calculations? Why?

f. In Richard Nixon's own words, what was he not?