Program to do augmented Dickey-Fuller test for unit root in real exchange rates (shown in table 7.2). PPP1.PGM. The program loads the data sets, SPOT.DAT and CPI.DAT. The program ADFPVAL1.PGM calculates p-values of the ADF test.
DF_1.PGM generates the distribution for the Dickey-Fuller studentized coefficient under the null hypothesis that the process follows a random walk. The user needs to set the following variables: a) the number of monte carlo simulations (NSIM), b) the number of starting values to drop in the simulations (NDROP), c) the number of time-series observations, (t1), and d) the value of the constant in the data generating process (alpha). The value assigned to alpha is the drift in the random walk.
Program to do augmented Dickey-Fuller test for unit root with a century of dollar-pound real PPI exchange rates (top panel of table 7.3) PPP5.PGM. Program loads in Lothian and Taylor’s data set LOT_TAY.CSV
Program to do augmented Dickey-Fuller test for unit root with a century of dollar-pound real CPI exchange rates (lower panel of table 7.3) PPP3.PGM. The program loads the data set STRATH.CSV.
LLPPPAS.PGM is the program to do Levin-Lin test for unit root (table 7.5). This program provides point estimates only. NICKEL.PGM does the bias adjustment for “rho” using Nickel’s formula. llpppbt.pgm is the program to do obtain the parametric and nonparametric bootstrap distribution for the Levin-Lin test statistics.
ipsppas.pgm is the program to get point estimates for the Im, Peseran and Shin test. ipsppbt.pgm is the program to obtain the parametric and nonparametric bootstrap distribution for the IPS test.
mwpppas.pgm (output file=MWPPAS.OUT) is the program to do Maddala-Wu’s Fisher test of PPP using (asymptotic) Dickey-Fuller p-values. To do the Fisher test, you must do the following (NOTE: I�ve already done this. Follow steps a) and b) below if you want to re-do it with different parameter settings).
Run df_1.pgm to get the Dickey-Fuller distribution. Do it without trend and with trend. Place the results in the file MWAS.IN. This will be a (nsim X 2) array. Put the distribution of the DF t-statistic without trend in the first column, (with trend in the second column) of MWAS.IN.
Set X in ADFDISTN[X, ,2]=MWAS.IN to the number of replications used in generating the DF distributions. If you used 20,000 replications, (if MWAS.IN is 20000 lines long), then set X=20000.
Now you are ready to run MWPPPAS.PGM.
df_c.csv is an Excel worksheet in CSV format with the Dickey-Fuller distribution based on 10000 Monte Carlo replications. Studentized coefficients are sorted in ascending order. A constant is included in the test equation.
df_ct.csv. Same as df_c.csv , except TREND is included in test equation. Note: These are pretty big files (about 1.3 MB each).