You can access the dataset over the web with the Stata command
use http://fmwww.bc.edu/ec-c/s2003/821/ec821aa.dta
The Stata datafile contains interest rate and price inflation series. In this file, the series y1, y2, ... y9 refer to interest rates on Treasury bills of one, two, ... nine months tenor, whereas the series pi1, pi2, ... pi9 refer to expected inflation measured over those same horizons.
Consider the 1, 2, and 3-month interest rates and inflation rates. Perform the Elliott et al. DFGLS unit root test on these series, and determine whether the null hypothesis of that test--that they are I(1), or unit root, processes, may be rejected.
Using these same series, perform the Kwiatkowski et al. KPSS unit root test on these series, and determine whether the null hypothesis of that test--that they are I(0), or stationary processes, may be rejected.
Consider the 1, 2, and 3-month interest rates and inflation rates. Use Phillips' MODLPR estimator on the level of each series, for power 0.50, 0.55, ... 0.80 and test the hypothesis that d=1.
Using the first differences of these same series, (D.varname rather than varname) use Robinson's ROBLPR estimator (over the same range of power values) to test the hypothesis that d=0 for the changes of the series.
Considering the totality of the evidence from these tests, what would you conclude about the timeseries behavior of interest rates and inflation? Do these series exhibit long memory behavior?
Use the multivariate form of Robinson's ROBLPR estimator to jointly estimate the d values for the first differences of each time horizon's interest rate and inflation series, and test the hypothesis that they have the same d (as would be implied by the Fisher hypothesis). Run this test (separately for the 1-month, 2-month, 3-month series) for power 0.70 and 0.80. Interpret your findings for each time horizon.
Use one of the panel unit root tests that we discussed in class (MADFULLER, LEVINLIN, IPSHIN, HADRILM) on the three interest rate series (taken as a panel), and interpret your findings.