Prasad V. Bidarkota and J. Huston McCulloch, "Optimal Univariate Inflation
Forecasting with Symmetric Stable Shocks", Journal of Applied
Econometrics, Vol. 13, No. 5, 1998, pp. 659-670.

The data are stored in 5 files. These files, which are in DOS (CR/LF) 
format, are zipped in the file bm-data.zip. 

(1). data1.txt :  The monthly CPI-U series taken from BLS publications. The
series runs from 1950:1 through 1994:12. The first column indicates the year.

(2). data2.txt :  The monthly CPI-X series. The series runs from 1967:1
through 1982:12. The first column indicates the year. 

(3). data3.txt : The monthly CPI-UX series that we construct using the
monthly CPI-U and CPI-X from (1) and (2) above. The series runs from
1950:1 through 1994:12. 

(4). data45.txt :

        First column :  Raw inflation rates constructed as first
differences of natural logarithms of the constructed CPIUX series from (3)
above and as described in Section I of the paper. Rates are annualized
percentages. They run from 1953:11 (Oct-Nov inflation)  to 1993:9 (Aug-Sep
inflation). 

        Second column :  Seasonally adjusted monthly inflation rates
constructed from the first column above. The seasonal adjustment procedure
is described in detail in Appendix III of the working paper version of the
paper (Ohio State Dept of Economics Working Paper No. 96-02). Rates are
annualized percentages. The series runs from 1953:11(Oct-Nov inflation) to
1993:9 (Aug-Sep inflation). 

(5). data6.txt :  Stable Filter Inflation Forecasts Forecasts: 
y-hat_t = E( x_t | Y_t ) = E( y_(t+j) | Y_t )  (Annual percentage rates)
and standard errors. The first column shows year and month.
These appear in Figure 1 in the paper along with the 2 standard error
bands and the seasonally adjusted inflation rates.