Julieta Fuentas, Pilar Poncela, and Julio Rodriguez, "Sparse Partial
Least Squares in Time Series for Macroeconomic Forecasting", Journal
of Applied Econometrics, Vol. 30, No. 4, 2015, pp. 576-595.

We use Stock and Watson's database. The original dataset consists of
132 monthly United States (U.S.) macroeconomic time series that span
the period from January 1959 through December 2003. Given that for
some variables the information is available only from 1960, the sample
period used in the paper begins in January 1960 for a total of T=528
observations. File: Database_2003.txt. The data set is the one used in

  Stock, J. and Watson, M. "Implications of Dynamic Factor Models for
  VAR Analysis", NBER Working Paper No. 11467 (2005),

in which you can find additional information on how to transform the
series to achieve stationarity by taking logs, first or second
differences as necessary. 

This database can also be downloaded from Mark Watson's web page

  http://www.princeton.edu/~mwatson/wp.html

The target  variables are:

1. Inflation (PUNEW) 
2. Industrial Production (IP)
3. Total Employment (EMT)
4. Personal Income (PI)
5. Retail Sales (RS)

The updated database contains 112 monthly macroeconomic time series,
and extends the time series of the original base through December 2010
for a total of T=610. 

File: Database_2010.txt

Target variable:
1. Inflation (PUNEW)

Both databases are ASCII text files in DOS format. They are zipped in
the file fpr-data.zip. Unix/Linux users should use "unzip -a".