Michael P. Clements and Ana Galvao, "Forecasting US Output Growth Using
Leading Indicators: An Appraisal Using MIDAS Models", Journal of
Applied Econometrics, Vol. 24, No. 7, 2009, pp. 1187-1206.

All data files are comma-separated ASCII files in DOS format. They are
zipped in the file cg-data.zip. Unix users should use "unzip -a".

The file gro.csv has quarterly output growth for 1959:1-2005:Q1. The first
column has real-time vintage data (first released) and the second has final
data. 

The file gdpaa.csv has quarterly vintages since 1965:Q4 of output growth
(159 vintages=159 columns) for the period of 1959:1-2005:Q1. Data of output
growth were obtained from the real-time dataset of the Philadelphia Fed.

The files dleq.csv and dlem.csv have the final data of the 10 indicators
(=10 columns) obtained from Stock and Watson data sets. The second file has
the transformed monthly data and the first one has transformed quarterly
data. Data are from 1959:M1 up to 2003:M12.  

The files dleqrtv.csv and dlemrtv.csv have the real-time vintage data of
(first-difference) hours and housing using monthly vintages since 1970.
Monthly data were obtained from the real-time dataset of the Philadelphia
Fed. Data is from 1959:M1 up to 2003:M12. Hours data are in the first column
and housing (building) in the second column.  

The files hours.csv and house.csv have monthly vintages of the
(first-difference) hours and housing starting with the 1992:M1 vintage (145
vintages=145 columns) for the period of 1959:M1 up to 2003:M12. This is for
the use of end-of-sample vintage data in the paper.