Vladimir Kuzin, Massimiliano Marcellino, and Christian Schumacher, 
"Pooling versus Model Selection for Nowcasting GDP with Many Predictors: 
Empirical Evidence for Six Industrialized Countries, Journal of Applied 
Econometrics, Vol. 28, No. 3, 2013, pp. 392-411.

The on-line appendix to the paper and the data files contain more
explanatory material. 


Coverage and Sources
--------------------

The dataset covers the countries US, UK, Japan, Germany, France, and
Italy. Each country contains quarterly GDP growth and a large set of
monthly indicators. The datasets have been collected from various
sources and differ with respect to the sample periods. 

For some countries, GDP data ends in 2009Q4, which is also the end of
the evalutation period, see the main text for details. The GDP data
for some countries end one period later. In order to harmonize the
evaluation period, the effective sample used from this country data
ends in 2009Q4.  For Italy, the GDP data ends in 2009Q3, thereby
limiting the evaluation sample for this country.

Also the starting periods differ between the countries: For the US,
the sample starts in 1982Q1, for the UK in 1980Q1. For Japan, Germany,
France, and Italy, the samples start in 1991Q1. In each country, the
monthly indicators are generally available in the months corresponding
to the GDP sample. However, there are differing missing values at the
end of the sample, reflecting the different publication lags of the
variables.

The data have been collected by economists from central banks, who
themselves took the data from national, mostly official, sources. For
the US, the dataset is an update of the dataset used in Giannone,
Reichlin, and Small (2008) The German data are updated from
Marcellino and Schumacher (2010). For the remaining countries, the
datasets have been collected by central bank economists for their
economies. The data coverage differs from country to country,
depending on availability and lack of harmonization of statistics. For
the US, we have 190 monthly indicators available, for the UK 60, Japan
71, Germany 113, France 167, and for Italy 150 series. 


Transformation
--------------

Stationarity was obtained by appropriately differencing the time
series after taking natural logarithms for all time series except
interest rates. Most of the time series taken from the above sources
are already seasonally adjusted. Remaining time series with seasonal
fluctuations were adjusted using Census-X12. Extreme outlier
correction for the indicators was done using a modification of the
procedure proposed by Watson (2003). Large outliers are defined as 
observations that differ from the sample median by more than six times
the sample interquartile range. The identified outlier observation is
set equal to the respective outside boundary of the interquartile
range.


Data files
------------

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

The data are collected in the following files: 

  US:      data_m_us.csv, data_q_us.csv
  UK:      data_m_uk.csv, data_q_uk.csv
  Japan:   data_m_jap.csv, data_q_jap.csv
  Germany: data_m_ger.csv, data_q_ger.csv
  France:  data_m_fra.csv, data_q_fra.csv
  Italy:   data_m_it.csv, data_q_it.csv

The files denoted as 'data_m_*' contain the monthly data, whereas
'data_q_*' contain GDP.

The top row in each country spreadsheet contains the short names of
each time series. The longer names of the time series can be found in
the appendix of the paper. The second row contains a consecutive
numbering of the series.

The third row denoted as 'sa' contains a flag indicating that seasonal
adjustment has been applied (=1) or not (=0) by Census X-12.

The fourth row denoted as 'log/diff (sw)' contains the transformation
codes used following Stock and Watson (2002). This code is: 1 = no
transformation, 2 = first difference, 3 = second difference, 4 = log,
5 = first difference of logged variables, 6 = second difference of
logged variables.

The same data may also be found in Excel ".xls" format in the file 
kms-xls.zip, which contains the files

  data_fra.xls  data_it.xls   data_uk.xls
  data_ger.xls  data_jap.xls  data_us.xls