Rob Luginbuhl and Siem Jan Koopman, "Convergence in European Time Series: A
Multivariate Common Converging Trend-cycle Decomposition", Journal of
Applied Econometrics, Vol. 19, No. 5, 2004, pp. 611-636.

The data for the real seasonally adjusted quarterly GDP for the 5 countries
comes from the OECD and can be obtained from their website

  http://www.sourceoecd.org/
  
We obtained (annual) population estimates for the 5 countries from the US
Census Bureau, which can be downloaded from their website

  http://www.census.gov/ipc/www/idbsprd.html
  
We interpolated the annual population data to quarterly data and then
calculated the quarterly GDP per capita in each country. The data are in
(real) local currencies, and therefore have arbitrary levels. For this
reason, the data are renormalized so that the year 1970 is the same for each
series. The final transformation we made to the data was to take (natural)
logarithms and multiply the result by 100.

The data we used are in the file convergence.data, an ASCII file in DOS
format that is zipped in lk-data.zip.

The data are organized in columns, each column of which consists of 125
quarterly observations starting with the first quarter of 1970 in the
second row. The first row consists of the label names of the data in each
column. Column 1 is the date (no label), Column 2 is the data for Germany
(label: LGermany), 3 for France (label: LFrance), 4 for Spain (label:
LSpain), 5 for Italy (label: LItaly), and 6 for the Netherlands (label:
LNL).

The programs logit3ItSpNl.ox and pvarISN.ox are OX programs which will only
run with the SsfPack package written by S.J. Koopman. Please see the webpage

  http://www.ssfpack.com/
  
for further details. The program pvarISN.ox produces MLE's for the
multivariate model in the paper without convergence. The program
logit3ItSpNl.ox produces the MLE's for the model with all three types of
convergence. Both programs are zipped in the file lk-programs.zip.