Factor Analysis / PCA with longitudinal data

Hello, I have a problem which I'm trying to get around... I am trying to conduct a principal component analysis which compares changes in labour market regulation in the EU pre and post financial crisis. The first problem I faced was a low sample size as my sample size initially concerned just 19 countries with a view to running a PCA with 10 variables. As this number of sample is inadequate, I decided/was advised to input data for each country year, treating each country year as a new participant (e.g. Austria 2005, Austria 2006, Austria 2007, Austria 2008...etc up until 2011 along with the other countries in this fashion). This solution has given me results which make sense and has allowed me to create scatter plots comparing year on year/country on country factor scores in a meaningful way for different slices of time on the same plot (e.g. I can then illustrate how labour market regulation has changed between 2005, 2008 and 2011 for Ireland and a few other interesting cases in the same diagram).

HOWEVER, I have now recently been warned that this method I've used is problematic as I am using longitudinal data which creates correlation problems. I had been using SPSS to look at case summaries and plot the changing REGR scores on each factor for different country years. Does this sound like a problem to anyone or is the method I've chosen actually adequate? I should point out that I went down the longitudinal route as a means of producing a valid PCA with enough cases, I am trying to produce something that stands up to a certain amount of scrutiny. - any help would be gratefully received! :)

Oh, I also wondered if the fact I'm using 'real data' changes the rules on conducting a PCA at all?!