Technical Issues in Providing Economic and Financial Data

Paul Bern
Data and Statistical Services Consultant
Princeton University

Deciding on what type of economic data you need and finding it are only
the first steps.  What are you going to do with it once you have it?  In
this presentation I will explore the various options and make some
recommendations on how to provide and analyze economic data.  First, we
must decide on how we are going to make the data available to our users.
Should we just hand them the file and say "Have fun?"   Many of these
files are very large and some are in formats not easily read by all
statistical packages, so you may need to process the file yourself
before making it available to your users.  Second, your users will need to
analyze these data.  What kind of statistical procedures should you use?
Most, if not all, economic analyses require some time component -
looking at wages over the last ten years, for example.  Panel studies -
surveying the same people several times - are common and require special
handling. Whatever statistical package you use, you must make sure that it
has the proper techniques and options available to do what you need.
Finally, what software should you use?  What's better, SAS, Stata, or
SPSS?  SAS is, by far, the best at file management, and has procedures for
reading CRSP, Compustat, OECD, IMF and many other data formats.  Stata, on
the other hand, is much better with time-series data and is much easier to
learn than SAS or SPSS.  SPSS has the best user-interface and good
graphics.  So maybe you'll use more than one.

---------------
Paul Bern is a Data and Statistical Services Consultant at Firestone
Library, Princeton University.  He holds a Ph.D. in Mass Communication
from Syracuse University where he also worked as a Research Consultant
and as Senior Analyst in the Office of Research and Evaluation.