As discussed previously, one of the three barriers to getting away from SAS in accounting research is getting the data into an alternative form.
The procedure I used to get data from SAS into an SQLite database involved the following steps:
1. Download the data from WRDS.
2. Convert the data from SAS to comma-separated values (CSV) using StatTransfer.
3. Use Perl to import the CSV data into an SQLite database and do a little work (i.e., indexing) to improve performance.
To illustrate this procedure, I use a subset of the CRSP monthly stock data:
1. Get SAS datasets. If I want to calculate stock returns including delisting returns and compare these returns to the CRSP value-weighted returns, I will need the following SAS datasets from the sm subdirectory of CRSP on WRDS.
A short UNIX shell script did the trick (for me…use your WRDS username in place of mine). Note that by using rsync, I only get files that I don’t have (or have outdated copies of).
2. Convert to CSV. Assuming that the only SAS files in the current directory are the four I downloaded above, I could simply issue the following command:
st "./*.sas7bdat" "./*.csv"
3. Use Perl to create an SQLite database. This is the most involved step (here is the script).