> Because actually creating the file on disk is the slowest part in this case, and creating a table with one field is a lot faster than creating a table with a lot of fields.

Excuse me for my big mistake
I have known that cursor file size obtained from query is effective on total execuation time.
The most important part of thread is wrong. Actually Query result of first is belong to 2. Query.

I have executed these two queries time and again. I have obtain same unexpected result.

Ali Altun