IMHO, merging the two files and then loading them using SQL*Loader would be a better option. It will be more scalable than your code as it performs array inserts and will provide comprehensive errors in it's log file as to why any inserts failed. Otherwise, you're re-inventing the wheel.

Also, aren't you concerned that the files you are receiving are logically one single table in two files (one column per file)? Can you be sure that the order in which the rows appear in each file matches exactly so that you are getting a proper row, and not actually two columns for two unrelated records. I personally wouldn't accept files constructed in this manner.