Which is quickest?

I have built a database for storing airlfare contracts... individual fares, the rules, seasons, penalties , and various conditions that apply to them.

Individual fares can amount to millions within the database spread amongst several hundered contracts and seasons routings etc.

The purpose of the database is to allow human entry of data and produce the correct data strings for use in an internet booking engine.

Data output is from a final table of 187 fields, with individual records possibly being in millions.

The database is built of various tables which must be combined into one flat database for the export process.

My question is about the most effective way of producing the final data export. The majority of the 187 fields on the export table (the export table is stored in a separate file from the actual database)are simply a matter of looking up data from related records... some require manipulation but most do not.... the majority of the maniplulations and calculations are done as data is entered.

So for the fields where there is no calculation needed... what is the best way to tackle it? Are there any speed differences from defining the fields as text with auto lookup or calculation, or setting them as calculation fields... and what about stored/unstored?

Export to the booking engine is once daily. Should I populate and update this export table as the user enters and edits data... or is it more efficient to simply run a script at the end of each day that purges and rebuilds the export file from scratch (speed during working is preferable ... if once dailly output would improve it)

Anybody got opinions/experience on these sort of issues?