There are occasions where a .qvd file can and should be larger in size that a .txt file containing the same data.
A .qvd file contains an .xml header for a start which takes up space so certainly at small volumes of data the text file will be smaller as it doesn't need to store this header (or other supporting XML data).
When it comes to larger datasets it all depends on the make up of the data, if the data can not be compressed due to the majority of the values in each field being unique then again the .qvd file can be as large as a corresponding .txt file.
To benefit from .qvd compression you need to have as many unique value in columns as possible (QV uses 'Column Compression'), the best way to do this is to split TimeStamps into seperate date and time fields and replace complex key fields with autonumbers - these are just 2 examples.
The .qvd file isn't necissarily about compressed storage it's main purpose is swift loading into other Qlikview documents; quite often if you save a .qvw full of data and compare it to teh total size of all its tables exported as .qvd the .qvw will be much smaller and compressed.
Hope that helps,
Matt - Visual Analytics Ltd
.png "Luminary Alumni"){kind=icon}