Technically BB since its origin didn't impose field lengths only record size lengths. This is because each field was delimited by a field separator character as opposed to having a preset position in the record.
This remains the same today meaning if you declare a record with say a company name field of 30 characters, your application could write 31 or more into the field as long as the total record length remained below the record size defined (basically stealing the extra bytes from the space set aside for other fields).
In PxPlus VLR files the record size is really just used to setup the buffers used to hold the records while being read/written. The active data in this record buffer is what it read from/written to the data blocks -- so over estimating the record size has no real impact on the actual file size. The file size is really impacted by the actual record size required to hold the data fields.
Now there is one 'caveat' to this. When using the DD and you plan to use ODBC to access you data, make sure you define the correct maximum field size. If you declared a field as 30 bytes long but actually wrote 35, most programs (e.g. Excel) will use the declared field length to define a buffer to hold the data and will get an error on the record where the data exceeds the defined length.
Lastly, if desired, you can enable data verification in PxPlus where it will verify the data you are writing to the file adheres to the field definition. This would include length, type (number/string), and even validation rulese.