On the contrary, there was lots of reason to suspect old code of being inefficient on new machines.
Much of that old code used clever techniques, highly rewarded when they were developed, to fit the software to the limitations of those ancient machines. When you have 48 K of core, and that is all you've got, you choose algorithms that can be written in tiny loops that will fit, and you use re-entrant techniques so that the code that is already in place for the date calculation can be re-used to calculate part of the return on investment, depending on the state of a one bit flag tucked into some other process. That could save seconds, or even minutes, by avoiding loading new code from tape. You optimize the size of the Hollerith card decks, to decrease the number of boxes that have to be hauled around on hand trucks, and the hours needed to read and compile the cards to tape.
It was much more important that the program could be compiled to tape in the 11 pm to 5:30 am time slot than how efficiently it would perform during the workday. Workday performance enhancements could be added in later revisions.