The scenario is hypothetical, and I target for conservative values. My personal experience, and impressions, matches better with your comment.

Do you know where I can extract exact numbers and measures of this?

The problem is that all the measures that I have are in story-points and are useless. My experience comparing teams in similar situations, I have seen that the ones using TDD correctly have speeds up to 100x faster after some years of development, sometimes even sooner. But that is my experience, and I do not have a proper study. Once, I suggested shrinking a "team" of 150 people to just five or ten persons. Yet, some studies [1] [2] seem to indicate that shrinking teams benefit for the reduction of the overhead of communication.

Btw, did you have noticed that, even in this so conservative scenario, going with Technical Debt produces a cost overrun after just six months?