Iirc the Yale study based its estimates on some sort of financial metadata and smaller revenue indicators that painted a very dire picture. I couldn't explain it though, I just have a vague recollection of it being based on actual empirical evidennce.