The explanation is simple. When you drag in from another application you are placing a smart object, not a regular layer.
Smart objects always honor the physical print dimensions in each file, not the native pixel grid as a normal copy/paste does. So when the two files have different resolutions (like one at 300 and the other at 72) you see this.
A smart object is an embedded full copy of the file that stays in its original state throughout. This means you can transform or scale repeatedly without any additional quality loss.
