Even intensity-only recordings have mono-compatibility
problems. A centered signal will rise in level 6dB, while
those to the extreme left or right will show no rise.

The cardioid mics in an X-Y array have an included angle of about 90
degrees. This is often varied in actual use, up to about 120 degrees.
Sounds
that are on-axis for the array are thus 45-60 degrees off-axis for either
mic. While cardioid mic coverage patterns vary, they are typically
something
like 3-8 dB down when the sound source is 45-60 degrees off-axis.
Therefore
sounds that are on the array's axis, are reduced by 3-8 dB in either mic
and
the sum varies between +2 and -2, more or less.

My partial goof. I was thinking of panned recordings. Obviously, most mics
in coincident arrays are turned so that "front center" sources fall off
their main axes.