They are identical mathematically. They'll vary slightly in actual implementation due to the inability to mount them so they are truly coincident, and also due to the pattern-response variations in real-world microphones. But those differences should be minor if done properly.
Let's assume a pair of well behaved 8's, mounted so they are as coincident as possible. The direct L/R output of those microphones when arranged in a 90 degree X/Y configuration will be very, very close to the matrix decoded L/R output of the same microphones if arranged Mid/Side, and decoded with a 50:50 ratio. Changing the ratio of that M/S decode is doing exactly same thing as changing the physical angle of the X/Y pair. Using all Mid signal matrixed with zero contribution from the Side signal (100:0 ratio) is the same as angling both X/Y 8's so they both point directly ahead. As far as the output signal is concerned, both situations are equivalent to a single forward facing figure 8 mic.
Likewise, using all Side signal with no Mid in the M/S matrix (0:100 ratio) is doing exactly the same thing as pointing the X/Y pair 180 degrees apart, with both mics facing directly to the sides.
Note: The behaviour described above only applies to two figure 8's. When the Mid/Side ratio is adjusted using any other Mid pattern than fig-8, not only will the angle between virtual microphones change, but the virtual polar pattern will also change, simultaneously. Crossed 8's are unique with regards to two channel M/S matrixing due to their front/back pattern symmetry. Their virtual patterns always remain 8's regardless of the Matrix ratio.
The L/R output from either a 90-degree X/Y or the equivalent 50:50 Mid/Side ratio will be a signal with the same (positive) polarity in both channels for all sounds arriving at the microphones from the quadrant in front. Sounds arriving from the rear quadrant will also produce signals with the same polarity in both channels, but polarity of those signals will be inverted (negative) relative to sounds arriving from the front. Sounds arriving from the side quadrants will produce signals with positive polarity in one channel and negative polarity in the other. No phase differences between the signals in the L/R channels will occur, regardless of angle of arrival.
Extra credit question: How are sounds handled which arrive from directly above and below a Blumlein pair of 8's when in X/Y? ..and when in M/S?
Lab credit: Take an Blumlein X/Y recording and run it through a L/R>M/S>L/R matrix and play around with the matrixing ratio. Can you get it to sound similar to the Blumlein M/S recordings you've expressed a preference for? The Voxengo MSED plugin (its free-
http://www.voxengo.com/product/msed/) is an good way to do that. It only does M/S matrixing so isn't overly complicated. Leave the pan controls centered and use it's "Inline" mode to do the L/R>M/S>L/R adjustment thing, which is two matrixing operations with your matrix ratio adjustment being done between them.