The issue of matched pairs of microphones often gets caught up in people's confusion between "ought" and "is."
If you want to compare two hi-fi amplifiers seriously, first you need to match their gain within some smallish fraction of a dB (certainly less than 0.5 dB)--otherwise, people tend to hear sonic differences that are real enough, but which disappear when the gains are matched more closely. Such small differences in loudness (or frequency response) are rarely perceived consciously in terms of loudness (or frequency response); instead, if people hear a difference, they perceive it in terms of sound quality. Experimentally this has occurred with loudness or frequency response differences as small as 0.1-0.2 dB in some cases. Of course, by then the supposed differences in sound quality are also quite subtle. But if you want to eliminate bias, these variables need to be controlled down to (as I said) smallish fractions of a dB.
Now take that well-understood set of circumstances and apply them to what we're talking about here. The very best manufacturers generally give tolerance limits of maybe ±1.5 dB for the frequency response of their microphones relative to their published curves, and a further ±1.5 dB or so for their sensitivity.
We're talking about devices that turn one kind of energy into another, not an amplifier which is purely electronic--so having such narrow tolerances is quite an accomplishment (and quite an improvement over, say, the 1960s or before). But such tolerances are a whole order of magnitude greater than the point at which differences cease to affect our impressions of sound quality.
This is especially relevant to those of us who use coincident and closely-spaced microphone setups, since the stereo impression is based on rather fine differences in amplitude, frequency and phase response between the two microphones of a pair. Human hearing is extremely sensitive to the differences between what our two ears pick up. There are sonic defects (e.g. phase distortion) that have little or no subjective impact as long as they're equal between the two channels of a system, but those same defects can become quite significant if they differ between the channels. We want the differences between the signals of the two microphones to result solely from their positions and angles within the sound field; we don't want them to result from response differences between the microphones. Such differences just throw garbage information into the stereo recording unless you can compensate for them somehow.
As a result, I really do recommend (and personally prefer) selected, matched pairs of capsules for two-microphone stereo recording, or for any recordings in which the main pickup is made with a coincident or closely-spaced pair. And personally I can hardly respect a manufacturer that claims to be so consistent in their production that special matching isn't needed, since such claims are simply not supported by the facts.
--best regards