It seems like the emissions of interest in EMC testing are radiated and conducted emissions. Is this because capacitive coupling is so weak it doesn't pose a threat to other devices?
I'm not sure why you think capacitive coupling is (always) weak. In the near field, capacitance is very effective at coupling high frequency power.
There are several methods of noise coupling that need to be considered for EMC:
- conductively coupled noise
- common impedance coupling (such as ground bounce/ground loops)
- near-field electric coupling
- near-field magnetic coupling
- far-field electromagnetic coupling (EM radiation)
These general areas of noise coupling are addressed through various Regulatory elements which must be met for EMC Compliance.
- IEC 61000-4-2 ESD
- IEC 61000-4-3 Radiated Immunity
- IEC 61000-4-4 EFT/Burst Immunity
- IEC 61000-4-5 Surge
- IEC 61000-4-6 Conducted Immunity
- CISPR 11 Class A Radiated Emission tests - (requires a 10m anechoic chamber)
Each of these tests specifies a test setup focused on one particular area of EMC. The radiated emission test doesn't address near field capacitive coupling effects, but that doesn't mean they aren't important; they just aren't regulated by that particular section of code.
Proper design for EMC would (among other techniques) use strategically placed capacitances to help route unwanted high-frequency noise away from sensitive internal nodes.
As @Hasan alattar mentioned in a comment, the Electrical Fast Transient (EFT) component of EMC testing, tests the DUT susceptibility to noise which is coupled through the capacitance between a DUT cable and a test cable. That's essentially testing for capacitively coupled noise susceptibility.
Capacitive coupling would also be involved in ESD protection between the equipment's outer case and internal circuitry; there may be parasitic capacitances that become important ESD pathways.
Switch-mode power supplies may also be subject to additional EMC regulations about harmonic generation, which would potentially impact other devices; in that case the DUT output could be capacitive coupled to nearby equipment. (For testing, a harmonic analyzer would be used.)