Brilliant question. I'm surprised nobody referred to the amazing JSWT User Documentation available at STScl-JWST. (This question was active again, somehow, so adding an answer.)
Antzi's answer was only regarding momentum management, stationkeeping is about orbit maintenance. But let's address both here.
From the JWST user documentation, it is clear that the solar torque although balanced out by reaction wheels is enough for momentum management, but for orbit maintenance around L2, frequent station-keeping maneuvers are necessary.
While orbits about the L2 point are inherently unstable, the orbit size is large and the orbital velocity is low (~1 km/s), so the orbit "decays" slowly. However, JWST's large sun shield, roughly the size of a tennis court, is subject to significant solar radiation pressure which results in both a force and a torque. The direction of solar force varies as the observatory's attitude changes from observation to observation. The solar torque is balanced by reaction wheels, but periodically, the accumulated momentum is dumped by firing thrusters. Because JWST operations are event-driven, the observatory attitude profile and momentum dumping cannot be accurately predicted months in advance. These two perturbations increase the acceleration of JWST from its orbit about L2, and necessitates more frequent orbit maintenance (station keeping) maneuvers than other Lagrange orbit missions (which are typically 3–4 times per year).
Thus, for station keeping JWST uses thrusters, while maintaining relative sun-pointing requirements, as follows:
Orbit perturbations along the Sun-L2 axis have the greatest impact on-orbit stability. Thrusters are mounted on the spacecraft bus on the side of the sun shield facing the Sun; those used for orbit correction are oriented as far away from the sun shield as possible, and the sun shield can support a larger sun-pitch angle1 for orbit correction than is allowed for science operations. This architecture allows thruster firing at angles up to 90° from the Sun consistent with Sun avoidance restrictions, which is sufficient to provide orbit correction in all cases.
For momentum management, JWST suffers from massive momentum buildup, as suggested in the question, described here.
During science observations, the observatory will be pointed at a target, in an orientation at which the sun shield center of pressure is not aligned with the observatory center of mass. As solar photons hit the large sun shield, they place a torque on the observatory as a whole. The attitude control subystem (ACS) counteracts this torque by appropriately changing the spin rate on the reaction wheels, with the consequence that angular momentum accumulates in the reaction wheels. Momentum accumulation depends on the solar pitch angle, the roll orientation of the telescope, and the visit duration at a particular pointing position. The angular momentum (spin rate) of the reaction wheels must be managed to be kept within operational limits.
Mission planners are creative in using a technique for momentum management that is passive:
Momentum changes can be managed at some level by the way a sequence of observations is planned; this is done by observing at an orientation that builds momentum in a particular reaction wheel, followed by an observation at an orientation that removes momentum from that wheel.
But, not always, as some need-based science require a quicker slew and orientation which overrides the above momentum management program, and hence "momentum dumping" is performed by unloading the wheels as required.
However, managing momentum is only one of a number of planning constraints. At some point, one or more wheels will need to be adjusted to stay within operational bounds. The planning and scheduling system inserts planned momentum unloads into the schedule as needed, based on the modeling of expected momentum buildup, currently expected to be 1–2 times per week. Each unload activity takes a few hours, in which the observatory slews to a particular orientation to minimize the impact on the orbit and then fires thrusters as needed to allow the spin rate of the reaction wheels to be adjusted.
P.S: What Steve Linton mentioned in the question comments, is also correct.
The orbit will be biased to compensate for mean outward forces associated with gravitation of the planets and radiation pressure on the sun shield.
Above quoted passages are all from the PDF JWST Cycle 1 Documentation for Telescope and Spacecraft found on this page