We model the response of spherical, nonrotating Milky Way (MW) dark matter and stellar halos to the Large Magellanic Cloud using the matrix method of linear response theory. Our computations reproduce the main features of the dark halo response from simulations. We show that these features can be well separated by a harmonic decomposition: the large-scale over/underdensity in the halo (associated with its reflex motion) corresponds to the ℓ = 1 terms, and the local overdensity to the ℓ ≥ 2 multipoles. Moreover, the dark halo response is largely dominated by the first-order forcing term, with little influence from self-gravity. This makes it difficult to constrain the underlying velocity distribution of the dark halo using the observed response of the stellar halo, but it allows us to investigate the response of stellar halo models with various velocity anisotropies: a tangential (respectively radial) halo produces a shallower (respectively stronger) response. We also show that only the local wake is responsible for these variations, the reflex motion being solely dependent on the MW potential. Therefore, we identify the structure (orientation and winding) of the in-plane quadrupolar (m = 2) response as a potentially good probe of the stellar halo anisotropy. Finally, our method allows us to tentatively relate the wake strength and shape to resonant effects: the strong radial response could be associated with the inner Lindblad resonance, and the weak tangential one with corotation.