The nucleation and rupture processes of deep-focus earthquakes have remained enigmatic ever since their discovery. These earthquakes occur mostly within the mantle transition zone where brittle failure is extremely unlikely due to the elevated pressures at these depths. Hence, other mechanisms have to be invoked to explain the occurrence of these events. To date, two main hypotheses have been put forward to explain deep focus earthquakes: transformational faulting (due to the polymorphic phase change of metastable olivine to either wadsleyite or ringwoodite) and thermal runaway (due to the conversion of deformational work to heat). More recently, it has been proposed that the feedback between those two mechanisms may explain the observed two-stage ruptures of large deep-focus earthquakes. To better understand the potential feedback between transformational faulting and thermal runaway, it is necessary to determine the stresses induced by the phase change due to i) the grain size reduction and corresponding viscosity reduction of the transformed material and ii) the volume reduction of the transformed phase. The former process triggers a stress transfer from the transformed material to the untransformed material, whereas the latter results in elevated stresses around the transformed phase. In this study, we employ numerical models with a viscoelastic compressible rheology to quantify the stress levels and patterns resulting from both processes. To gain a better understanding of the parameters controlling the stress transfer from transforming regions to the surrounding matrix, we employ simplified numerical models where transforming regions are approximated using elliptical inclusions. In a second step, more realistic model geometries are used to additionally study the effect of the morphology of transformed regions on stress levels and heterogeneities. Results show that both processes result in significantly different stress evolution upon a phase transition. Whereas a phase transition affecting only the viscosity of the transformed material results in moderate stress increases which occur on relatively long timescales, a phase transition affecting both viscosity and density results in significantly larger stresses, which also exhibit a significantly faster build-up. In both cases, the attained stress are sufficiently large to activate additional ductile weakening mechanisms that could trigger ductile ruptures. The higher stress levels resulting from the combined effect of a viscosity and density change would likely result in stronger weakening effects and faster occurrence of ductile failure