We predict the evolution of giant clumps undergoing star-driven outflows in high- z gravitationally unstable disc galaxies . We find that the mass loss is expected to occur through a steady wind over many tens of free-fall times ( { t _ { ff } } \sim 10 { Myr } ) rather than by an explosive disruption in one or a few { t _ { ff } } . Our analysis is based on the finding from simulations that radiation trapping is negligible because it destabilizes the wind ( 43 ; 44 ) . Each photon can therefore contribute to the wind momentum only once , so the radiative force is limited to L / c . When combining radiation , protostellar and main-sequence winds , and supernovae , we estimate the total direct injection rate of momentum into the outflow to be 2.5 L / c . The adiabatic phase of supernovae and main-sequence winds can double this rate . The resulting outflow mass-loading factor is of order unity , and if the clumps were to deplete their gas the timescale would have been a few disc orbital times , to end with half the original clump mass in stars . However , the clump migration time to the disc centre is on the order of an orbital time , about 250 { Myr } , so the clumps are expected to complete their migration prior to depletion . Furthermore , the clumps are expected to double their mass in a disc orbital time by accretion from the disc and clump-clump mergers , so their mass actually grows in time and with decreasing radius . From the 6-7 giant clumps with observed outflows , 5 are consistent with these predictions , and one has a much higher mass-loading factor and momentum injection rate . The latter either indicates that the estimated outflow is an overestimate ( within the 1- \sigma error ) , that the SFR has dropped since the time when the outflow was launched , or that the driving mechanism is different , e.g . supernova feedback in a cavity generated by the other feedbacks .