Successful recoveries of households in the aftermath of extreme weather events are key to avoid long-term poverty implications. Yet, in frequently hit regions, there may not always be enough time for households to recover in-between recurrent events. There is a common narrative that the resulting incomplete recoveries aggravate adverse impacts, but there may also be counteracting mechanisms where a cluster of events leads to less destruction than a series of well-separated events because, after the first events, there are less assets left that can be destroyed by the subsequent events. To develop a systematic quantitative understanding for the interplay of the different mechanisms, we extend an agent-based household model to recurrent floods and study their welfare effects in the Philippines in dependence of household exposure and income. We find that incomplete recoveries increase cumulative consumption and well-being losses across the study period 2000-2018 by 50%. While low-income households suffer the highest well-being losses, lower-middle income households experience the largest relative increase in well-being (240%) and consumption losses (120%) due to incomplete recoveries. Our results show that the impacts of recurrent extreme weather events on households are not additive. In consequence, the well-being and consumption losses can be critically underestimated when concluding from the poverty implications of an individual event on the implications of recurrent events, as usually done in conventional disaster risk management. Thus, accounting for incomplete recoveries may allow to develop more effective risk management strategies and better prepare societies for an intensification of these events under global warming.