In this paper I investigate the retirement-consumption puzzle in Italy for the period 2010-2016, using SHIW data. In order to address the endogeneity of the retirement decision, I estimate the effect of retirement by exploiting the exogeneity of pension eligibility in an instrumental variable approach; the IV regression is then applied in a regression discontinuity design where only households close to the eligibility point are considered. The eligibility-instrument is found to be a strong predictor of the retirement decision, and the estimated non-durable consumption drop is equal to 12.3%. When households are distinguished according to the gender of the household head, female-led households are found to undergo a consumption decline that is more than double that estimated for households with male heads. The data and the literature on the subject indicate that this large difference is likely related to the gender pay-gap that translates into a gender pension-gap. Moreover, the consumption decline appears to be concentrated in households in the lower part of the wealth distribution. Nonetheless, households in the lowest wealth quintile, do not show a significant consumption decline. The data suggests that this might be due to the impossibility for these households to further reduce their consumption at retirement, as they are mostly composed of essential expenditures.