This study analyses the distributional effects of corruption on public spending in developing countries. It hypothesized that public expenditures with long and complex budgetary procedures are more prone to corruption. However, the new instrumental variables method proposed by Norkute et al. (J Economet 10.1016/j.jeconom.2020.04.008, 2021), was used to correct for the endogenous nature of corruption and the cross-sectional dependence bias of the panel units. The empirical analysis involved data from a sample of 40 countries, observed over the period 2005–2018. The main results show that the bias induced by corruption on the allocation of public expenditure depends as much on the opportunity offered by the expenditure in terms of payment of bribes as on the recipient of this expenditure. Investment spending with complex procedures is favored by corrupt bureaucrats over current spending. Wages and salaries are favored by corruption because they increase the financial benefits of bureaucrats. National and international anti-corruption institutions need to pay particular attention to the channels through which these public expenditure components are processed to establish greater transparency. Supplementary Information The online version contains supplementary material available at 10.1007/s43546-023-00452-1.