What should be the role of a public development bank (PDB)? Which projects/firms should the PDB target? What can theory say about the types of loans and ways of delivering them that PDBs around the world use? We analyze these questions in the context of a model where screening is costly to banks. Underprovision of credit results from the inability of banks to (i) appropriate the full benefits of projects they finance, more pronounced for high value projects; and (ii) internalize the benefits of screening in terms of aggregate lending. PDB intervention naturally addresses inefficiencies originating in failures in the private provision of credit. Though lending to commercial banks at subsidized rates or providing credit guarantees are valid alternatives, guarantees are less effective than subsidies at equivalent cost. PDB lending is particularly important in recessions, when liquidity/capital shortages are likely, and risk profiles deteriorate potentially leading to further credit underprovision. Subsidized lending is further preferred to guarantees when banks are facing a liquidity shortage, while a credit guarantees program has additional benefits when banks are undercapitalized.