Financing consumption of the elderly in the face of the projected increase in life expectancy is a key challenge for economic policy. Moreover, standard structural models with fully rational agents suggest that about 50-60 percent of old-age consumption is financed with voluntary savings, even in the presence of a fairly generous public pension system. This is clearly inconsistent with either the data, or the alarming simulations of old-age poverty in the years to come. Old-age saving (OAS) schemes are widely used policy instruments to address this challenge, but structural evaluations of such instruments remain rare. We develop a framework with incompletely rational agents: lacking financial literacy and experiencing commitment difficulties. We study a broad selection of OAS schemes and find that they raise welfare of financially illiterate agents and to a lesser extent improve welfare of agents with a high degree of time inconsistency. They also reduce the incidence of poverty at old age. Unfortunately, these instruments are fiscally costly, induce considerable crowd-out and direct fiscal transfers mostly to those agents, who need it the least.