This paper studies the political sustainability of programs that are targeted toward the poor. Given that the poor to whom these programs cater do not constitute a majority, we show that for their own good it pays to let the middle class benefit from them in a random way. This approach mimics the actual institutional arrangements whereby middle-class individuals feel that they can successfully apply to the programs. We consider a two stage decision process: first a Rawlsian government chooses the probability at which the middle class is allowed to benefit from a given program; then, majority voting determines the level of benefit and the rate of contribution. At the first, constitutional stage, the government cannot commit to a specific level of taxes and benefit but anticipates that these are set by majority voting in the second stage.
Cremer, H., Klimaviciute, J., & Pestieau, P. (2021). A political economy of loose means-testing in targeted social programs. Economics Letters, 202, 109810. https://doi.org/10.1016/j.econlet.2021.109810 (Original work published 2021)