A model of theft and bribery
用一个简单的共同代理模型同时刻画盗窃和贿赂两种腐败行为，发现政府收入存在一个阈值，低于阈值时官员只收贿不盗款，高于阈值才开始盗款；在政治竞选中，候选人的贿款与盗款互为替代。
Abstract Two of the most pervasive kinds of public corruption are theft and bribery. I show how a simple common agency model can simultaneously depict both. In one setting, a public official (agent) who has discretion over government revenues can steal some of it and allocate the rest to public spending towards various sectors (principals), some of whom can pay bribes in exchange. I find that there is a threshold level of revenues at and below which the official obtains bribes but does not steal revenues. She only starts stealing revenues above the threshold. In another setting, there are two political candidates (agents), each promising to allocate public spending to various groups of voters (principals), some of whom can pay bribes in exchange. Each candidate then uses the bribes and any government revenues that she expects to steal in office to buy some votes. I find that the candidate who would steal less revenues in office will obtain larger bribes, and vice versa, which suggests that bribe money and stolen revenues are substitute sources of (illicit) campaign funds.