算法歧视的两副面孔及其法律规制

我国现行算法歧视的法律治理存在两个弊端，分别是算法歧视规制范围过宽和算法歧视规制工具的凌乱。弊端的消除应当追溯至歧视概念本身，通过区分歧视的法律维度和道德维度，呈现算法歧视的两副面孔。第一副面孔是“算法偏见”，在语义上采用法律维度的歧视，对其治理旨在解决算法当中的历史偏见问题，适用反歧视法的规制原理，对算法设计者和算法应用者施加基于社会法的结果义务；第二副面孔是“算法不平等”，在语义上采用道德维度的歧视，对其治理旨在解决信息时代的资源分配问题。对于“算法不平等”，既要尊重私法自治又要削弱潜在歧视风险，法律规制除了事后司法救济外还可以采取事中方案，通过价值维度和信任维度，对算法设计者和算法应用者施以适度的负担。     

Do men and women discriminate against women for the same reason? Evidence from China

In this study, we examine whether men and women form gender discrimination for the same reason. To do that, we build an experimental Chinese labor market in which employers evaluate the productivity of workers who perform a real-effort task. Before evaluation, the employer observes the worker’s personal and group information. The personal information contains gender identity and a signal of productivity. The group information reveals the productivity distributions of some other male and female workers who do not differ in average productivity. However, it shows more male workers at the very top productivity levels and more female workers at the very bottom productivity levels in one treatment than in the other. According to the belief-based theory, there will be a greater degree of discrimination against female workers in the former. We find that, however, only male employers’ evaluations are well predicted by this approach. Female employers behave oppositely: their degree of gender discrimination is smaller in the treatment emphasizing men’s advantage in the tails of the productivity distributions. To explain female employers’ evaluations, we adopt the preference-based approach. Our findings suggest that employers of different genders can have different motivations for gender discrimination, and thus call attention to the theoretical foundation of gender discrimination and policy measures aimed at reducing gender discrimination.     

政府推动的就近城镇化能够消除工资歧视吗

随着大中城市城镇化空间的逐渐饱和以及农民工回流现象的出现,就近城镇化成为推进我国新型城镇化进程的重要模式。文章基于2017年中国流动人口动态监测调查(CMDS)数据,研究政府推动的就近城镇化对农村劳动力工资收入的影响发现:政府通过征地、户口改革方式推动的就近城镇化并不能够消除工资歧视,实现就近城镇化的农村劳动力在其他条件相同的情况下其工资水平仍然低于城镇本地职工,这一结论在纠正自选择偏差和样本选择偏差后仍然成立。机制分析发现,城镇劳动力市场针对实现就近城镇化的农村劳动力在就业机会方面的歧视与隔离已不明显,同工不同酬成为实现就近城镇化者面临的主要歧视形式,而城乡劳动力之间教育回报率的差异是导致同工不同酬的主要原因。进一步讨论发现,就近城镇化对于农村劳动力参保城镇职工保险具有显著的提升作用,而自我雇佣的就业形式则有助于打破城镇劳动力市场对农村劳动力的收入歧视。文章拓展研究了就近城镇化对农村劳动力工资收入的影响,为政府进一步推进以县域城镇化为主要形式的就近城镇化提供了研究参考。未来应持续推进城乡基本公共服务均等化,进一步缩小城乡教育质量差异,加强农村劳动力"同工同酬"的保护力度,稳步推进农村劳动力的就近城镇化。