AI 中文总结
本文利用1980-2024年美国49州578场州长选举的县级面板数据，采用堆叠双重差分法发现，2016年后共和党提名女性州长会使县级得票率降6.5个百分点，民主党无此效应，且初步支持需求侧歧视的可能。
AI 中文摘要
当一个政党提名女性担任行政职务时，该政党的得票率会发生什么变化？关于女性候选人的证据主要集中在立法选举，且研究设计通常将各政党和不同时期合并为单一的平均效应，而该效应通常为零。本文构建了一个新的美国州长选举县级面板数据集——涵盖1980年至2024年49个州的578场选举，通过原始资料重建并审核——采用堆叠双重差分法，估计政党特定、随时间变化的提名效应，该方法围绕州内从男性提名者向女性提名者的转变展开。平均为零的效应背后隐藏着明显的不对称性：2016年以来共和党转向女性提名者后，县级层面的得票率下降了6.5个百分点，该结果在队列、加权和推断检验中均具有稳健性；而民主党提名者未检测到类似的平均差异。选票诊断结果与需求侧歧视相符，但无法确定该渠道。
英文摘要
What happens to a party's vote share when it nominates a woman for executive office? The evidence on female candidates is dominated by legislative races and by designs that pool parties and periods into a single average effect, which is typically null. This paper builds a new county-level panel of US gubernatorial elections - 578 races in 49 states from 1980 to 2024, reconstructed and audited from primary sources - and estimates party-specific, time-varying nomination effects with a stacked difference-in-differences design around within-state transitions from male to female nominees. The average null conceals a sharp asymmetry. Republican transitions to a female nominee since 2016 are followed by a county-level vote-share decline of 6.5 percentage points, robust across cohort, weighting and inference checks. No comparable average differential is detected for Democratic nominees. Survey-ballot diagnostics are compatible with demand-side discrimination but cannot identify that channel.
