Intertwined with the persisting effects of the COVID-19 pandemic on the world economy, the price of WTI crude oil futures became negative on April 20, 2020. This anomalous incident has drawn much attention within the literature. This paper attempts to investigate the origins and specific impacts of the negative pricing event on the price discovery of WTI futures by employing a bivariate VECM-DCC-GARCH-SNP model, incorporating Legendre polynomials, where the dynamics of major information share measures at high-frequency time intervals are uncovered. Time-varying patterns of information share are identified across the period surrounding negative WTI prices. In particular, price discovery effects steadily abate after a sharp shock during the eight weeks before the negative pricing event. Peak price discovery differentials then re-occur within the negative-pricing event, before once again abating. Such results verify CFTC concerns surrounding the peculiarity of WTI futures trading conditions, that is, the conditions for the negative pricing event were well-established in the weeks before April 2020. Our results shed light on stylised evidence relating to the information efficiency of the international crude oil market more generally.

与 COVID-19 大流行对世界经济的持续影响交织在一起，WTI 原油期货价格于 2020 年 4 月 20 日变为负值。这一异常事件在文献中引起了广泛关注。本文试图通过采用二元 VECM-DCC-GARCH-SNP 模型，结合 Legendre 多项式，研究负定价事件对 WTI 期货价格发现的起源和具体影响，其中主要信息共享的动态测量在高频率时间间隔被发现。在 WTI 负价格期间，确定了随时间变化的信息共享模式。特别是价格发现效应在负定价事件发生前的八周内经历了剧烈震荡后稳步减弱。然后在负定价事件中再次出现峰值价格发现差异，然后再次减弱。这样的结果证实了 CFTC 对 WTI 期货交易条件特殊性的担忧，即负定价事件的条件在 2020 年 4 月之前的几周内已经确立。我们的结果揭示了与国际信息效率相关的程式化证据原油市场较为普遍。