講座題目：Textual Analysis and Momentum Spillover 文本大數(shù)據(jù)與動量溢出效應

時間：2023年11月23日15:00-17:00

地點：博學樓925

主講人：李少然

主講人簡介：

李少然，北京大學經(jīng)濟學院金融系助理教授、博導。2021年獲劍橋大學博士學位。主要研究方向為金融計量，資產(chǎn)定價，投資組合管理，機器學習。已在Journal of Econometrics，Journal of Business & Economic Statistics、Journal of the Royal Statistical Society等期刊發(fā)表論文。主持國家自然科學基金青年項目一項。

講座內(nèi)容簡介：

中文摘要：本項目由兩篇文章組成，旨在探究A股市場的“動量溢出”效應與投資者注意力之間的聯(lián)系。我們通過百萬條財經(jīng)新聞數(shù)據(jù)構(gòu)建了兩個數(shù)據(jù)庫，即“新聞共現(xiàn)數(shù)據(jù)庫”與“文本領(lǐng)先滯后網(wǎng)絡(luò)數(shù)據(jù)庫”。其中，第一個數(shù)據(jù)庫產(chǎn)生的關(guān)聯(lián)網(wǎng)絡(luò)可以主導現(xiàn)階段學界業(yè)界其他常用的A股個股關(guān)聯(lián)網(wǎng)絡(luò)，更好地刻畫市場上存在的“動量溢出”效應。而第二個數(shù)據(jù)庫讓我們有機會更深入地了解動量溢出效應產(chǎn)生的原因和投資者獲取信息的特點。

Abstract:This project compares the momentum spillover effects detected by different linkage networks in the A Market and concludes that most connectivity documented in the past literature can be unified by the cross-firm momentum driven by the news-comention linkage. In other words, like the analyst momentum in the US market, a series of momentum spillover effects lose their predictive power after controlling for the media-implied momentum. We further separate the undirected news-comention network into directed sub-networks, including Lead-follower and Peer-effect networks. We argue that directed network has lower information processing cost for investors. Empirical evidence shows significantly contrasting behavior of sub-networks' momentum spillover, depending on leaders or peers to follow. We finally analyze the mechanism behind these phenomena.