[金融学]大家一起读Paper7 Yuan, Yu. 2011. “Attention and trading.” Unpublished-经管之家官网！

1 论文标题
Attention and Trading
2 作者信息
Yu Yuan
3 出处和链接（比如，NBER working paper No.11000）
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1105532
4 摘要
This study empirically explores the effects of attention levels on investors'
trading behavior and on market price dynamics. Specifically, we analyze the
ability of market-wide attention-grabbing events - record-breaking events of
the Dow index and front page articles about the stock market - to predict the
trading behavior of investors and market returns. The empirical results show
that the impact of attention is pervasive across the market. High attention
causes individual investors to reduce their stock holdings dramatically when
the market level is high and to increase their stock holdings modestly when
the market level is low. The aggressive selling by individual investors induces
institutional investors to trade and has a negative impact on market prices,
reducing market returns by 19 basis points on days following attention-grabbing
events.
这哥们上海交大的本科，威斯康星麦迪逊的双硕士，沃顿的金融学博士，一个学术新星，毕业后去了爱荷华大学，最近又回到沃顿做visiting AP，最新的消息是他已经从爱荷华辞职了。2007年的时候在大连的CICF上有过一面之交，吉林人，说起话来感觉更像是一个领导，可能这就是东北人身上特有的气质。那年他获得了最佳论文奖，有趣的是，他当时以学生身份注册的，就没有招待酒会的票，而这个奖正是在这个酒会上颁发的，结果他后来才知道自己得了奖。
说说这篇文章吧，主要有以下几个方面可以借鉴：
第一，它很好的处理了attention-grabbing event和information content之间的关系。他选用道琼斯指数达到历史高点作为event（一个dummy var），同时控制了市场过去一年的收益率，这样，dummy var所度量的就是attention的净影响，从而解决了与information content的内生性问题。这是这篇文章的一个很大的贡献，因此也花了很多篇幅来进行验证；
第二，用了三套数据得到了同样的结论，其中一套关于新闻头条的数据是coding出来的，足见下的功夫有多大。给人的感觉是，结论不信也得信；
第三，为了便于解释回归系数，变量进行了标准化处理。
我的读书笔记见下：
5.1 Purpose
1. Do attention levels affect the trading behavior of individual and institutional investors?
2. Do attention levels influence the level of the market prices?
5.2 Issues
1. An obstacle to examine the impact of attention is that attention-grabbing events normally coincide with the release of economically meaningful information. Thus, a good candidate event for our test should attract investors' attention but not contain much economic content. Furthermore, the remaining economic information should be controllable in the tests. We propose that Dow record events and news events fit these criteria. Because
(a) From the alternative hypothesis that posits record events are related to economic fundamentals, we expect record events of the broader market indices, the NYSE and the S&P, to show strong empirical patterns. However, significant empirical patterns emerge only for the highly visible indices (the Dow and the Nasdaq), while no patterns emerge for the economically meaningful indices (the NYSE and the S&P). These results strongly suggest that attention rather than economic information drives our empirical findings.
(b) while Dow record events can only happen when the market price level is high, news events span the periods associated with high price levels (good times) and low price levels (bad times).
5.3 Brief Conclusions
1. we find that attention has a strong impact on the trading behavior of individual investors. High attention causes individual investors to reduce dramatically their stock positions in good times and to increase modestly their stock positions in bad times.
2. high attention also influences the aggregate market price level. In a 75- year sample, Dow record events predict the next-day return of the value-weighted NYSE-AMEX index to be 19 basis points lower. Furthermore, when the Dow reaches 17 “milestones” (hundred marks when the Dow is below 1000 and thousand marks when the Dow is over 1000) for the first time, the next-day market returns experience an additional 28 basis points drop. The results imply that aggressive selling places considerable pressure on market prices and lowers next-day returns.
3. attention raises the trading intensity of institutional investors in good times. This result suggests that attention indirectly influences the trading behavior of institutional investors through individual-investor trading behavior and price changes.
4. The empirical results above also show that attention has a stronger impact on individual investors than on institutional investors. Individual investors are more occupied with other activities and have weaker analyzing capacity than institutional investors. Attention-grabbing events should significantly raise the level of attention individual investors give to financial activities and hence have a stronger direct impact on their trading behavior.
5. We find strong evidence that the disposition effect together with high attention drives individual-investor abnormal selling behavior; we also find some support for the rebalancing hypothesis.
5.4 Miscellaneous
1. All series except Dowt are normalized to have unit variance. Note: how to interpret coefficient in this case.