Investor Attention in Financial Markets
Investor Attention in Financial Markets
2020 SAIF‐CAFR Financial Research Summer Camp
Zhi Da University of Notre Dame Caveats and logistics
This is a VERY incomplete literature survey on investor attention A partial collection of papers can be downloaded at https://www.nd.edu/~zda/PhD_Papers.rar Check my website later this year for a survey paper with Wei Xiong and Lin Peng Email me ([email protected]) if you are aware of other interesting related papers (including your own) Format: 9:05 – 10:20: lecture 10:20 –10:35: Break 10:35 – 11:45: lecture 11:45 – 12:00: Q&A
2 Traditional Asset Pricing Models
Information
Trading
Price, volatility, etc.
3 Traditional Asset Pricing Models
assume that information is incorporated into prices with lightning speed
4 The Role of Investor Attention
Information
Investor Attention
Trading
Price, volatility, etc.
5 Limited Investor Attention
Attention is a scarce Information cognitive resource (Kahneman, 1973)
“a wealth of information creates a poverty of attention and a need to allocate that attention efficiently among the overabundance of information sources that might consume it,” Herbert Simon, Nobel Laureate in Economics
6 Road Map
Theoretical
Measurement
Empirical
Future Research?
7 Theoretical Analysis
8 Conceptual framework
One can broadly divide attention effects into two large categories, motivated by Kahneman’s systems 1 and 2 thinking:
Ex‐ante attention: deliberate attention under constraints Focus of today
Ex‐post attention: fast attention response to external triggers Salience, Sensational news, Ostrich effect, etc…
9 Deliberate attention
Rational Inattention Representative agent is aware of her attention constraint and optimally allocates attention under the constraint Merton (1987), Sims (2003), Peng (2005), Peng and Xiong (2006), Kacperczyk, van Nieuwerburgh and Veldkamp (2016)…
“Neglective” Inattention Inattentive investors are not aware of their attention constraints and do not learn from prices Continuum of agents: prices are jointly determined by inattentive and fully attentive investors Hirshleifer and Teoh (2003), Tetlock (2011), Hirshleifer, Lim, and Teoh (2011), Da, Gurun and Warachka (2014)…
10 Merton (1987, JF)
Attention is costly and investors will not have perfect information on all stocks at all time
Extreme Attention Constraint: Investors include a stock to her portfolio only if she knows the stock
Thus they hold “suboptimal” portfolios
11 Merton (1987, JF)
Expected return is higher when Beta is higher Idiosyncratic volatility is higher Firm is larger Shareholder base is smaller
12 Attention and Entropy
Sims (2003, JME) Information reduces the uncertainty (entropy) associated with a random variable: H(X) = ‐E log[f(X)] Investor inattention places an upper bound on the reduction in entropy
Peng (2005, JFQA) Investors learn about fundamental factors subject to the attention constraint as in Sims (2003) Amount of information and asset prices are jointly determined in equilibrium
13 Peng and Xiong (2006, JFE)
A representative investor solves two optimization problems each period: 1. Optimally allocate her limited attention to (1) market factor; (2) sector factors; (3) firm‐specific factors
attention reduces uncertainty
1. Based on the processed information, she then solves the standard consumption Bellman equation
14 Peng and Xiong (2006, JFE)
Limited attention results in categorical learning Investors allocate more attention to market‐ and sector‐level factors than to firm‐specific factors In severely constrained cases, the investor allocates all attention to market‐ and sector‐level information and ignores all the firm‐specific data
Limited attention could acerbate the impact of behavioral biases on asset prices Excessive co‐movement
15 Kacperczyk, van Nieuwerburgh, Veldkamp (2016, Econometrica) Mutual fund managers allocate attention to make profitable portfolio choices
Their information choices are not observable
Need a theory to link observable variable to information and portfolio choices
16 Model Setup
17 Time 1: Choose Attention Allocation
18 Equilibrium
19 Neglective Inattention Inattentive investors are not aware of their constraints Convenient toy model to motivate empirical studies
Consider a one‐period model. A stock pays a liquidating dividend (d) at the end of the period, where d is a random variable with a variance of Var(d). For simplicity, we assume the stock has net zero supply and borrowing and lending rates are zero.
There are two types of agents in the market: rational or attentive investors (labelled as r) and constrained or limited attention investors (labelled as l). Limited attention investors account for a fraction m of the economy, while rational investors account the remaining 1‐m.
Both types of investors have the CARA utility over their end‐of‐period consumption: Ui(Ci) = Ei(Ci) –0.5γVar(Ci), i= r or l, where γ is the coefficient for risk‐aversion. They have different expectations of the dividend.
What’s the equilibrium stock price today? 20 Solution
Assume xi is the stock demand of investor i, i= r or l. The utility is therefore: xi [Ei(d)‐p] –0.5γVar(xid).
First‐order condition gives the optimal demand function as xi = [Ei(d)‐ p]/[Avar(d)].
Equating aggregate demand to aggregate supply (which is assumed to be zero) gives:
m[El(d)‐p]+(1‐m) [Er(d)‐p] = 0 =>
p = mEl(d) + (1‐m) Er(d)
In other words, the equilibrium price today is the weighted average of investors’ cash flow expectations.
21 Three Examples (I)
Tetlock (2011, RFS) S1 and S2 are two different signals in two periods Investors with limited attention perceive kS1 + S2 as the new signal in the second period Price reacts to stale news
Da, Gurun, and Warachka (2014, RFS) Frog‐in‐the‐pan (FIP) investors overlook small signals (|Si| < k) during the first period A series of small changes attracts less attention and results in under‐reaction and price momentum
22 Three Examples (II)
Hirshleifer, Lim, and Teoh (2011, RAPS) Two types of inattentive investors Type I pays no attention Type II pays attention to total earnings (e) only Rational investor pays attention to components of earnings (e = a + c) as well: cash flow (c) and accrual (a)
Type I investor post‐earnings announcement drift
Type II investor over‐reaction to accruals
23 Measurement
24 Measuring investor attention
Measuring attention empirically is tricky Trading volume: Gervais, Kaniel, and Mingelgrin (2001); Barber and Odean (2008); Hou, Peng, and Xiong (2008) Extreme returns: Barber and Odean (2008) Up/down markets: Hou, Peng, and Xiong (2008) Firms’ advertising expense: Grullon, Kanatas, and Weston (2004), Chemmanur and Yan (2009), Lou (2014) The existence of news: Barber and Odean (2008) Repeated news stories: Tetlock (2011)
Turnover and returns are noisy and “catch‐all” proxies and news coverage and advertising expense capture the supply of attention or passive attention
25 Da, Engelberg, and Gao (2011, JF)
Google’s Search Volume Index (SVI)
26 Da, Engelberg, and Gao (2011, JF)
Consider “AAPL” and “MSFT” (shown below) We focus on innovations in SVI of each search term
27 Da, Engelberg, and Gao (2011)
We show that our attention measure is capturing retail attention Intuitively it should be individual, retail investors Barber and Odean (2008) “… individual investors are more likely to buy rather than sell those stocks that catch their attention. … this is so because attention affects buying—where investors search across thousands of stocks— more than selling—where investors generally choose only from the few stocks that they own. While each investor does not buy every single stock that grabs his attention, individual investors are more likely to buy attention‐grabbing stocks than to sell them. (pg 786)” Increased retail attention positive price pressure
Preferences determine choices after attention has determined the choice set 28 ASVI and Retail Trading Madoff NYSE/ARCH Comparison Δ Order Δ Turnover Δ Order Δ Turnover Δ Order Δ Turnover (1) (2) (3) (4) (5) (6) ASVI (t-1, t) 0.264*** 0.297*** 0.0920*** 0.104*** 0.166*** 0.204*** (0.0317) (0.0355) (0.0105) (0.0132) (0.0218) (0.0256) ASVI X Madoff 0.109*** 0.0951** (0.0328) (0.0374) Madoff 0.000440 0.0223*** (0.00223) (0.00253) Log(Market Cap) (t-1) -0.0117*** -0.0122*** -0.00889*** -0.0129*** -0.00411*** -0.00841*** (0.00202) (0.00207) (0.000641) (0.000713) (0.00132) (0.00152) RET (t) 0.154*** 0.0772* 0.0999*** 0.00647 0.0418 -0.0875*** (0.0372) (0.0437) (0.0173) (0.0199) (0.0284) (0.0331) |RET(t)| 1.299*** 1.570*** 1.001*** 1.418*** 1.244*** 1.622*** (0.0528) (0.0622) (0.0271) (0.0338) (0.0405) (0.0493) Chunky News Dummy (t) 0.0658*** 0.0915*** 0.0936*** 0.125*** 0.0768*** 0.0991*** (0.00997) (0.0121) (0.00301) (0.00364) (0.00678) (0.00841) Advert. Expense / Sales (t) -0.104* -0.0954 0.00255 -0.0328*** -0.0713 -0.0568 (0.0630) (0.0642) (0.00643) (0.00636) (0.0610) (0.0658) Constant 0.255*** 0.251*** 0.175*** 0.229*** 0.0570* 0.119*** (0.0480) (0.0492) (0.0148) (0.0167) (0.0303) (0.0349) Control Variables YES YES YES YES YES YES Month Fixed Effect YES YES YES YES YES YES Observations 35,280 35,280 103,253 103,253 52,837 52,837 Number of Clusters (Stock) 1,358 1,358 2,743 2,743 962 962 29R2 0.131 0.127 0.299 0.291 0.173 0.191 ASVI and Price Pressure
Week 1 Week 2 Week 3 Week 4 Week 5-52 (1) (2) (3) (4) (5) ASVI 18.742*** 14.904** 3.850 -1.608 -28.912 (7.000) (7.561) (6.284) (6.903) (17.162) Log Market Cap * ASVI -21.182*** -15.647** -4.710 4.290 16.834 (6.508) (6.768) (6.516) (6.398) (88.624) Log Market Cap 2.653 3.858 3.144 3.575 -39.229 (3.023) (3.160) (3.063) (3.186) (67.405) Percent Dash-5 Volume * ASVI 3.552** 1.904 1.687 -2.744 16.258 (1.639) (1.522) (1.612) (1.717) (23.822) Percent Dash-5 Volume 1.607 1.351 1.486 0.364 119.901*** (1.644) (1.652) (1.659) (1.711) (31.765) APSVI -2.532*** -1.379 -0.701 -0.704 2.286 (0.930) (0.990) (0.808) (0.639) (9.909) Absolute Abnormal Return 1.314 -2.389 -1.128 -0.463 -1.510 (1.879) (1.979) (1.563) (1.405) (28.505) Advertising Expense / Sales -4.012* -4.686** -3.959* -4.153* -162.210*** (2.237) (2.228) (2.172) (2.234) (52.414) Log(1 + # of analysts) -3.747** -4.547*** -3.961** -4.120** -173.875*** (1.548) (1.741) (1.769) (1.769) (29.683) Log(Chunky News Last Year) -5.157 -5.549* -4.349 -5.409 -14.999 (3.370) (3.272) (3.292) (3.558) (80.730) Chunky News Dummy 3.610* 1.378 -3.825 -0.058 32.466 (2.025) (2.424) (2.483) (1.910) (28.441) Abnormal Turnover 2.398** 2.309** 2.022 0.316 10.531 (1.204) (1.144) (1.404) (1.098) (10.109)
30 Observations per week 1499 1498 1497 1496 1414 R-Squared 0.0142 0.0119 0.0112 0.0111 0.0170 First‐day IPO Return
20.00%
18.00% Mean IPO First-day Return 16.98%
Median IPO First-day Return 16.00%
14.00%
12.00% 12.00% 10.90%
10.00%
8.00%
6.07% 6.00%
4.00%
2.00%
0.00% Low Pre-IPO ASVI High Pre-IPO ASVI
31 Post‐IPO Return [5w, 52w], High vs. Low ASVI
10.00% 7.55%
5.00%
1.39%
0.00%
-1.56%
-5.00%
-10.00%
Mean Size and B/M Portfolio Adj. Post-IPO Return -15.00%
Median Size and B/M Portfolio Adj. Post-IPO Return
-20.00% -19.51%
-25.00% High First-day IPO Return, Low Pre-IPO ASVI High First-day IPO Return, High Pre-IPO ASVI
32 Institutional vs. Retail Attention?
Attention Allocation Institutional attention: deliberate, proactive, endogenously determined Retail attention: fast, reactive, triggered by salient events and media coverage
Asset Pricing Institutional attention: Immediate, permanent price impact Retail attention: Delayed, positive and temporary price pressure (Barber and Odean, 2008)
33 Institutional Attention
Ben‐Rephael, Da, and Israelsen (2017) examine abnormal institutional attention (AIA) using ticker search on 34 Bloomberg “WWE” EXAMPLE: ABNORMAL INSTITUTIONAL ATTENTION & NEWS
Earnings AnalystAnnouncements Recommendation Changes
News about Executives
News about Litigation
News about Competitors
Other News
35 Post‐earnings‐announcement drift Only when institutional investors fail to pay attention
36 Fedyk (2018): Front page News on Bloomberg
FP
NFP
37 Fedyk (2018)
Establish causality using random front‐page news positioning on the Bloomberg terminals Front Page News vs. Non Front Page News: High vs. Low attention
More attention more trading More attention faster price adjustment
38 Financial Attention
Sicherman, Loewenstein, Seppi, and Utkus (RFS, 2015) Use daily investor online account logins to measure financial attention Strong ostrich effect: more logins in up market than in down markets, more so for males and for wealthy investors High VIX less logins; More media coverage more logins
39 Online Brokerage Account Activities
Gargano and Rossi (RFS, 2018) New brokerage account dataset covering 10,000 random selected accounts over 2013/01 ~ 2014/06 Access to their entire online activity on the trading platform
Different types of attention Overall attention Categorical attention (Balance and Positions; Research; Trading…) Stock attention (time spent on each ticker)
Differentiate deliberate vs. fast attention
40 41 Click Data: Attention to Specific News Benamar, Foucault, and Vega (2020, RFS) Use Bitly data on clicks by individuals on news articles URL: https://blogs.wsj.com/economics/2016/01/07/why-december- private-payrolls-arent-a-great-predictor-of-the-jobs-report/ SURL created using Bitly is: http://on.wsj.com/2oJQ2py
Write down a model of endogenous attention allocation where more attention is allocated to more uncertain events # clicks before an event is therefore a measure of uncertainty associated with that event Predicts stronger announcement effect
42 Attention to Macro Announcements
43 Big data allow us to measure attention more directly and more precisely
44 Empirical Evidence
45 Outline
Anecdotal Evidence
Asset pricing Limited attention to different types of information
Corporate finance and others
Identification
46 MCI‐MCIC Massively Confused Investors Making Conspicuously Ignorant Choices (Rashes, 2001) MCI is an NYSE‐listed closed‐end fund. MCIC is MCI Communications When MCI Communications was in acquisition talks with WorldCom, these two securities had very similar trading patterns Excessive price co‐movement during 94‐97
47 More recent example
Shares of Tweeter Home Entertainment Group surged briefly after investors apparently confused the stock for that of Twitter. Its ticker now changed from “TWTRQ” to “THEGQ” 48 The wrong Hathaway?
≠
But students in investment theory class at KU found the news coverage of Anne Hathaway is highly correlated with the trading volume in Berkshire Hathaway’s stock
49 Time‐varying attention constraint
Corwin and Coughenour (2008, JF) NYSE specialists allocate effort toward their most active stocks during periods of increased activity, resulting in less frequent price improvement and increased transaction costs for their remaining assigned stocks
Identification: distraction due to MSFT is plausible exogenous (negative) attention shock to F
50 A Piece of NYSE History at Mendoza
51 Time‐varying attention constraint
Hirshleifer, Lim, and Teoh (2009, JF) Investor distraction hypothesis: More post‐earnings announcement drift following days with more announcements
DellaVigna and Pollet (2009, JF) More post‐earnings announcement drift following Friday announcement
52 Limited attention to long‐term information
DellaVigna and Pollet (2007, AER) Demographic shocks today lead to predictable shifts in consumption and industry profitability in 5 to 10 years Investors have limited attention to information beyond 5 years, resulting in return predictability
53 DellaVigna and Pollet (2007, AER) Future population can be accurately predicted
54 DellaVigna and Pollet (2007, AER) Demographic change affects future industry CF in a highly predictable way
55 DellaVigna and Pollet (2007, AER) But investor limited attention may cause a delayed price adjustment
56 Limited attention to long‐term information
Da and Warachka (2009, JFE) Equity analysts issue both short‐term and long‐term earnings forecasts Investors pay more attention to the short‐term forecasts than to the short‐term forecasts Disparity in the forecasts predicts future return
57 Da and Warachka (2009, JFE)
Disparity LTG and ISTG reflects limited attention to long‐term implication of information
58 Limited attention to economic links
Cohen and Frazzini (2008)
Supplier’s stock price reacts to shocks to its customer with a delay
59 Cohen and Frazzini (2008)
60 Cohen and Frazzini (2008)
61 Limited attention to stale information
Tetlock (2011, RFS) “People everywhere confuse what they read in the newspaper with news.” –A. J. Liebling Identify stale news using textual similarity measure Price reacts to stale news immediately, followed by reversal
Gilbert et. al. (2012, MS) The US Leading Economic Index (LEI) is a summary statistic of previously released inputs Yet investors react to its announcement Smart money front runs
62 Limited attention to continuous information
Da, Gurun, and Warachka (2014, RFS)
63 64 65 Corporate Finance
Kempf, Manconi, and Spalt (RFS, 2016) “Distracted” shareholders make “worse” corporate decisions Diversifying, value‐destrorying acquisition Cut dividends Less likely to fire bad‐performing CEOs
Causality Use exogenous shocks to unrelated parts (e.g. different industry) of institutional shareholders’ portfolio
66 Identification
67 68 Merger & Acqusition
Lie (JFQA, 2019) In certain acquisitions, target shareholders can choose to receive either cash or acquirer shares Ecolab acquisition of Nalco in 2011: “Nalco stockholders may elect to receive either 0.7005 shares of Ecolab common stock or $38.80 in cash, without interest, per share of Nalco common stock.”
Define Value wedge as log(stock value / cash value) Ignore taxes, take stock if Value wedge>0; otherwise take cash
69 Payment Choices
100%
10% partially inattentive
0% -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Value wedge
Fraction stock Fraction cash
70 No Election
30%
7% inattentive
0% -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Value wedge
71 Insider Trading
Alldredge and Cicero (JFE, 2015) Insider trading profit may come from insiders pay close attention to public info when outside investors are relatively inattentive Setting: supplier insider trade on public info on customers Extending Cohen and Frazzini (2008) to insider trading
72 IPO
Bajo, Chemmanur, Simonyan, and Tehranian ( JFE, 2016)
Lead underwriters who are central in the network of investment bankers are better in attracting investor attention to the IPOs More attention more information extraction better IPOs
73 Center of Attention
74 CEO Equity Vesting
Edmans, Goncalves‐Pinto, Groen‐Xu, and Wang (2018, RFS) CEOs regular sell stocks after they are vested They strategically release 20% more news (not always positive) during scheduled vesting months Why? Identification issues? Corr(sales, news release)>0
75 2SLS
76 77 Causal Evidence in the Existing Literature
Distraction by unrelated shocks Multiple earnings announcements on the same day: Hirshleifer, Lim, and Teoh (2009, JF) Earnings announcements on Fridays: DellaVigna and Pollet (2009, JF) NYSE specialist making market on multiple stocks: Corwin and Coughenour (2008, JF), Schmit (2019) Distracted shareholder: Kempf, Manconi, and Spalt (2016, RFS) Competing special news on TV: Liu, Sherman, and Zhang (2014, MS), Peress and Schmit (2019, JF)
78 Causal Evidence in the Existing Literature
Distraction by unrelated shocks Shocks in the family (marriage / divorce / parental loss): Lu, Ray and Teo (JFE, 2016), and Shu, Sulaeman and Yeung (2017)
79 Causal Evidence in the Existing Literature
Exogenous events Weather shocks: Engelberg and Parsons (2011, JF), Gargano and Rossi (2018, RFS) Newspaper strike: Peress (2014, JF) College football Bowl games: Mayer (2018)
80 Causal Evidence in the Existing Literature
Positioning / salience of information Discontinuity around WSJ “Category Kings” ranking: Kaniel and Parham (2017, JFE)
Record‐setting days in Dow (NASDAQ) but not in S&P500 (NYSE Composite) : Yuan (JFE, 2015) Stock ranking variation due to rounding: Wang (2017, WP)
81 Three elements of an interesting attention paper
New application Make sure it has not been done before
Identification: Pin down causality (in)attention outcomes
Ex‐ante surprising results It usually is not enough to just document that investors did not pay attention What are some unexpected consequences of their inattention?
82 Conclusion Investors are likely to have limited attention
Limited investor attention affects financial market outcomes both theoretically and empirically
Measuring attention and pinning down causality are challenging tasks
83