Guiding Seminar 1A (2020)

Question 1

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

Why do consumers turn to fellow customers when choosing among products instead of relying on expert advice?

Answer 1

Because of:

1. The emergence of social media platforms;
2. The increase in creation, popularity & consumption of user-generated content.

Question 2

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the 2 Research Questions of this paper?

Answer 2

1. Do peer opinions include value-relevant news or do they merely constitute “random chatter” (i.e. the task should be left to professional analysts)?
2. Are some users attempting to intentionally spread false “information” and mislead fellow market participants?

Question 3

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the goals of this study?

Answer 3

1. To assess the performance of investors-turned-advisors

2. To test whether investors can turn to their peers for genuine, useful investment advice.

Question 4

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What is the main source of the data in this study, and how is it measured?

Answer 4

“Seeking Alpha (SA)” - one of the biggest investment-related social media websites in the US.

The data is measured in the frequency of negative words used in an article.

Question 5

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the two channels in SA for investors to voice their opinion?

Answer 5

1. Through opinion articles which are reviewed by a panel and are subject to editorial changes;
2. Through commentaries written in response to the opinion articles.

Question 6

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the key variables used to make a regression in this study?

Answer 6

X1 & X2:
1. NegSAi,t – an average fraction of NEGATIVE WORDS across all single-ticker ARTICLES published on SA about company i on day t

2. NegSA-Commenti,t – an average fraction of NEGATIVE WORDS across all SA COMMENTS posted in an article about company i on day t

Y:
3. ARet - abnormal returns - “the difference between RAW RETURNS minus RETURNS on a value-weighted portfolio OF FIRMS WITH SIMILAR SIZE, book-to-market ratio, and past returns”.

Question 7

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What is the main finding of this study?

Answer 7

“The fraction of negative words in SA articles and the fraction of negative words in SA comments both negatively predict stock returns over the ensuing three months”.

As per our dear Kostja:
If negative words in articles/comments are up by 1%,
the future abnormal returns are lower by 0.332%/0.194%.

Question 8

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the possible reasons (channels) behind the main finding of the study?

Answer 8

Possible reasons behind findings:

1. Predictability Channel: SA articles & commentaries contain pieces of value-relevant information which are not fully factored into the stock price as of the article publication date. As investors learn from the SA view, prices gradually adjust.
   - -If this reason is true:
   - —SA views indeed predict future stock market performance;
   - —Social media outlets are a useful source of value-relevant advice
2. Clout Channel: SA views reflect false information and cause investors to trade in the direction of the underlying articles and comments and move prices accordingly” (i.e. exploiting naïve investors)
   - -This reason is less likely to be true due to:
   - —SA followers’ insufficient capital to cause real & sufficient market movements.

Question 9

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What is an Earnings Surprise?

Answer 9

The difference between:

• -the REPORTED earnings-per-share (EPS) AND;
• -the average of financial ANALYSTS’ EPS FORECASTS issued within 30 days prior to the earnings announcement.

Question 10

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

How do the authors prove if the information in SA is value-relevant (Predictability Channel) or false (Clout Channel)?

Answer 10

They regress the number of negative words in articles/comments (the same X1 & X2 as before) on subsequent earnings surprises (new Y).

If they find that negative words predict earnings surprises: the information in SA is value-relevant and is not fully factored into the stock price as of the article publication date.

If they find that negative words do NOT predict earnings surprises: the information in SA is false as users cannot affect the company’s earnings (no relationship between X1, X2 & Y).

Question 11

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

Do the authors find that the SA articles & commentaries contain pieces of value-relevant or false information?

Answer 11

They find that:

“The fraction of negative words in SA articles & comments strongly predict earnings surprises,” which suggests that the opinions expressed in SA articles & comments.
indeed provide value-relevant information.

Question 12

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

Why should a SA user share their insights of value-relevant information publicly?

Answer 12

1. Popularity: (users want attention & recognition they receive from posting opinions that subsequently are
   confirmed by the stock market. Sometimes, SA articles
   are referred to and discussed in Forbes, WSJ, etc. - many users strive to become online celebrities)
2. Monetary Compensation: (each SA contributor earns $10 per 1,000-page views & potentially more depending on the article quality)
   - -is TESTED on this paper
3. Immediate Feedback: (users correct bad
   articles, which, on the assumption that the crowd is educated, increases the informativeness of social media)
   –is TESTED on this paper
4. Convergence to Fundamental Value: (if SA users have some price impact they can expect the direction of market prices to go to what authors perceive to be the
   fair fundamental value)

Question 13

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

The authors test whether Monetary Compensation and Immediate Feedback drives users to post more value-relevant information on the paper. What do they conclude?

Answer 13

1. Monetary Compensation Findings
   - -“The number of page views AND the number of times an article is read-to-end are increasing with the author’s historical level of consistency.“
   - -SA followers can “differentiate between authors that offer historically good VS bad advice;
   - -SA followers give more “popularity” to the good than the bad advice given.
2. Immediate Feedback:
   - -The authors test “to what degree SA commentaries are of a different tone than the underlying SA article”
   - -They find that “followers disagree with authors more when the authors’ articles have been inconsistent.”
   - -They find that if “the authors’ articles are inconsistent, it is the tone of COMMENTS that more reliably predicts subsequent stock market performance.”

Question 14

“Wisdom of Crowds: The Value of Stock Opinions Transmitted Through Social Media” (Chen, H., De, P., Hu, Y. J., Huang, B., (2014))

What are the key takeaways from the article?

Answer 14

• -Peer-based advice is Useful & Value-Relevant
• -Social media outlets enable direct & immediate interaction among users which, combined with intelligence of a “crowd,” are one of the primary reasons social media platforms are able to produce value-relevant content.

Question 15

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

Why should we have tougher criteria today for discovering a factor that supposedly explains expected returns?

Answer 15

Context:
There have been hundreds of papers trying to explain the cross-section (factors) of expected returns since one of the first CAPM tests forty years ago. Given this extensive data mining, the usual t-statistics cut-off is insufficient for establishing significance. Moreover, most of the sample factors have been proposed over the last ten years.

Why should we have tougher criteria?

–The rate of discovering a true factor has likely decreased.

–There is a limited amount of data

–The cost of data mining (examining databases to gain new information) has dramatically decreased (aka it is easy to get data)

Question 16

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

What are Type I & Type II Errors in Statistics?

Answer 16

Type I and Type II errors
• Type I error:
–a false positive;
–the error of rejecting a null hypothesis when it is actually true;
–plainly speaking, we are observing a difference when in truth there is none.

• Type II error:
- a false negative;
–the error of not rejecting a null hypothesis when the alternative hypothesis is true.
–plainly speaking, it occurs when we are failing to
observe a difference when in truth there is one.
.

Question 17

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

What is Multiple Testing in Statistics?

Answer 17

Multiple Testing:
–the potential increase in Type I error that occurs
when statistical tests are used repeatedly.

What does that mean?

Suppose we consider the efficacy of a drug in terms of the reduction of any one of a number of disease symptoms. As more (=multiple) symptoms are considered on the same sample, it becomes increasingly likely that the drug will appear to be an improvement (=more false positives) over existing drugs in terms of at least one symptom.

Question 18

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

What is a Bonferroni Correction?

Answer 18

Bonferroni Correction:

Depicts for how much the size of the allowable error (usually 0.05/0.10) should be reduced to reject null hypothesis (depending on the number of tests done).

A bit deeper for topscorers:

If the significance level for a given experiment is 0.05, the experiment-wise significance level will increase exponentially (significance decreases) as the number of tests increases.

Thus, in order to retain the same overall rate of false positives in a series of multiple tests, the standards for each test must be more stringent.

To obtain the usual alpha of 0.05 with ten tests, requiring an alpha of 0.005 (0.05/10) for each test can be demonstrated to result in an overall alpha which does not exceed 0.05.

Question 19

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

What do the researchers do in their experiment?

Answer 19

• They use Multiple Testing Framework for the 316 supposedly “discovered” factors to determine whether they are still applicable if put under a more “strict” cut-off.
• They use Bonferroni, Holm & BHY corrections to determine what the “cut-off” for rejecting null hypothesis should have been for the factors.
• They compute adjusted t-statistics from 1965 to 2032 using the three adjustments and plot the sample factors against the new significance levels.

Question 20

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

What are the results of their experiment?

Answer 20

“A much higher t-statistic threshold is needed when multiple testing is taken into account.”

“A newly discovered factor today should have a t-statistic that exceeds 3.0 (a p-value of 0.27%). (a usual t-statistic of 2.0 has a p-value of 5%)”

Of the 296 published significant factors, over 100 would be considered false discoveries under Holm, BHY & Bonferroni (Bonferroni claimed the highest - 158 factors as false discoveries!)

Question 21

“…and the Cross-Section of Expected Returns”
(Harvey, C. R., Liu, Y., Zhu, H. (2016))

Which of the factors survive all three adjustments and new “cut-offs”?

Answer 21

1. Book-to-Market ratio
2. Momentum factor
3. Durable Consumption Goods (goods that do not quickly wear out; goods that yield utility over time)
4. Short-Run Volatility
5. Market Beta

…are considered good factors when measuring expected returns.

Question 22

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

How do traditional and behavioral models in economics differ in practice, according to Thaler?

Answer 22

Behavioral approach offers the opportunity to develop better models of economic behavior by incorporating insights from other social science disciplines.

Traditional models claim:
–there is a set of factors that will have NO effect on economic behavior, i.e. supposedly irrelevant factors or SIFs.

Behavioral models claim:
–that SIFs matter in economic behavior, and in some situations, a SIF might be the single most important determinant of economic behavior.

Question 23

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What is a Homo Economicus and why is it considered problematic in the world of behavioral economics?

Answer 23

Homo Economicus is a person that:
– has well-defined preferences;
–has unbiased beliefs and expectations;
–makes optimal choices based on these beliefs and preferences (not choosing what is momentarily
tempting)
–has infinite cognitive abilities (are as smart as the smartest economist)
–their primary motivation is self-interest (there is no acting “for another person’s well-being”, etc.)

All the traditional economic models are based on the Homo Economicus - the rational decision-maker. Behavioralists argue that in the real world, people do not portray the traits of a Homo Economicus; therefore, the traditional economic models are faulty and should be adjusted to fit the average human.

Question 24

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What are the 3 most important concepts of Behavioral Economics?

Answer 24

1. Overconfidence (in economics - about investments; often combined with myopia - “only seeing the short term” and being confident that this is a great investment and not looking at the possible long term returns)
2. Loss Aversion (in economics - not wanting to sell a declining stock - trying to avoid a loss; and quickly selling an outperforming stock, thinking more about “what if it sinks?” more than “what if it goes up even more?”)
3. Self-control

Question 25

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What are explainawaytions?

Answer 25

Shortly - defensive reactions by economists.

Thaler claims that economists react to questions about the realism of basic economic models by either denying the existence of irrationality in DMP or by claiming that it just does not matter all that much.

Question 26

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Please, refute this argument/explainawaytion:
“People should judge economic theories based on not whether they describe behavior patterns but whether they predict behavior.”

Answer 26

Refute:

Economic theories do not predict behavior of a human being, because:

–Sometimes, even experts are unable to optimize their choices when the problems are difficult, let alone non-experts (which are the majority of the world) who are assumed to optimize their choices as well as experts in the economic models.

–e.g., the lifecycle hypothesis is intended to be a theory of how the typical citizen saves for retirement, not just those with MBAs.

Question 27

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Please, refute this argument/explainawaytion:
“The errors of humans are randomly distributed with the mean of zero, canceling out when individual observations are aggregated; non-experts just have more noise (higher variance)”.

Answer 27

Refute:

No, the errors of human decision-making do not cancel out, because:

• -humans make judgments that are systematically biased;
• -e.g. options with unlimited gain are more appealing than options with unlimited loss even if the gains might be suuuuper unrealistic

Question 28

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Please, refute this argument/explainawaytion:
“If you raise the stakes, people will take the questions more
seriously and choose in a manner more consistent with optimization. Also, if given a chance to learn, people will get it right.”

Answer 28

Refute:

• -As a rule, the higher the stakes, the less often we get to do something - therefore, an opportunity to learn may be rare.
• -Even where there are multiple opportunities to learn, people may not make the best of those situations (Thaler cites a couple of experiments who have proven that average people do not learn…mostly.).

Question 29

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Please, refute this argument/explainawaytion:
“When people interact in a market environment, any tendencies to misbehave will be defeated (Thaler calls it “the invisible handwave - “just let the market do its work”)

Answer 29

Refute:

• -it is easier and more profitable for businesses to cater to biases than to eradicate them
• -e.g. gym fees, pension plans

–there is no magic market potion that miraculously turns Humans into Econs; in fact, the opposite pattern is more likely to occur - that markets will exacerbate behavioral biases by catering to those preferences

Question 30

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Why do Traditional Economists believe that Financial Markets are Efficient?

Answer 30

Intellectual underpinning of the efficient market hypothesis (EMH) is that:
–even if most investors are fools, the activities of “smart money” arbitrageurs (the ones “who balance it out”) can assure that markets behave “as if” everyone were smart.

Question 31

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What are the 2 components of Efficient Markets that Thaler talks about?

Answer 31

1. “No free lunch” - you cannot beat the market on a risk-adjusted basis.
2. “Price is right” - stock prices include all information - past & present, public & private; therefore, stock prices are right.

Question 32

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

How can you refute the 2 components of Efficient Markets that Thaler talks about?

Answer 32

Two EMH components:

1. ‘No free lunch’ - can’t beat the market on a risk-adjusted basis.
   - -You CAN beat the market if you identify irrational behavior of investors (herding - selling/buying stocks because everyone does it; myopia - investing in a short-term trend even though the data says that you will lose in the long-term)
2. “Price is right”
   - -of course, it is difficult to test if the stock prices are right as their fundamental value is not known;
   - -but there are practical examples like the CUBA closed-end mutual fund which had reacted to some political actions of the country of Cuba and significantly changed in value even though the only fund’s connection to the country was the fund’s name and nothing else.

Question 33

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What does Thaler fear about if the economists keep acting that the Financial Markets are efficient?

Answer 33

1. If policymakers think that Efficient Market Hypothesis is true; and, therefore, that bubbles are impossible, then they may not make appropriate regulations to dampen them.
2. We might expect that financial markets are the most efficient of all markets. Yet if financial markets can be off sometimes, how much confidence should we have that prices in other markets are good measures of value? (E.g. labour markets and the presumptions that income differences reflect productivity)

Question 34

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

Traditional economists claim that people make decisions according to Expected Utility theory, Intertemporal Choice and by putting Self-Interest first. How do people make their decisions, according to Thaler?

Answer 34

People make their decisions according to:

1. Prospect Theory instead of Expected Utility theory.
   - -in Prospect Theory utility is derived from changes in wealth relative to some reference point, and this reference point goes on that losses weigh more heavily than gains
2. Myopia instead of Intertemporal Choice
   - -short-term vision - the discount rate between “now” and “later” is much higher than between “later” and “even later” - people show time-inconsistent behavior.
3. “Other-regarding preferences” rather than self-interest (sometimes):
   - -humans are not completely selfish, even to strangers. their first instinct is to cooperate as long as they expect others to do likewise.

Question 35

“Behavioural Economics: Past, Present, and
Future” (Thaler, R. H. (2016,))

What are the 2 models that Thaler proposes to use in economics?

Answer 35

1. Two-self, or “two-system” approach
   - -model which deals with the inconsistency between the patient long-run self and myopic short-run self
2. “Beta-delta” model
   - -delta is the standard exponential discount rate, BUT;
   - -beta measures short-term impatience