Econ 440 Lecture 4, Texas A&M University, 2024 PDF

Econ 440: Lecture 4 Prof. Ragan Petrie Texas A&M University August 29, 2024 Research Group preferences ▶ Submit research group preferences on Canvas by 8pm Sept 1 ▶ Groups announced next class More on Biases and Heuristics More on Biases and Heuristics  More on Biases and Heuri...

Econ 440: Lecture 4 Prof. Ragan Petrie Texas A&M University August 29, 2024 Research Group preferences ▶ Submit research group preferences on Canvas by 8pm Sept 1 ▶ Groups announced next class More on Biases and Heuristics More on Biases and Heuristics  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Hot Hand Fallacy and Gambler’s Fallacy  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Motivation ▶ A fair coin is flipped 10 times, and each time it has come up heads. Which of the following is correct? 1. The next flip is more likely to be heads than tails. That is, the coin flipper is “on a run”. 2. The next flip is less likely to be heads than tails. That is, the flipper is “due for tails”. 3. The next flip is equally likely to be heads or tails. Definition The hot hand fallacy is the belief that once an event has occurred several times in a row, it is more likely to occur again, even though the events are independent. Definition The gambler’s fallacy is the belief that once an event has occurred several times in a row, it is less likely to occur again, even though the events are independent.  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Evidence: Gilovich, Vallone, and Tversky (1985) ▶ Data: shots during home games of the Philadelphia 76ers (basketball team) during one season ▶ Look at accuracy of next shot after streaks of misses or makes ▶ What pattern would we expect if we believed in hot hand?  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Gilovich, Vallone, and Tversky results ▶ Results: Streak next shot make percentage 3 misses 56% 2 misses 53% 1 miss 54% 1 make 51% 2 makes 50% 3 makes 46% ▶ Does evidence support existence of hot hand? ▶ Any concerns about this research design? Source: Gilovich, Vallone, and Tversky (1985)  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Hot hand: Counter evidence ▶ Statistical argument of selected samples ▶ Percent of makes following a streak of makes not necessarily the same as following a streak of misses ▶ Miller and Sanjurjo (2018) example, coin flips ▶ Take flips after sequence of HHH ▶ Claim: tails more likely than heads ▶ Why? HHHT restricts the sample more than HHHH, so tails more likely ▶ Data from Gilovich et al (1985) paper shows roughly same percentage makes after all sequences. This is unlikely (statistically). Evidence of 11 percentage points more makes (hot hand). Source: Miller and Sanjurjo, 2018, “Surprised by the hot hand fallacy? A truth in the law of small numbers,” Econometrica and The Conversation  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Do Baseball players get the hot hand? ▶ Unlike in basketball, harder for defense to adjust to “hot” hitters ▶ Basketball teams can guard better opponents more closely ▶ As a result, they have shooting percentages that are not better than league average ▶ In baseball, however, only one player bats at a time ▶ Lots more data: authors look at over 2 million player batting appearances, so less a problem of statistical power ▶ Results: ▶ Recent high performance for a batter predicts high performance on next at-bat ▶ For example: being “hot” on home runs predicts 20% more likely to hit home run on next plate appearance ▶ True for a variety of statistics (batting average, home runs, on base percentage) ▶ Pitchers might believe hitter has “hot hand” and underperform Source: Green and Zweibel, 2018, ”The Hot Hand Fallacy: Cognitive Mistakes or Equilibrium Adjustments? Evidence from Baseball” Management Science  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Evidence: Terrell (1994) ▶ Data: betting on daily drawings for New Jersey’s parimutuel lottery ▶ Choose a number between 000 and 999 ▶ Winning number is drawn uniformly random ▶ If multiple people pick same number, lottery prize divided evenly ▶ What behavior would we see if bettors believe the gambler’s fallacy?  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Terrell Results ▶ Results Winning number repeated Average payout per person within last week $349.06 1 to 2 weeks ago $349.44 2 to 3 weeks ago $307.76 3 to 8 weeks ago $301.03 not within last 8 weeks $260.11 all winners $262.79 ▶ Is evidence consistent with people having gambler’s fallacy? Source: Terrell (1994)  More on Biases and Heuristics Hot Hand Fallacy and Gambler’s Fallacy Where do These Fallacies Come From? ▶ Both fallacies can be explained by one heuristic ▶ Which one is it? ▶ Consider how both fallacies view coin flips HHH ▶ Hot hand: > Observing HHH convinces decision-maker that is representative of the population average > Then HHHH seems more likely than HHHT > The fallacy comes from assuming that small samples are representative of the whole ▶ Gamblers: > Knows that the true underlying probability is equal weight on heads and tails > HHHT is more representative of equal probabilities than HHHH, so it must be more likely > The fallacy comes from assuming that more representative strings are more likely  More on Biases and Heuristics Narrow Framing and Mental Accounting Narrow Framing and Mental Accounting  More on Biases and Heuristics Narrow Framing and Mental Accounting Bounded Rationality ▶ Bounded rationality is the concept that people have cognitive or computational limits that prevent them from fully evaluating the consequences of their decisions ▶ For example, when you decide what to buy for lunch, you are probably not looking at your bank account and the stock market to calculate your future expected income ▶ More likely you are using a heuristic ▶ For lunch, for example, maybe you just choose the best item that is under $10 ▶ These heuristics are often helpful in simplifying a complex problem  More on Biases and Heuristics Narrow Framing and Mental Accounting Narrow Framing ▶ People engage in narrow framing when they consider only a small set of options for a decision problem rather than optimizing globally ▶ Back to the lunch example: ▶ Suppose there are two options on the menu: a chicken sandwich for price pc and a steak sandwich for price ps ▶ You have amount m in your wallet ▶ The narrow frame compares the “minimal” bundles: (chicken sandwich, m − pc ) vs (steak sandwich, m − ps ) ▶ In theory the bundles are much “larger” than that: (chicken sandwich, x2 , x3 , x4 ,... , x1000 ) vs (steak sandwich, y2 , y3 , y4 ,... , y1000 ) ▶ That is, you should consider how your choice of sandwich affects what you’ll get for dinner, whether you’ll watch a movie tonight, how much you’ll save for retirement, when you get a job, etc  More on Biases and Heuristics Narrow Framing and Mental Accounting Mental Accounting ▶ How do we determine the size of the frame? ▶ One possibility: people divide certain purchase decisions into different mental accounts or mental budgets ▶ E.g. a separate budget for lunches, a separate budget for dinners, a separate budget for movies, and so on ▶ Another possible type of accounting is temporal, e.g. daily or weekly budgets ▶ Since money is fungible, these budgets are totally artificial ▶ We call the act of assigning a consumption decision to a certain mental account booking ▶ E.g. when you buy the steak sandwich, you book it to your lunch budget  More on Biases and Heuristics Narrow Framing and Mental Accounting Example: Lost Tickets ▶ Consider the following vignettes: ▶ Problem A: Imagine that you have decided to see a play where admission is $10 per ticket. As you enter the theatre you discover that you have lost a $10 bill. Would you still pay $10 for a ticket to the play? ▶ Problem B: Image that you have decided to see a play and paid the admission price of $10 per ticket. As you enter the theatre you discover that you have lost the ticket. The seat was not marked and the ticket cannot be recovered. Would you pay $10 for another ticket? ▶ How many people say yes to buying a ticket? ▶ Problem A: 88% ▶ Problem B: 56% ▶ Moblab: Problem A: 72.2% ▶ Moblab: Problem B: 61.1% Source: Kahneman and Tversky, ”The framing of decisions and the psychology of choice,” Science (1981)  More on Biases and Heuristics Narrow Framing and Mental Accounting Mental Accounting is consistent with Tickets Vignettes ▶ Note that in either case you have to pay $10 to see the play, and your total wealth is the same ▶ In Problem A, the lost $10 does not get booked to the entertainment budget, for example ▶ In this case, still have room in that budget to buy the ticket ▶ In Problem B, the originally purchased ticket may have maxed out the entertainment budget ▶ In this case, no room in that budget to buy a second ticket  More on Biases and Heuristics Framing and Presentation Effects: Coherent Arbitrariness Framing and Presentation Effects: Coherent Arbitrariness  More on Biases and Heuristics Framing and Presentation Effects: Coherent Arbitrariness Framing and Presentation Effects: Coherent Arbitrariness ▶ Framing (in the context of narrow framing) is used to mean how individuals presented the information to themselves ▶ There is another meaning for the word framing: how information is presented to the subject by an outside party (e.g. an experimenter or an advertiser) ▶ Here, bounded rationality still plays a role ▶ A different heuristic is used: the decision-maker looks for clues or shortcuts in the information provided ▶ Can lead to bias when some of the information at hand is totally irrelevant  More on Biases and Heuristics Framing and Presentation Effects: Coherent Arbitrariness Poetry Workshop ▶ Ariely, Loewenstein, and Prelec (2006) ran experiment to elicit student’s willingness to pay to attend a poetry workshop ▶ Started by writing down the last digit of their social security number (call this digit n) ▶ If n is odd, asked “Would you attend the poetry reading if you were paid $n?” - “bad” ▶ If n is even, asked “Would you pay $n to attend the poetry reading?” - “good” ▶ Additionally, willingness to attend elicited for both groups in same way: price list from being paid $10 to attend to paying $10 to attend  More on Biases and Heuristics Framing and Presentation Effects: Coherent Arbitrariness Poetry Workshop: Results ▶ Both treatment groups require payment to attend on average ▶ But the odd group, which was asked initially if they would attend for payment, has a much more negative valuation ▶ Authors propose that these results are due to coherent arbitrariness ▶ Value of an experience is determined somewhat arbitrarily (e.g. by last digit of SSN) ▶ Once value is established, however, subsequent valuations are coherent with first Source: Ariely, Loewentstein, and Prelec, ”Tom Sawyer and the construction of value,” JEBO (2006)  More on Biases and Heuristics Framing and Presentation Effects: Coherent Arbitrariness Methodological Aside: Price Lists ▶ When trying to determine how much a participant values something, we often ask them a series of questions where we systematically vary the price: Would you pay $9 to attend the poetry reading? Yes No Would you pay $8 to attend the poetry reading? Yes No Would you pay $7 to attend the poetry reading? Yes No etc... ▶ This is called a price list ▶ Note that subjects should switch from No to Yes at most once on this list ▶ Price lists are a specific example of the strategy method ▶ Elicit decision (i.e. “strategy”) from subject for many possible outcomes ▶ Only one outcome will actually be implemented More on Biases and Heuristics The Decoy Effect The Decoy Effect  More on Biases and Heuristics The Decoy Effect Motivating Experiment ▶ In a lab experiment, 153 students were asked to make hypothetical choices between objects in several choice categories ▶ E.g. cars, TVs, restaurants ▶ Treatment variable: two or three options in choice set ▶ Two options: target and competitor, where neither clearly dominates the other ▶ E.g., 35-inch TV for $400 or 27-inch TV for $300 ▶ Three options: add a decoy option, which is dominated by target option ▶ E.g., add 29-inch TV for $450 as third option ▶ Results: Target Competitor Decoy Two options 51.5% 48.5% – Three options 65.3% 32.7% 2.0% Source: Huber, Payne, and Puto (1982)  More on Biases and Heuristics The Decoy Effect Moblab exercise on the Decoy effect ▶ Suppose you find out that you have won a free vacation to Paris over spring break in a sweepstakes. They offer you the list of options below. Which option do you choose? ▶ Two options: (a) 8 night stay at a three-star hotel or (b) 4 night stay at a five-star hotel ▶ Three options: (a) 8 night stay at a three-star hotel or (b) 4 night stay at a five-star hotel (c) 6 night stay at a two-star hotel ▶ Which option in “three options” is the target, competitor and decoy and why? Target Competitor Decoy Two options 42.8% 57.1% – Three options 64.3% 35.7% 0%  More on Biases and Heuristics The Decoy Effect What is Going On Here? ▶ Classically, adding a third option should not make the purchase frequency of other options go up ▶ Authors propose a decoy effect ▶ Participants have difficulty making comparison directly between target and competitor ▶ However, can clearly see that target is better than decoy ▶ Thus they presume that target is likely to be better deal overall Source: Huber, Payne, and Puto (1982)  More on Biases and Heuristics The Decoy Effect Decoy Effect