Not exactly cutting edge idea:

• “For most things are differently valued by those who have them and by those who wish to get them: what belongs to us, and what we give away, always seems very precious to us.” Aristotle

source: https://www.quora.com/What-is-the-best-LOTR-meme

In other words:

• Willingness to pay to acquire good is significantly lower than compensation demanded to give up good.
• It is as if merely possessing good makes it more valuable.
• Looks more like the behaviour of Homer-sapiens than Homo-economicus.

Source: https://produto.mercadolivre.com.br/

Relevance for environmental economics:

• Environmental quality is not bought and sold: market data cannot be used to estimate demand.
  • In environmental economics contingent valuation (CV) is most widely applicable method: ask people hypothetical questions about how much they value the environment.
  • These valuations are very sensitive to the framing of the question:
    • Positive frame: how much are you willing to pay for a slight improvement in environmental quality (small WTP).
    • Negative frame: how much compensation would you demand for a slight decrease in environmental quality (large WTA).

Theoretical importance of Endowment Effect:

• Welfare economics either not ok or not not ok.

Source: Schitts Creek

Not ok:

• “Any endowment effect that is substantial and ubiquitous could make [the tools of welfare economics] virtually useless” Hovenkamp (1991).

• Suppose that individual has to choose between bundle A and B.
  • Flips a coin, chooses A. Once endowed with A, strictly prefers A to B.
    • Thus indifference curves do not exist.

Not not ok:

• Standard preferences consistent with WTA>>WTP for quantity changes when \(\xi=\frac{\eta}{\sigma_0}>>0\), where \(\eta\) is income elasticity and \(\sigma_0\) is the aggregate Allen-Uzawa elasticity of substitution, Hanemann (1991).
  • If either \(\eta=0\) (no income effects) or \(\sigma_0=\infty\) (one or more other good is a perfect substitute for the good in question) then WTP=WTA, otherwise not.

Source: https://www.reddit.com/user/kdubzubz/

Practical importance of Endowment Effect:

source: The Gimp

Jumping off point:

• Four explanations of endowment effect of which I am aware:

source: http://j4h.net/base-jumping/

1) Reference-dependent preferences:

Kahneman, Knetsch, and Thaler (1990), Kőszegi and Rabin (2006), Medvec, Madey, and Gilovich (1995).

source: https://www.today.com/health/only-silver-why-second-place-feels-worse-bronze-2D12092210

Possible that third feels better than first?

A.K.A. loss aversion:

2) Subject Confusion:

• regarding mechanism used to elicit preferences: Cason and Plott (2014), Bull et al. (2019).
• In particular when confronted with a second price auction (or BDM) use a first price heuristic.

source: https://www.picgifs.com/reaction-gifs/reaction-gifs/confused/confused048.gif

3) Subject Indifference:

• Flat optimum critique of incentive compatible mechanisms used to elicit preferences: Harrison (1992)

source: https://giphy.com/gifs/i-dont-care-dumb-and-dumber-idc-cpUarZKcIVcT6

4) Common rather than private values:

• In a second price common value auction subjects should bid less than their signal anticipating the “Winner’s Curse.”
• Winning the auction means that all other bidders think the object is worth less than you do: bad news.
• Optimal to bid your value conditional this bad news.
• An implication is that as the number of bidders goes up bids should be less aggressive.

source: https://dismalscientist86.substack.com/p/chart-of-the-day-the-winners-curse

Approach taken:

• 20 rounds, one of which determines payoffs.
• In each round:

1. Subjects participate in a second price sealed bid auction for a chocolate bar: WTP?
2. Subjects are asked to estimate their probability of winning auction in part 1): reference point.
3. Not knowing if they purchased a chocolate bar in 1) subjects then participate in a second price sealed ask auction to sell a chocolate bar: WTA?

Very demanding test of endowment effect:

• low probability of endowment.
• subjects unaware if they are actually endowed when making their asks.
• very little opportunity to form an attachment to the chocolate bar.

source: https://giphy.com/gifs/chocolate-food-baby-ITyJtH0hoQ1nG

Reasonable to expect a mixture of subject types:

• Some subjects have private values (intent to eat).
• Some subjects have common values (no intent to eat).
• Tan and Xing (2011) shows that in second price auctions with a mixture of private and common value agents “private-value bidders truly bid their values; and each common-value bidder acts as she is pivotal, bidding the expected common value given her own signal and that the highest bid among all others, equals her bid.”

2*2 Experiment Design:

1. A group size manipulation:

• For half of the subjects they participate in second price sealed bid/ask auctions (SPSBA) with one other player. (random rematch each round)

• For the other half of the subjects they participate in SPSBA with 9 other players. (same set of players each round)

• If subjects have private values and standard preferences, weakly dominant strategy to bid and ask one’s value regardless of number of participants.

2nd price auction: optimal bid=value.

2nd price procurement auction: optimal ask=forgone value.

But…

• Some subjects likely have common values and thus bid less and ask more than their signal of value.
  • This discrepancy should be wider in larger groups but shrink over time.
    
• In a 10 person auction expected payoff function is flat over a large range of choices.
  • serves as a baseline where we know that there is no real reason for bids and asks to coincide with values.
• Increasing group size lowers the reference point of a loss-averse subject: lower bids and asks: Rosato and Tymula (2019).

Expected Payoff with private values:

• 2 person auctions:
  • On-margin subjects (their value within \(\frac12\) a standard deviation from the mean of the other bids or asks) have some incentive to bid and ask close to their value.
  • Off-margin subjects (their value outside of \(\frac12\) a standard deviation from the mean of the other bids or asks) should be essentially indifferent regarding all bids and asks weakly less than their value.
    
• 10 person auctions:
  • For all subjects any \(B \leq V \leq A\) is “optimal”: random choice mimics Endowment Effect?

Expected Payoffs with private values:

• 2 person auctions:
  • On-margin subjects (their value within \(\frac12\) a standard deviation from the mean of the other bids or asks) have some incentive to bid and ask close to their value.
  • Off-margin high value subjects (their value outside of \(\frac12\) a standard deviation from the mean of the other bids or asks) should be essentially indifferent regarding all bids and asks weakly greater than their value.
    
• 10 person auctions:
  • For all subjects any \(B \leq V \leq A\) is “optimal”: random choice mimics Endowment Effect?

2*2 Experiment Design:

2. A nudge manipulation:

• Half of the subjects are given hints:
• Before bidding: “Hint: in a second price auction bidding your value maximizes your expected payoff.”
• Before asking: “Hint: In a second price procurement auction asking your forgone value maximizes your expected payoff.”
• If endowment effect is artifact of subject confusion then nudges should reduce.

https://www.reactiongifs.us/wp-content/uploads/2015/08/oh_i_get_it_chris_farley.gif

The UI:

The UI:

The UI:

Hypotheses:

1. The subjects’ belief of becoming endowed should decrease with group size (first stage).

2. Ask-Bid should decrease with group size (reduced form, intention to treat). If endowment effect is due to subject confusion, ask-bid difference should decrease in the “hints” treatments.

3. Ask-Bid should be increasing in the treatment group average belief of endowment (IV).

But various reasons exclusion restriction might not hold:

4. If endowment effect is due to the flat optimum problem, ask-bid should either be invariant to or increase with group size.

5. If endowment effect is due to common values, ask-bid should increase with group size.

6. Reference dependent preferences implies that bids \(\downarrow\) and asks \(\downarrow\downarrow\) with group size. Rosato and Tymula (2019)

H1: Endowment belief should decrease with group size.

• Subjects were asked to estimate their probability of winning the first auction, i.e. becoming endowed.

H2: Ask-Bid by treatment:

• Ask-Bid should decrease with group size (reduced form, intention to treat).

• If endowment effect is due to subject confusion, ask-bid difference should decrease in the “hints” treatments.

H2: Same thing, different view.

H2: Same thing, different view.

H3: Ask-Bid should be increasing in the treatment-group average belief of endowment (IV)

Naive view of (endogenous) probability of endowment vs. ask-bid :

Possible failures of exclusion restriction?

4. If endowment effect is due to the flat optimum problem, ask-bid should either be invariant to or increase with group size.

5. If endowment effect is due to common values, ask-bid should increase with group size.

6. Reference dependent preferences implies that both the bids and asks should decrease with group size. Rosato and Tymula (2019)

7. The ability of subjects to collude might differ between treatment: This could cut either way because subjects in the small groups were randomly re-matched every round, whereas the large groups were fixed.

H6: Bids by treatment and round:

• Do subjects derive utility from “winning” an auction irrespective of the consequences (final consumption bundles)?
• Is bidding high and asking low just a way to alleviate boredom for off-margin subjects?

H6: Asks by treatment and round:

Subject comments…

• “I was in the 10 person group with hints treatment. In retrospet, I think I was not rational, depiste being given hints, one possible reason is that I did not quite understand the logic behind the hints at that time, another reason was that my mind at that moment was quite occupied somehow by the idea of how I could win -getting the chocolate bar which is by biding a high price, and selling the chocolate bar by which is by asking low. I did not put much thought about maximuzing my expected profit.”

• “In the experiment I was in the treatment where I bid with one other person and was given hints. I decided to place a high bid for the chocolate bars, usually around 7.5 with low asking prices, usually around 3.5. My goal wasn’t to buy chocolate bars, as I didn’t get much enjoyment from just buying the chocolate bars. My goal was to buy the chocolate bars for a lower price and sell them for a higher price to make a ‘profit’ as I found that more fun. After the first 2 rounds I saw the pattern of the person I was paired with, and I noticed that they were risk averse as their bids were very low, with a higher ask price. I focused on bidding to have fun and enjoy the game aspect of the experiment more than the bidding and winning chocolate bars aspect.”

H7: Difference in small group payoffs due to collusion?

Treatment Network Plots:

H7: Difference in small group payoffs due to collusion?

Next Semester:

• Use BDM rather than second price auctions to elicit valuations.
  • Common Values no longer a problem: Winning in a BDM is not bad news regarding the value of the good.
  • Eliminate possibility of collusion.
  • No number of participants effect like in Rosato and Tymula (2019).
  • But how to manipulate the probability of endowment?

Alter the BDM price distribution as a treatment.

• More likely exclusion restriction satisfied.

Expected profit from bidding:

Expected profit from asking:

• Random price \(p\sim U[0,10]\) for all treatments.

Take away message:

• Off-margin subjects do not choose randomly within the flat region of their expected payoff function.
• These subjects are willing to incur losses in order to move towards the margin.
• I did not make it explicit that I would not be taking money from students, but it was probably anticipated.
• A show up fee from which losses could be deducted would deal with this problem.