Empty-talk equilibrium

Suppose that the seller's signaling strategy is ( $S_{{\theta }_H}=0$ , $S_{{\theta }_L}=0$ ). The buyer's best response is to offer $P_L$ , allowing both types to earn 50. To ensure that both types have no incentive to deviate, we must specify the buyer's response to $S=1$ . If this response is $P_H$ , ${\theta }_H$ earns 100, which exceeds his payoff from sending $S=0$ . Therefore, we must have $r_1\in [0,\frac{3}{4})$ to ensure that the buyer's best response to $S=1$ is $P_L$ . Then, both types have no reason to deviate, as both earn $50-\omega <50$ .

Thus, there exists a pooling equilibrium where $(S_{{\theta }_H}=0,S_{{\theta }_L}=0)$ . The buyer's best response is always $P_L$ , while the seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ when $S=1$ and $P=P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_L$ regardless of the values of S and P. Regarding the buyer's belief, we must set $r_0=p\left({\theta }_H\right)$ and $r_1\in [0,\frac{3}{4})$ .

Limited sincerity pooling equilibrium

Suppose that the seller's signaling strategy is $(S_{{\theta }_H}=1,S_{{\theta }_L}=1)$ . The buyer's best response depends on $p\left({\theta }_H\right)$ , as $r_1=p\left({\theta }_H\right)$ .

If $p\left({\theta }_H\right)\in [0,\frac{3}{4})$ , the buyer's best response is to offer $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can earn $50-\omega$ . However, both types can earn 50 if they deviate to $S=0$ . Therefore, this equilibrium does not exist if $p\left({\theta }_H\right)\in [0,\frac{3}{4})$ .

If $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , the buyer's best response is to offer $P_H$ so that ${\theta }_H$ and ${\theta }_L$ can earn 100 and $200-{\theta }_L-\omega$ , respectively. To ensure that both types have no incentive to deviate, we need to specify the buyer's response to $S=0$ . As the buyer's best response is always $P_L$ , allowing both types to earn only 50, there is no incentive to deviate.

Thus, if $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , there exists a pooling equilibrium where $(S_{{\theta }_H}=1,S_{{\theta }_L}=1)$ . The buyer's best response is $P_H$ , given $S=1$ , and $P_L$ otherwise. The seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ when $S=1$ and $P=P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_L$ regardless of S and P. Regarding the buyer's belief, we must set $r_1=p\left({\theta }_H\right)$ but place no restriction on $r_0$ .

We omit the discussion of other equilibria, as they do not exist.

Empty-talk equilibrium

Suppose that the seller's signaling strategy is $(S_{{\theta }_H}=0,S_{{\theta }_L}=0)$ . The buyer's best response is to offer $P_L$ , allowing both types to earn 50. To ensure that neither type has an incentive to deviate, we must specify the buyer's response to $N\ge S>0$ . If the buyer's response to any $N\ge S>0$ is $P_H$ , then ${\theta }_H$ earns 100, which exceeds his payoff from sending $S=0$ . Accordingly, we must set $r_{\alpha }\in [0,\frac{3}{4})$ for $\forall \alpha$ in $N\ge \alpha >0$ to ensure that the buyer's best response to $S=\alpha$ is $P_L$ . Then, neither type has a reason to deviate, as both types earn $50-\omega <50$ .

Thus, there exists a pooling equilibrium where $(S_{{\theta }_H}=0,S_{{\theta }_L}=0)$ . The buyer's best response is always $P_L$ , while the seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ when $N\ge S>0$ and $P=P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_L$ regardless of S and P. Regarding the buyer's belief, we must set $r_0=p\left({\theta }_H\right)$ and $r_{\alpha }\in [0,\frac{3}{4})$ for $\forall \alpha$ in $N\ge \alpha >0$ .

Limited sincerity pooling equilibria

Suppose that the seller's signaling strategy is $(S_{{\theta }_H}=\alpha ,S_{{\theta }_L}=\alpha )$ , where $N\ge \alpha >0$ . The buyer's best response depends on $p\left({\theta }_H\right)$ , as $r_{\alpha }=p\left({\theta }_H\right)$ .

If $p\left({\theta }_H\right)\in [0,\frac{3}{4})$ , the buyer's best response is to offer $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can earn $50-\omega$ . However, both types can earn 50 if they deviate to $S=0$ . Therefore, this type of equilibrium does not exist if $p\left({\theta }_H\right)\in [0,\frac{3}{4})$ .

If $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , the buyer's best response is to offer $P_H$ so that ${\theta }_H$ and ${\theta }_L$ can earn 100 and $200-\alpha ·{\theta }_L-\omega$ , respectively. To ensure that neither type has an incentive to deviate, we must specify the buyer's response to other messages. When $S=0$ , the buyer's best response is always $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can only earn 50. Neither type has an incentive to deviate. The buyer's best response to sincerity other than $S=\alpha$ and $S=0$ depends on his off-equilibrium belief. Here, we need not restrict the off-equilibrium belief for any $N\ge S>\alpha$ because neither type can earn more through deviation, regardless of the buyer's best response. However, we must restrict the off-equilibrium belief for $\alpha >S>0$ such that the buyer's best response is always $P_L$ , given these messages. Therefore, we must set $r_S\in [0,\frac{3}{4})$ for $\forall$ S in $\alpha >S>0$ .

Thus, if $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , there exist pooling equilibria where $(S_{{\theta }_H}=\alpha ,S_{{\theta }_L}=\alpha )$ , $N\ge \alpha >0$ . The buyer's best response is always $P_H$ , given $S=\alpha$ , or $P_L$ , given $\alpha >S\ge 0$ . Her price decision depends on $r_S$ when $N\ge S>\alpha$ . The seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ when $N\ge S>0$ and $P=P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_L$ regardless of S and P. Regarding the buyer's belief, we must set $r_{\alpha }=p\left({\theta }_H\right)$ and $r_S\in [0,\frac{3}{4})$ for $\forall$ S in $\alpha >S>0$ but place no restriction on other off-equilibrium beliefs.

We omit the discussion of other equilibria, as they do not exist.

Limited sincerity pooling equilibria

Suppose that the seller's signaling strategy is $(S_{{\theta }_H}=\alpha ,S_{{\theta }_L}=\alpha )$ , where $k-1\ge \alpha >0$ . The buyer's best response depends on $p\left({\theta }_H\right)$ , as $r_{\alpha }=p\left({\theta }_H\right)$ .

If $p\left({\theta }_H\right)\in [0,\frac{3}{4})$ , the buyer's best response is to offer $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can earn $50-\omega$ . However, the buyer's best response to $N\ge S\ge k$ is always $P_H$ , and both types can earn 100. Thus, there is incentive to deviate, and this type of equilibrium does not exist.

If $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , the buyer's best response is to offer $P_H$ so that ${\theta }_H$ and ${\theta }_L$ can earn 100 and $200-\alpha ·{\theta }_L-\omega$ , respectively. To ensure that neither type has an incentive to deviate, we must specify the buyer's response to other messages. When $S=0$ , the buyer's best response is always $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can only earn 50. Neither type has an incentive to deviate. Similarly, when $N\ge S\ge k$ , the buyer's best response is always $P_H$ so that both types can earn 100; again, there is no incentive to deviate. The buyer's best response to sincerity other than $S=\alpha$ , $S=0$ , and $N\ge S\ge k$ depends on his off-equilibrium belief. Here, we need not restrict the off-equilibrium belief for any $k-1\ge S>\alpha$ because neither type can earn more through deviation, regardless of the buyer's best response. However, we must restrict the off-equilibrium belief for $\alpha >S>0$ such that the buyer's best response is always $P_L$ , given these messages. Therefore, we must set $r_S\in [0,\frac{3}{4})$ for $\forall$ S in $\alpha >S>0$ .

Thus, if $p\left({\theta }_H\right)\in \left(\frac{3}{4},,,1\right)$ , there exist pooling equilibria where $(S_{{\theta }_H}=\alpha ,S_{{\theta }_L}=\alpha )$ , $k-1\ge \alpha >0$ . The buyer's best response is always $P_H$ , given $N\ge S\ge k$ or $S=\alpha$ , or $P_L$ , given $\alpha >S\ge 0$ . Her price decision depends on $r_S$ when $k-1\ge S>\alpha$ . The seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ given $N\ge S>0$ and $P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_H$ given $N\ge S\ge k$ and $P_H$ , and $Q_L$ otherwise. Regarding the buyer's belief, we must set $r_{\alpha }=p\left({\theta }_H\right)$ and $r_S\in [0,\frac{3}{4})$ for $\forall$ S in $\alpha >S>0$ but place no restriction on other off-equilibrium beliefs.

High sincerity pooling equilibria

Suppose that both ${\theta }_H$ and ${\theta }_L$ pool on a sufficiently sincere statement, where $N\ge S_{{\theta }_H}=S_{{\theta }_L}=\alpha \ge k$ . The buyer's best response to $\alpha$ is to offer $P_H$ , allowing both types to earn 100. To ensure that neither type has an incentive to deviate, we must specify the buyer's response to other messages. When $S=0$ , the buyer's best response is always $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can only earn 50. Neither type has an incentive to deviate. When $k-1\ge S>0$ , the buyer's best response must always be $P_L$ to ensure that type ${\theta }_L$ has no incentive to deviate. Thus, we must set $r_S\in [0,\frac{3}{4})$ for $\forall$ $k-1\ge S>0$ . When $S\in [k,\alpha )\cup (\alpha ,N]$ , the buyer's best response is always $P_H$ . Both types can earn 100 and have no incentive to deviate.

Thus, there exist pooling equilibria where $(S_{{\theta }_H}=\alpha$ , $S_{{\theta }_L}=\alpha )$ , $N\ge \alpha \ge k$ . The buyer's best response is $P_H$ , given $N\ge S\ge k$ , and $P_L$ otherwise. The seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ given $N\ge S>0$ and $P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_H$ given $N\ge S\ge k$ and $P_H$ , and $Q_L$ otherwise. Regarding the buyer's belief, we must set $r_{\alpha }=p\left({\theta }_H\right)$ and $r_S\in [0,\frac{3}{4})$ for $\forall$ S in $k-1\ge S>0$ but place no restriction on other off-equilibrium beliefs.

High sincerity separating equilibria

Suppose that ${\theta }_H$ and ${\theta }_L$ separate on sufficiently sincere statements, where $N\ge S_{{\theta }_H}=\alpha \ge k$ , $N\ge S_{{\theta }_L}=\beta \ge k$ and $\alpha \ne \beta$ . The buyer has an updated belief, namely $r_{\alpha }=1$ , $r_{\beta }=0$ , and her best responses corresponding to these beliefs are $P_H$ and $P_H$ , respectively. Accordingly, both types can earn 100. To ensure that neither type has an incentive to deviate, we must specify the buyer's responses to other messages. When $S=0$ , the buyer's best response is always $P_L$ so that both ${\theta }_H$ and ${\theta }_L$ can only earn 50. Neither type has an incentive to deviate. When $k-1\ge S>0$ , the buyer's best response must always be $P_L$ to ensure that seller type ${\theta }_L$ has no incentive to deviate. Therefore, we must set $r_S\in [0,\frac{3}{4})$ for $\forall$ $k-1\ge S>0$ . If seller type ${\theta }_L$ chooses $S=\alpha$ , the buyer's best response is $P_H$ so that type ${\theta }_L$ can earn 100 and has no incentive to deviate. Likewise, if seller type ${\theta }_H$ chooses $S=\beta$ , the buyer's best response is $P_H$ so that type ${\theta }_L$ can earn 100 and has no incentive to deviate. The buyer's best response to sincerity other than $S=\alpha$ , $S=\beta$ , and $k-1\ge S\ge 0$ remains to be determined. In this case, the buyer's best response is always $P_H$ , regardless of the off-equilibrium beliefs, allowing both types to earn 100 and removing their incentive to deviate.

Thus, there are separating equilibria where $(S_{{\theta }_H}=\alpha$ , $S_{{\theta }_L}=\beta )$ , with $N\ge \alpha \ge k$ , $N\ge \beta \ge k$ , and $\alpha \ne \beta$ . The buyer's best response is $P_H$ , given $N\ge S\ge k$ , and $P_L$ otherwise. The seller's optimal quality decision is as follows: type ${\theta }_H$ offers $Q_H$ , given $N\ge S>0$ and $P_H$ , and $Q_L$ otherwise, while type ${\theta }_L$ offers $Q_H$ , given $N\ge S\ge k$ and $P_H$ , and $Q_L$ otherwise. Regarding the buyer's belief, we must set $r_{\alpha }=1$ , $r_{\beta }=0$ and $r_S\in [0,\frac{3}{4})$ for $\forall$ $k-1\ge S>0$ but place no restriction on other off-equilibrium beliefs.

We omit the discussion of other equilibria, as they do not exist.

The experimental procedure

You will play a 10-round game with the same rules. At the beginning of each round, you will be randomly paired with another participant in this room. In each round, one member of the pair will be randomly selected as the seller, while the other will be selected as the buyer.

The buyer first decides which price to choose. If a high price is chosen, the buyer must pay 200 tokens to the seller; if a low price is chosen, the buyer must pay 50 tokens. The seller can observe the buyer's choice and then determine the quality of the product. For a high-quality product, the cost of production is 100 tokens, and the buyer will earn 300 tokens; for a low-quality product, the cost of production is 0, and the buyer will earn 100 tokens.

Rules of Profit Calculation:

Buyer's revenue = Value of the product - Buyer's bid

Seller's revenue = Buyer's bid - Cost of production

In other words, if the buyer chooses a high price and the seller chooses high quality, both will obtain 100 tokens. If the buyer chooses a high price and the seller chooses low quality, the buyer will lose 100 tokens and the seller will obtain 200 tokens. If the buyer chooses a low price and the seller chooses high quality, the buyer will obtain 250 tokens and the seller will lose 50 tokens. If the buyer chooses a low price and the seller chooses low quality, both will obtain 50 tokens.

When the decisions are settled, we will provide feedback on the results to each participant at the end of each round. This feedback will include your assigned role (buyer/seller), the price and quality decisions, and the revenue obtained by you and your paired participant. During the game, your private trading history will be available at the bottom left corner of the screen, which will include all of your feedback given in previous rounds.

Payoff calculation

We will randomly draw one round to calculate your payoff. For every 5 tokens you obtain, you will receive 1 RMB.

When the 10-round game ends, you will need to complete a follow-up questionnaire, which will collect your personal information. You will also earn some tokens when you complete the questionnaire. If you complete both parts successfully, you will receive a 15 RMB show-up fee and an extra 20 RMB for the 10-round game to cover your potential losses.

When the experiment is over, please wait patiently in your seat. The experimenter will distribute the earnings separately to each participant based on your identification number.

For all treatments with communication

Welcome to the Smith Experimental Economics Research Center at Shanghai Jiao Tong University. You are going to participate in an economics experiment on decision-making. You can be paid during the experiment. Your earnings will depend on the choices that you and other participants make during the experiment, as well as some random factors. At the end of the experiment, each participant will be paid privately.

During the experiment, it is important that you do not talk with the other participants, and you will need to follow our given instructions. If you have any questions, please raise your hand; our experimenter will come to you to answer your questions.

We promise that the information collected during the experiment will be kept strictly confidential, and the data will only be used for academic research.

The procedure of the experiment

You are going to play a 10-round game with the same rules. At the beginning of each round, you will be randomly paired with another participant in this room. In each round, one member of the pair will be randomly selected as the seller, while the other will be selected as the buyer. The paired participants will make a transaction.

Restricted threat treatment only

Before the transactional decisions are made in each round, the seller can send a message to the buyer. In this stage, the seller can choose between the following two options:

1. 1. “If you pay a low price, I’ll choose low quality.”

2. 2. “ ” (An empty message)

Restricted promise treatment only

Before the transactional decisions are made in each round, the seller can send a message to the buyer. In this stage, the seller can choose between the following two options:

1. 1. “If you pay a high price, I’ll choose high quality.”

2. 2. “ ” (An empty message)

Message-select treatment only

Before the transactional decisions are made in each round, the seller can send a message to the buyer. In each round, the seller can choose between 10 options and an empty message. In each round, the options are randomly drawn (without replacement) from a corpus collected during previous experiments. The contents of the previous experiments and the origin of the corpus will be reviewed in a subsequent section of these instructions.

Free-form treatment only

Before the transactional decisions are made in each round, the seller can send a message to the buyer through a text box. An empty message is allowed. Please do not include any personal information in the message.

For all treatments with communication

The buyer will receive the seller's message and decide which price to choose. If a high price is chosen, the buyer must pay 200 tokens to the seller; if a low price is chosen, the buyer must pay 50 tokens. The seller can observe the buyer's choice and then determine the quality of the product. For a high-quality product, the cost of production is 100 tokens and the buyer will earn 300 tokens; for a low-quality product, the cost of production is zero and the buyer will earn 100 tokens.

Rules of Profit Calculation:

$\text{Buyer's}\text{revenue}=\text{Value}\text{of}\text{the}\text{product}-{\text{Buyer}}^{\prime }\text{s}\text{bid}$

$\text{Seller's}\text{revenue}=\text{Buyer's}\text{bid}-\text{Cost}\text{of}\text{production}$

In other words, if the buyer chooses a high price and the seller chooses high quality, both will obtain 100 tokens. If the buyer chooses a high price and the seller chooses low quality, the buyer will lose 100 tokens and the seller will obtain 200 tokens. If the buyer chooses a low price and the seller chooses high quality, the buyer will obtain 250 tokens and the seller will lose 50 tokens. If the buyer chooses a low price and the seller chooses low quality, both will obtain 50 tokens.

When the decisions are settled, we will provide feedback on the results to each participant at the end of each round. This feedback will include your assigned role (buyer/seller),the price and quality decisions, and the revenue obtained by you and your paired participant. During the game, your private trading history will be available at the bottom left and bottom right corner of the screen, which will include all of your feedback given in previous rounds, as well as the messages sent/received.

Payoff calculation

We will randomly draw one round to calculate your payoff. For every 5 tokens you obtain, you will receive 1 RMB.

When the 10-round game ends, you will need to complete a follow-up questionnaire, which will collect your personal information. You will also earn some tokens when you complete the questionnaire. If you complete both parts successfully, you will receive a 15 RMB show-up fee and an extra 20 RMB for the 10-round game to cover your potential losses.

When the experiment is over, please wait patiently in your seat. The experimenter will distribute the earnings separately to each participant.

Message-select treatments only

The Origin of the Corpus

As mentioned above, the options provided to sellers during the message transmission stage were collected during a previous experiment. Actually, the designs of the current and previous experiments are almost identical; the only difference lies in the message transmission stage. In the previous experiment, the seller could write a message to the buyer through a text box (empty messages were allowed), and the only restriction on the message was that no personal information should be revealed.

We integrated all of the messages sent in the previous experiment and obtained a corpus of 355 sentences (after excluding five empty messages). In each round, the options available to the seller will be randomly drawn from this corpus by the computer. Some of the messages may be similar or even identical, as some participants follow a consistent strategy when sending messages in multiple rounds.

The procedure of the experiment

In this experiment, you will be required to code some messages. These messages were all generated during a previous experiment. Before classifying each message, please read these instructions carefully. The details and a description of the previous experiment are provided below. This information will help you better understand the role these messages played in the previous experiment.

Experimental procedure of the previous experiment

In the previous experiment, the participants were required to play a 10-round game with the same rules. At the beginning of each round, each participant was randomly paired with another. In each round, one member of each pair was randomly selected as the seller, while the other was selected as the buyer.

The paired participants were required to make a transaction. Before the transactional decisions were made in each round, the seller could send a message to the buyer through a chat box. An empty message was allowed. The seller was not allowed to identify himself/herself in the message.

The buyer received a message from the seller and then decided which price to choose. If a high price was chosen, the buyer needed to pay 200 tokens to the seller; if a low price was chosen, the buyer needed to pay 50 tokens. The seller could observe the buyer's choice and then determine the quality of the product. For a high-quality product, the cost of production was 100 tokens, and the buyer earned 300 tokens; for a low-quality product, the cost of production was 0, and the buyer earned 100 tokens.

Rules of Profit Calculation:

$\text{Buyer`s ,revenue}=\text{Value, of, the, product}-\text{Buyer`s, bid}$

$\text{Seller`s, revenue}=\text{Buyer`s ,bid}-\text{Cost, of, production}$

In other words, if the buyer chose a high price and the seller chose high quality, both would obtain 100 tokens. If the buyer chose a high price and the seller chose low quality, the buyer would lose 100 tokens and the seller would obtain 200 tokens. If the buyer chose a low price and the seller chose high quality, the buyer would obtain 250 tokens and the seller would lose 50 tokens. If the buyer chose a low price and the seller chose low quality, both would obtain 50 tokens.

When the decisions were settled, feedback on the results was provided to each participant at the end of each round. This feedback included the assigned role (buyer/seller), the price and quality decisions, and the revenue obtained by the seller and buyer in the group. Only one round was randomly drawn to calculate the participants’ payoffs. For every 5 tokens, the participant received 1 RMB.

Rules of this message classification experiment

In this experiment, the messages you will classify were generated by sellers during the above-mentioned experiment.

For each message, you will need to make a binary choice (Y/N) in the following four dimensions:

1. 1. Is this message a promise?

2. 2. Is this message a threat?

3. 3. Does this message include the concept of honesty?

4. 4. Does this message include humor?

Payoff calculation

When the experiment is over, we will randomly select three messages to calculate your final payoff. Your profit will depend on whether your classifications are consistent with the most popular evaluation.

For each dimension, we will follow the principle of a simple majority: if your choice is consistent with that of more than 50 $\%$ of the participants, you will receive a reward of 7 RMB for that dimension; otherwise, you will receive no reward. If there is a tie, the computer will randomly select one of the options (Y/N) to calculate the final payoff.

Besides the payoff for message categorization, you will receive a 5 RMB show-up fee if you successfully complete the experiment. When the classification ends, you will need to complete a follow-up questionnaire, which will collect your personal information.

After the experiment is over, please wait patiently in your seat. The experimenter will distribute the earnings separately to each participant.

The procedure of the experiment

In this experiment, you will be required to rate some messages. These messages were all generated during a previous experiment. Before rating each message, please read these instructions carefully. The details and a description of the previous experiment are provided below. This information will help you better understand the role these messages played in the previous experiment.

Experimental procedure of the previous experiment

In the previous experiment, the participants were required to play a 10-round game with the same rules. At the beginning of each round, each participant was randomly paired with another. In each round, one member of each pair was randomly selected as the seller, while the other was selected as the buyer.

The paired participants were required to make a transaction. Before the transactional decisions were made in each round, the seller could send a message to the buyer through a chat box. An empty message was allowed. The seller was not allowed to identify himself/herself in the message.

The buyer received a message from the seller and then decided which price to choose. If a high price was chosen, the buyer needed to pay 200 tokens to the seller; if a low price was chosen, the buyer needed to pay 50 tokens. The seller could observe the buyer's choice and then determine the quality of the product. For a high-quality product, the cost of production was 100 tokens, and the buyer earned 300 tokens; for a low-quality product, the cost of production was 0, and the buyer earned 100 tokens.

Rules of Profit Calculation:

$\text{Buyer`s ,revenue}=\text{Value, of, the, product}-\text{Buyer`s, bid}$

$\text{Seller}`\text{s}\text{revenue}=\text{Buyer`s ,bid}-\text{Cost, of, production}$

In other words, if the buyer chose a high price and the seller chose high quality, both would obtain 100 tokens. If the buyer chose a high price and the seller chose low quality, the buyer would lose 100 tokens and the seller would obtain 200 tokens. If the buyer chose a low price and the seller chose high quality, the buyer would obtain 250 tokens and the seller would lose 50 tokens. If the buyer chose a low price and the seller chose low quality, both would obtain 50 tokens.

When the decisions were settled, feedback on the results was provided to each participant at the end of each round. This feedback included the assigned role (buyer/seller), the price and quality decisions, and the revenue obtained by the seller and buyer in the group. Only one round was randomly drawn to calculate the participants’ payoffs. For every 5 tokens, the participant received 1 RMB.

Task in this experiment

In this experiment, the messages you will rate were generated by sellers during the above-mentioned experiment. For each message, you will need to evaluate its “sincerity” on a scale from 0 to 10.

Payoff calculation

When the experiment is over, we will randomly select six messages to calculate your final payoff. Your earning for each message (counted in RMB) is calculated as follows: