Submission 150
Information bias, communication networks, and financial markets
PS5-G07-03
Presented by: Jordi Brandts
We present the results from laboratory experiments designed to study how asset markets are affected by the presence of two important phenomena that have recently gained prominence: (1) traders communicating through social networks, and (2) traders receiving different types of information from potentially biased sources.
Social media have become an important source of information for trading in financial markets. They have drastically changed the way in which individuals share relevant (and irrelevant) information with each other. This affects financial markets (witness for example the case of Game Stop or the recent demise of Silicon Valley Bank). It has been argued that most of the effects of social media in finance occur due to a social transmission bias, either because the signal transmitted is distorted - either in its sign or intensity- or there is a selection bias of the information that is transmitted. Furthermore, there is evidence that political polarization in beliefs affects financial markets. Our experiments are designed to shed some light on these issues in a controlled environment.
There are eight traders in a financial market and the timing of the interaction is the following. First, participants receive a private signal about the value of a financial asset that will be traded at the end of the time sequence. Then, there is a communication stage between traders. Finally, all eight traders can trade the financial asset in a market organized as a continuous double auction for a fixed number of minutes. We have six treatments. They are organized in a 2x3 design in which we vary the type of information traders receive and the way in which communication between traders takes place. The two types of information differ in whether all market participants receive unbiased information from the same source or whether subgroups of traders receive biased from different sources. The three levels of free-form communication are no communication, segregated information (two islands) and non-segregated communication (one island). The experiments were run in December 2024 at the LINNEX laboratory in Valencia. A preliminary analysis of our results shows that allowing for free form communication among traders increases the forecast accuracy of beliefs regarding the fundamental value of the asset. This is particularly strong when signals are unbiased. In treatments where all traders communicate, biased information reduces the forecast accuracy of beliefs regarding the value of the fundamental. Trading volume is lower in markets where all traders communicate with each other compared to markets with no communication or segmented communication. The result is stronger for unbiased information treatments. Overall, free form communication increases forecast accuracy of beliefs and price informativeness. Price informativeness strongly increases with communication in biased information treatments. In treatments where all traders communicate, biased information reduces forecast accuracy of beliefs and price informativeness.
Social media have become an important source of information for trading in financial markets. They have drastically changed the way in which individuals share relevant (and irrelevant) information with each other. This affects financial markets (witness for example the case of Game Stop or the recent demise of Silicon Valley Bank). It has been argued that most of the effects of social media in finance occur due to a social transmission bias, either because the signal transmitted is distorted - either in its sign or intensity- or there is a selection bias of the information that is transmitted. Furthermore, there is evidence that political polarization in beliefs affects financial markets. Our experiments are designed to shed some light on these issues in a controlled environment.
There are eight traders in a financial market and the timing of the interaction is the following. First, participants receive a private signal about the value of a financial asset that will be traded at the end of the time sequence. Then, there is a communication stage between traders. Finally, all eight traders can trade the financial asset in a market organized as a continuous double auction for a fixed number of minutes. We have six treatments. They are organized in a 2x3 design in which we vary the type of information traders receive and the way in which communication between traders takes place. The two types of information differ in whether all market participants receive unbiased information from the same source or whether subgroups of traders receive biased from different sources. The three levels of free-form communication are no communication, segregated information (two islands) and non-segregated communication (one island). The experiments were run in December 2024 at the LINNEX laboratory in Valencia. A preliminary analysis of our results shows that allowing for free form communication among traders increases the forecast accuracy of beliefs regarding the fundamental value of the asset. This is particularly strong when signals are unbiased. In treatments where all traders communicate, biased information reduces the forecast accuracy of beliefs regarding the value of the fundamental. Trading volume is lower in markets where all traders communicate with each other compared to markets with no communication or segmented communication. The result is stronger for unbiased information treatments. Overall, free form communication increases forecast accuracy of beliefs and price informativeness. Price informativeness strongly increases with communication in biased information treatments. In treatments where all traders communicate, biased information reduces forecast accuracy of beliefs and price informativeness.