Final Report Summary - LVFM (Learning and volatility in financial markets: theory, experiments and empirics)
This research is an investigation of how learning occurs in financial markets. The analysis has been conducted by developing economic models that have then been tested in the laboratory and/or by using data from the stock market. Some highlights of the research are the following.
We have introduced a completely novel methodology to identify and measure the extent to which traders herd (i.e. engage in imitative behavior) in the financial market. In the existing literature, the only measures of herding were a-theoretical, statistical measures of how much traders cluster their decisions. Of course, a serious drawback of these measures is that the fact that traders make the same decisions (for instance, to sell a security) may or may not be due to herding. In our work, instead, we start from a model of traders’ behaviour. The model shows that traders can rationally decide to herd (because it is the most profitable choice) after a sufficiently large number of traders have made the same decision. We estimate this model using data from the New York Stock Exchange. Given the model estimates, we are able to detect the periods during a day of trading in which traders are herding. Our results show that herding is present in the market and has a significant effect on stock prices. This methodology can also be used to study other problems such the impact of a financial transaction tax on market efficiency, something also done in the project. Moreover, it can be used by other researchers to investigate similar phenomena not in the financial market but in other economic and social environments in which imitation is potentially relevant. Outside the academic world, the work can also be used by practitioners in the financial market to price assets, and by regulators and policy makers to understand the market activity. These issues are important not only for the scientific community but for the society at a whole, given that a well functioning financial market is crucial for economic growth and prosperity.
In the experimental analysis of herding, we have studied the behavior of financial professionals in a controlled laboratory experiment. The results are in line with our theories of herd behavior in financial markets. Traders in the laboratory engaged in herding when it was rational to do so, but otherwise did not exhibit a tendency to imitate. We observed, instead, a higher tendency to go against the market (acting as “contrarians”).
Finally, at a theoretical level, we have studied how the propensity of people to imitate others is affected by bounded rationality. We have shown that when agents have only a limited understanding of the relation between the actions that other agents take and the payoff they receive, history matter more. In other words, if you consider a sequence of agents taking an action one after the other (for instance, whether to adopt a new technology or not), the actions taken by the first people in the sequence affect the decisions of others more than they do according to a model of rational decision making. Although eventually the population learns the correct action (if the technology is good, it is adopted), this may take much longer than a model of full rationality would predict. This work contributes to a vibrant debate in the scientific community about the relevance and implications of bounded rationality.
We have introduced a completely novel methodology to identify and measure the extent to which traders herd (i.e. engage in imitative behavior) in the financial market. In the existing literature, the only measures of herding were a-theoretical, statistical measures of how much traders cluster their decisions. Of course, a serious drawback of these measures is that the fact that traders make the same decisions (for instance, to sell a security) may or may not be due to herding. In our work, instead, we start from a model of traders’ behaviour. The model shows that traders can rationally decide to herd (because it is the most profitable choice) after a sufficiently large number of traders have made the same decision. We estimate this model using data from the New York Stock Exchange. Given the model estimates, we are able to detect the periods during a day of trading in which traders are herding. Our results show that herding is present in the market and has a significant effect on stock prices. This methodology can also be used to study other problems such the impact of a financial transaction tax on market efficiency, something also done in the project. Moreover, it can be used by other researchers to investigate similar phenomena not in the financial market but in other economic and social environments in which imitation is potentially relevant. Outside the academic world, the work can also be used by practitioners in the financial market to price assets, and by regulators and policy makers to understand the market activity. These issues are important not only for the scientific community but for the society at a whole, given that a well functioning financial market is crucial for economic growth and prosperity.
In the experimental analysis of herding, we have studied the behavior of financial professionals in a controlled laboratory experiment. The results are in line with our theories of herd behavior in financial markets. Traders in the laboratory engaged in herding when it was rational to do so, but otherwise did not exhibit a tendency to imitate. We observed, instead, a higher tendency to go against the market (acting as “contrarians”).
Finally, at a theoretical level, we have studied how the propensity of people to imitate others is affected by bounded rationality. We have shown that when agents have only a limited understanding of the relation between the actions that other agents take and the payoff they receive, history matter more. In other words, if you consider a sequence of agents taking an action one after the other (for instance, whether to adopt a new technology or not), the actions taken by the first people in the sequence affect the decisions of others more than they do according to a model of rational decision making. Although eventually the population learns the correct action (if the technology is good, it is adopted), this may take much longer than a model of full rationality would predict. This work contributes to a vibrant debate in the scientific community about the relevance and implications of bounded rationality.