Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ARP: Behavioural Finance (February 2024)

(1) “Literature Review”: Behavioural Finance – The Big Picture

You will need to read the literature to build up a fuller knowledge of the field.  As a starting point, refer back to the material you covered in the first year modules I taught, but I will expect you to have a much better understanding of the field than in Y1, so you will need to read around widely.

Your initial literature review should give a concise overview of BF research.  Don’t just build up a list of different behavioural effects that have been identified. Are there common themes? How is this research being used? Consider the different methods used in BF research. What has BF research achieved, and where is BF research heading?

In part 2 (below) you should also include sections considering in more depth the specific literatures relating to the effects that you are aiming to detect, and how your research contributes to these literatures.

You may use generative AI tools to help find sources in the literature. Add a paragraph in an Appendix explaining how you have used AI.  Be careful: (i) check your sources (AIs can ‘hallucinate’); (ii) you can use AI to help yourself build up a good understanding of the subject, but do not use it to construct a text without getting a good understanding - this would become apparent in the viva (below); (iii) never copy AI-generated text into your report. Instead you need to describe what you have found in your own words. More generally, be careful to avoid plagiarism and cite your sources carefully.

(2) Risk Taking and Behavioural Effects among Gamblers

You will each be given a dataset of the behaviour of a gambler playing a form of roulette. This shows the gambles taken in a long sequence of games: (i) the amount gambled, (ii) the outcome bet upon (red/black/green), and (iii) the resulting gain or loss. The gambler thus makes two decisions each round: how much to gamble and which colours to bet upon. A bet on red or black pays $2 for every $1 staked (a net gain of $1). A bet on green pays $14 (a net gain of $13).  Red and black each occur with probability 46.65%, green 6.7%. Note that the gambler may place two bets in each round (we can discuss how to handle this).

Construct an econometric model of how the gambler’s risk-taking varies in response to their experience in previous rounds. As a minimum, you need to construct a well-specified model of the amount bet each round as a function of previous betting levels and previous wins/losses.

Consider your modelling approaches very carefully, using the tools available in your econometrics software. It is very easy to produce bad econometric models! Explain the choices that you made to arrive at your preferred model (what potential problems did you identify and how were you able to deal with them?).  Discuss the strengths and weaknesses of your preferred model(s), and interpret what your results tell us about the behaviour of this gambler.

Now extend your analysis of the dataset to derive further interesting results and test further hypotheses. For example, you could construct a model of the level of upside risk taken each round (this will be different from the downside risk if the gambler bets on green). Compare this with your model of downside risk. Which appears to represent a more satisfactory model of gambler behaviour, and why?

You can now move onto other models:

· Does the passage of time since the last bet change behaviour? This would be an indication of the increased salience of more recent experience.

· Does your roulette gambler show evidence of the gambler’s fallacy? This would require a discrete choice model of the colours chosen.

· Now that you understand the dataset, you might think of other models you could construct which would test for other behavioural effects.

· Specifically (although this could be challenging!) you could consider the gambler making simultaneous decisions about (i) which colour(s) to bet on (in particular, high-risk (green) vs. low-risk (red or black); and (ii) how much to bet.  This would suggest modelling these two variables as part of a single system, such as a Vector AutoRegression (VAR).

After you have handed in your final version, we will have a short individual discussion of your work (a “viva”). We can discuss your literature review, and it will also give you the opportunity to clarify anything that I found unclear in your write-up.

Enjoy your projects!