Inferring Investor Utility from Trading Decisions
Summary
The document asks whether an investor’s utility function can be inferred from observed trades when the asset’s return distribution is known and stationary and the investor’s wealth is available. It frames the task as selecting a suitable functional form and estimating its parameters from decisions, rather than assuming a particular preference specification from the outset.
The responses describe a two-stage approach: use observed choices to identify a utility model, then estimate its risk parameters. They point to prior research on model selection and parameter estimation, but give no equations, dataset, or empirical results, so the practical method is only sketched. When individual trade data are unavailable, the document mentions a distinct route: infer preferences from the pricing kernel under the assumption that prices reflect the marginal investor. This alternative depends on that market-level assumption and is not developed further.
Key ideas
- Observed investment choices can be used to investigate an investor’s utility function.
- The proposed process first identifies a functional form and then estimates its risk parameters.
- The document cites earlier studies for utility model selection and parameter estimation but does not explain their procedures.
- Without individual trade records, pricing-kernel inference is suggested under a marginal-investor assumption.
Tags
Full text
# Estimating investor's utility from the trades data # Estimating investor's utility from the trades data Is it possible to infer investor's utility function from the set of decisions she is making? Let's assume for simplicity that the market consists of a single traded asset whose return distribution is stationary and known to the agents. We are also given a set of trades made by a particular investor in this market. We also know the wealth of an investor. How do we estimate the investor's utility function from this data? ## Answer by vonjd (score 1) https://quant.stackexchange.com/a/8392 The problem is to find the best functional form of the utility function plus estimate its parameters. A good starting point is the following draft chapter from an upcoming book which gives a good intuition and many examples: Preferences by Andrew Ang ## Answer by T123 (score 1) https://quant.stackexchange.com/a/66362 Yes, this has been done by Hackethal, Meyer and Jakusch. If you have a single traded asset or a set of trades from traders, you could use those stated decisions to infer the form of the utility functions first and then find the risk parameters once you identified the utility function. There is a bunch of papers from some these guys who just did that. This paper is on a utility model selection approach: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2845866 This one on finding the correct risk parameters: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2845338 ..and this one is about how it's done: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2845871 However, if you don't have data on individual trades, you can use the assumption that prices are driven by the marginal investor and infer utility functions from the pricing kernel: Blackburn and Ukhov share some work on this: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=890592 Hope that helps. Thomas
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.