In economics and psychology, a random utility model,[1][2] also called stochastic utility model,[3] is a mathematical description of the preferences of a person, whose choices are not deterministic, but depend on a random state variable.

Background

A basic assumption in classic economics is that the choices of a rational person choices are guided by a preference relation, which can usually be described by a utility function. When faced with several alternatives, the person will choose the alternative with the highest utility. The utility function is not visible; however, by observing the choices made by the person, we can "reverse-engineer" his utility function. This is the goal of revealed preference theory.

In practice, however, people are not rational. Ample empirical evidence shows that, when faced with the same set of alternatives, people may make different choices.[4][5][6][7][8] To an outside observer, their choices may appear random.

One way to model this behavior is called stochastic rationality. It is assumed that each agent has an unobserved state, which can be considered a random variable. Given that state, the agent behaves rationally. In other words: each agent has, not a single preference-relation, but a distribution over preference-relations (or utility functions).

The representation problem

Block and Marschak[9] presented the following problem. Suppose we are given as input, a set of choice probabilities Pa,B, describing the probability that an agent chooses alternative a from the set B. We want to rationalize the agent's behavior by a probability distribution over preference relations. That is: we want to find a distribution such that, for all pairs a,B given in the input, Pa,B = Prob[a is weakly preferred to all alternatives in B]. What conditions on the set of probabilities Pa,B guarantee the existence of such a distribution?

Falmagne[10] solved this problem for the case in which the set of alternatives is finite: he proved that a probability distribution exists iff a set of polynomials derived from the choice-probabilities, denoted Block-Marschak polynomials, are nonnegative. His solution is constructive, and provides an algorithm for computing the distribution.

Barbera and Pattanaik[11] extend this result to settings in which the agent may choose sets of alternatives, rather than just singletons.

Uniqueness

Block and Marschak[9] proved that, when there are at most 3 alternatives, the random utility model is unique ("identified"); however, when there are 4 or more alternatives, the model may be non-unique.[11] For example,[12] we can compute the probability that the agent prefers w to x (w>x), and the probability that y>z, but may not be able to know the probability that both w>x and y>z. There are even distributions with disjoint supports, which induce the same set of choice probabilities.

Some conditions for uniqueness were given by Falmagne.[10] Turansick[13] presents two characterizations for the existence of a unique random utility representation.

Models

There are various random utility models, which differ in the assumptions on the probability distributions of the agent's utility, A popular random utility model was developed by Luce[14] and Plackett.[15] They assume that the random utility terms are generated according to Gumbel distributions with fixed shape parameter. In the Plackett-Luce model, the likelihood function has a simple analytical solution, so maximum likelihood estimation can be done in polynomial time.

The Plackett-Luce model was applied in econometrics,[16] for example, to analyze automobile prices in market equilibrium.[17] It was also applied in machine learning and information retrieval.[18] It was also applied in social choice, to analyze an opinion poll conducted during the Irish presidential election.[19] Efficient methods for expectation-maximization and Expectation propagation exist for the Plackett-Luce model.[20][21][22]

Azari, Parkes and Xia[23] extend the Plackett-Luce model: they consider random utility models in which the random utilities can be drawn from any distribution in the Exponential family. They prove conditions under which the log-likelihood function is concave, and the set of global maxima solutions is bounded for a family of random utility models where the shape of each distribution is fixed and the only latent variables are the means.

Application to social choice

Random utility models can be used not only for modeling the behavior of a single agent, but also for decision-making among a society of agents.[23] One approach to social choice, first formalized by Condorcet's jury theorem, is that there is a "ground truth" - a true ranking of the alternatives. Each agent in society receives a noisy signal of this true ranking. The best way to approach the ground truth is using maximum likelihood estimation: construct a social ranking which maximizes the likelihood of the set of individual rankings.

Condorcet's original model assumes that the probabilities of agents' mistakes in pairwise comparisons are independent and identically distributed: all mistakes have the same probability p. This model has several drawbacks:

  • It ignores the strength of agents' expressed preferences. An agent who prefers a "much more than" b and an agent who prefers a "a little more than b" are treated the same.
  • It allows for cyclic preferences. There is a positive probability that an agent will prefer a to b, b to c, and c to a.
  • The maximum likelihood estimator - which is the Kemeny–Young method - is hard to compute (it is -complete).[24]

Random utility models provides an alternative: there is a ground-truth vector of utilities; each agent draws a utility for each alternative, based on a probability distribution whose mean value is the ground-truth. This model captures the strength of preferences, and rules out cyclic preferences. Moreover, for some common probability distributions (particularly, the Plackett-Luce model), the maximum likelihood estimators can be computed efficiently.

Generalizations

Walker and Ben-Akiva[25] generalize the classic random utility model in several ways, aiming to improve the accuracy of forecasts:

  • Flexible Disturbances: allowing a richer covariance structure, estimating unobserved heterogeneity, and random parameters;
  • Latent Variables: explicitly representing the formation and effects of unseen constructs, such as perceptions and attitudes;
  • Latent Classes: capturing hidden segmentation in terms of taste parameters, choice sets, and decision protocols;
  • Combining Revealed Preferences and Stated Preferences: to combine advantages of these two data types.

Blavatzkyy[26] studies stochastic utility theory based on choices between lotteries. The input is a set of choice probabilities, which indicate the likelihood that the agent choose one lottery over the other. The desired output is a stochastic utility representation: a writing of the choice probabilities as a non-decreasing function of the difference in expected utilities of the lotteries. He proves that choice probabilities admit a stochastic utility representation iff they are complete, strongly transitive, continuous, independent of common consequences, and interchangeable.

References

  1. ↑ Manski, Charles F. (July 1977). "The structure of random utility models". Theory and Decision. 8 (3): 229–254. doi:10.1007/BF00133443. S2CID 120718598. ProQuest 1303217712.
  2. ↑ Cascetta, Ennio (2009). "Random Utility Theory". Transportation Systems Analysis. Springer Optimization and Its Applications. Vol. 29. pp. 89–167. doi:10.1007/978-0-387-75857-2_3. ISBN 978-0-387-75856-5.
  3. ↑ Manski, Charles F. (August 1975). "Maximum score estimation of the stochastic utility model of choice". Journal of Econometrics. 3 (3): 205–228. doi:10.1016/0304-4076(75)90032-9.
  4. ↑ Camerer, Colin F. (April 1989). "An experimental test of several generalized utility theories". Journal of Risk and Uncertainty. 2 (1): 61–104. doi:10.1007/BF00055711. S2CID 154335530.
  5. ↑ Starmer, Chris; Sugden, Robert (June 1989). "Probability and juxtaposition effects: An experimental investigation of the common ratio effect". Journal of Risk and Uncertainty. 2 (2): 159–178. doi:10.1007/BF00056135. S2CID 153567599.
  6. ↑ Hey, John D.; Orme, Chris (1994). "Investigating Generalizations of Expected Utility Theory Using Experimental Data". Econometrica. 62 (6): 1291–1326. doi:10.2307/2951750. JSTOR 2951750. S2CID 120069179.
  7. ↑ Wu, George (1994). "An empirical test of ordinal independence". Journal of Risk and Uncertainty. 9 (1): 39–60. doi:10.1007/BF01073402. S2CID 153558846.
  8. ↑ Ballinger, T. Parker; Wilcox, Nathaniel T. (July 1997). "Decisions, Error and Heterogeneity". The Economic Journal. 107 (443): 1090–1105. doi:10.1111/j.1468-0297.1997.tb00009.x. S2CID 153823510.
  9. 1 2 Block, H. D. (1974). "Random Orderings and Stochastic Theories of Responses (1960)". Economic Information, Decision, and Prediction. pp. 172–217. doi:10.1007/978-94-010-9276-0_8. ISBN 978-90-277-1195-3.
  10. 1 2 Falmagne, J.C. (August 1978). "A representation theorem for finite random scale systems". Journal of Mathematical Psychology. 18 (1): 52–72. doi:10.1016/0022-2496(78)90048-2.
  11. 1 2 Barberá, Salvador; Pattanaik, Prasanta K. (1986). "Falmagne and the Rationalizability of Stochastic Choices in Terms of Random Orderings". Econometrica. 54 (3): 707–715. doi:10.2307/1911317. JSTOR 1911317.
  12. ↑ https://scholar.harvard.edu/files/tomasz/files/lisbon32-post.pdf%5B%5D%5B%5D
  13. ↑ Turansick, Christopher (July 2022). "Identification in the random utility model". Journal of Economic Theory. 203: 105489. arXiv:2102.05570. doi:10.1016/j.jet.2022.105489. S2CID 231861383.
  14. ↑ Luce, R. Duncan (2012). Individual Choice Behavior: A Theoretical Analysis. Courier Corporation. ISBN 978-0-486-15339-1.
  15. ↑ Plackett, R. L. (1975). "The Analysis of Permutations". Applied Statistics. 24 (2): 193–202. doi:10.2307/2346567. JSTOR 2346567.
  16. ↑ McFadden, Daniel (1974). "Conditional Logit Analysis of Qualitative Choice Behavior". In Zarembka, Paul (ed.). Frontiers in Econometrics. Academic Press. pp. 105–142. ISBN 978-0-12-776150-3.
  17. ↑ Berry, Steven; Levinsohn, James; Pakes, Ariel (1995). "Automobile Prices in Market Equilibrium". Econometrica. 63 (4): 841–890. doi:10.2307/2171802. JSTOR 2171802.
  18. ↑ Liu, Tie-Yan (2007). "Learning to Rank for Information Retrieval". Foundations and Trends® in Information Retrieval. 3 (3): 225–331. doi:10.1561/1500000016.
  19. ↑ Gormley, Isobel Claire; Murphy, Thomas Brendan (June 2009). "A grade of membership model for rank data". Bayesian Analysis. 4 (2). doi:10.1214/09-BA410. hdl:10197/7121. S2CID 53559452.
  20. ↑ Caron, François; Doucet, Arnaud (January 2012). "Efficient Bayesian Inference for Generalized Bradley–Terry Models". Journal of Computational and Graphical Statistics. 21 (1): 174–196. arXiv:1011.1761. doi:10.1080/10618600.2012.638220. S2CID 42955305.
  21. ↑ Hunter, David R. (February 2004). "MM algorithms for generalized Bradley-Terry models". The Annals of Statistics. 32 (1). doi:10.1214/aos/1079120141.
  22. ↑ Guiver, John; Snelson, Edward (2009). "Bayesian inference for Plackett-Luce ranking models". Proceedings of the 26th Annual International Conference on Machine Learning. pp. 377–384. doi:10.1145/1553374.1553423. ISBN 978-1-60558-516-1. S2CID 16965626.
  23. 1 2 Azari, Hossein; Parks, David; Xia, Lirong (2012). "Random Utility Theory for Social Choice". Advances in Neural Information Processing Systems. Curran Associates, Inc. 25. arXiv:1211.2476.
  24. ↑ Hemaspaandra, Edith; Spakowski, Holger; Vogel, Jörg (December 2005). "The complexity of Kemeny elections". Theoretical Computer Science. 349 (3): 382–391. doi:10.1016/j.tcs.2005.08.031.
  25. ↑ Walker, Joan; Ben-Akiva, Moshe (July 2002). "Generalized random utility model". Mathematical Social Sciences. 43 (3): 303–343. doi:10.1016/S0165-4896(02)00023-9.
  26. ↑ Blavatskyy, Pavlo R. (December 2008). "Stochastic utility theorem" (PDF). Journal of Mathematical Economics. 44 (11): 1049–1056. doi:10.1016/j.jmateco.2007.12.005.
    This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.