Bibliography
Download 7.23 Mb.
|
- Navigate this page:
- PT falsified
- Christiane, Veronika I
| §6 (p.F32) brings up a very strange exaggerated accusation of experimental economics: “a common failure by EE’s [experimental economists] to assign subjects randomly to different treatments.”
I agree with the criticism in §1 that experimental economists have not been well aware of issues of internal-external validity for a long time, and the present popularity of field studies is an unbalanced counterswing to catch up with what other social sciences have routinely known for longer times. %} Loewenstein, George F. (1999) “Experimental Economics from the Vantage-Point of Behavioural Economics,” Economic Journal 109, F25–F34. {% On visceral factors. %} Loewenstein, George F. (2000) “Emotions in Economic Theory and Economic Behavior,” American Economic Review, Papers and Proceedings 90, 420–432. {% The quantitatively-oriented tradeoff approach of classical decision theory can be applied to only a limited number of real-life decisions. This paper describes many reasons for why other factors can play a role in real-life decisions. The enthusiasm of the author appears from sentences such as “Decision research is currently in ferment, the most intellectually vibrant period that I have witnessed since joining the field in the mid-1980s” (p. 504, closing comment), and the importance of his own contributions to the field appears from the discussions of those. %} Loewenstein, George F. (2001) “The Creative Destruction of Decision Research,” Journal of Consumer Research 24, 499–505. {% Loss aversion: let people predict for which price they will sell; some minutes later they have to sell, and ask higher prices. %} Loewenstein, George F. & Daniel Adler (1995) “A Bias in the Prediction of Tastes,” Economic Journal 105, 929–937. {% The authors kind of implicitly equate libertarian paternalism with asymmetric paternalism, implicitly arguing that if you do no coerce people then you will not make deliberate rational people go wrong. %} Loewenstein, George F., Troyan Brennan & Kevin G. Volpp (2007) “Asymmetric Paternalism to Improve Health Behaviors,” Journal of the American Medical Association 298, 2415–2417. {% time preference; p. 203: refs on free-will/determinism intertemporal separability criticized %} Loewenstein, George F. & John Elster (1992) “Choice over Time.” Russell Sage Foundation, New York. {% %} Loewenstein, George F., Ted O’Donoghue, & Matthew Rabin (2003) “Projection Bias in Predicting Future Utility,” Quarterly Journal of Economics 118, 1209–1248. {% %} Loewenstein, George F. G. & Emily Haisley (2007) “The Economist as Therapist: Methodological Issues Raised by ‘Light’ Paternalism.” In Andrew Caplin & Andrew Schotter (2008, eds.), Perspectives on the Future of Economics: Positive and Normative Foundations, Vol. 1. in the Handbook of Economic Methodologies, 210–248, Oxford University Press, Oxford, England. {% Source dependence means a different thing here than in Tversky’s sense of sources of uncertainty being collections of events in decision under uncertainty. Here it means a preference for a good when self chosen than when given by someone else. They demonstrate this. It is a problem in Ellsberg-urn studies of ambiguity if subjects can choose the color to gamble on, the most common way to control for suspicion (suspicion under ambiguity). %} Loewenstein, George F. & Samuel Issacharoff (1994) “Source Dependence in the Valuation of Goods,” Journal of Behavioral Decision Making 7, 157–168. {% time preference; preferring streams of increasing income; P. 350: intertemporal additivity has never been viewed as normatively compelling Preferences over sequences; argue for violations of intertemporal separability; more extensive version is, apparently, Loewenstein & Prelec (1993) Psych. Rev. 100, 91–108. %} Loewenstein, George F. & Drazen Prelec (1991) “Negative Time Preference,” American Economic Review, Papers and Proceedings 81, 347–352. {% time preference; DC = stationarity (p. 575 top & p. 592 3rd para & p. 595 3rd para; also that they call the (constant) discounted utility model normative.); Eq. (21): Horst & I can do state-dept. EU also for time context. The distributional condition that the authors state on top of p. 579 is a special case of Savage’s (1954) P4. The authors refer to K&T (1979 p. 290) where it indeed also appears. %} Loewenstein, George F. & Drazen Prelec (1992) “Anomalies in Intertemporal Choice: Evidence and an Interpretation,” Quarterly Journal of Economics 107, 573–597. {% time preference %} Loewenstein, George F. & Drazen Prelec (1993) “Preferences for Sequences of Outcomes,” Psychological Review 100, 91–108. {% %} Loewenstein, George F., Daniel Read, & Roy F. Baumeister (2003, eds.) “Time and Decision: Economic and Psychological Perspectives on Intertemporal Choice.” Russell Sage Foundation, New York. {% time preference. dominance violation by pref. for increasing income: assume that two wage profiles give the same sum of money over time but the first, at each time point, gives a higher total up to that time point than the second (e.g., first decreases, second increases). Then in fact by no more than monotonicity (if money is the only relevant attribute), one should prefer the first profile. Discounting only adds to that. However, the majority of participants prefers the second profile. They explained the issue to the participants, also mentioning psychological arguments for why one might still want to prefer the second profile. A little more than half of the participants adhered to a preference over profile 1. This suggests that it is not irrationality, but people deliberately have their utility depend on other things than absolute level of money. intertemporal separability criticized: sequence effects P. 71: “An important question concerns whether violations of present value maximization (and therefore dominance) should be treated as errors in decision making or as rational manifestations of a preference function that includes arguments other than absolute levels of consumption. This question is analogous to the debate over the status of Savage’s independence axiom.” P. 82: “Whether the observed preference for increasing payments is treated as rational or as a mistake depends on whether we are willing to accept a more complex utility function than has generally been assumed.” %} Loewenstein, George F. & Nachum Sicherman (1991) “Do Workers Prefer Increasing Wage Profiles?,” Journal of Labor Economics 9, 67–84. {% time preference; cite data of very high discount rates, exceeding 25%. P. 184 seems to write: In this study, and some others described here, the questions asked were hypothetical. Of course, all things being equal it would be better to study actual choices. However, there are serious trade-offs between hypothetical and real money methods. Using hypothetical questions one can ask subjects to consider options that incorporate large amounts of money, both gains and losses, and delays of a year or more. In studies using real choices, the experimenter must reduce the size of the stakes and the length of the delay, and it is difficult to investigate actual losses. A lso, in a hypothetical question, one can ask the subject to assume that there is no risk associated with future payments, while in experiments using real stakes, subjects must assess the experimenter’s credibility. %} Loewenstein, George F. & Richard H. Thaler (1989) “Anomalies: Intertemporal Choice,” Journal of Economic Perspectives 3 no. 4, 181–193. {% This paper expresses many subjective opinions about paternalism, and I agree with all of them. §2 describes the history of the ordinal revolution, and the new developments of economics opening up more to nonrevealed preference inputs as propagated by Kahneman and others, which fully agrees with the history described in §§2-3 of Abdellaoui, Barrios, & Wakker (2007). paternalism/Humean-view-of-preference: conservation of influence: p. 1796: “The strong preference people show for the default option suggests that more than rational self interest is at work.” P. 1797 “the main problem with experienced utility is its failure to incorporate non-hedonic aspects of experience, such as meaning and capabilities (even if such capabilities are not used) that are important to people but have little impact on their subjective happiness.” P. 1797: “Given the limitations of measures of welfare based either on decision utility or on experience utility, is there any hope for coming up with a normatively compelling welfare criterion? In Section 6, we argue that no simple criterion based on either concept can surmount these problems. Instead, evaluations of welfare will inevitably have to be informed by a combination of both approaches, patched together in a fashion that will depend on the specific context.” P. 1804, §5.7: “In this section we have argued that a major—indeed perhaps fatal—problem with experience utility is its failure to incorporate dimensions of experience other than simple happiness that people justifiably care about. To some extent, we may be able to overcome these flaws by expanding and improving the measures we include as part of experience utility. It is theoretically possible to capture people's experience of meaning and purpose in their lives, independent from their moment to moment affect. But we expect that this will not address all the problems we have raised with experience utility. Instead, we believe that there are circumstances that matter to people independent of their influence on moment to moment experience. Despite other patent flaws, decision utility has the advantage of capturing these values in a way that experience utility does not—e.g., if an individual cares about meaning, he or she can incorporate that concern into their choices.” P. 1805, §6.1.2 is on debiasing, citing studies by Ubel et al. trying to debias the overweighting of small probabilities. Several sentences show the enthusiastic style of Loewenstein. P. 1798 l. 4-5: “an issue of growing importance in an age of increasing income inequality.” P. 1804 l. -9: “Such a policy ignores the problems raised by the phenomenon of hedonic adaptation,”: hedonic adaptation is not the problem, but one of many symptoms of the problem, being that people have no anchor for the scales offered to them so that interpersonal comparisons (and even between- over time, as with adaptation) are problematic, as often in between-subject studies. P. 1805 4th para, suggests that, in order to investigate if a decision of nonsafe sex was wise, the ultimate criterion should come from investigating neural processes in the brains of the people involved. “If we could investigate the brain waves of each partner”. %} Loewenstein, George F. & Peter A. Ubel (2008) “Hedonic Adaptation and the Role of Decision and Experience Utility in Public Policy,” Journal of Public Economics 92, 1795–1810. {% inverse-S: p. 276 argues for it, with the reason that people are not sufficiently sensitive towards probabilities. %} Loewenstein, George F., Elke U. Weber, Christopher K. Hsee, & Edward S. Welch (2001) “Risk as Feelings,” Psychological Bulletin 127, 267–286. {% %} Loewenstein, Yonatan, Drazen Prelec, & H. Sebastian Seung (2009) “Operant Matching as a Nash Equilibrium of an Intertemporal Game,” Neural Computation 21, 2755–2773. {% standard-sequence invariance: P and Q are jazz records. (A,P) designates receiving $A and P. An additive model is assumed. It is pointed out that (A,P) ~ (B,Q) and (A',P) ~ (B',Q) imply that the utility difference between A and B is the same as between A' and B'. It is studied with five subjects and four such indifferences. %} Loewenton, Edward & R. Duncan Luce (1966) “Measuring Equal Increments of Utility for Money without Measuring Utility Itself,” Psychonomic Science 6, 75–76. {% foundations of statistics; criticizes hypotheses testing on six points: (1) The usual “point” H0 is impossible beforehand; you don’t want to know it is untrue, you want to know by how much it is untrue; sufficiently large samples always become significant (I think that has been called “statistically significant but not psychologically/economically/medically meaningful”). (2) Often H0 is rejected but it is not clear what the alternative is (“A depends on B” can be just anything). (3) With H1 vague, also power is vague. Power is especially important if H0 is not rejected. (4) Artificial dichotomy reject/not-reject: if one study accepts H0 and the other not that does not mean a contradiction! (5) One routinely assumes linear relations (regression) or normal distributions. A nice example can be found in Fig.1 about high-learned people who forget as low-learned but with a delay of two days, but this relationship is not detected by common methods. (6) The Bayesian point that p-value does not consider the relevant conditioning. Then four remedies are given: (1) Plots are clearer than tables. (2) Confidence intervals help to describe power (can be depicted in plots around the estimations). (3), about normalization in meta-analyses, I will not discuss here. (4) “Contrasts” (I think, specify alternative hypotheses and see how well they fit data) Conclusion: “Hypothesis testing provides appearance of objectivity ... only illusion of insight.,” suggesting different preferable frequentist methods (confidence intervals, plots, etc.). Gives many references to discussions of hypothesis testing. %} Loftus, Geoffrey R. (1996) “Psychology Will Be a Much Better Science when We Change the Way We Analyze Data,” Psychological Science 7, 161–171. {% %} Loomes, Graham (1988) “Further Evidence of the Impact of Regret and Disappointment in Choice under Uncertainty,” Economica 55, 47–62. {% %} Loomes, Graham (1988) “When Actions Speak Louder than Prospects,” American Economic Review 78, 463–470. {% %} Loomes, Graham (1989) “Predicted Violations of the Invariance Principle in Choice under Uncertainty,” Annals of Operations Research 19, 103–113. {% Asks participants to choose x to optimize (p,x; q,20x; r,0) and also in (p',x; q',20x; r',0) with p'/p = q'/q. EU predicts same x. This is not found. EV predicts x = 20 or x = 0, but there were remarkably many deviations. PT falsified: regarding inverse-S: for RDU, his evidence cannot be reconciled with an inverse-S weighting function (p. 104) but it can neither be with a convex (p. 1-3). Uses 2p3 3p2 + 2p as inverse-S weighting function. %} Loomes, Graham (1991) “Evidence of a New Violation of the Independence Axiom,” Journal of Risk and Uncertainty 4, 92–109. {% %} Loomes, Graham (1993) “Disparaties between Health State Measures: An Explanation and Some Implications.” In Bill Gerrard (ed.) The Economics of Rationality, 149–178 (Ch. 9), Routledge, London. {% %} Loomes, Graham (1995) “The Myth of the HYE,” Journal of Health Economics 14, 1–7. {% A more elaborate paper is Dubourg, Jones-Lee, & Loomes (1997, Economica). %} Loomes, Graham (1997) “Valuing Health and Safety: Some Economic and Psychological Issues.” In Robert F. Nau, Erik Grnn, Mark J. Machina, & Olvar Bergland (eds.) Economic and Environmental Risk and Uncertainty, 3–32, Kluwer, Dordrecht. {% Christiane, Veronika & I Considers risky choices for monetary stakes, and risky choices where the stake is a probability of gaining a prize (there is only one fixed prize, and one neutral outcome). The two quantities give similar phenomena. In the case of probabilities of gaining a prize, RCLA trivially prescribes all choices through stoch. dom. So, the data violate RCLA with only two outcomes! The data suggest that participants simply do numerical heuristics. %} Loomes, Graham (1998) “Probabilities vs Money: A Test of Some Fundamental Assumptions about Rational Decision Making,” Economic Journal 108, 477–489. {% Argues that the violations of EU are not caused by what the nonEU theories describe but by fundamental issues such as participants not even having prefs but just using heuristics to produce answers. %} Loomes, Graham (1999) “Some Lessons from Past Experiments and Some Challenges for the Future,” Economic Journal 109, F35–F45. {% error theory for risky choice; Best core theory depends on error theory: seems to be %} Loomes, Graham (2005) “Modelling the Stochastic Component of Behaviour in Experiments: Some Issues for the Interpretation of Data,” Experimental Economics 8, 301–323. {% The author considers the probability triangle, with probability distributions over three fixed outcomes x3 x2 x1. Then prospects can be characterized as S = (p1, p2, p3) and R = (q1, q2, q3). Nontrivial choices will have p1 < q1 & p3 < q3 (then R is more risky than S). Under EU, S is preferred iff (q1p1)/(q3p3) (U(x3)U(x2))/(U(x2)U(x1)). The author proposes a generalization (P) (X) where P is a measure depending only on the probabilities and X one depending only on the outcomes. This model is called PRAM (perceived relative argument model), with P the perceived relative argument due to probabilities and X the one due to outcomes. It entails a kind of separability between probabilities and outcomes. The most salient aspect is that the model violates transitivity, somewhat like regret theory but now with a similar thing in the probability dimension. The author considers special cases of the functions, pointing out that they can accommodate known paradoxes and preference cycles, with some forms in Eqs. 11-13 adding only one or two parameters to EU. (P. 910: the version with one parameter is violated by common consequence.) Some limitations: I cannot imagine how this model could in any tractable way be extended beyond the probability triangle. Further, intransitive models are hard to extend beyond binary choices. It would be interesting to pin down more precisely what the implications of the model are; it has some separability of outcomes and probabilities, with may be the possibility to build in rank dependence. A detail: p. 903 Eq. 2 on RDU is not correct because q1 and p1 should be handled as worst ranks, with 1w(1p1) rather than w(p1) the weight of p1 for instance. This affects the following analysis in details but not in substantive manners. P. 906 footnote 8 erroneously thinks that in modern 1992 PT (called CPT by the author) there would be no more certainty effect. P. 913: “CPT (taken to be the flagship of non-EU models)” The experimental evidence in §5 does not test the basic model, but only qualitative add-on predictions (such as testing risk aversion when supposedly testing EU). The RIS is used to incentivize. %} Loomes, Graham (2010) “Modelling Choice and Valuation in Decision Experiments,” Psychological Review 117, 902–924. {% utility elicitation; relates SG to TTO; p. 305 top of 2nd column explicitly leaves it open if patient utilities or community utilities are to be used, in agreement with what I think, and deviating from the unfortunate viewpoints of Gold et al. (1996). intertemporal separability criticized: p. 303 (quality of life depends on past and future health) risk seeking for small-probability gains: p. 305, bottom of 2nd column, points out that even people who are generally risk averse can be risk seeking for treatments with low-probability-high-effects, such as for heart and liver transplants, and coronary and neonatal intensive care units. P. 306 middle of 2nd column: “Given that decisions have to be made, and cannot be postponed until researchers have perfected the decision tools, the use of QALYs at their present stage of development may be defended as being no worse than any alternative measures.” Then warns that we should not be too quick. P. 307 first column suggests that nonEU theories be used in utility measurement. %} Loomes, Graham & Linda McKenzie (1989) “The Use of QALYs in Health Care Decision Making,” Social Science and Medicine 28, 299–308. {% N = 234 volunteers all individually interviewed at their homes! Are asked some simple statistical questions (prob of picking diamond card for instance), some public-risk questions (new monarch next year), and some private risks (you lose wallet in next X days). First asking statistical questions lowers other probability estimates, and “insensitivity to temporal scope” (burglary in your house next X years too independent of X), mostly for personal risks, then for public risks, then for statistical) Findings and tests are thin given the experimental investment. %} Loomes, Graham & Judith Mehta (2007) “The Sensitivity of Subjective Probability to Time and Elicitation Method,” Journal of Risk and Uncertainty 34, 201–216. {% error theory for risky choice: test EU and prospect theory/RDU, with error-theories added. Their footnote 11 points out that they do not consider losses, so that RDU is the same as PT. Watch out: they do old-fashioned bottom-up RDU integration, with w around 0 relevant to worst outcomes and w around 1 relevant to best outcomes. P. 104, next-to-last para: “expected utility theory and with its most prominent rival, rank-dependent theory.” P. 115, beginning of §6: “In part, we made this choice in recognition of the prominence of RD [rank-dependent utility] in the literature: it is probably the most widely-used non-EU theory. But we were also influenced by the properties of the data.” P. 119, next-to-last para: “these results establish that RD [RDU] model has significantly greater explanatory power than the EU model.” They find (p. 123) that deviations from EU decrease as subjects get more experienced (more repeated choices). Conclusion will claim convergence to EU inverse-S & risk seeking for small-probability gains: they find and model overweighting of the best outcome (called “bottom-edge effect”) and, remarkably, not of the worst (see their p. 115 last para, and p. 116 between Eq. 11b and 12a); (EU+a*sup+b*inf). It implied that the Prelec one-parameter family performed worse than the simple overweighting of best outcome. equate risk aversion with concave utility under nonEU: Unfortunately, in their writing they often equate utility with risk attitude, which is not correct for rank-dependent utility. Endnote 12 points out that non-cumulative weighting theories (they say it for Viscusi’s prospective reference theory) cannot treat overweighting of good outcomes differently than of bad outcomes. They also test which probabilistic choice model works best. parametric fitting depends on families chosen: seem to point that out %} Loomes, Graham, Peter G. Moffat, & Robert Sugden (2002) “A Microeconometric Test of Alternative Stochastic Theories of Risky Choice,” Journal of Risk and Uncertainty 24, 103–130. {% They propose a model of consumer preference with loss aversion, explaining the discrepancy between WTP and WTA. They assume that consumers are uncertain about what their true preferences are (reminding me of Kreps' 92 work on it). For instance an owner of a mug, when exchanging it for a chocolate, may just be uncertain whether the exchange is a gain or loss. Then the usual loss aversion can come into play, with status quo bias and so on. P. 121 end of §1 describes it clearly. They do the Sugden extension of allowing the reference point to be random (what I like to call random reference theory). For multiattribute outcomes such as commodity bundles, it is well known that one can do loss aversion in two ways. One is attribute-wise, having within each attribute a reference level, and maybe gains in some attribute levels and losses in others, such as in Tverky & Kahneman (1991, QJE). The other is global, taking one indifference class of multiattribute outcomes as reference level, and all preferred outcomes as gains, and the dispeferred ones as losses. In the latter case, being a gain or a loss is a holistic property. The latter was the approach of, for instance, Wakker & Tversky (1993, JRU) in which outcomes can even be from connected topological spaces, which includes commodity bundles as special case and works globally. The authors call the former, attribute-wise, approach dimension-based, and the holistic approach they call taste uncertainty approach. The attribute-wise approach has only been considered in the literature in combination with additive separability across attributes, and the authors go at great length to emphasize the empirical failures of it. They also compare extensively with Köszegi-Rabin (2006), where reference point is endogenous and not exogenous as in this paper. Also there is a function in K-R applying only to m differences (m something like basic utility) so that absolute m levels then do not affect degree of loss aversion. In this paper, the degree of loss aversion can depend entirely on the wealth level and the authors emphasize this much. P. 118 end of 2nd para is interesting: one can measure the degree of loss aversion by finding sequences of exchanges, all much disliked, that end where they started, and finding a net compensation required to implement the cycle. I liked §4, which discusses how experience can reduce uncertainty and, hence, loss aversion, and discrepancy between WTP and WTA. But p. 131 is strange in claiming that attribute-wise models of loss aversion cannot accommodate reduction of loss aversion by learning. What they mean to say is that these models do not consider learning explicitly in their model. Of course everone using them will say that, if learning is incorporated, then it will reduce loss aversion. %} Loomes, Graham, Shepley Orr, & Robert Sugden (2009) “Taste Uncertainty and Status Quo Effects in Consumer Choice,” Journal of Risk and Uncertainty 39, 113–135. {% PT falsified: measure certainty equivalents of prospects, allowing for choice errors. Find violations of PT, and suggest that a similarity theory may fit better. The authors are negative on PT (which they call CPT): “If CPT is to justify its current status as the front runner among alternatives to EUT, it should be able to organise the data from our CREPROBS treatment; but it cannot do so,” (p. 209). The main purpose of the paper is to argue for the use of error theories. %} Loomes, Graham & Ganna Pogrebna (2014) “Testing for Independence while Allowing for Probabilistic Choice,” Journal of Risk and Uncertainty 49, 189–211. {% Participants chose x to optimize (p+,x; p,Tx,12p,K) with the other parameters fixed, > 0. K = T/2 (so that certainty results with x = T/2) or K = 0 was chosen. Under EU’s second-order risk aversion, x > T/2, under 1st order risk aversion x = T/2 can occur. The authors indeed found x = T/2 for several participants. Unfortunately, no statistical analysis is given so it is not clear if the data can result from merely noise. Because the common outcome K was displayed as such, participants may have ignored it. %} Loomes, Graham & Uzi Segal (1994) “Observing Different Orders of Risk Aversion,” Journal of Risk and Uncertainty 9, 239–256. {% Typical of their early experimental papers. They find more preference cycles in direction predicted by regret theory than the other way around. They argue that preference reversals may reflect genuine intransitivities, as predicted by regret theory. Later papers by (some of) these authors will argue that event-splitting effects rather than intransitivities may explain the early findings of regret theory. %} Loomes, Graham, Chris Starmer, & Robert Sugden (1989) “Preference Reversal: Information-Processing Effect of Rational Non-Transitive Choice?,” Economic Journal 99, Supplement, 140–151. {% Find intransitivities while ruling out choice-matching discrepancy and some other biases. %} Loomes, Graham, Chris Starmer, & Robert Sugden (1991) “Observing Violations of Transitivity by Experimental Methods,” Econometrica 59, Supplement, 425–439. {% Show that preference violate monotonicity in a way predicted by regret theory. %} Loomes, Graham, Chris Starmer, & Robert Sugden (1992) “Are Preferences Monotonic: Testing Some Implications of Regret Theory,” Economica 59, 17–33. {% Shaping hypothesis: because agents are uncertain about what their preferences are, they let them be influenced by market prices observed in previous rounds. So, the market shapes preferences. Then, if anomalies disappear in repeated markets, it may not be because of increased rationality but just by the shaping hypothesis. The issue is investigated experimentally. They find convergence of WTA to WTP, which in itself does not make clear if it is the shaping hypothesis or a convergence to true preference. Some other anomalies, less clearly visible to subjects, such as overbidding, however, remain, as does a large variance in preference (not suggesting convergence to true preference). Hence, the authors suggest that the shaping hypothesis is more plausible than a convergence to true preference. %} Loomes, Graham, Chris Starmer, & Robert Sugden (2003) “Do Anomalies Disappear in Repeated Markets?,” Economic Journal 113, C153–C166. {% In repeated markets WTP-WTA disparaties are reduced, but preference reversals are not. %} Loomes, Graham, Chris Starmer, & Robert Sugden (2010) “Preference Reversals and Disparities between Willingness to Pay and Willingness to Accept in Repeated Markets,” Journal of Ecomic Psychology 31, 374–387. {% risky utility u = strength of preference v (or other riskless cardinal utility, often called value): p. 807: assume experienced utility, called “choiceless.” Say it is Bernoullian Download 7.23 Mb. Share with your friends:
|