Bibliography
Download 7.23 Mb.
|
- Navigate this page:
- Risk averse for gains, risk seeking for losses
- Christiane, Veronika I
CE bias towards EV: a process analysis showed that 9 of 24 participants used an EV heuristic in CEs (certainty equivalents). %} Johnson, Eric J. & David A. Schkade (1989) “Bias in Utility Assessments: Further Evidence and Explanations,” Management Science 35, 406–424. {% %} Johnson, Eric J., Michael Schulte-Mecklenbeck, & Martijn C. Willemsen (2008) “Process Models Deserve Process Data: Comment on Brandstätter, Gigerenzer, and Hertwig (2006)” Psychological Review 115, 263–272. {% %} Johnson, Eric J., Michael Schulte-Mecklenbeck, & Martijn C. Willemsen (2008) “Postscript: Rejoinder to Brandstätter, Gigerenzer, and Hertwig (2008)” Psychological Review 115, 272–273. {% Risk averse for gains, risk seeking for losses: find the predictions of prospect theory for below-target banks confirmed for data from 142 banks. P. 86, “Theoretically, if the utility functions of bank managers do contain convex segments below target, models of the banking firm that assume universal risk aversion or risk neutrality are improperly specified. The results of this study suggest that the concepts of target outcome and distance below target should be incorporated into models that rely on risk preference assumptions. The target return is the point of inflection of the utility function and outcomes below target may induce significantly different levels of risk tolerance. Furthermore, the distance below target can affect the degree of change in risk tolerance. It is clear that models of the banking firm may be at best imprecise without considering the possibility of convex segments of the utility function below target.” PT, applications: different risk attitude for gains than for losses. %} Johnson, Hazel J. (1994) “Prospect Theory in the Commercial Banking Industry,” Journal of Financial and Strategic Decisions 7, 73–89. {% real incentives/hypothetical choice: for time preferences; N=6 participants, screened for a history of psychiatric disorder. Choices until an indifference point was reached. Choices between immediate reward and delayed reward. Immediate reward was adjusted. Delayed rewards were between $10 and $250. Every subject answered the same set of questions. Both hypothetical and real rewards were done for each of the four amounts. One of the choices in the session for each of the four amounts was paid in the real treatment. (Despite adaptive experiment, but subjects cannot notice.) Thus, subjects received four real payments. Random incentive system but 4 times so still income effect. Delays ranged from 1 day to 6 months. In the hypothetical treatment the delays of 1 year, 5 years and 25 years were added, along with the rewards $1000 and $2500. Session lasted for about 2.5 hrs with two 5 mins breaks in between. Mazur discounting, exponential discounting. Linear utility. Magnitude effect was found. Statistical analysis may be weak. They tested whether there was correlation between real and hypothetical treatment, but did not test whether this correlation is 1. %} Johnson, Matthew W., & Warren K. Bickel (2002) “Within-Subject Comparison of Real and Hypothetical Money Rewards in Delay Discounting,” Journal of the Experimental Analysis of Behavior 77, 129–146. {% %} Johnson, Norman L. & Samuel Kotz (1970) “Continuous Univariate Distributions” 2. Wiley, New York. {% %} Johnson, Richard M. (1974) “Trade-off Analysis of Consumer Values,” Journal of Marketing Research 11, 121–127. {% probability elicitation: applied to experimental economics; proper scoring rules: consider, more generally, incentive compatibility, with proper scoring rules as a special case. Assume risk neutrality throughout. P. 877, condition TR (truth revelation, referring to Myerson 1982 for it) means there is a one-to-one relation between types and answers. Incentive compatibility can be achieved, under some assumptions, if center’s info depends—perhaps solely through messages—stochastically, however slightly, on all relevant private info. Note that the payments scheme need not observe the types in the end. In this sense it may be related to Prelec (2004). %} Johnson, Scott, John W. Pratt, & Richard J. Zeckhauser (1990) “Efficiency despite Mutually Payoff-Relevant Private Information: The Finite Case,” Econometrica 58, 873–900. {% Review of subjective probability measurements in the medical literature, primarily based on direct judgments, but citing Winkler, Savage, and others. %} Johnson, Sindhu R., George A. Tomlinson, Gillian A. Hawker, John T. Granton, & Brian M. Feldman (2010) “Methods to Elicit Beliefs for Bayesian Priors: A Systematic Review,” Journal of Clinical Epidemiology 63, 355-369. {% risky utility u = transform of strength of preference v, latter doesn’t exist: paper seems to argue for ordinal approach. %} Johnson, William E. (1913) “The Pure Theory of Utility Curves,” Economic Journal 23, 483–513. {% adaptive utility elicitation; p. 220: health states with negative utility were given utility 0 .... %} Johnston, Katharine, Jackie Brown, Karen Gerard, Moira O’Hanlon, & Alison Morton (1998) “Valuing Temporary and Chronic Health States Associated with Breast Screening,” Social Science and Medicine 47, 213–222. {% foundations of statistics %} Johnstone, David J. (1988) “Hypothesis Tests and Confidence Intervals in the Single Case,” British Journal for the Philosophy of Science 39, 353–360. {% proper scoring rules: shows that in betting market proper scoring rules better classify analysts than their monetary consequences. %} Johnstone, David J. (2007) “Economic Darwinism: Who Has the Best Probabilities,” Theory and Decision 62, 47–96. {% proper scoring rules; People in proper scoring rules are better off, a.o. in view of concave utility, if they do it jointly as a group and share their profits afterwards. Can be related to hedging in CAPM. %} Johnstone, David J. (2007) “The Value of Probability Forecast from Portfolio Theory,” Theory and Decision 63, 153–203. {% Maximum likelihood probability estimate is equivalent to maximization of log utility. The paper examines how several kinds of risk aversion utility functions impact probability estimations, and optimal collections of info. %} Johnstone, David J. (2012) “Economic Interpretation of Probabilities Estimated by Maximum Likelihood or Score,” Management Science 57, 308–314. {% foundations of statistics %} Johnstone, David J. & Dennis V. Lindley (1995) “Bayesian Inference Given Data ‘Significant at ’: Tests of Point Hypothesis,” Theory and Decision 38, 51–60. {% DOI: http://dx.doi.org/10.1214/12-STS408 Discuss the history of Borch’s argument that mean-variance analyses will always lead to violations of stochastic dominance. It can be escaped by restricting the payoff domain, or by restricting the probability distributions considered (restricting to normal is popular for this purpose). %} Johnstone, David & Dennis Lindley (2013) “Mean–Variance and Expected Utility,” Statistical Science 28, 223–237. {% Christiane, Veronika & I: if German people had to judge on salaries or prices in their own home-country, then they treated € too much as if DM, so went by numerical effects not just by value. If people had to judge on foreign currencies or prices in € in a foreign country, they did not do this. %} Jonas, Eva, Tobias Greitemeyer, Dieter Frey, & Stefan Schulz-Hardt (2002) “Psychological Effects of the Euro—Experimental Research on the Perception of Salaries and Price Estimations,” European Journal of Social Psychology 32, 147–169. {% %} Jones, Martin & Robert Sugden (2001) “Positive Confirmation Bias in the Acquisition of Information,” Theory and Decision 50, 59–99. {% Paper presented in Oslo. %} Jones-Lee, Michael W. & Graham Loomes (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. {% Critically discuss the applications of behavioral economics by the UK government. %} Jones, Rhys, Jessica Pykett, & Mark Whitehead (2011) “Governing Temptation: Changing Behaviour in an Age of Libertarian Paternalism,” Progress in Human Geography 35, 483–501. {% On Behavioral insights team installed in the UK by Cameron. %} Jones, Rhys, Jessica Pykett, & Mark Whitehead (2013) “On the Rise of the Psychological State.” Edward Elgar Publishers, Cheltenham, UK. {% Consider infinite streams of outcomes, and consider preference orders that are anonymous (which is not so easy for infinite streams). Consider discounted utility with discount factor going to 1, thus generalizing the overtaking criterion and others. %} Jonsson, Adam & Mark Voorneveld (2018) “The Limit of Discounted Utilitarianism,” Theoretical Economics 13, 19–37. {% %} Joore, Manuela A., Danielle Brunenberg, Horst Zank, Hans van der Stel, Lucien Anteunis, Gijs Boas, & Hans J.M. Peters (2002) “Development of a Questionnaire to Measure Hearing-Related Health State Preferences Framed in an Overall Health Perspective,” International Journal of Technology Assessment in Health Care 18, 528–539. {% History of St. Petersburg paradox. %} Jorland, Gérard (1987) “The Saint Petersburg Paradox 1713–1937.” In Lorenz Krüger, Lorraine J. Daston & Michael Heidelberg (eds.) The Probabilistic Revolution: Vol. 1, Ideas in History, 157–190, MIT Press, Cambridge, MA. {% proper scoring rules; Show that in a mathematical sense scoring rules amount to the same as optimizing particular utility functions in decision situations and to measures of entropy. They take the family of utility with risk tolerance (reciprocal of Pratt-Arrow index of risk aversion) linear in money x. The slope is the power of power utility and is index of risk aversion. Exponential utility is when slope is 0. So level of absolute risk aversion does not count. Eq. 1: I guess that the capital delta, described as the gradient of V(r,r) w.r.t. r (also denoted as V(r) or as V by the authors), should be the linear function p --> V(r,p) V(r,r) (which is its own gradient). %} Jose, Victor Richmond R., Robert F. Nau, & Robert L. Winkler (2008) “Scoring Rules, Generalized Entropy, and Utility Maximization,” Operations Research 56, 1146–1157. {% Imagine we want an agent to reveal his quantile xL of a probability distribution over the reals. That is, for a random variable X, P(X xL) = . Then we ask him to state xL´ and, after observing X, we pay him xL´ (xL´X)1[XxL´]. Under EV, the optimal answer is xL´ = xL. A nice result! A dual to proper scoring rules that was much needed, and was only invented in 2009. Congratulations to the authors. %} Jose, Victor Richmond R. & Robert L. Winkler (2009) “Evaluating Quantile Assessments,” Operations Research 57, 1287–1297. {% %} Jose, Victor Richmond R., Robert F. Nau, & Robert L. Winkler (2009) “Sensitivity to Distance and Baseline Distributions in Forecast Evaluation,” Management Science 55, 582–590. {% %} Jouini, Elyès, Jean-Michel Marin, & Clotilde Napp (2010) “Discounting and Divergence of Opinion,” Journal of Economic Theory 145, 812–829. {% %} Jouini, Elyès & Clotilde Napp (2003) “Comonotonic Processes,” Insurance: Mathematics and Economics 32, 255–265. {% On representative agent. Take an otherwise standard Arrow-Debreu model but deviate from representative agent by considering heterogeneous beliefs, which introduce a kind of extra risk. The same equilibrium results with homogeneous agents with “consensus” probabilities, that may be more optimistic or more pessimistic depending on the degree of risk aversion in the utility function. Use Ito to analyze. %} Jouini, Elyès & Clotilde Napp (2007) “Consensus Consumer and Intertemporal Asset Pricing with Heterogeneous Beliefs,” Review of Economic Studies 74, 1149–1174. {% Investigate how changes in individual risk tolerance can affect the aggregate risk tolerance, which is not always monotonically. %} Jouini, Elyès, Clotilde Napp, & Diego Nocetti (2013) “Collective Risk Aversion,” Social Choice and Welfare 40, 411–437. {% %} Jouini, Elyès, Walter Schachermayer, & Nizar Touzi (2008) “Optimal Risk Sharing for Law Invariant Monetary Utility Functions,” Mathematical Finance 18, 269–292. {% %} Journal of Behavioral Decision Making 20, Issue 5, 2007: Special Issue: Decision Making and the Law. {% proper scoring rules: Seems to bring in epistemic criterion (closeness to true state of nature I guess) besides behavioral (“pragmatic”) criteria, and get impossibility results for sets of priors. %} Joyce, James M. (1998) “A Nonpragmatic Vindication of Probabilism,” Philosophy of Science 65, 575–603. {% utility = representational: seems to write: “decision theory must throw off the pragmatist / behaviourist straitjacket that has hindered its progress for the past seventy years” (p. 254). %} Joyce, James M. (1999) “The Foundations of Causal Decision Theory.” Cambridge University Press, Cambridge. {% For subjective probabilities, makes the well known distinction between balance and weight. Then there is a third dimension, specificity. It apparently means something like whether all pieces of info that led to the probability assessment supported that probability assessment, or if some pieces supported higher probability assessments and others supported lower ones. Probably similar to expert aggregation where a difference is made between imprecise and conflicting expert judgments. In the author’s approach if no probability measure is known then it must be a set of probability measures (as with people who always exclusively think in terms of sets of priors). Then specificity for some event is maximal if all probability measures in the set of priors assign the same probability to that event. I do not really see that this would be a new dimension apart from balance and weight. The paper assumes that if your credal state is not reflected by one probability measure, then it is by a set of probability measures. (I did not see it refer to higher-order beliefs with 2nd-order probabilities over those probability measures.) It does not look much into alternatives. P. 154 claims to show that it can only be this. The paper also takes Bayesianism not to assume completeness of preference and, hence, not one unique probability measure (§2 l. 2). The paper uses the term bias not in the sense of mistake, but in the sense of subjective info. P. 168: U4 is a case of an urn with colored balls with total absence of info on the composition, and the author really does not want the principle of insufficient reason then (“it is clearly wrong in the fourth,” middle of p. 168). Sentences such as that subjective probabilities accurately reflect total evidence are fine if reflect means the weak depend on, reckon with. They are off if reflect means that they completely capture everything relevant. %} Joyce, James M. (2005) “How Probabilities Reflect Evidence,” Philosophical Perspectives 19, 153–178. {% A.o., p. 653 reviews discussions of the game that convinced me of forward induction. §6, p. 658 etc discusses small worlds. They suggest that Savage’s model be “partition-dependent.” I do not see this but didn’t study it in detail. %} Joyce, James M. & Allan Gibbard (1998) “Causal Decision Theory.” In Salvador Barberà, Peter J. Hammond, & Christian Seidl (eds.) Handbook of Utility Theory, Vol. 1, Principles, 627–666, Kluwer Academic Publishers, Dordrecht. {% Application of ambiguity theory; Assume repeated decisions at time points 1, 2, …, where at each time point the smooth model of KMM holds, and a recursive model is used. They emphasize that they get a clear separation between risk attitude (vNM utility), ambiguity (the 2nd order probability distribution of the smooth model), ambiguity aversion (through the second-order utility function of the smooth model), and intertemporal preference. Many models in the literature are special cases of their general setup. They take a tractable version of their model and use it to analyze dynamic asset-price phenomena, where they can accommodate many phenomena. A problem may be that the model is very general. P. 560 top cites puzzles in asset markets/macroeconomics. P. 561 Footnote 3 cites ambiguity/robustness for finance. Pp. 563-564 hits the nail on the head when explaining that the smooth model of ambiguity is popular for being tractable, allowing to analyze ambiguity attitude as traditional risk attitude. (I add: using the familiar utility curvature.) %} Ju, Nengjiu & Jianjun Miao (2012) “Ambiguity, Learning, and Asset Returns,” Econometrica 80, 559–591. {% Christiane, Veronika & I %} Juliusson, Asgeir, Amelie Gamble, & Tommy Gärling (2006) “Learning Unit Prices in a New Currency,” International Journal of Consumer Studies 30, 1–7. {% Christiane, Veronika & I; examines factors influencing how quickly people learn to think in terms of a new unit of money (the Euro). %} Juliusson, Asgeir, Amelie Gamble, & Tommy Gärling (2006) “Learning the Value of a New Currency from Prices,” Journal of Experimental Psychology: Applied 11, 45–52. {% PT, applications, loss aversion; Presented in Chantilly, 1997; Consider data of 10 years of horse race betting in UK. Note that this concerns a population that is more risk seeking than average. So, for instance, the certainty effect typically should not be expected to occur; it indeed didn’t. They observe what the betting odds are for many races. This and the results of the races is the only data they use, and they do not use data about the stakes bet on various horses. They assume one representative agent, and assume that the betting odds are such that the agent is indifferent between all horses. This follows from market equilibrium: if one horse was better, betting on it would increase and, hence, its prices. From this assumption alone (their Eq. 1), they can derive both the probabilities of horses winning and the (risk-)preference functional of the bettors. It works as follows. First, for each preference functional given, the indifference between all horses gives n1 equations, enough to get the n probabilities (that add to 1). Then, for each preference functional, a proper scoring rule is calculated relative to the actual winning horses. Finally, the preference functional is chosen with the best scoring rule. Find that RDU does not improve on EU, but PT does. They cannot incorporate loss aversion (utility more steep for losses than for gains) because the data do not permit. The better performance of PT results from different probability weighting for gains than for losses. Weighted utility does not seem to fit the data well (p. 528). The data do not suggest inverse-S. PT estimations suggest convex (pessimistic) w for gains, concave for losses (also pessimistic, because of dual integration for losses that PT does). For losses they seem to find risk aversion, for gains a little risk seeking. This is contrary to the common empirical findings although their footnote 17 suggests that it is in agreement with common findings. This population of betters can obviously not be expected to agree with general findings. %} Jullien, Bruno & Bernard Salanié (2000) “Estimating Preferences under Risk: The Case of Racetrack Bettors,” Journal of Political Economy 108, 503–530. {% %} Jullien, Bruno, Bernard Salanié, & François Salanié (2007) “Screening Risk-Averse Agents under Moral Hazard: Single-Crossing and the Cara Case,” EconomicTheory 30, 151–169. {% game theory for nonexpected utility: The fixed-point reasoning leading to Nash equilibrium can be extended to ambiguity without expected utility. %} Jungbauer, Thomas & Klaus Ritzberger (2011) “Strategic Games beyond Expected Utility,” Economic Theory 48, 377–398. {% Estimate concavity of utility under EU from agricultural data, and find so much concavity that they say it can’t be. So, nonEU is desirable. They confirm Rabin's (2000) calibration idea. %} Just, David R. & Hikaru Hanawa Peterson (2003) “Diminishing Marginal Utility of Wealth and Calibration of Risk in Agriculture,” American Journal of Agricultural Economics 85, 1234–1241. {% Distinguish between standard risk aversion, which concerns final wealth, and marginal risk aversion, which concerns taking a prospect as reference point and evaluating changes from there. So, exactly Sugden's (2003) random reference theory. The authors' approach has also been studied under the heading of background risks, as in Barberis, Huang, & Thaler (2006, AER). 10 interviewers interviewed 290 households in India, asking about real decisions made first, then about hypothetical seeding decisions that were presented as objective probability distributions over outcomes. One sentence (p. 618 last one) says that payment was performance-based, but I did not find how and if it was really real-incentive. The authors consider probability weighting but it is not clear if for three-outcome prospects as considered in their experiment they do rank-dependent or separate outcome transformation. They do not seem to consider loss aversion, only different utility and probability weighting for losses, only mentioning once that they find no “discrete loss aversion” (p. 624 just above Conclusion) without specifying what it means. They measure risk aversion as preference for increasing variance. Risk averse for gains, risk seeking for losses: p. 620 3rd para reports risk seeking for the only loss prospect they consider (relative to the reference prospect). %} Just, David R. & Travis J. Lybbert (2009) “Risk Averters that Love Risk? Marginal Risk Aversion in Comparison to a Reference Gamble,” American Journal of Agricultural Economics 91, 612–626. {% %} Just, Richard E. (1974) “An Investigation of the Importance of Risk in Farmers Decisions,” American Journal of Agricultural Economics 56, 14–25. {% %} Just, Richard E. (1975) “Risk Response Models and Their Use in Agricultural Policy Evaluation,” American Journal of Agricultural Economics 57, 836–843. {% Proposition 1 seems to show that revealed preference data cannot identify utility and subjective probability, but I do not understand. I do not see what domain is assumed. Surely, with rich enough domains, revealed preference can uniquely identify utility and subjective probability. %} Just, Richard E. & David R. Just (2011) “Global Identification of Risk Preferences with Revealed Preference Data,” Journal of Econometrics 162, 6–17. {% questionnaire versus choice utility: negative conclusions on predicting actual behavior from verbal expressions of expectations. %} Juster, F. Thomas (1964) “Anticipations and Purchases: An Analysis of Consumer Behavior.” Princeton University Press, Princeton NJ. {% %} Kaas, Rob, Jan Dhaene, & Marc J. Goovaerts (2000) “Upper and Lower Bounds for Sums of Random Variables,” Insurance: Mathematics and Economics 27, 151–168. {% DOI: http://dx.doi.org/10.1152/jn.00177.2009 Seems that pattern of increasing/constant/decreasing impatience was not affected by adding front-end delays. %} Kable, Joseph W. & Paul W. Glimcher (2010) “An “As soon as Possible” Effect in Human Inter-Temporal Decision Making: Behavioral Evidence and Neural Mechanisms,” Journal of Neurophysiology 103, 2513–2531. {% For variability of quantity of food, animals are risk averse. But for variability of delay time they are risk seeking. %} Kacelnik, Alex & Melissa Bateson (1996) “Risk Theories—The Effects of Variance on Foraging Decisions,” American Zoologist 36, 402–434. {% inverse-S; real incentives/hypothetical choice, discussion of it on p. 1121; ask certainty equivalents; Seems that for Canadian students with one group they paid out exactly, and for another group they took 100 times higher payments in the experimental questions but in implementation of incentives divided them by 100. Download 7.23 Mb. Share with your friends:
|