Bibliography
Download 7.23 Mb.
|
| §2 has a nice general discussion of preference axioms, pointing out for instance that first-order axioms cannot axiomatize most decision models which is why we need technical axioms such as continuity. P. 163 top argues against the use of auxiliary objective probabilities as, for instance, in the Anscombe & Aumann (1963) model. P. 164 points out that the Archimedean axiom is not first-order.
P. 167 3rd para, that axioms be about the appropriate degree of crudeness, comes out of the blue and is apparently an attempt to sell his axioms yet to come. The axioms consider the case of n equally likely events with crisp probability 1/n for calibration, which are used to provide upper and lower bounds for the probabilities of the other events in the obvious way. As regards the axiomatization, this is not very interesting, but the resulting model and discussion are interesting. §4 compares to geometry and quantum mechanics. An argument that can be advanced against upper and lower probability models (as against multiple priors) is that not only about probabilities, but also about anything else such as length, we can have uncertainties, so if we should work with upper and lower probabilities should we then not just as well work with upper and lower lengths instead of deterministic lengths? Suppes argues that subjective probabilities are to be treated differently than physical length, and that subjective probabilities should rather be treated as physical scales in quantum mechanics, where often locations and so on are not deterministic but probabilistic. Points out that the source of uncertainty, that any measurement will distort the location, holds also for subjective probability, where each measurement will distort it. Unlike most social scientists, Suppes does not start writing silly and exaggerated comparisons with quantum mechanics but keeps control and credibility, writing on top of p. 172: “I do not mean to suggest that the exact theoretical ideas of quantum mechanics carry over in any way to the measurement of belief, but I think the general conceptual situation does.” I personally do not believe that the analogy holds, and that the measurement of beliefs through certainty equivalents and so on does not meet the fundamental impossibility of quantum mechanics to reach high degrees of precision, but this is a matter of taste. Suppes is in fact favoring the constructive view of preference here!!! Nice. P. 174, final para of paper, compares the indeterminacy of subjective probability to the impossibility to do perfect meteorological measurements (which cannot be done because of complexity, which is a different point than for the indeterminacy in quantum mechanics). He ends, poetically, with: “Our beliefs, it seems to me, are rather like the leaves on a tree. They tremble and move under even a minor current of information. Surely we shall never predict in detail all of their subtle and evanescent changes.” Argues that probabilities are not so much as determinate as physical quantities in Newtonian mechanics. %} Suppes, Patrick (1974) “The Measurement of Belief,” Journal of the Royal Statistical Society 36, 160–191. Reprinted in Patrick Suppes (1993) “Models and Methods in the Philosophy of Science: Selected Essays,” Ch. 14, pp. 181 ff., Kluwer Academic Publishers, Dordrecht. {% foundations of probability, foundations of statistics %} Suppes, Patrick (1976) “Testing Theories and the Foundations of Statistics.” In William L. Harper & Clifford.A. Hooker (eds.) “Foundations of Probability Theory, Statistical Inference, and Statistical Theories of Science;”Vol. II, 437–455, Reidel, Dordrecht. {% foundations of quantum mechanics %} Suppes, Patrick (1980, ed.) “Studies in the Foundations of Quantum Mechanics.” Philosophy of Science Association, East Lansing, Michigan. {% %} Suppes, Patrick (1980) “Limitations of the Axiomatic Method in Ancient Greek Mathematical Sciences.” In K. Jaako J. Hintikka, C. David Gruender, & Evandro Agazzi (eds.) Proceedings of the 1978 Pisa Conference on the History and Philosophy of Science, Vol. I, 197–213. Reidel, Dordrecht. {% foundations of probability %} Suppes, Patrick (1983) “The Meaning of Probabilistic Statements,” Erkenntnis 19, 397–403. {% foundations of probability; expresses views of Bayesianism; contains bibliography, of which my archive has copy. %} Suppes, Patrick (1984) “Probabilistic Metaphysics.” Wiley, New York. (1st edn. 1974, published by the Philosophical Society and the Dept. of Philosophy, University of Uppsala.) {% I have, read, and learned much from written text, which he presented 1989 in Nijmegen and I attended. %} Suppes, Patrick (1988) “Determinism, Computation and Free Will,”, Ernest Nagel Memorial Lecture. {% foundations of probability; foundations of quantum mechanics. Ch. 5 on general criteria for axiomatizations seems to be very interesting. %} Suppes, Patrick (1993) “Models and Methods in the Philosophy of Science: Selected Essays.” Kluwer Academic Publishers, Dordrecht. {% utility = representational? %} Suppes, Patrick (2005) “Where Do Bayesian Priors Come from?,” Stanford University. {% %} Suppes, Patrick & Colleen Crangle (1990) “Robots that Learn: A Test of Intelligence,” Revue Internationale de Philosophie 44, 5–23. {% Crr 11: survey; Ch 12: vector space, affine geometry, Ch 13: ordered line, betweenness, projective planes; Ch 14: proximity measurement; multi-dimensional representation; Ch 15: color and force measurement, Grassman structure (seems to concern convex cones rather than convex sets); Ch. 16: representations with thresholds; Ch. 17: Survey on probabilistic choice. %} Suppes, Patrick, R. Duncan Luce, David H. Krantz, & Amos Tversky (1989) “Foundations of Measurement, Vol. II. (Geometrical, Threshold, and Probabilistic Representations).” Academic Press, New York. {% strength-of-preference representation: representation uses absolute differences though. All attempts to make strength of preference observable from actual decisions that I know are a special case of the following: We consider two-attribute (x1,x2) and assume additive representation V1(x1) + V2(x2). Under minimal continuity assumptions, V1 and V2, and their sum, are interval scales, and their ordering of differences is meaningful. We can then for instance observe: (a1,G2) ~ (b1,g2) and (c1,G2) ~ (d1,g2) to conclude that the strength of preference of a1 over b1 is as big as that of c1 over d1, with V1 differences correspondingly. That is, improving [a1 into b1] offsets the same gauge [improving g2 into G2] as improving [c1 into d1]. The additive representation means that there is no interaction between first and second coordinate, and this is necessary for things to work. The authors consider on p. 260 the special case where the second coordinate x2 refers to money, g2 = 0, and G2 is a positive side payment. The authors next consider the special case of a housewife who chooses between combinations of applyances. Say, starting from (a1,a2), that (b1,a2) is a better improvement than (a1,b2). Can we conclude that [from a1 to b1] is a better improvement than [from a2 to b2]? One again needs absence of interaction between the 1st and 2nd coordinate goods to derive strength of preference. If interactions then the improvement [from (a1,a2) to (b1,a2)] can be different than the improvement [from a1 to b1] (which we interpret as the improvement [from (a1,0) to (b1,0)]). The improvement [from (a1,a2) to (a1,b2)] can be different than the improvement [from a2 to b2] (which we interpret as the improvement [from (0,a2) to (0,b2)]). We could try to give more status to the improvement [from (a1,a2) to (b1,a2)] by assuming that (0,a2) iso (0,0) is the initial endowment, and we could give more status to the improvement [from (a1,a2) to (b1,a2)] by assuming that (a1,0) iso (0,0) is the initial endowment, but the two cannot be combined into one consistent initial endowment. On p. 259 they consider the special case where (x1,x2) concerns a gamble yielding x1 under one event and x2 under its complement. Absence of interaction between the two coordinates holds under expected utility and is needed here. On difficulty to disentangle different parameters, they write: “The interaction between probability and utility makes it difficult to make unequivocal measurements of either one or the other. The recent Mosteller and Nogee experiments (1951) may be interpreted as measuring utility if objective probabilities are assumed or as measuring subjective probabilities if utility is assumed linear in money.” (p. 259) P. 259 2nd para points out that measurements of utility under risk are distorted by interaction with probability weighting (they use the term subjective probability to indicate probability weighting), using this as argument to use introspective-based strengths of preferences. questionnaire versus choice utility: p. 261 penultimate para of §1: “It is also our opinion that many areas of economic and modern statistical theory do not warrant a behavioristic analysis of utility. In these domains, there seems little reason to be ashamed of direct appeals to introspection. For example, in welfare economics there are sound arguments for adopting a subjective view which would justify the determination of utility differences by introspective methods.” %} Suppes, Patrick & Muriel Winet (1955) “An Axiomatization of Utility Based on the Notion of Utility Differences,” Management Science 1, 259–270. {% probability intervals; deal only with prospects that are sums of indicator functions of events, meaning they are simple prospects taking only nonnegative integer values. %} Suppes, Patrick & Mario Zanotti (1989) “Conditions on Upper and Lower Probabilities to Imply Probabilities,” Erkenntnis 31, 323–345. {% A formal exposition of measurement theory, fundamental versus derived measurement, meaningfulness, and other things. The presentation is abstract and the examples are not very interesting I found. The definition of scale types in §1.3 p. 11 is not very accurate. %} Suppes, Patrick & Joseph L. Zinnes (1963) “Basic Measurement Theory.” In R. Duncan Luce, Robert R. Bush, & Eugene Galanter (eds.) Handbook of Mathematical Psychology, Vol. I, 1–76, Wiley, New York. {% violation of objective probability = one source: they investigate this. They confirm that affect-rich outcomes give more pronounced insensitivity (inverse-S). On one point my interpretation is different than the authors’. I think that probability neglect is an extreme form of insensitivity, and not something different as the authors think, expressed in their title (“versus”), and what they have as a central theme throughout their paper. Figure 7.1.1, p. 205, of Wakker (2010) shows the point, with to the left perfect sensitivity, in the middle partial sensitivity, and to the right extreme insensitivity which means probability neglect. Thus, what the authors take as evidence against inverse-S, in my opinion is strong support. They also find higher elevation of probability weighting for affect-rich outcomes. It was not clear to me from the text and the formulas if higher elevation was coupled with more or with less risk aversion. Also, with only one nonzero outcome, elevation may be determined only up to one joint power for utility and probability weighting. This need not affect inverse-S but it does affect elevation. Adding assumptions about (the power of0 utility makes the power of probability weighting also indentifiable. %} Suter, Renata, Thorsten Pachur, & Ralph Hertwig (2016) “How Affect Shapes Risky Choice: Distorted Probability Weighting Versus Probability Neglect,”Journal of Behavioral Decision Making 29, 437–449. {% Show that category rating scales have been subject to the same internal inconsistencies as the standard gambl in lotteries with one nonzero outcome %} Sutherland, Heather J., Virginia Dunn, & Norman F. Boyd (1983) “Measurement of Values for States of Health with Linear Analogue Scales,” Medical Decision Making 3, 477–487. {% Seems to have introduced MET (maximum endurable time) %} Sutherland, Heather J., Hillary A. Llewelynn-Thomas, Norman F. Boyd, James E. Till (1982) “Attitudes toward Quality of Life: The concept of “Maximal Endurable Time”,” Medical Decision Making 2 299–309. {% small probabilities: this paper explains, and references (p. 183 last para), that people can as well overweight unlikely events as fully ignore them. The latter is referred to as the low probability, high consequence events bias (the paper, unfortunately, never defines the latter, but p. 186 following Eq. 6 states it casually). They investigate how house prices react to tornado risk. A 1/million extra annual chance of dying increases the house price by 3%. %} Sutter, Daniel & Marc Poitras (2010) “Do People Respond to Low Probability Risks? Evidence from Tornado Risk and Manufactured Homes,” Journal of Risk and Uncertainty 40, 181–196. {% Shows that in Gneezy & Potters (1997 QJE) the myopic loss aversion is reduced if people work in teams. %} Sutter, Matthias (2007) “Are Teams Prone to Myopic Loss Aversion? An Experimental Study on Individual versus Team Investment Behavior,” Economics Letters 97, 128–132. {% N = 661 children aged 10-18. real incentives/hypothetical choice: for time preferences: use real incentives for time preferences, as for all preferences. In a school they pay on a pre-arranged future time. They used choice lists to observe indifferences. They estimate risk aversion from one observed CE (certainty equivalent) of a fifty-fifty prospect, referring to the known Ellsberg urn. For ambiguity aversion, they observed the CE for the unknown two-color Ellsberg urn, and took the normalized difference between the risky and ambiguous CEs (certainty equivalents) as index of ambiguity aversion. gender differences in risk attitudes: women more risk averse than men. (P. 517.) Gender is only demographic that correlates with risk aversion. For example, age does not. No demographic variable correlates with ambiguity aversion (gender differences in ambiguity attitudes: ). Time preference: they fix a near and remote time point, fix the payment at the near time point, and determine the remote payment to generate indifference. Did so 4 times, where two have early time right now and two have early time later (upfront delay). Find mostly constant impatience, but once decreasing impatience. correlation risk & ambiguity attitude: there is a negative relation, but it is not written in the paper. Is pointed out in survey chapter by Trautmann & van de Kuilen (2015). P. 510 cites seven studies that relate risk/time preferences to actual behavior. This paper does it for 661 children age 10-18. More impatient children smoke more, drink more, have higher BMI (body-mass index), save less, violate more school codes, and have lower maths grades. Risk and ambiguity aversion do not correlate with much. Risk averse subjects have lower BMI, ambiguity averse smoke less. P. 511 cites literature that children are more risk averse and more impatient than adults. More risk aversion then more patience. P. 527 mentions that intertemporal attitude correlates better with other things than risk/ambiguity attitudes, in agreement with what has been found more often I think. A little bit this may also be because they used four questions to measure intertemporal attitudes, and only one to measure risk attitude and only one to measure ambiguity attitude. %} Sutter, Matthias, Martin G. Kocher, Daniela Glätzle-Rüetzler, & Stefan T. Trautmann (2013) “Impatience and Uncertainty: Experimental Decisions Predict Adolescents’ Field Behavior,” American Economic Review 103, 510–531. {% %} Sutter, Matthias, Martin G. Kocher, Sabine Strauss (2003) “Bargaining under Time Pressure in an Experimental Ultimatum Game,” Economic Letters 81, 341–347. {% revealed preference %} Suzumura, Kotaro & Yongsheng Xu (2003) “Recoverability of Choice Functions and Binary Relations: Some Duality Results,” Social Choice and Welfare 21, 21–37. {% Ch 4 and p. 41 seem to be on probability. %} Svennilson, Ingvar (1938) “Ekonomisk Planering.” Almqvist & Wiksell, Uppsala. {% Kirsten&I: shows that for the countably-infinite consumption streams of Koopmans (1960) symmetry (such as in zero discounting) is possible in combination with continuity if the topology w.r.t. which continuity should hold is taken sufficiently coarse. %} Svensson, Lars-Gunnar (1980) “Equity among Generations,” Econometrica 48, 1251–1256. {% Seems to be: decision under stress, with models of rational decision. %} Svenson, Ola & A. John Maule (1993, eds.) “Time Pressure and Stress in Human Judgment and Decision Making.” Plenum, New York. {% “An optimist is just a misinformed pessimist.” %} Svidler, Peter (1998) “,” New in Chess, 1998 no. 7. {% On behavioral economists and experimental economists, Vernon Smith, Plott, Kahneman, and others, how they discussed and how some hostilities came. Sidney Siegel initiated, independently of and simultaneously with Smith, the principles of experimental economics, emphasizing real incentives and no deception. Unfortunately, Siegel suddenly died at young age. The author’s writings on deception are shaky. P. 279 writes: “In general, deception in experiments occurs when the actual purpose of an experiment differs from the purpose announced to the test subjects.” This is not the definition of deception commonly accepted. It is usually taken as giving false, untrue, information to subjects. For one thing, this is broader than just about the purpose of the experiment. For another, it allows for incomplete info. Often, subjects are not given complete info on an experiment and the purposes of the experimenter, e.g., “proving that theory X is superior to theory Y,” or “showing that subjects overestimate probabilities.” P. 290 l. 4 erroneously writes: “the former group [behavioral] advocated allowing deception and hypothetical choices in economic experiments; the latter [experimental economists] avoided such experiments.” I do not think that behavioral decision theorists just allow for deception. I don’t remember that Kahneman & Tversky ever wrote about it, but I also do not remember any case where they used deception. Although I do not remember ever discussing deception with Tversky, I would be very surprised if he would not have thought that it should be avoided. Another strange claim is on p. 288: “the emerging behavioral economics became less and less reliant on experimentation and was equally embracing other empirical as well as modeling approaches.” I do not understand in which sense behavioral economics would care less about experiments. May be the author thinks that psychology-type experiments are not to be called experiments? %} Svorenčík, Andrej (2016) “The Sidney Siegel Tradition: The Divergence of Behavioral and Experimental Economics at the End of the 1980s,” History of Political Economy 48, 270–294. {% Bibliographic info about the issue of the journal is essential, because each issue renumbers from zero. Nice, enthusiastic, empirical study of utility functions, very well suited for students to understand what utility measurement is about. Use CEs (certainty equivalents) of 50-50 gambles to measure utility, for both gains and losses. P. 128, 2nd para brings the known claim of those days that choices from paradoxes (Ellsberg in this case) are exceptional laboratory findings, not very relevant to practical applications. concave utility for gains, convex utility for losses: pp. 132-133: utilities nicely exhibit the prospect-theory shapes of concave for gains, convex for losses, loss aversion, underlying prospect theory. These were incorporated in Fishburn & Kochenberger (1979). They are, however, not representative because they were a subselection chosen by the authors according to choice criteria not specified. P. 134, 4th para, finds clear loss aversion. P. 135, penultimate para, that utilities for losses are more erratic. %} Swalm, Ralph O. (1966) “Utility Theory. Insights into Risk Taking,” Harvard Business Review 44, Issue 6, 123–136. {% A nice intermediate between compensatory and noncompensatory tradeoffs. Subjects set thresholds but, then, violations of the thresholds are allowed and are evaluated smoothly as losses of utility. It looks a bit like prospect theory with several reference points. %} Swait, Joffre D. (2001) “A Non-Compensatory Choice Model Incorporating Attribute Cutoffs,” Transportation Research Part B 35, 903–928. {% Existence of God is derived using Bayesian reasoning. %} Swinburne, Richard (1986) “The Coherence of Theism.” Oxford University Press, New York. {% Existence of God is derived using Bayesian reasoning. %} Swinburne, Richard (2004) “The Existence of God.” Clarendon Press, Oxford, 2004. {% small risks overinsured; Point out that according to traditional EU analyses, the commonly found insurance decisions regarding deductibles for home insurance would imply absurd degrees of risk aversion. They have real data on these insurance decisions. P. 183 penultimate para: by taking $1000 deductible iso $250 or $500 deductible, people could on average have saved $100 per year! The price people pay extra for having $500 deductible iso $1000 is five times its average value! P. 187 bottom: $4.8 billion per year can the saved by all house-owners in the US by taking $1000 deductble. One individual can on average gain $6,300 until age 65. P. 184 mentions consumer inertia, of people keeping insurance even though price has become much worse. Hence better to estimate only for new customers (p. 189 end 3rd para). P. 192 ff: for measuring relative risk aversion, proper level of initial wealth is discussed in detail. P. 193 penultimate para: people have to overestimate probability of loss by factor 5 (18.3 iso 3.7) to come to single-digit relative risk aversion index. P. 195-196: common degrees of probability weighting thus neither can explain it well. Traditional loss aversion plays no role because insurance is all about losses. P. 196: If we take premium paid as reference point (which is psychologically plausible), then loss aversion can explain it. The Köszegi-Rabin (2006) model also leads to this. %} Sydnor, Justin (2010) “(Over)insuring Modest Risks,” American Economic Journal: Applied Economics 2, 177–199. {% foundations of probability %} Synthese 55, 1983, special issue on theory of knowledge. {% foundations of probability & foundations of statistics: special issue dedicated to the memory of Henry Kyburg. %} Synthese 186, 2012, Number 2. {% Shows that every partial order can be extended to an order, which is an easy application of Zorn’s lemma. An accessible (English) account seems to be in Joseph G. Rosenstein (1982) “Linear Orderings, Pure and Applied Mathematics,” 98, Academic Press, New York. %} Szpilrajn, Edward (1930) “Sur l’Extension de l’Ordre Partiel,” Fundamenta Mathematicae 16, 386–389. {% %} Szpiro, George G. (1985) “Optimal Insurance Coverage,” Journal of Risk and Insurance 52, 704–710. {% utility elicitation?; decreasing ARA/increasing RRA: seem to find constant RRA (consequently, decreasing absolute). %} Szpiro, George G. (1986) “Measuring Risk Aversion: An Alternative Approach,” Review of Economics and Statistics 68, 156–159. {% %} Szpiro, George G. (1987) “Optimal Insurance Coverage: Reply,” Journal of Risk and Insurance 54, 813–815. {% %} Szpilrajn, Edward (1930) “Sur l’Extension de l’Ordre Partiel,” Fundamentà Mathematicae 16, 386–389. {% %} Tadelis, Steven (2013) “Game Theory: An Introduction.” Princeton University Press, Princeton, NJ. {% equity-versus-efficiency %} Tadenuma, Koichi (1996) “Trade-off between Equity and Efficiency in a General Economy with Indivisible Goods,” Social Choice and Welfare 13, 445–450. {% equity-versus-efficiency %} Tadenuma, Koichi (2002) “Efficiency First or Equity First? Two Principles and Rationality of Social Choice,” Journal of Economic Theory 104, 462–472. {% Tests many discount families, both for group average and individual. Finds that generalized hyperbolic is best, with unit invariance second. Assumes linear utility. %} Takahasi, Taiki (2005) “Loss of Self-Control in Intertemporal Choice May be Attributable to Logarithmic Time-Perception,” Medical Hypotheses 65, 691–693. {% nonconstant discount = nonlinear time perception; Eq. 6 proposes the unit invariance discounting family, with the nice interpretation that this is constant discounting with, however, Stevens-type power perception of time. %} Takahasi, Taiki (2006) “Time-Estimation Error Following Weber–Fechner Law May Explain Subadditive Time-Discounting,” Medical Hypotheses 67, 1372–1374. {% DOI: http://dx.doi.org/10.1016/j.physa.2007.11.047 %} Takahasi, Taiki, Hidemi Oono, & Mark H.B. Radford (2008) “Psychophysics of Time Perception and Intertemporal Choice Models,” Physica A: Statistical Mechanics and its Applications 387, 2066–2074. {% Let people choose, hypothetically, between an amount received immediately with certainty, and a risky amount received with delay. With general probability weighting one then cannot determine the power, but they assume EU and use a random-choice model with constant discounting and power utility to fit data. They find usual powers of utility (around 0.8) and usual discount rates (around 6%). They correlate with smoking, drinking, and two kinds of gambing. Smokers and gamblers are more impatient and less risk averse. For drinkers it is, overall, opposite. Bu the opposite is only for moderate drinkers (p. 615 bottom). Extreme drinkers are again more impatient and less risk averse. The authors defend rationality of moderate drinking (p. 615, jokingly: "Sake is the best medicine"). The writing and self-praising is sometimes naïve, with English-language limitations as likely excuse. %} Takanori, Ida & Rei Goto (2009) “Interdependency among Addictive Behaviours and Time/Risk Preferences: Discrete Choice Model Analysis of Smoking, Drinking, and Gambling,” Journal of Economic Psychology 30, 608–621. {% Seems to retest book-making tests of Tversky & Kahneman (1981), showing that it disappears if subjects have to justify. %} Takemura, Kazuhisa (1993) “The Effect of Decision Frame and Decision Justification on Risky Choice,” Japanese Psychological Research 35, 36–40. {% Seems to retest book-making tests of Tversky & Kahneman (1981), showing that it disappears if subjects have to justify, adding in this paper that it also gets less if they get more decision time. %} Takemura, Kazuhisa (1994) “Influence of Elaboration on the Framing of Decision,” Journal of Psychology 128, 33–39. {% On endogenous state spaces. %} Takeoka, Norio (2007) “Subjective Probability over a Subjective Decision Tree,” Journal of Economic Theory 136, 536–571. {% nonconstant discount = nonlinear time perception: this point was stated nicely in the working paper version but, unfortunately, as the author explained to me in personal communication, a referee had him take it out in the published version. decreasing/increasing impatience: finds counter-evidence against the commonly assumed decreasing impatience and/or present effect. First part of paper tests stationarity qualitatively as often done before, which can be called utility free because it needs not know utility. Second part first uses decision under risk and the standard gamble method to measure utility, assuming expected utility, and then measures discounting in utility rather than in money. The author suggests that this part does not measure utility at all (p. 460, §2.2, 2nd sentence), but measuring the standard gamble probabilities is equivalent to measuring utility. All of this conditional on assuming expected utility, which the author does. Similar things have been done by Andersen et al. (2008, Econometrica) and partly by Chapman (1996). The author calls his method utility-free because it works, given his assumptions, whatever utility is. The idea to pay in probability and then under EU have linear utility has been used before (Allen 1987; Anscombe & Aumann 1963; Berg et al. 1986 QJE; Roth & Malouf 1979; Cedric Smith 1961). Its drawbacks are that EU is extensively violated, with Selten, Sadrieh, & Abbink (1999) finding that the deviations from EU are bigger than those from linear utility, and that cardinal utility under risk need not be the same as cardinal intertemporal utility, as established after the ordinal revolution (Baumol 1959). Given the assumptions made, the author can in fact measure a model D(t,x)u(x), with discounting D(t,x) outcome dependent, as he points out on p. 457. The experiment finds quite some future bias. P. 471 “When does the future really start?” (Italics from original.) %} Takeuchi, Kan (2011) “Non-Parametric Test of Time Consistency: Present Bias and Future Bias,” Games and Economic Behavior 71, 456–478. {% PT, applications: nonadditive measures, sunspot equilibria %} Tallon, Jean-Marc (1998) “Do Sunspots Matter when Agents Are Choquet-Expected-Utility Maximizers,” Journal of Economic Dynamics and Control 22, 357–368. {% Using nonadditive measures and belief interpretations of those. Knowing E negative means that Ec has belief zero but E need not have belief one. %} Tallon, Jean-Marc (1998) “Asymmetric Information, Nonadditive Expected Utility, and the Information Revealed by Prices: An Example,” International Economic Review 39, 329–342. {% Games with incomplete information %} Tan, Tommy Ch.-Ch. (1988) “The Bayesian Foundations of Solution Concepts of Games,” Journal of Economic Theory 45, 370–391. {% real incentives: average payment was $11, roughly 7-day labor wage for casual unskilled labor. Random incentive system with one choice played for real. Use prospect theory, power utility and 1-parameter Prelec weighting function, and loss aversion, with same parameters for gains and losses. So then the unit of payment assumed does not matter for the definition of loss aversion. Choice stimuli: no sure prospects. Find indifference by choice list: 400.3010 ~ x0.105; 400.9030 ~ x0.705. The third choice list was more complex, with losses involved for both options. So, basically, three indifferences are used to fit three parameters. They use the first two indifferences to elicit utility power and probability weighting, and the third, given the first two, to elicit loss aversion. Find power 0.61 and weighting-function parameter 0.74. real incentives/hypothetical choice: for time preferences: they implemented using again random incentive system. Future payments for subjects were left to one of the subjects, a specially chosen “trusted agent,” who was asked to deliver the money on the days promised. I find it hard to believe that this would work well. Actually, I think that it would be immoral for the trusted agent NOT to deliver the money immediately. He is then causing money (interest and opportunities) to be lost for the people in his village just because some American told him so, with no use for the research (already over) or anything else, other than tribute to an abstract ethical principle of “never break a promise also if completely useless and to someone you will never see again.” The stimuli for intertemporal choice concerned immediate rewards versus rewards delayed by 3 days up to 3 months. For discounting they use a 3-parameter discount function, combining generalized hyperbolic discounting with also presence-effect à la quasi-hyperbolic. I regret that the two parameters besides exponent overlap in generating decreasing impatience, but they cannot fit increasing impatience which will surely be found for part of the subjects. It is like fitting risky data allowing only for risk aversion for every individual. The families by Bleichrodt, Rohde, & Wakker (2009, GEB) can handle increasing impatience. Subjects invited had participated in a demographic study 3 years before, so that things could be correlated. Richer villages are less loss averse and more patient. Richer households are more patient but no risk attitude effects. %} Tanaka, Tomomi, Colin F. Camerer, & Quang Nguyen (2010) “Risk and Time Preferences: Linking Experimental and Household Survey Data from Vietnam,” American Economic Review 100, 557–571. {% value of information %} Taneja, Harish C. & Sanju Sihmar (1994) “An Axiomatic Characterization of the Quantitative-Qualitative Measure of Information Improvement,” Information Sciences 78, 209–214. {% Discuss interpretations of loss aversion. Put forward the most common interpretation, that losses are felt more intensively than gains. One aspect of this they question in a way that I did not understand. They say that, contrary to the common view that gains reduce loss aversion and losses increase it (this I already do not understand), gains and losses may work in the same direction and both increase loss aversion. They seem to instead favor a sort of holistic evaluation. Peeters & Czapinski (1990) is a nice discussion of different interpretations of loss aversion. %} Tang, Hui, Zhe Liang, Kun Zhou, Gui-Hai Huang, Li-Lin Rao & Shu Li (2016) “Positive and Negative Affect in Loss Aversion: Additive or Subtractive Logic?,” Journal of Behavioral Decision Making 29, 381–391. {% foundations of probability & conservation of influence: discusses teleological theories of belief, and the role of objective and subjective probabilities in those. %} Tang, Weng Hong (2014) “Intentionality and Partial Belief,” Synthese 191, 1433–1450. {% %} Tännsjö, Torbjörn (2002) “Why We Ought to Accept the Repugnant Conclusion,” Utilitas 14, 339–359. {% doi:10.1093/lpr/mgv008 foundations of statistics: pp. 6-7 takes the subjective view of probability and discusses other views. This paper argues in fact for the likelihood principle, where statistical info is completely captured by the likelihood ratio. It argues against p-value-type info. It does all these things in the legal context. There are two comments and a rejoinder in this issue of the journal. %} Taroni, Franco, Silvia Bozza, Alex Biedermann, & Colin Aitken (2016) “Dismissal of the Illusion of Uncertainty in the Assessment of a Likelihood Ratio,” Law, Probability and Risk 15, 1–16. {% foundations of quantum mechanics %} Tarozzi, Gino & Alwyn van der Merwe (1988) “The Nature of Quantum Paradoxes.” Kluwer, Dordrecht. {% %} Taylor, Kimberley A. (1995) “Testing Credit and Blame Attributions as Explanation for Choices under Ambiguity,” Organizational Behavior and Human Decision Processes 54, 128–137. {% cognitive ability related to risk/ambiguity aversion: Subjects of high cognitive ability are more risk seeking in hypothetical choice than with real incentives. For others it makes no difference. Overall, there is no significant difference between risk aversion in real and hypothetical choice. The author seems to think that Holt & Laury (2002) invented the price list to measure risk aversion, citing a handful of studies that used it after in footnote 8, and not citing the many that used it before. %} Taylor, Matthew P. (2013) “Bias and Brains: Risk Aversion and Cognitive Ability across Real and Hypothetical Settings,” Journal of Risk and Uncertainty 46, 215–246. {% Seems to survey studies of optimism. %} Taylor, Shelley E. (1989) “Positive Illusions: Creative Self-Deceptions and the Healthy Mind.” Basic Books, New York. {% intuitive versus analytical decisions; replicate findings of Snijders, Tazelaar, & Batenburg (2003); add puzzling finding: purchasing managers predict worse the more experienced they are; %} Tazelaar, Frits & Chris Snijders (2004) “The Myth of Purchasing Professionals’ Expertise. More Evidence on whether Computers Can Make Better Procurement Decisions,” Journal of Purchasing & Supply Management 10, 211–222. {% %} Teigen, Karl H. (1983) “Studies in Subjective Probability III: The Unimportance of Alternatives,” Scandinavian Journal of Psychology 24, 97–105. {% EU+a*sup+b*inf: uses Choquet expected utility with this model. Leads to recommendations for negligence and against liability in unilateral accident cases. %} Teitelbaum, Joshua C. (2007) “A Unilateral Accident Model under Ambiguity,” Journal of Legal Studies 36, 431–477. {% conditional probability %} Teller, Paul (1973) “Conditionalization and Observation,” Synthese 26, 218–258. {% Gathered 154 quality of life measurements, %} Tengs, Tammy O. & Amy Wallace (2000) “One Thousand Health-Related Quality-of-Life Estimates,” Medical Care 38, 583–637. {% Measures utility, assuming EU, through hypothetical choices under risk, conditional on having two legs paralized or being healthy. This is entirely state-dependent utility à la Karni, with Anscombe-Aumann too. %} Tengstam, Sven (2014) “Disability and marginal utility of income: Evidence from Hypothetical Choices,” Health Economics 23, 268–282. {% Deviations from subgame perfect Nash equilibrium are independent of size of stake, and are of an omission-commission type. The errors do increase with the difficulty of the task. In my words, this means that cognitive rather than motivational factors cause the deviation from rationality here. (cognitive ability related to likelihood insensitivity (= inverse-S)) %} Tenorio, Rafael & Timothy N. Cason (2002) “To Spin or not to Spin? Natural and Laboratory Experiments from THE PRICE IS RIGHT,” Economic Journal 112, 170–195. {% Considers evaluation of prospect (act) if there is only a probability measure on some subalgebra and the prospect is not measurable with respect to it, using a model for this by Lehrer, taking either expected utilty of Choquet expected utility as point of departure. It considers such a preference for each time point and then analyzes continuity properties with time going to infinity, which is called time continuity. %} Teper, Roee (2009) “Time Continuity and Nonadditive Expected Utility,” Mathematics of Opertions Research 34, 661–673. {% %} Teper, Roee (2010) “On Comparison of Non-Bayesian Experts,” Mathematical Social Sciences 60, 119–122. {% %} Terlouw, Pieter (1989) “Subjective Probability Distributions: a Psychometric Approach.” Ph.D. Dissertation, University of Groningen. {% proper scoring rules: for many years he interviewed many politicians etc., asking them for probability judgments. Then he evaluated it all through proper scoring rules. Much in the spirit of Hofstee (1988). The book also shows that specialists do not perform better than others because specialists want to impress using bold estimates. %} Tetlock, Philip E. (2005) “Expert Political Judgment.” Princeton University Press, Princeton, NJ. {% %} Tetlock, Philip E., Ferdinand M. Vieider, Shefali V. Patil, & Adam Grant (2013) “Accountability and Ideology: When Left Looks Right and Right Looks Left,” Organizational Behavior and Human Decision Processes 122, 22–35. {% Many nice real-world examples about endowment effect, e.g. pp. 45-46. P. 50 suggests that Weber-Fechner law says that just noticeable difference is proportional to the absolute value, leading to logarithmic evaluation. ratio-difference principle: people do more effort to save $4 on a $25 radio, than on a $500 tv. P. 51 footnote 15 describes nice add where man takes $37 from $5000, says “It may not seem like a lot here” pointing to the pile of $5000, and then says “but it will feel like a lot here” pointing to his wallet. Many more on precommitment, billiard player who subconsciously follows sophisticated mathematical laws. %} Thaler, Richard H. (1980) “Towards a Positive Theory of Consumer Choice,” Journal of Economic Behavior and Organization 1, 39–60. {% dynamic consistency; time preference; seems to also find sign-dependence of discounting, with smaller discounting for losses than for gains. %} Thaler, Richard H. (1981) “Some Empirical Evidence of Dynamic Inconsistency,” Economics Letters 8, 201–207. {% Argues in favor of value function of prospect theory, for one reason because it captures the psychophysics of quantity. P. 201: “… captures the basic psychophysics of quantity. The difference between $10 and $20 seems greater than the difference between $110 and $120, irrespective of the signs of the amounts in question.” The paper distinguishes between acquisition utility (intrinsic utility) and transaction utility (process utility). %} Thaler, Richard H. (1985) “Mental Accounting and Consumer Choice,” Marketing Science 4, 199–214. {% real incentives/hypothetical choice: p. 96 seems to suggest that there is little improvement of rationality when real monetary rewards are introduced. %} Thaler, Richard H. (1987) “The Psychology and Economics Conference Handbook: Comments on Simon, on Einhorn and Hogarth, and on Tversky and Kahneman.” In Robin M. Hogarth & Melvin W. Reder (eds.) “Rational Choice: The Contrast between Economics and Psychology,” 95–100, University of Chicago Press. {% P. 138 writes: “illusions demonstrate the need for rulers” %} Thaler, Richard H. (1991) “Quasi Rational Economics.” Russell Sage Foundation, New York. {% %} Thaler, Richard H. (2015) “Misbehaving: The Making of Behavioral Economics.” W. W. Norton & Company, New York. {% A general discussion arguing for the importance of behavioral economics. Unfortunately, the author desires too much to show that other researchers are dumb and wrong, with the implicit implication that he himself is more clever. And, unfortunately, he does not try to properly position views other than his own, but he tries to make them look ridiculous using puns (p. 1579: “explainawaytions”), which does not advance communication and exchange of ideas, even if primitive readers (those who also enjoy violent movies) may enjoy it. It is good in writing and for clarity to skip some nuances, but this paper does it too much. P. 1579 beginning of §II: “In the process of making economics more mathematically rigorous after World War II, the economics profession appears to have lost its good intuition about human behavior.” P. 1579 footnote 1 is characteristic of the sense of humor of the author. In the beginning of the paper, and in several other places (p. 1578 middle: “Indeed, Ashraf, Camerer, and Loewenstein (2005) convincingly document that Adam Smith, often considered the founder of economics as a discipline, was a bona fide behavioral economist.”), the author tries to argue that the behavioral approach means simply returning to the pre-ordinal period. Unfortunately, he never uses the term ordinal or refers to the ordinal revolution, but this is the crucial dividing line. P. 1580 seems to confuse the ordinal and the marginal revolution, apaprently putting the marginal revolution in the 1940s. The marginal revolution was in the 1870s. He sometimes refers to “after World War II” I disagree with his idea for the same reason that I disagree with the idea expressed in “Back to Bentham” (elsewhere). The ordinal revolution added much good, giving a clear and firm basis to economics. The behavioral revolution (using this term, also sometimes used in this paper) does not mean throwing these ideas away. It means extending these ideas, keeping the formal concepts but extending the empirical domain (a) by incorporating irrational phenomena studied before in psychology; (b) relaxing the restriction to revealed-preference data. Those extensions should be linked to the firm basis thanks to the ordinal revolution. Fortunately, in one place the author puts this rightm, being p. 1592 1st para: “A second general point is that we should not expect some new grand behavioral theory to emerge to replace the neoclassical paradigm. We already have a grand theory and it does a really good job of characterizing how optimal choices and equilibrium concepts work. Behavioral theories will be more like engineering, a set of practical enhancements that lead to better predictions about behavior. So far, most of these behavioral enhancements focus on two broad topics: preferences and beliefs.” Unfortunately, in the conclusion p. 1597 the author returns to the unnuanced statement: “Rather, behavioral economics should be considered simply a return to the kind of open-minded, intuitively motivated discipline that was invented by Adam Smith and augmented by increasingly powerful statistical tools and datasets.” P. 1581, l. -4 presents EU as normative: “Prospect theory was intended to be a descriptive alternative to von Neumann and Morgenstern’s (1947) expected utility theory, which is rightly considered by most economists to characterize how a rational agent should make risky choices.” P. 1582 l. -3 lists Thaler’s 1980 paper together with the work of Kahneman & Tversky. P. 1583 2nd para shows how the desire to show others wrong (end of 3rd para: “So critics can’t have it both ways. Either the real world is mostly high stakes or it offers myriad opportunities to learn—not both.”) blinds the author: his point that decisions with large stakes usually cannot be repeated much is irrelevant and in no way weakens the argument that both large stakes and repetition/learning increase rationality. P. 1585 end of 2nd para: “In the nearly 40 years since Grether and Plott’s seminal paper, I do not know of any findings of “cognitive errors” that were discovered and replicated with hypothetical questions but then vanished as soon as significant stakes were introduced.” Many studies, also some with me as co-author, find more noise with hypothetical choice (and less risk aversion). This usually means that any pattern is weakened and, hence, also violations of preference conditions. Still, it is clear that real incentives, other things equal, gives higher quality of data. P 1585, §C, can be briefly summarized as: “market mechanisms will often but not always reduce irrationalities.” The play with words of “invisible handwave” p. 1585 3rd para is typical of this paper’s style. P. 1591 bottom: theory and empirics ALWAYS go hand in hand, so thing are way more universal than in the following citation: “Some might worry about basing theories on empirical observation, but this methodology has a rich tradition in science. The Copernican revolution, which placed the sun at the center of our solar system rather than the earth, was based on data regarding the movement of the planets, not on some first principles.” P. 1592 footnote 9: utility of income has more to do with reference dependence than with mental accounting. P. 1597 last para: “If economics does develop along these lines the term “behavioral economics”will eventually disappear from our lexicon.” The ambitious idea is that everyone will be doing behavioral, so no more need to use the adjective. %} Thaler, Richard H. (2016) “Behavioral Economics: Past, Present, and Future,” American Economic Review 106, 1577–1600. {% The authors got some firms to implement a program, called SMarT, to automatically make their employees save each month, in a percentage that they could influence. It led to considerably more savings. People save too little (p. 166 2nd para: as can be inferred from their answers if asked. A lternative explanation of their answer can be social desirability.). Four biases are advanced to be underlying this (summarized and listed briefly on p. 170 2nd para (“In summary … these households.” )): 1. Bounded rationality. People cannot calculate what is optimal for them. 2. Lack of self-control (time inconsistency/hyperbolic discounting). 3. (Much like 2): procrastination. 4. Loss aversion (the authors also involve money illusion). This lead to the following aspects of the SMarT program (§III pp. 170-1st para of 171): Because of 1, SMarT does not ask the clients but determines itself to what level it tries to make clients increase payment, and then stop there. Because of 2 and 3, clients are asked to commit to payment way before the first payment comes. Because of 4, let payment be raised only after salary rises. Further loss aversion and the implied inertia (which will be generated much by incompleteness of preference rather than loss aversion) should serve to imply that clients do not opt out of the program once being in. Relying on this, clients at each stage had the possibility to opt out if they wanted. paternalism/Humean-view-of-preference: all actions stay within the boundaries of libertarian paternalism, of not doing anything people do not want by their gut feelings. P. 167 last para: DC = stationarity; P. 169 penultimate para: loss aversion underlies inertia which, in turn, underlies why people don’t save enough. P. 185: “One reason why the SMarT plan works so well is that inertia is so powerful.” P. 170 end of 1st para: the authors suggest that a 7 percent wage cut under no inflation should be as fair as a 5% salary raise under 12% inflation. This is not correct because 12% inflation means that the economy is doing badly, making it more “fair” to get worse off by oneself. paternalism/Humean-view-of-preference: conclusions on p. 185 ff. discuss it. Refer to Thaler & Sunstein (2003) on libertarian paternalism. P. 186: “we plead guilty to the charge of trying to be paternalistic. ... we have used behavioral principles to design a plan to increase savings rates and tested the ideas in the real world.” %} Thaler, Richard H. & Shlomo Benartzi (2004) “Save More tomorrow: Using Behavioral Economics to Increase Employee Saving,” Journal of Political Economy 112, S164–187. {% Subsections 5.1 and 5.2: house money effect: A prior gain increases the willingness to accept gambles, as long as they do not risk loosing the entire recent winnings. So it is a kind of decreasing ARA (absolute risk aversion). (In a casino you are then gambling with the money you already won so with the “house money.”) A prior loss decreases the willingness to gamble (so again decreasing ARA), except if it can generate breaking even (or turn losses to gains). Subsection 6.1 discusses some alternative explanations. They give evidence against the isolation effect; i.e., prior gains etc. can matter. It’s a kind of income effect. real incentives/hypothetical choice: p. 652 beginning of Subsection 4.1: “However, an experiment in which subjects can lose money creates some ethical dilemmas.” P. 653: participants who lose money can pay by hours of clerical work, if they want. utility concave near ruin: seems that they have a quasi-hedonic editing rule that suggests this. %} Thaler, Richard H. & Eric J. Johnson (1990) “Gambling with the House Money and Trying to Break Even: The Effects of Prior Outcomes on Risky Choice,” Management Science 36, 643–660. {% Christmas and diet clubs to help self-control %} Thaler, Richard H. & Hersh M. Shefrin (1981) “An Economic Theory of Self-Control,” Journal of Political Economy 89, 392–410. {% paternalism/Humean-view-of-preference: libertarian paternalism means not trying to change preference held by clients. Only in situations where it is all the same to the client and the client has no preference (as with situations where default has so much impact), libertarian paternalisms takes it the way the analyst thinks best for the client. So libertarian paternalism plays in the space left by incomplete preferences. paternalism/Humean-view-of-preference: “we clearly do not always equate revealed preference with welfare.” %} Thaler, Richard H. & Cass R. Sunstein (2003) “Libertarian Paternalism,” American Economic Review, Papers and Proceedings 93, 175–179. {% P. 6 seems to write, defining a nudge: “an aspect of choice architecture that alters people’s behavior in a predictable way, without forbidding any options or significantly changing their economic incentives” Here the “without forbidding” part expresses libertarian. Changing economic incentives can trivially change people’s behavior but is not a nudge. Later on the book seems to write that a nudge should “make people better off as judged by themselves.” %} Thaler, Richard H. & Cass R. Sunstein (2008) “Nudge: Improving Decisions about Health, Wealth, and Happiness.” Yale University Press, New Haven. {% %} Thaler, Richard H. & Amos Tversky (1996) “Myopic Loss Aversion in Financial Investment,” unpublished manuscript, University of Chicago. {% PT, applications, loss aversion decreasing ARA/increasing RRA: use power utility. %} Thaler, Richard H., Amos Tversky, Daniel Kahneman, & Alan Schwartz (1997) “The Effect of Myopia and Loss Aversion on Risk Taking: An Experimental Test,” Quarterly Journal of Economics 112, 647–661. {% Opening sentence: “Economics can be distinguished from other social sciences by the belief that most (all?) behavior can be explained by assuming that agents have stable, well-defined preferences and make rational choices consistent with those preferences in markets that (eventually) clear.” Discuss biases in bets and lotteries, where sometimes one can even have positive expectation if knowing the biases. inverse-S: the favorite-longshot bias in horse racing: people underestimate the winning probabilities if they are high and overestimate them when they are low. So, they bet too much on outsiders and too little on favorites, to the extent even that for favorites with 0.7 probability or more of winning the expectation of gambling is positive. P. 171 Reason 5 lists that people gamble on horses for reasons such as name etc., unrelated to the winning chances. This looks like likelihood insensitivity. P. 172: lotteries only became popular when New Jersey let people choose their own numbers, speculating on illusion of control. Dutch book: p. 167 discusses and references cross-track gambling where different bookmakers had dramatically-different odds. In lotto 6/49, they list numbers that are overchosen (7 most) and those that are underchosen. P. 170 discusses the problems of the Friedman & Savage (1948) utility curve. %} Thaler, Richard H. & William T. Ziemba (1988) “Parimutual Betting Markets: Racetracks and Lotteries,” Journal of Economic Perspectives 2 no. 2, 161–174. {% Alfabetisch onder “T” Describes the result of Rabin & Thaler (2001, JEP 15), arguing against expected utility and in favor of loss aversion. %} The Economist (2001) “Economics Focus Averse to Reality,” Economist, August 11, p. 61. {% On loss aversion. %} The Economist (2003) “To Have and to Hold,” Economist, August 30, p. 56. {% Bayes formula. Describes research by Griffiths & Tenenbaum on updating. Text is overly simplistic about Bayes formula simply working well with negative statements about frequentists. %} The Economist (2006) “Bayes Rules,” Economist, January 7, p. 70–71. {% %} The Economist (2008) “Anti-Terrorist Spending: Feel Safer now?,” Economist, March 8. p. 69. {% That Keynes and Knight pointed out that uncertainty is really different than risk. Then goes into rent policies when market does bad. %} The Economist (2009) (written by Chris F. Masse) “Bribing the Markets; The Impossible Task of Eliminating Uncertainty,” Economist, November 11. {% foundations of statistics %} Thomas, Hoben (2000) “What Statistical Evidence Is and What it Is Not,” Book Review of: Richard Royall (1997) Statistical Evidence: A Likelihood Paradigm, Chapman & Hall, New York; Journal of Mathematical Psychology 44, 480–487. {% %} Thomsen, Gerhard (1927) “Un Teorema Topologico sulle Schiere di Curve e una Caratterizzazione Geometrica delle Superficie Isotermo-Asintotiche,” Bolletino della Unione Matematica Italiana 6, 80–85. {% Detailed discussion of many aspects of axiomatizations for game theory and resource allocation. The paper is mostly oriented towards applications in other economic theories, so with theoretical requirements such as continuity, and less towards empirical or practical prescriptive applications, in which continuity plays no role. There are some comments on operationalism in §10.1, and p. 372 point 4 of §12.2 has a nice discussion. Download 7.23 Mb. Share with your friends:
|