Ambiguity Made Operational by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint with Mohammed Abdellaoui & Aurélien Baillon) ESA, Tucson, Oct.

Slides:



Advertisements
Similar presentations
From risk to opportunity Lecture 11 John Hey and Carmen Pasca.
Advertisements

Paradoxes in Decision Making With a Solution. Lottery 1 $3000 S1 $4000 $0 80% 20% R1 80%20%
Representations for KBS: Uncertainty & Decision Support
Lecture 4 Environmental Cost - Benefit - Analysis under risk and uncertainty.
Non-Probability-Multiple-Prior Models of Decision Making Under Ambiguity: new experimental evidence John Hey a and Noemi Pace b a University of York, UK.
1 Lower Distribution Independence Michael H. Birnbaum California State University, Fullerton.
Evaluating Non-EU Models Michael H. Birnbaum Fullerton, California, USA.
Integration of sensory modalities
Model assessment and cross-validation - overview
Using Prospect Theory to Study Unknown Probabilities ("Ambiguity") by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint with Mohammed Abdellaoui.
Making Rank-Dependent Utility Tractable for the Study of Ambiguity Peter P. Wakker, June 16, 2005 MSE, Université de Paris I Aim:Make rank-dependent utility.
Tutorial on Risk and Uncertainty Peter P. Wakker Part 1: Introduction into Prospect Theory. Part 2: Using Prospect Theory to Better Describe and Prescribe.
Topic: Subjective chance estimates, e.g.: - chance of rain tomorrow; - chance of strategy choice of opponent. Application considered first: grading of.
Decision-Principles to Justify Carnap's Updating Method and to Suggest Corrections of Probability Judgments Peter P. Wakker Economics Dept. Maastricht.
Topic: Our chance estimates of uncertain events; e.g.: Hillary Clinton next president of the US. Adapting Proper Scoring Rules for Measuring Subjective.
Ambiguity will concern Dow Jones & Nikkei indexes today: Topic: Quantitative measurement of capacities under ambiguity. Question: How do people perceive.
½U(1000) + ½U(x 1 ) = ½U(8000) + ½U(x 0 ) _ ( U(8000)  U(1000) ) Tradeoff (TO) method for EU x2x x 1 ~ ½ ½ ½ ½ x4x x 3 ½ ½ ~ ½ ½.
UNCERTAINTY AVERSION VS. COMPETENCE: AN EXPERIMENTAL MARKET STUDY Carmela Di Mauro Università di Catania, Italy.
Tractable Quantifications of Ambiguity Attitudes by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint with Mohammed Abdellaoui & Aurélien Baillon)
I’m not overweight It just needs redistribution Michael H. Birnbaum California State University, Fullerton.
Example, on Dow Jones & Nikkei indexes today: Peter P. Wakker & Enrico Diecidue & Marcel Zeelenberg Domain: Individual Decisions under Ambiguity (events.
Lecture 05 Rule-based Uncertain Reasoning
On Dow Jones & Nikkei indexes today: Peter P. Wakker & Enrico Diecidue & Marcel Zeelenberg This file will be on my homepage on coming Monday. Domain: Decisions.
Combining Bayesian Beliefs and Willingness to Bet to Analyze Attitudes towards Uncertainty by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint.
Uncertainty RUD, June 28, 2006, Peter P. Wakker Paper: Chapter for New Palgrave. Its didactical beginning is not optimal for this specialized audience.
Adapting de Finetti's Proper Scoring Rules for Measuring Subjective Beliefs to Modern Decision Theories of Ambiguity Gijs van de Kuilen, Theo Offerman,
Using ranking and DCE data to value health states on the QALY scale using conventional and Bayesian methods Theresa Cain.
Stochastic Dominance Michael H. Birnbaum Decision Research Center California State University, Fullerton.
Making Choquet Expected Utility Tractable for the Study of Ambiguity Peter P. Wakker, December 20, 2005 Tel-Aviv University Aim:Make Choquet-expected utility,
Reconciling Introspective Utility with Revealed Preference: Experimental Arguments Based on Prospect Theory Peter P. Wakker ( & Abdellaoui & Barrios; Ecole.
PSY 5018H: Math Models Hum Behavior, Prof. Paul Schrater, Spring 2005 Rational Decision Making.
A new tool is presented for measuring beliefs/likelihoods under uncertainty. It will simplify:  preference axiomatizations;  quantitative measurements;
Combining Bayesian Beliefs and Willingness to Bet to Analyze Attitudes towards Uncertainty by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint.
Behavior in the loss domain : an experiment using the probability trade-off consistency condition Olivier L’Haridon GRID, ESTP-ENSAM.
Ambiguity will concern Dow Jones & Nikkei indexes today: Peter P. Wakker & Enrico Diecidue & Marcel Zeelenberg Lecture + paper are on my homepage Domain:
CSM/KSS'2005 Knowledge Creation and Integration for Solving Complex Problems August 29-31, 2005, IIASA, Laxenburg, Austria ______________________________________.
Adapting de Finetti's Proper Scoring Rules for Measuring Bayesian Subjective Probabilities when Those Probabilities Are not Bayesian Peter P. Wakker (&
On the smooth ambiguity model and Machina’s reflection example Robert Nau Fuqua School of Business Duke University.
New Views on Risk Attitudes Peter P. Wakker Economics University of Amsterdam € 100 € 0€ 0 ½ ½ or € 50 for sure What would you rather have? Such gambles.
Proper Scoring Rules and Prospect Theory August 20, 2007 SPUDM, Warsaw, Poland Topic: Our chance estimates of Hillary Clinton to become next president.
Stochastic choice under risk Pavlo Blavatskyy June 24, 2006.
RISK, AMBIGUITY, GAINS, LOSSES Sujoy Chakravarty Department of Humanities and Social Sciences Indian Institute of Technology, Hauz Khas, New Delhi ,
LOGISTIC REGRESSION David Kauchak CS451 – Fall 2013.
Modern Ways to Model Risk and Uncertainty Peter P. Topic: prospect theory (  classical expected utility) for modeling risk/uncertainty/ambiguity.
A Stochastic Expected Utility Theory Pavlo R. Blavatskyy June 2007.
LANGUAGE MODELS FOR RELEVANCE FEEDBACK Lee Won Hee.
Choice Behavior, Asset Integration and Natural Reference Points Steffen Andersen, Glenn W. Harrison & E. Elisabet Rutström.
Lecture 3 on Individual Optimization Uncertainty Up until now we have been treating bidders as expected wealth maximizers, and in that way treating their.
Axiomatic Theory of Probabilistic Decision Making under Risk Pavlo R. Blavatskyy University of Zurich April 21st, 2007.
Measuring the impact of uncertainty resolution Mohammed Abdellaoui CNRS-GRID, ESTP & ENSAM, France Enrico Diecidue & Ayse Onçüler INSEAD, France ESA conference,
Fundamentals of Decision Theory Chapter 16 Mausam (Based on slides of someone from NPS, Maria Fasli)
Allais Paradox, Ellsberg Paradox, and the Common Consequence Principle Then: Introduction to Prospect Theory Psychology 466: Judgment & Decision Making.
Regression Analysis1. 2 INTRODUCTION TO EMPIRICAL MODELS LEAST SQUARES ESTIMATION OF THE PARAMETERS PROPERTIES OF THE LEAST SQUARES ESTIMATORS AND ESTIMATION.
Adapting de Finetti's proper scoring rules for Measuring Subjective Beliefs to Modern Decision Theories ofAmbiguity Peter P. Wakker (& Gijs van de Kuilen,
Machine Learning CUNY Graduate Center Lecture 6: Linear Regression II.
Scoring Rules, Generalized Entropy and Utility Maximization Victor Richmond R. Jose Robert F. Nau Robert L. Winkler The Fuqua School of Business Duke University.
Organizing the Data Levin and Fox Elementary Statistics In Social Research Chapter 2.
1 Chapter 4 Interpolation and Approximation Lagrange Interpolation The basic interpolation problem can be posed in one of two ways: The basic interpolation.
1 BAMS 517 – 2011 Decision Analysis -IV Utility Failures and Prospect Theory Martin L. Puterman UBC Sauder School of Business Winter Term
Outline Historical note about Bayes’ rule Bayesian updating for probability density functions –Salary offer estimate Coin trials example Reading material:
Probability and statistics - overview Introduction Basics of probability theory Events, probability, different types of probability Random variable, probability.
Updating Ambiguity Averse Preferences Eran Hanany Peter Klibanoff Tel Aviv UniversityNorthwestern University Faculty of EngineeringKellogg School of Management.
ETHEM ALPAYDIN © The MIT Press, Lecture Slides for.
American Association of Wine Economists (AAWE)
Adapting Proper Scoring Rules for Measuring Subjective Beliefs
Ascertaining certain certainties in choices under uncertainty
Peter P. Wakker & Enrico Diecidue & Marcel Zeelenberg
to Prospect Theory Peter P. Wakker Theo Offerman Joep Sonnemans
Integration of sensory modalities
Mathematical Foundations of BME Reza Shadmehr
Presentation transcript:

Ambiguity Made Operational by Peter P. Wakker, Econ. Dept., Erasmus Univ. Rotterdam (joint with Mohammed Abdellaoui & Aurélien Baillon) ESA, Tucson, Oct. 19 '07 Keynes 1921 & Knight 1921: probabilities unknown. Long time not studied, simply because no models. Only in 1989: Choquet-expected utility & multiple priors (Gilboa & Schmeidler); 1992: Prospect theory (Tversky & Kahneman); other theories: Robust control (Hansen & Sargent 2001); Smooth (Klibanoff, Marinacci & Mukerji 2005); Variational (Maccheroni, Marinacci, Rustichini 2006).

These theories not implemented quantitatively. Empirical studies are as yet qualitative. We make ambiguity quantitatively operational by introducing source functions. We get exact ambiguity premium etc. 2

1. Introduction First some words on risk (so no ambiguity yet). Most descriptive (nonEU) theories: x  y  0; xpy  w(p)U(x) + ( 1–w(p) ) U(y); Relative to EU: one more graph … 3

4 inverse-S, (likelihood insensitivity) p w expected utility pessimism extreme inverse-S ("fifty-fifty") prevailing finding pessimistic "fifty-fifty" Common graphs found:

Now to Uncertainty (unknown probabilities); x-axis has events. So, no nice graphs … 5

2. Theory Most ambiguity theories: x  y  0; xEy  W(E)U(x) + ( 1–W(E) ) U(y). For rich state space, such as continuum, general W is too complex. Machina & Schmeidler (1992) considered tractable special case. 6

Machina & Schmeidler (1992), probabilistic sophistication (in a "global" sense): W(. ) = w(P(. )) for a probability P. So, xEy  w ( P(E) ) U(x) + ( 1–w ( P(E) ) ) U(y). Then still get nice x-axis for uncertainty. W.r.t. Bayesian, still "one more graph." Tractable and OK. However, 7

Ellsberg (1961) 2-color paradox: 2 different urns with irreconcileable prob.beliefs; no global probabilistic sophistication; no p's exist. No x-axis … Life remains difficult … Now to our operationalization of ambiguity (we will get back an x-axis and probabilistic sophistication but in a "local" sense): 8

Theory for our operationalization of ambiguity: 9 For x  y  0: xEy  w S ( P(E) ) U(x) + ( 1–w S ( P(E) ) ) U(y). Like Machina & Schmeidler, but with S added. Explanation: Not all events alike. Divide events into different sources S. Source S is set of events referring to same "mechanism generating uncertainty"; Define uniform sources: its events have uniform degree of ambiguity = within such a source we have probabilistic sophistication ("local" prob. soph.); w S is the source function. We get back P & x-axis & nice graphs!

Wait a minute … One graph for every source!? Many graphs!? More complex than Machina-Schmeidler!? Yes it is. Ambiguity more complex than risk or Machina-Schmeidler "global" probabilistic sophistication. No free lunch … Still, major simplification wrt general W or general set of priors. 10

3. Rubber Meets the Road: An Experiment Data: 11 4 sources: 1.CAC40; 2.Paris temperature; 3."Foreign" temperature; 4.Risk.

12 Method for measuring choice-based probabilities: EEEEEE Figure 6.1. Decomposition of the universal event a 3/4 E a 1/2 a 1/4 a 1/8 a 3/8 E b1b1 a 5/8 a 7/8 b0b0 a 3/4 a 1/2 a 1/4 EE b1b1 b0b0 E E a 1/2 E b1b1 b0b0 E E = S b1b1 b0b0 The italicized numbers and events in the bottom row were not elicited. repeated bisection into equally likely events of probability 2 –n.

Median choice-based probabilities (real incentives) Real data over 1900  Figure 7.2. Probability distributions for Paris temperature Median choice-based probabilities (hypothetical choice) 0.0 Median choice-based probabilities (real incentives) Real data over the year 11 22 3 Figure 7.1. Probability distributions for CAC40 Median choice-based probabilities (hypothetical choice) Results for choice-based probabilities Uniformity confirmed 5 out of 6 cases.

Certainty-equivalents of prospects. Fit power utility with w(0.5) as extra unknown Hypothetical Real Figure 7.3. Cumulative distribution of powers Method for measuring utility Results for utility

15 Results for uncertainty ("ambiguity?")

16 These were within-person comparisons Source functions for participant 2 Fig. a. Raw data and linear interpolation Paris temperature; a = 0.78, b = 0.12 foreign temperature; a = 0.75, b = 0.55 risk: a = 0.42, b = 0.13 choice-based probabilities wSwS (source function weights)

17 participant 2; a = 0.78, b = * Fig. a. Raw data and linear interpolation. * Source functions for Paris temperature participant 48; a = 0.21, b = 0.25 Between-person comparisons:

Example of predictions [Homebias; Within- Person Comparison; subject lives in Paris]. Consider investments. Foreign-option: favorable foreign temperature: $40000 unfavorable foreign temperature: $0 Paris-option: favorable Paris temperature: $40000 unfavorable Paris temperature: $0 Assume in both cases: favorable and unfavo- rable equally likely for subject 2; U(x) = x Under Bayesian EU we’d know all now: no homebias. NonEU: need source function; we have them! 18

19 Paris temperature Foreign temperature decision weight expectation certainty equivalent uncertainty premium risk premium ambiguity premium –3662 Within-person comparisons:

20 Subject 2, p = decision weight expectation certainty equivalent uncertainty premium risk premium ambiguity premium –3099 Subject 48, p = Subject 2, p = Subject 48, p = –39 –4034 – Between-person comparisons; Paris temperature

Conclusion: By carrying out quantitative measurements of (a) probabilities (subjective), (b) utilities, 21 we make ambiguity operational at a quantitative level. Bayesian (c) source functions (uncertainty attitudes), non-Bayesian (ambiguity)

The end 22

23 Many advanced theories; CEU (Gilboa 1987; Schmeidler 1989) PT (Tversky & Kahneman 1992) Multiple priors (Gilboa & Schmeidler 1989) Endogenous definitions (Epstein, Zhang, Kopylov, Ghirardato, Marinacci) Smooth (KMM; Nau) Variational model (Maccheroni, Marinacci, Rustichini) Biseparable (Ghirardato & Marinacci 2001) Choice-based but intractable Nice graphs!? x-axis-problem: choice- less probabilities We connect Many tractable empirical studies; Curley & Yates 1985 Fox & Tversky 1995 Kilka & We- ber 2001 Cabantous 2005 di Mauro & Maffioletti 2005 Einhorn & Hogarth 1985 (over 400 citations) "Anchor probabilities" on x-axis (c.f. Hansen & Sargent). Others take midpoint of probability interval.

24 Results for measuring ambiguity attitudes

* * * * * * * Figure 8.3. Probability transformations for participant 2 Fig. a. Raw data and linear interpolation Fig. b. Best-fitting (exp(  (  ln(p))  ))  CAC40; a = 0.80; b = 0.30 risk: a = 0.42, b = 0.13 Paris temperature; a = 0.78, b = 0.12 foreign temperature; a = 0.75, b = 0.55 CAC40;  = 0.15;  = 1.14 risk:  = 0.47,  = 1.06 Paris temperature;  = 0.17,  = 0.89 foreign temperature;  = 0.21,  = 1.68

26

@ 27

knalwit, minder fel wit, nog minder fel wit knalgeel, geel, zachtgeel/groen groen blauw bruin, 24 pnt.nog goed bij slideshow rood oranje lila grijs-achtig zwart if it works 28 next p.