EFFICIENT, WEAK EFFICIENT, AND PROPER EFFICIENT PORTFOLIOS

Similar documents
ON THE TRUNCATED COMPOSITE WEIBULL-PARETO MODEL

Choice under Uncertainty

Ph.D. Preliminary Examination MICROECONOMIC THEORY Applied Economics Graduate Program June 2016

ИЗМЕРЕНИЕ РИСКА MEASURING RISK Arcady Novosyolov Institute of computational modeling SB RAS Krasnoyarsk, Russia,

Competitive Equilibria in a Comonotone Market

Consumer theory Topics in consumer theory. Microeconomics. Joana Pais. Fall Joana Pais

Weak* Axiom of Independence and the Non-Expected Utility Theory

Department of Economics The Ohio State University Final Exam Questions and Answers Econ 8712

The Nash Bargaining Solution vs. Equilibrium in a Reinsurance Syndicate

This is designed for one 75-minute lecture using Games and Information. October 3, 2006

WORKING PAPER MASSACHUSETTS ALFRED P. SLOAN SCHOOL OF MANAGEMENT. c.x CAMBRIDGE, MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE

Increases in Risk Aversion and the Distribution of Portfolio Payoffs

Entropic Selection of Nash Equilibrium

Assortative Matching in Two-sided Continuum Economies

Increases in Risk Aversion and Portfolio Choice in a Complete Market

Notes on Recursive Utility. Consider the setting of consumption in infinite time under uncertainty as in

General equilibrium with externalities and tradable licenses

Elimination of Arbitrage States in Asymmetric Information Models

Are Probabilities Used in Markets? 1

COOPERATIVE GAME THEORY: CORE AND SHAPLEY VALUE

Near-Potential Games: Geometry and Dynamics

Near-Potential Games: Geometry and Dynamics

Contracts under Asymmetric Information

4 Lecture Applications

Microeconomics, Block I Part 2

Representation of TU games by coalition production economies

Market Equilibrium and the Core

1 General Equilibrium

The Value of Information Under Unawareness

Von Neumann Morgenstern Expected Utility. I. Introduction, Definitions, and Applications. Decision Theory Spring 2014

Graduate Macroeconomics 2 Problem set Solutions

Notes on General Equilibrium

Confronting Theory with Experimental Data and vice versa. Lecture I Choice under risk. The Norwegian School of Economics Nov 7-11, 2011

Notes on Supermodularity and Increasing Differences. in Expected Utility

When does aggregation reduce risk aversion?

CORVINUS ECONOMICS WORKING PAPERS. Young's axiomatization of the Shapley value - a new proof. by Miklós Pintér CEWP 7/2015

Recitation 7: Uncertainty. Xincheng Qiu

Incremental Risk Vulnerability 1

Incremental Risk Vulnerability

20 GUOQIANG TIAN may cooperate and in others they may not, and such information is unknown to the designer. (2) This combining solution concept, which

Course Handouts: Pages 1-20 ASSET PRICE BUBBLES AND SPECULATION. Jan Werner

Economics 201B Second Half. Lecture 12-4/22/10. Core is the most commonly used. The core is the set of all allocations such that no coalition (set of

Coalitional Structure of the Muller-Satterthwaite Theorem

Comonotonicity and Maximal Stop-Loss Premiums

ECO 317 Economics of Uncertainty Fall Term 2009 Slides to accompany 13. Markets and Efficient Risk-Bearing: Examples and Extensions

Lecture 1. History of general equilibrium theory

The Ohio State University Department of Economics. Homework Set Questions and Answers

The newsvendor problem with convex risk

Problem Set Suggested Answers

Fundamentals in Optimal Investments. Lecture I

NOTES ON COOPERATIVE GAME THEORY AND THE CORE. 1. Introduction

Indeterminacy and Sunspots in Macroeconomics

1 Uncertainty. These notes correspond to chapter 2 of Jehle and Reny.

Pareto Optimal Allocations for Law Invariant Robust Utilities

Incremental Risk Vulnerability 1

Almost Transferable Utility, Changes in Production Possibilities, and the Nash Bargaining and the Kalai-Smorodinsky Solutions

Introduction. 1 University of Pennsylvania, Wharton Finance Department, Steinberg Hall-Dietrich Hall, 3620

Department of Agricultural Economics. PhD Qualifier Examination. May 2009

Output contingent securities and efficient investment by

Final Examination with Answers: Economics 210A

General Equilibrium. General Equilibrium, Berardino. Cesi, MSc Tor Vergata

The Max-Convolution Approach to Equilibrium Models with Indivisibilities 1

Economics 200A part 2 UCSD Fall quarter 2011 Prof. R. Starr Mr. Troy Kravitz1 FINAL EXAMINATION SUGGESTED ANSWERS

First Welfare Theorem

FINANCIAL OPTIMIZATION

Unlinked Allocations in an Exchange Economy with One Good and One Bad

Informed Principal in Private-Value Environments

Majority Decision Rules with Minority Protections: Cost Assignments for Public Projects

Multi Objective Optimization

14.12 Game Theory Lecture Notes Theory of Choice

Variational Inequalities. Anna Nagurney Isenberg School of Management University of Massachusetts Amherst, MA 01003

1 Second Welfare Theorem

* Professor of Finance at INSEAD, Boulevard de Constance, Fontainebleau Cedex, France.

Moral Hazard in Teams

Rational Expectations Equilibrium in Economies with Uncertain Delivery

On the Maximal Domain Theorem

Definitions and Proofs

1 Axiomatic Bargaining Theory

Optimal Risk Sharing with Different Reference Probabilities

Rational Expectations Equilibrium in Economies with Uncertain Delivery

EC476 Contracts and Organizations, Part III: Lecture 2

Pareto optimal allocations and optimal risk sharing for quasiconvex risk measures

Jaroslav Pazdera, Johannes Schumacher and Bas Werker Intra-group Risk Sharing under Financial Fairness

A Characterization of Polyhedral Convex Sets

Introduction to General Equilibrium

Klaus Kultti Hannu Vartiainen Bargaining with Many Players: A Limit Result. Aboa Centre for Economics

Birgit Rudloff Operations Research and Financial Engineering, Princeton University

KIER DISCUSSION PAPER SERIES

Noncooperative Games, Couplings Constraints, and Partial Effi ciency

Interval-Valued Cores and Interval-Valued Dominance Cores of Cooperative Games Endowed with Interval-Valued Payoffs

Game Theory and Economics of Contracts Lecture 5 Static Single-agent Moral Hazard Model

Homework #6 (10/18/2017)

Could Nash equilibria exist if the payoff functions are not quasi-concave?

Introduction to game theory LECTURE 1

Nash Bargaining in Ordinal Environments

Characterization of Upper Comonotonicity via Tail Convex Order

Worst Case Portfolio Optimization and HJB-Systems

A Characterization of Robust Sunspot Equilibria

Risk Aversion over Incomes and Risk Aversion over Commodities. By Juan E. Martinez-Legaz and John K.-H. Quah 1

1: PROBABILITY REVIEW

Transcription:

EFFICIENT, WEAK EFFICIENT, AND PROPER EFFICIENT PORTFOLIOS MANUELA GHICA We give some necessary conditions for weak efficiency and proper efficiency portfolios for a reinsurance market in nonconvex optimization problems. AMS 2000 Subect Classification: 62P05, 91B30, 90A12. Key words: reinsurance market, portfolio, efficiency. 1. INTRODUCTION The reinsurance problem appears at first sight to be a problem which can be analyzed in terms of classical economic theory, if the obectives of the companies have been formulated in an operational manner by means of Bernoulli s utility concept: the expected gain must not be maximized, but the expected utility of the gain [3]. However, closer investigations show that the economic theory is relevant only part of the way. Then the problem becomes a problem of cooperation between parties that have conflicting interests and are free to form and break any coalitions which may serve their particular interests [12]. Classical economic theory is powerless when it comes to analyze such problems. In the last decades it was shown that there are many possibilities to study and explain the apparently chaotic situation by using game theory or convex analysis. In this paper we use and extend results from convex analysis established by Rockafellar [13] and Kaliszewski [11]. A basic result of convex analysis is the Fundamental Theorem on Convex Functions [4], [13] which states that an efficient solution of a convex vector optimization problem necessarily minimizes a linear combination of obective functions. Kaliszewski characterized efficient solutions without assuming convexity. All these results help us to establish some new results for portfolios in a reinsurance market in nonconvex optimization problems. If we see N = {1, 2,..., n} as a group of n reinsurers, having preferences i, i N, over a suitable set of random variables denoted by R, or gambles with realizations (outcomes) in some A R, we represent these preferences MATH. REPORTS 10(60), 4 (2008), 309 315

310 Manuela Ghica 2 by von Neumann-Morgenstern expected utility, meaning that there is a set of continuous utility functions u i : R R such that X i Y if and only if Eu i (X) Eu i (Y ), where E stands for the mean operator. We assume monotonic preferences, and risk aversion, so that we have u i (w) > 0, u i (w) 0 for all w in the relevant domains [5]. In some cases we shall also require strict risk aversion, meaning strict concavity for some u i. For a better understanding we assume that each agent is endowed with a random variable payoff X i called initial portfolio. More precisely, there exists a probability space (Ω, K, P ) such that we have the payoff X i (ω) when ω Ω occurs and both expected values and variances exist for all these initial portfolios, which means that all X i L 2 (Ω, K, P ) [6]. Because every agent can negotiate any affordable contracts, we will have a new set of random variables Y i, i N, representing the final portfolios. We say if n i=1 Z i n i=1 X i = X N then the allocation Z = (Z 1, Z 2,..., Z n ) is said to be feasible [1]. The paper is organized as follows. In Section 2 we give an alternative theorem for a basic result in the nonconvex framework. In Sections 3 and 4 we characterize efficiency, weak efficiency and proper efficiency of portfolios for a reinsurance market in nonconvex optimization problems by different necessary conditions related to the alternative theorem given in Section 2. 2. AN ALTERNATIVE TYPE THEOREM In this section we extend a fundamental theorem on convex functions to the case when the convexity assumption does not hold. This alternative type theorem enables us to characterize the efficiency of portfolios for a reinsurance market. Theorem 1. Let Z be the set of feasible allocations. Then one and only one of the following alternative holds: (a) Z i Z such that Eu i (Z i ) < 0, i N; (b) for any negative numbers δ 1, δ 2,..., δ n there exist positive numbers λ 1, λ 2,..., λ n such that maxλ i (Eu i (Z i ) δ i ) 1, Z Z. i Proof. Suppose (a) holds. Let Z Z such that Eu i (Z i ) < 0 for i N. Take δ i R, i N, with δ i max Eu i (Z i ). i Thus, Eu i (Z i ) δ i < 0, for any i 1, n, and so (b) cannot hold. Now, suppose (a) does not hold. Let Z Z. Then there exists at least one N such that Eu (Z ) 0. There fore, we have max Eu (Z ) 0.

3 Efficient, weak efficient, and proper efficient portfolios 311 max Let δ 1, δ 2,..., δ n be any negative numbers and λ i = (δ i ) 1, i N. Hence λ i Eu i (Z i ) 0. This inequality is equivalent to max [ λ i (Eu i (Z i ) δ i ) 1] 0, i.e., max λ i (Eu i (Z i ) δ i ) 1. Thus, (b) holds. Consider the vector problem (1) min Z Z (Eu 1(Z 1 ), Eu 2 (Z 2 ),..., Eu n (Z n )), where min denotes the operation of deriving all efficient portfolios. 3. EFFICIENT AND WEAK EFFICIENT PORTFOLIOS In this section we first define efficiency and weak efficiency of portfolios for a reinsurance market in nonconvex optimization problems. Then we give a necessary condition for weak efficiency and show that it is a consequence of Theorem 1. Let Z be the set of feasible allocations. Definition 2. Y Z is said to be an efficient portfolio (or a Pareto portfolio) if for Z Z, it follows from Eu i (Z i ) Eu i (Y i ), i N, that Eu i (Z i ) = Eu i (Y i ), i N. Definition 3. Y Z is a weakly efficient portfolio (or a Slater portfolio) if there is no Z Z such that Eu i (Z i ) < Eu i (Y i ), i N. Theorem 4. Let Y Z be a weakly efficient portfolio. Then it solves the problem min ϕ(z), Z Z where ϕ(z) = max λ i (Eu i (Z i ) yi ), with λ i = (Eu i (Y i ) yi i N ) 1, i N and yi is any real number such that λ i > 0 for every i N. Proof. Let Y Z be a weakly efficient portfolio. Then the system of inequalities Eu i (Z i ) < Eu i (Y i ), i N, Z Z has no solution. Now, by Theorem 1, for any negative numbers ε 1, ε 2,..., ε n we have maxλ i (Eu i (Z i ) Eu i (Y i ) δ i ) 1, Z Z, i where λ i = ( δ i ) 1, i N. Take δ 1, δ 2,..., δ n and y1, y 2,..., y n such that δ i = yi Eu i (Y i ) < 0, i N. Then λ i = (Eu i (Y i ) yi ) 1 > 0 and max λ i (Eu i (Z i ) yi ) 1 for every Z Z. i N

312 Manuela Ghica 4 We see that λ i (Eu i (Y i ) yi ) = 1 for any i N. Hence max λ i (Eu i (Z i ) yi ) max λ i (Eu i (Y i ) yi ), Z Z, i N i N i.e., Y solves the problem min max λ i (Eu i (Z i ) yi ) Z Z i N and the theorem is proved. 4. PROPERLY EFFICIENT PORTFOLIOS In this section we define properly efficient portfolios and then give two characterization theorems. First we prove that a properly efficient portfolio is a solution of problem (4), as a consequence of Theorem 1. The last theorem, closely related to Theorem 5, cannot be derived from Theorem 1, but follows directly from Lemma 6. Definition 5. Y Z is said to be a properly efficient (or a Geoffrian portfolio) portfolio for (1) if Y is an efficient portfolio for (1) and there exists a real number M > 0 such that for each i N we have Eu i (Y i ) Eu i (Z i ) Eu (Z ) Eu (Y ) M for some such that Eu (Y ) > Eu (Z ) whenever Z Z. Lemma 6. Let Y be a properly efficient portfolio for (1). Then the system of inequalities (2) α i Eu i (Z i ) + ρ N Eu (Z ) < α i Eu i (Y i ) + ρ N Eu (Y ), where α 1, α 2,..., α n R +, has no solutions in Z for some ρ > 0. Proof. Suppose Y is a properly efficient portfolio for (1). Then Y also is a weakly efficient portfolio for (1). Hence the system of inequalities Eu i (Z i ) < Eu i (Y i ), i N, Z Z, has no solution. Let α 1, α 2,..., α n R +. Then the system α i Eu i (Z i ) < α i Eu i (Y i ), i N, Z Z has no solution. Suppose that system (2) is consistent. Let Ẑ Z, Ẑ Y with α i Eu i (Ẑi) + ρ i We consider two cases. Eu i (Ẑi) < α i Eu i (Y i ) + ρ i Eu i (Y i ).

5 Efficient, weak efficient, and proper efficient portfolios 313 Case 1. Eu (Y ) Eu (Ẑ). If we compare the last two inequalities, we deduce that Eu i (Ẑi) + ρ N Eu (Ẑ) < Eu i (Y i ) + ρ NEu (Y ), i N has no solutions. Case 2. Eu (Y ) > Eu (Ẑ). We have an efficient portfolio Y. Then there exists l, 1 l n, such that Eu l (Y l ) Eu l (Ẑl) and Eu l (Ẑl) Eu l (Y l ) = max (Eu i(ẑi) Eu i (Y i )). Because i N Y is a properly efficient portfolio, there exists a number M > 0 such that for any i, 1 i n, we have Eu i (Y i ) > Eu i (Ẑi), Eu i (Y i ) Eu i (Ẑi) M ( Eu (Ẑ) Eu (Y ) ) for some such that Eu (Ẑ) > Eu (Y ). When Eu i (Y i ) Eu i (Ẑi), we have Eu i (Y i ) Eu i (Ẑi) M ( Eu l (Ẑl) Eu l (Y l ) ). Therefore, ( Eui (Y i ) Eu i (Ẑi) )( Eu l (Ẑl) Eu l (Y l ) ) 1 M(k 1), where is taken over all i such that Eu i (Y i ) Eu i (Ẑi) > 0. Let 0 < ρ ρ 0, [( ) 1. where ρ 0 = min α i M(k 1)] Then we have 1 i n [ ) ( ) ] 1 1 ρ α l (Eu i (Y i ) Eu i (Ẑi) Eu l (Ẑl) Eu l (Y l ) i.e., ) < α l (Eu [ ( ) ] 1 l (Ẑl) Eu l (Y l ) Eu (Y ) Eu (Ẑ), ρ ( ) ( ) Eu (Y ) Eu (Ẑ) < α l Eu l (Ẑl) Eu l (Y l ) and α l Eu l (Y l ) + ρ Eu (Y ) < α l Eu l (Ẑl) + ρ Eu (Ẑ). This means that the system of inequalities (3) α i Eu i (Ẑi) + ρ Eu (Ẑ) < α i Eu i (Y i ) + ρ (Y ) N NEu [( ) 1. is inconsistent for 0 < ρ min α i M(k 1)] 1 i n

314 Manuela Ghica 6 Theorem 7. Let Y be a properly efficient portfolio. Then Y is an optimal solution for the problem (4) min Z Z ϕ(z) (Eui for some ρ > 0, where ϕ(z) = max i{ λ (Z i ) yi i N with real numbers yi such that { (Eui λ i = (Z i ) yi ) ( + ρ Eu (Z ) y ) } 1 ) ( + ρ Eu (Z ) y ) } > 0, i = 1, k. Proof. For a properly efficient portfolio Y and α 1 = α 2 = = α n = 1, by Lemma 6, system (2) has no solution for some ρ > 0. Hence, by Theorem 1, for any negative numbers ε 1, ε 2,..., ε n we have max λ i {Eu i (Z i ) Eu i (Y i ) + ρ [(Eu (Z ) Eu (Y )) δ i ] for all Z Z, where λ i = ( δ i ) 1, i = 1, n, and { δ i = yi Eu i (Y i ) + ρ ( Eu (Y ) y ) } 1, i = 1, n. Hence ϕ(z) 1 for all Z Z. Since for any i = 1, n { (Eui λ i (Y i ) yi ) ( + ρ Eu (Y ) y ) } = 1, we obtain ϕ(z) ϕ(y ) for all Z Z, i.e., Y is an optimal solution for (4). } 1, Theorem 8. If Y is a properly efficient portfolio for (1) then Y solves (4) for some ρ > 0, where λ i = (Eu i (Y i ) y i ) 1, i = 1, k, and y i are real numbers such that λ i > 0, i = 1, n. Proof. Let α i = (Eu i (Y i ) yi ) 1, i = 1, n. Since systems (2) and (3) have no solutions and Ẑ is arbitrary we have ϕ (Z) ϕ (Y ), Z Z, where [ ϕ (Z) = max α i (Eu i (Z i ) yi ) + ρ ( Eu (Z ) y ) ], since ( Eu (Y ) y ). ϕ (Y ) = 1 + ρ The theorem holds with λ i = α i, i N.

7 Efficient, weak efficient, and proper efficient portfolios 315 REFERENCES [1] K.K. Aase, Equilibrium in a reinsurance syndicate: existence, uniqueness and characterization. ASTIN Bull. 22 (1993), 185 21. [2] K.K. Aase, Perspectives of risk sharing. Scand. Actuar. J. 2002, 2, 73 128. [3] K.J. Arrow, The theory of risk-bearing: small and great risks. J. Risk Uncertainty 12 (1996), 103 111. [4] C. Berge and A. Ghouila-Houri, Programming, Games and Transportation Networks, Wiley, New York, 1965. [5] K. Borch, Recent developments in economic theory and their application to insurance. ASTIN Bull. 2 (1963), 322 341. [6] K. Borch, The theory of risk. J. Roy. Statist. Soc. 29 (1967), 432 453. [7] H. Gerber, On additive premium calculation principles. ASTIN Bull. 7 (1974), 215 222. [8] M. Ghica, A risk exchange model with a mixture exponential utility function. An. Univ. Bucureşti Mat. Inform. 55 (2006), 169 176. [9] M. Ghica, The core of a reinsurance market. Math. Rep. (Bucur.) 10(60) (2008), 155 164. [10] M. Ghica, Optimal Portfolios in a Reinsurance Market. Ph. Thesis, Bucharest Univ., 2008. [11] I. Kaliszewski, A theorem on nonconvex functions and its application to vector optimization. European J. Oper. Res. 80 (1995), 439 449. [12] V. Preda, Teoria deciziilor statistice. Ed. Academiei Române, Bucureşti, 1992. [13] T. Rockafellar, Convex Analysis. Princeton Univ. Press, Princeton, N.Y., 1970. Received 12 May 2008 Spiru Haret University Faculty of Mathematics and Computer Science manuela.ghica@gmail.com