OR 2006 Karlsruhe Welfare Economy under Rough Sets Information Takashi Matsuhisa Ibaraki National College of Technology Ibaraki 312-8508, Japan E-mail: mathisa@ge.ibaraki-ct.ac.jp September 8, 2006 1 Background Economy under uncertainty consists of Economy: Trader set, Consumption set, Utility functions Uncertainty 1. By Exact set information: Partition structure on a state-space, or equiv. Knowledge structure. 2. By Rough sets information: Non-partition structure on a state-space, or equiv. Belief structure. 2 Aim and Scope Economy under Exact Sets Information 1. Core equivalence theorem: There is no incentive among all traders to improve their equilibrium allocations. Fundamental Theorem for Welfare Economy: Each Pareto optimal allocation is an equilibrium allocation. Others; e.g., No trade theorem: There is no trade among traders if the initial endowments are an equilibrium. 2. 3. Economy under Rough Sets Information Can we extend these results into the economy There are a few extensions of “No trade.” We extend the welfare theorem. 3 Purpose 1. “Rough sets” information structure induced from a belief structure 2. Economy with belief structure and expectation equilibria in belief 3. Characterization of the extended equilibria by Ex-post Pareto optimal allocations in traders. 4 Chronicle of Extensions Author(s) Aumann Result (1962) Core equiv Geanakoplos (1989) No Trade Einy et al (2000) Core equiv Matsuhisa and Core equiv Ishikawa (2005) Matsuhisa (2005) Welfare Economy Uncertainty Information sets × Pt ()(Exact set) ○ Pt := non Partition ○ Pt := Partition(Exact set) ○ Pt := non Partition ○ (Ref, Trn: Rough set) (Ref: Rough set) Pt := non Partition(None: Rough set) 5 Outline Belief structure and Rough sets information Economy on belief Expectations equilibrium in Belief Fundamental Theorem for Welfare Remark 6 Economy under Uncertainty 〈T, S,m,W, e, (Ut)t∈T, (πt)t∈T, (Pt)t∈T,〉 l : the number of commodities R+l : the consumption set of trader t T: a finite set of traders t∈T e : T×W R+l : an initial endowment Ut : R+l×W→R : t’s utility function πt : subjective prior on W for t∈T Pt : partition on Wwhich represents trader t’s uncertainty 7 Economy on Belief 〈T, W, e, (Ut)t∈T, (πt)t∈T, (Bt)t∈T, (Pt)t∈T〉 l : the number of commodities R+l : the consumption set T : a finite set of traders t e : T×WR+l : an endowment Ut : R+l×W→R : t’s utility function initial πt : subjective prior on Wfor t∈T 〈 W, (Bt)t∈T, (Pt) t∈T 〉: the belief structure 8 Belief structure 〈 W, (Bt)t∈T, (Pt)t∈T 〉 W : a non-empty finite set of states 2 W∋E : an event T : a set of traders E ∋: “E occurs at ” 9 Belief structure 〈 W, (Bt)t∈T, (Pt)t∈T 〉 t’s belief operator Bt : 2 W → 2 W Bt E ∋ : “t believes E at ” t’s possibility operator Pt : 2 W → 2 W,E → Pt(E):= W∖ Bt (W∖ E) Pt E ∋ : “E is possible for t at ” Pt():= Pt({) : t’s information set at 10 Livedoor v.s. Fuji TV Japan L F 11 L-F Example T = { L, F } 1 = L does not commit the injustice W= {1 , 2 } 2 = L commits the injustice Belief structure: E BL E φ φ {1} {1} {2} {2} W W BF E φ {1} φ W 12 L-F Example T = { L, F } The possibility operators E PL E φ φ {1} {1} {2} {2} W W PF E φ W {2} W The Information Sets: Pt()= Pt({}) 1 2 PL 1 2 PF 13 Rough Set Theory An event E is exact if Pt(E) = Bt (E) An event E is rough if Pt(E) ≠ Bt(E) If 〈W, (Bt )〉 is the Kripke semantics for Modal logic S5 then {Pt()|∈W} is a partition of W,and every Pt() is exact. Our interest is the case that Pt() does not make a partition, and so Pt() is rough in general. 14 Economy on Belief 〈 T, S, m, W, e, (Ut)t∈T, (πt)t∈T, (Bt)t∈T, (Pt)t∈T〉 (A-1) Se (t, ) ≩ 0 t∈T Dom (Pt) := { | Pt(ω) ≠ φ } = the domain of Pt (A-2) For ∀t, Dom (Pt) = Dom (Ps) ≠ φ 15 Allocations An assignment x : T×W R+l An allocation a : T×WR+l Sa (t, ) ≦ S e (t, ) t ∈T t ∈T 16 Price and Budget Price system p : WR+l ≠0 ⊿(p) = the partition of W induced by p; ⊿(p)() = { x | p(ξ) = p() } = “the information given by p at ” Budget set of t at Bt(, p) = { x | p()x ≦ p()e(t, ) } 17 Expectations in belief t’s interim expectation Et[Ut(x (t, * )) | ⊿(p)∩Pt ]() := ∑Ut(x (t, x),x))πt({x} | ⊿(p) ()∩Pt()) x∈Dom( Pt ) t’s ex-ant expectation Et[Ut(x (t, * )]():= ∑Ut(x (t, x),x))πt({x}) x∈Dom( Pt ) 18 Expectation equilibrium in belief (p, x) : = an expectations equilibrium in belief (EE1) x(t, ) ∈ Bt(, p) (EE2) y(t, ) ∈ Bt(, p) if ⇒ Et[Ut(x(t, *))|⊿(p)∩Pt]() ≧ Et[Ut(y(t, *))|⊿(p)∩Pt]() (EE3) Sx(t, ) = Se(t, ) t∈T t∈T 19 Existence Theorem for EE Trader t is risk averse if: (A-3) Ut(x , ∙) = ‘‘strictly increasing, quasi concave on R+l, etc’’ Measurability of Utility: (A-4) Ut(x , ∙) = ‘‘measurable for the finest field generated by Pt for all t ∈ T ’’ Theorem 1: Economy on belief with (A-1), (A-2), (A-3) and (A4). There exists an expectation equilibrium. 20 Question Question : What’s characteristics of the expectations equilibrium in belief? Answer 1. Welfare theorem: The expectations equilibrium is an ex-ante Pareto-optimal Answer 2. Core equivalence: The expectations equilibrium is a core allocation, and vice versa. 21 Pareto Optimality An allocation a = ‘‘ex-ante Pareto optimal’’ if there is no allocation x such that (1) ∀t∈T Et[Ut(x (t, * ))] ≧ Et[Ut(a (t, * ))] (2) ヨs∈T Es[Us(x (s, * ))] >Es[Us(a (s, * ))] 22 Welfare Theorem Economy with belief structure: (A-1), (A-2), (A-3), and (A-4) An allocation a = ‘‘ex-ante Pareto optimal’’ For ∃p = price, ⇔ (p, a) = ‘‘an expectations equilibrium’’ for some initial endowments. 23 Concluding remark Propose an extended economy under rough sets information. Emphasize with the epistemic aspect of belief of the traders Remove out: Partition structure of traders’ information. Extend Fundamental Theorem for Welfare. Bounded rationality point of view :The relaxation of the partition structure for player’s information can potentially yield important results in a world with imperfectly Bayesian agents 24 Thank you! Danken ! 25