Introduction The Model Equilibrium Welfare Long-Run Equilibrium Advanced Microeconomics Partial and General Equilibrium Giorgio Fagiolo giorgio.fagiolo@sssup.it http://www.lem.sssup.it/fagiolo/Welcome.html LEM, Sant’Anna School of Advanced Studies, Pisa (Italy) Part 2 Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Industry Partial Equilibrium Analysis Studying a small part of the overall economy One good (G) vs other (L − 1) markets Industry of commodity (G) is small Basic Assumptions Consumers spend a small (negligible) part of their total income for G When consumer wealth increases, demand for G does not increase (no wealth effect) Substitution effects are negligible and dispersed: when pG increases, no effects on demand for other commodities Prices of other commodities can be considered as given (L − 1) goods as composite commodity (money or numeraire) Formally: Consumers have quasi-linear preferences Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Quasi-Linear Preferences Consider an economy with L = 2 commodities. The representative consumer holds a preference relation with associated utility function: u(x, m) = φ(x) + m where x ≥ 0 is the consumption of commodity 1 and m ∈ R is the consumption of commodity 2. We also assume that φ0 > 0, φ00 < 0 and that φ(0) = 0. Commodity 1 is a consumption good, while commodity 2 can be defined as ’everything else’, i.e. money left apart by the consumer for purchasing all other goods, after having made the optimal choice for commodity 1. The latter can be referred to as the ’consumption good’, while commodity 2 is the ’numeraire’. Thus, we normalize prices so that px = p > 0 and pm = 1. Notice that we allow ’money’ to be negative. Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Quasi-Linear Preferences: No Wealth Effects The most important feature of the quasi-linear preference is that an increase in consumer’s income has no effect on the demand of the consumption good. To see that let’s solve the consumer program: max φ(x) + m, s.t. px + m = y x≥0,mR for an income’s level y . The Lagrangean is: L(x, m, λ) = φ(x) + m − λ[y − px − m] and FOCs for an interior solution are: φ0 (x(p, y )) ≡ p p · x(p, y ) + m(p, y ) ≡ y Notice that FOCs are also sufficient because φ00 < 0. Introduction The Model Equilibrium Welfare Long-Run Equilibrium Quasi-Linear Preferences: No Wealth Effects Differentiating FOCs with respect to y gives: ∂x(p, y ) ∂y ∂x(p, y ) ∂m(p, y ) p· + ∂y ∂y φ00 (x(p, y )) · = 0 = 1 which hold if and only if: ∂m(p, y ) ∂x(p, y ) = 0 and = 1. ∂y ∂y Hence an increase in y has no effect on the demand of x and all wealth increase has a 1:1 effect on the numeraire. Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Quasi-Linear Preferences and UPS Consider an economy with I consumers, each one holds a quasilinear utility function ui (xi , mi ) = φi (xi ) + mi , i = 1, ..., I, an initial endowment of the numeraire ωmi : Σi ωmi = ωm > 0 and an initial endowment of the consumption good ωxi : Σi ωxi = ωx > 0. Then the utility possibility set (UPS) is defined as: X X U = {(u1 , ..., uI ) ∈ RI : ui ≤ φi (xi ) + mi , mi = ωm , xi = ωx } i i that is the set of all attainable utility levels by consumers given resource constraints. Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Quasi-Linear Preferences and UPF By summing up the inequalities ui ≤ φi (xi ) + mi over all consumers and by using the resource constraint, one gets: Σi ui ≤ Σi φi (xi ) + ωm Hence, given any feasible allocation, the utility possibility frontier (UPF) reads: UPF = {(u1 , ..., uI ) ∈ RI : Σi ui = max{Σi φi (xi ) + ωm , Σi xi = ωx }} is an hyperplane in the utility space and all points in this boundary are associated with consumption allocations that differ only in the distribution of the amount of the numeraire ωm among the I consumers. Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Quasi-Linear Preferences: UPS and UPF Example: I = 2 and φi (xi ) = log xi . Then: U = {(u1 , u2 ) ∈ R2 : u1 ≤ log x1 + m1 , u2 ≤ log x2 + m2 , m1 + m2 = ωm , x1 + x2 = ωx } This implies that x2 = ωx − x1 , m2 = ωm − m1 and m1 ≤ log(ωx − x1 ) + ωm − u2 . Hence: U = {(u1 , u2 ) ∈ R2 : u1 + u2 ≤ log[x1 (ωx − x1 )] + ωm , 0 ≤ x1 ≤ ωx } whose boundary is the line u1 + u2 = log[x1 (ωx − x1 )] + ωm . Thus, by changing the demand of the consumption good by consumer 1, one simply shifts the boundary in a parallel manner. It is intuitive that when the allocation is Pareto Efficient, the set U will be extended as far out as possible. In this case, as both consumers have the same preferences over the consumption good, the Pareto efficient allocation is (ωx /2, ωx /2), which is the value of x1 which maximizes log[x1 (ωx − x1 )]. Hence, the boundary of U (i.e. the UPF) in this particular example is given by: Bd(U) = {(u1 , u2 ) ∈ R2 : u1 + u2 = log[ωx2 /4] + ωm }. Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions The Model Two-good economy Good 1: Commodity under study `; price: p` = p; agent i’s consumption: xi Good 2: Numeraire (money) m; price: pm = 1; agent i’s consumption: mi Consumers: i = 1, . . . , I Utility: ui (xi , mi ) = φi (xi ) + mi , mi ∈ R, xi ∈ R+ Assumptions: φ0i > 0, φ00 i < 0, φi (0) = 0 Firms: j = 1, . . . , J Employ good m to produce `. Since pm = 1, if firm j uses zj units of m to produce `, then its cost is zj Production sets: Yj = {(−zj , qj ) : qj ≤ fj (zj ), qj ≥ 0} = {(−zj , qj ) : zj ≥ cj (qj ), qj ≥ 0}, where fj0 > 0, cj0 > 0 and cj00 > 0 (strict convexity) NB: Since J is fixed, we are in the SR! Endowments and Shares Good `: no endowments ω` = 0 (must be produced) Good m: ωmi , i = 1, . . . , I, ωm = Σi ωmi Each consumer i owns a share θij of firm j, Σi θij = 1 Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Supply Behavior artial Equilibrium Competitive Analysi Competitive equilibria: � At price p ∗ , firm j’s equilibrium output level qj∗ must solve max p ∗ qj − cj (qj ) qj ≥0 with necessary and sufficent FOC: p ∗ ≤ cj� (qj ), with equality if qj∗ > 0 � consumer i’s equilibrium consumption vector (mi∗ , xi∗ ) must solve max m ∈R mi + φi (xi ) Introduction with necessary and sufficent FOC:Welfare The Model Equilibrium Long-Run Equilibrium Conclusions Demand Behavior p ∗ ≤ cj� (qj ), with equality if qj∗ > 0 � consumer i’s equilibrium∗ consumption vector (mi∗ , xi∗ ) must Each consumer i given p and πj∗ = p∗ qj∗ − cj (qj∗ ) solves: solve max mi ∈R mi + φi (xi ) xi ∈R+ � J ∗ ∗ ∗ )) mi + i ≤ ωmi + Partial s.t. Equilibrium Competitive j=1 θij (p qj − cj (qjAnalysis p∗x Competitive equilibria: � . . . since the budget constraint must hold with equality: J � max φi (xi ) − p ∗ xi + ωmi + θij (p ∗ qj∗ − cj (qj∗ )) xi ≥0 j=1 with necessary and sufficent FOC: φ�i (xi∗ ) ≤ p ∗ , with equality if xi∗ > 0 Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Equilibrium It is an allocation (x1∗ , . . . , xI∗ ; q1∗ , . . . , qJ∗ ) and a price (scalar) p∗ such that: 1 All firms maximize profit given technology and p∗ , i.e. cj0 (qj∗ ) ≥ p∗ (= if qj∗ > 0) 2 All consumers maximize utility s.t. budget constraint given p∗ and firms’ optimal profits, i.e. φ0i (xi∗ ) ≤ p∗ (= if xi∗ > 0) Market for good ` clears: Σi xi∗ = Σj qj∗ 3 Remarks Since all utilities satisfy non-satiation, all budget constraints are satisfied with equality and therefore all markets must clear (with equality) Check that mi∗ and zj∗ can be recovered using BC and production function Does equilibrium condition imply that market for numeraire also clears? Check that using budget constraints. . . X X X X X zj∗ = ωmi mi∗ + qj∗ = xi∗ ⇒ j i i j i Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Existence Does an interior equilibrium always exist? No! Sufficient condition: If maxi φ0i (0) > minj cj0 (0) then an interior equilibrium Σi xi∗ = Σj qj∗ > 0 does exist Why? Suppose not, then there exists an equilibrium such that Σi xi∗ = Σj qj∗ = 0, meaning that xi∗ = qj∗ = 0 for all i, j. From conditions (1) and (2) we have: φ0i (0) ≤ cj0 (0). This implies: maxi φ0i (0) ≤ minj cj0 (0), which contradicts the assumption. Consequences In any interior solution: φ0i (xi∗ ) = p∗ = cj0 (qj∗ ). Thus equilibrium price: is equal to firm marginal benefit in selling one additional unit of good (i.e. firm marginal cost in producing one additional unit) is equal to consumer marginal cost in buying one additional unit of good (i.e. consumer marginal benefit in consuming one additional unit of good) Conditions 1-3 not affected by the distribution of endowments and ownership shares (because of quasi-linearity of utility functions) Introduction The Model Equilibrium Welfare Long-Run Equilibrium Aggregate Supply and Aggregate Demand What does all that have to do with the traditional demand=supply scheme? Consumer behavior: xi∗ = xi (p) for any given level p Firm behavior: qj∗ = qj (p) for any given level p Market clearing condition ⇒ aggregate demand = aggregate supply Aggregate Supply and Aggregate Demand S(p) = Σj qj (p), increasing in p if p > minj cj0 (0) D(p) = Σi xi (p), decreasing in p if p < maxi φ0i (0) Conclusions Introduction Aggregate Supply Function for Long-Run Good � The Model Equilibrium Welfare Equilibrium AS vs. AD: Graphical Construction Aggregate Demand Function for Good � Conclusions Introduction The Model AS vs. AD: Equilibrium Equilibrium Welfare Long-Run Equilibrium Conclusions Introduction The Model Equilibrium Welfare Partial Equilibrium: When does it fail to exist? See exercise! Long-Run Equilibrium Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Comparative Statics How does a change in underlying conditions affect the equilibrium outcome? exogenous parameters α ∈ R M affecting consumer’s preferences: φi (xi , α) exogenous parameters β ∈ R S affecting firm’s technology: cj (qj , β) exogenous tax and subsidy parameters t ∈ R K affecting price: p̂i (p, t) (paid by consumers) and p̂j (p, t) (received by firms) Typical exercise: the equilibrium allocation and price are functions of (α, β, t) if functions are differentiable, then the implicit function theorem can be used to derive the marginal change in equilibrium allocation and price in response to a differential change in parameters Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Comparative statics effects of a sales tax Suppose to introduce a sales tax such that consumers must pay t ≥ 0 for each unit of good ` Price received by producers is p(t) Price paid by consumers is p(t) + t Equilibrium condition: D(p(t) + t) ≡ S(p(t)) Differentiating and solving for p0 (t): d(p(t)) D 0 (p(t) + t) = p0 (t) = − 0 ∈ [−1, 0) dt D (p(t) + t) − S 0 (p(t)) d(p(t) + t) = p0 (t) + 1 ∈ [0, 1) dt Total quantities produced and consumed fall But how much this change is felt by consumers and producers mainly depends on the steepness of supply function: the steeper S, the more rigid production to prices, the more producers feel the burden of the sale tax Introduction The Model Equilibrium Welfare Long-Run Equilibrium Comparative statics effects of a sales tax: Three cases The larger |S 0 (p(t))|, the flatter the S schedule If |S 0 (p(t))| very large (i.e. supply very elastic, i.e. horizontal) then S is flat and all the burden is felt by consumers If |S 0 (p(t))| is very small (i.e. supply very rigid, i.e. vertical) then S is vertical and all the burden is felt by producers Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Main questions 1 Suppose that for a given endowment distribution and owner shares the market mechanisms reaches an equilibrium allocation (x1∗ , . . . , xI∗ ; q1∗ , . . . , qJ∗ ) and a price (scalar) p∗ . Is that allocation Pareto optimal? 2 Suppose we start from a Pareto optimal allocation (x1o , . . . , xIo ; q1o , . . . , qJo ). Is there a redistribution of initial endowments that allows to sustain that allocation as a market equilibrium? I.e. is there a way to redistribute initial endowments in such a way that a given PO allocation will be reached by market forces without any additional intervention? ∗ ∗ The allocation (x )i , (y )j maximizes social welfare for some choice of weights λ∗ Convexity The price-allocation {p∗ , (x∗ )i , (y ∗ )j } is a competitive equilibrium ? ∗ ∗ The allocation (x )i , (y )j is a Pareto optimum Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions The Utility Possibility Set (UPS) The UPS for the partial-equilibrium economy is defined as U = {(u1 , . . . , uI ) : I X i=1 ui ≤ I X i=1 φi (xi ) + ωm − J X cj (qj )} j=1 for P any given P feasible allocation (x1 , . . . , xI ; q1 , . . . , qJ ), i.e. such that i xi = j qj . Recall: The boundary of the UPS (i.e. the UPF) is in a 1:1 relation with all suchPthat u ∈ PIPO allocations. PI Therefore all allocations PJ P{(u1 , . . . , uI ) : i=1 ui = max{ i=1 φi (xi ) + ωm − j=1 cj (qj ), i xi = j qj }}, are PO. Recall: When consumer preferences are quasilinear, the boundary of the economy’s utility possibility set is linear, and all points in this boundary are associated with consumption allocations that differ only in the distribution of the numeraire (ωi1 , . . . , ωIm ) s.t. ωm = Σi ωim Introduction The Model Equilibrium The Utility Possibility Set (UPS) Welfare Long-Run Equilibrium Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Partial Equilibrium Analysis: First Welfare Theorem Theorem If the price p∗ and the allocation (x ∗ , q ∗ ) = (x1∗ , . . . , xI∗ , q1∗ , . . . , qJ∗ ) constitutes a competitive (Walrasian) equilibrium, then (x ∗ , q ∗ ) is Pareto optimal Proof. We prove it for the case of an interior allocation. An allocation (x1+ , . . . , xI+ , q1+ , . . . , qJ+ ) is Pareto optimal if given the set U, it lies in the boundary of the UPS, i.e. if it maximizes I X φi (xi ) + ωm − i=1 subject to feasibility constraints, i.e. J X cj (qj ) j=1 P i xi = P j qj . Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Partial Equilibrium Analysis: First Welfare Theorem Theorem If the price p∗ and the allocation (x ∗ , q ∗ ) = (x1∗ , . . . , xI∗ , q1∗ , . . . , qJ∗ ) constitutes a competitive (Walrasian) equilibrium, then (x ∗ , q ∗ ) is Pareto optimal Proof (Cont’d). FOCs for this problem read (in an interior solution): φ0i (xi+ ) = µ = cj0 (qj+ ) X i xi+ = X qj+ j Note that FOCs are sufficient because φ00i < 0 and cj00 > 0. Thus an allocation that solves FOCs is PO. But FOCs are equivalent to those for an interior competitive equilibrium when µ = p∗ . Therefore, if p∗ and (x1∗ , . . . , xI∗ , q1∗ , . . . , qJ∗ ) satisfy FOCs for µ = p∗ , they also satisfy sufficient conditions for PO. Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Partial Equilibrium Analysis: Second Welfare Theorem Theorem For any Pareto optimal levels of utility (u1∗ , . . . , uI∗ ) ∈ PUPF , there are transfers of the numeraire commodity (T1 , . . . , TI ) satisfying i Ti = 0, such that a competitive equilibrium reached from the endowments (ωm1 + T1 , . . . , ωmI + TI ) yields precisely the utilities (u1∗ , . . . , uI∗ ) Argument: consumption levels, production levels and firms’s profits are unaffected by changes in consumers’ wealth levels; thus, ex-ante transfers of the numeraire causes each equilibrium consumption of the numeraire to change by exactly the amount of the transfer; hence, they allow to reach any utility vector in the boundary of the utility possibility set Note: convexity assumptions about technologies and preferences are central to show this result! Introduction The Model Equilibrium Welfare Long-Run Equilibrium Equilibrium and Welfare ∗ ∗ The allocation (x )i , (y )j maximizes social welfare for some choice of weights λ∗ Convexity The price-allocation {p∗ , (x∗ )i , (y ∗ )j } ∗ ∗ The allocation (x )i , (y )j is a Pareto optimum Co nv ex ity is a competitive equilibrium Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Free-Entry and Long-Run Competitive Equilibria So far: Short run competitive equilibria Number of firms J is fixed Firms hold heterogeneous technologies and cannot change them Long run: free entry/exit model Let J (number of firms) in the market be endogenous: firms can freely enter/exit in response to π opportunities: a firm will enter if it can earn a positive profit upon entry; it exits if it gets negative profits at the current price p for any q > 0 There is a potentially infinite set of firms all having access to technology c(q) s.t. c(0) = 0 (no sunk costs in the LR) Firms are price takers: in the equilibrium each firm must earn zero profits (why?) Let the aggregate demand be X (p), with X 0 < 0 Long run equilibrium. It is a triple (p∗ , q ∗ , J ∗ ) such that: 1 2 3 π-max: q ∗ solves maxq≥0 {p∗ q − c(q)} AD=AS: X (p∗ ) = J ∗ · q ∗ Free entry: π(p∗ ) = p∗ q ∗ − c(q ∗ ) = 0 Introduction The Model Equilibrium Welfare Long-Run Equilibrium Long-Run Aggregate Supply (LR-AS) Correspondence Define LR-AS as: ∞ if π(p) > 0 {Q ≥ 0 : Q = Jq for some J ≥ 0 and q ∈ q(p)} if π(p) = 0 Q(p) = 0 if π(p) < 0 Intuition: If π(p) > 0 then every firm wants to supply a strictly positive amount of product, therefore AS is ∞ If π(p) = 0 and for some integer J we have Q = Jq(p), then there are J firms in the market each producing q(p) at price p. All other firms stay out and produce 0 (which is a profit maximizing choice as c(0) = 0) If π(p) < 0 all firms produce 0 LR equilibrium: alternative characterization The price level p∗ is a LR competitive equilibrium if and only if X (p∗ ) = Q(p∗ ) Proof: See MWG, p. 336, footnote 27 Conclusions Free-Entry and Long-Run Competitive Equilibria Example 1: LR Equilibrium with CRTS technology Introduction The Model Equilibrium Welfare Constant returns to scale Q(p) = ∞ [0, ∞) 0 if p > c if p = c if p < c Long-Run Equilibrium Conclusions Equilibria Introduction The Model Equilibrium Welfare Long-Run Equilibrium Example 2: LR Equilibrium with Strictly DRTS technology Non existence of long run competitive equilibrium with strictly convex costs Q(p) = � ∞ 0 if p > c � (0) if p ≤ c � (0) Conclusions Free-Entry and Long-Run Competitive Example 3: LR Equilibrium with U-shaped Costs Equilibria Introduction The Model Equilibrium Welfare Long-Run Equilibrium Average costs exhibit a strictly positive efficient scale ∞ if p > c̄ {Q ≥ 0 : Q = J q̄ for some integer J ≥ 0 if p = c̄ Q(p) = 0 if p < c̄ Conclusions Introduction The Model Equilibrium Welfare Long-Run Equilibrium Conclusions Conclusions Partial Equilibrium Approach From the positive side, it allows to determine the equilibrium outcome in one market in isolation from all other markets From the normative side, it allows to prove the 1st and 2nd welfare theorems (a sort of formal expression of Adam Smith’s “invisible hand”) Problems Prices of all other goods remain fixed There are no wealth effects in the market under study What about issues that are inherently general-equilibrium ones? It is crucial to consider a setup where all prices are simultaneously determined so as to take into account economy-wide feedbacks! Questions: Do welfare theorems still hold?