Handout 13th Lecture, ECON 5200 Uncertainty and time Kjell Arne Brekke November 12, 2010 1 1.1 Application of general equilibrium Lindahl equilibrium We consider an economy with a private and a public good X x2 = q = f (z) is the public good X xi1 + z = ! i market clearing in the private good is i’s utility Now, this does not correspond to a Walrasian equilibrium. Suppose however that the …rm produces an public good for each consumer q1 = q2 = ::: = qI = q = f (z) and with market clearing in each market xi2 = q2 with individualized prices s.t. p1 xi1 + pi2 xi2 max ui (xi1 ; xi2 ) = wi and pro…t maximizing production of the public good X max( pi2 )f (z) p1 z z This is called a Lindahl equilibrium. In a sence it presumes that each consumer in fact pays his willingness to pay (at the margin) for the public good. We note that if this was the case then X ( pi2 )f 0 (z) = p1 That is the total willingness to pay for an additional unit of the public good should equal the cost. This solution is of course Pareto e¢ cient. 1 If individuals do in fact pay equally much for each unit, then increasing production at a point where X ( pi2 )f 0 (z) < p1 would not be a Pareto improvement. Some will bene…t and some will lose. 2 Uncertainty and Arrow Debreu equilibrium We start with a straightforward extension of the general equilibrium model. Suppose that there are a list of state of the world f! 1 ; ::::; ! S g = fsnow, rain, sunnyg as well as L commodities. We also have two periods, where all consumption happend in period 1 but markets are only open in period 0. We now introduce a new set of state.contingent commodities xls is a delivery of xls units of commodity l contingent on the state being ! s . The true state is not known in period 0 when the market is open but will be revealed before consumption in period 1. Now, with a seperate market for each state dependent commodity, there will be equilibrium prices pls for each state dependent commodity. The …rst and second welfare theorem applies as this formally is just an ordinary Walrasian equilibrium. 2.1 Sequential trading - Radner We do not today trade "appels tomorrow if it snows". We trade those apples tomorrow after we have observed the state of the world. On the other hand: We have two kind of decisions: How do we allocate our wealth within each state and how do we allocate our entire wealth between di¤erent states. In some states we may want to have more resources than in others. Suppose that there is only one commodity where the markets are open in period 0, and that all other markets open …rst after the true state is revealed. Now, when consumers trades in period 0, markets in period 1 are not yet open, hence prices are not yet known. But we assume that consumers have perfect foresight and correctly predicts all prices. The market in period 0 are called forward markets, and the markets that opens in period 0 are sport markets. The resulting equilibrium is called an Radner Equilibrium. Proposition 19.D.1 The allocation emerging from an Arrow-Debreu equilibrium can also be supported as a Radner equilibrium. As only relative prices in spot markets matter we may have to transform prices such that ps = 2 s ps 2.2 State prices We need only forward trading in one commodity since what we need is to transfer wealth between states. This is essentially what …nancial instruments does. As microeconomic theory does not have a good theory of money we simply make a leap here, study the single commodity traded between states and call it money, and ignore the other goods - those markets are not yet open. To gain some insight consider …rst the case where there is a contingent claim for each state. The contingent claim 1s pays a $ in period 0 if state s emerges otherwise nothing. And there is one asset for each state s = 1; :::; S. The price of each asset is denoted s . Now consider a more standard asset with an uncertain return, it pays xs in state s. Note that this is equivalent to buying a portfolio of rs units of the contingent claim 1s for P each state. This will cost x0 = x s s : which should be the price of r in period 0. Note that if we buy one unit of each contingent claim we get 1 for sure in period 1, which is 1 in period 0. But uisng the value formula above we see that worth 1+r F X 1 1 X = qs s = 1 + rF 1 + rF qs = (1 + rF ) s X qs = 1 a kind of probabilities Note that with these probabilities x0 = P qs x s Eq (x) = 1 + rF 1 + rF Thus expected return with these probabilities is equal to the risk free rate of interest. THis is the basis for a technique used in …nanace. Adjust the expected return to assets to the risk free rate, keep the volatility and value e.g. an option using the risk free rate of return. 2.3 No arbitrage If there are no such contingent claims 1s ? Well if there are two states s = 1; 2 and two stocks with return (2,1) and (1,1) we get a (1,0) by owning one (2,1) and sitting short (1,1). And (0,1) by owning 2 (1,1) and sitting short (2,1). Thus state prices should follow from asset prices. In general let there be K assets with return rks and with prices in period 0 equal to qk If a portofolio is for free: q 0 but pays something rs 0 for all s, with strict inequality for at least one 3 Then this is an arbitrage. Proof. There is no arbitrage if and only if there exist state prices such that X qk = s rks for all k 2.4 Incomplete markets Note that if there are many states and few assets, we may not be able to generate a 1s for all s. So some of the state prices are ambigous. We cannot trade on the asset market to get rid of all the oil risk Norway is facing. That there are commodities for which there is no separate market is thus a real phenomena here. 3 Intertemporal equilibrium As pointed out, it is trivial to extend general equilibrium theory to several time periods. Disregaring uncertainty now, we simply label each commodity with its time period, and open one market for each. Some issues arises about perfect forsight, but this is as discussed for uncertainty. There are some special issues involed however, with intertemporal consistency. What would be the problem of assuming for three periods U (c1 ; c2 ; c3 ; ::::) = X1 t u(ct ) There are two issues: Separability Given separability we may consider the choice between c2 and c3 independently, but the rankin changes over time. Plans are inconsistent. 4