MACROECONOMIC THEORY Lecture Notes Alexander W. Richter! Department of Economics ‘Auburn University April 2014 ‘Correspondence: Department of Economics, Auburn University, Auburn, AL 36849, USA. Phone: +1(334)844-8638, E-mail: [email protected],Contents Chapter 1 Ll 12 13 14 1s Chapter 2 24 22 Introduction to Discrete Time Models Basic Robinson Crusoe Economy 1.1.1 Euler Equation 1.12 Dynamic Programming 1.13 Lagrange Method 1.1.4 Intuitive Derivation of the Euler Equation 1.15 Graphical Sotution 1.1.6 Stability and Saddlepath Dynamics Extensions (o the Basic Robinson Crusoe Economy 1.2.1 Endogenous Labor Supply 1,22. Investment Adjustment Costs Competitive Economy 13.1 Consumer's Problem 13.2 Firm's Problem 13.3 Competitive Equilibrium Solution Methods 1.4.1 Method of Undetermined Coefficients 1.42. Value Function Iteration 143 Euler Equation Iteration 1.4.4 Howard's Improvement Algorithm Stochastic Economy 15.1 Example: Stochastic Technology Linear Discrete ‘Time Models Analytical Solution Methods 2.1.1 Model Setup 2.1.2 Log-Linear System 2.1.3 Solution Method I: Direct Approach 2.14 Lucas Critique 2.1.5 Solution Method Il: MSV Approach Numerical Solution Method 2.2.1 Introduction to Gensys 2.2.2 Model Setup 22.3 Deterministic Steady State 2.24 Log-Linear System 2.2.5 Mapping the Model into Gensys Form 2.2.6 Gensys and Moving Average Components 2.2.7 Gensys and Forward/Lag Variables 2.2% Gensys: Behind the Code Cee nnnenne a 10 u u 12 12 7 18 RERRA.W. Richter CONTENTS Chapter 3 Real Business Cycle Models 40 3.1 Basic Facts about Economic Fluctuations 40 3.2 Campbell: Inspecting the Mechanism 40 2.1 Model 1: Fixed Labor Supply . 4 3.2.2. Model 2: Variable Labor Supply 47 Chapter 4 Money and Policy 33 4.1 Fiat Currency in a Lucas Tree Model 53 4.2 Fiscal and Monetary Theories of Inflation 55 4.2.1 Policy Experiments. 37 4.2.2 Stationary Equilibrium 37 4.2.3 Monetary Doctrines 59 4.2.4 Summary 2 43 Monetary and Fiscal Policy Interactions 2 4.3.1 Region I: Active Monetary and Passive Fiscal Policies 64 4.3.2. Region Il: Passive Monetary and Active Fiscal Policies 64 4.3.3. Region III: Passive Monetary and Passive Fiscal Policies 65 44 Classical Monetary Model 65 4.5 Basic New Keynesian Model 68 4.5.1 Households 68 4.5.2. Firms 68 4.5.3. Aggregate Price Dynamics R 454 Equilibrium... . : ” 4.5.5 Effects of a Monetary Policy Shock 7 4.5.6 Effects of a Technology Shock 9 4.5.7 Rotemberg Quadratic Price Adjustment Costs 80 Bibliography aeChapter 1 Introduction to Discrete Time Models 1.1 Basic Robinson Crusoe Economy It is a common in dynamic macroeconomic models to assume agents live forever. One ean think of agents as family dynasties, where those members of the dynasty who are alive today take into account the welfare of all members of the family, including those of generations not yet born. We begin by studying the basic dynamic general equilibrium closed economy model, which assumes that there is only one individual (or a social planner) who makes consumption-savings decisions each period. The social planner chooses {ce, kr+1}2o to maximize lifetime utility, given by, DY siuler), aa) = where u(:) is the instantaneous utility function, which is increasing (u'(-) > 0), but at a decreasing rate (w"(.) < 0). The planner values consumption more today than inthe future; hence, the discount factor, 0 < 8 < 1. The planner’s choices are constrained by atu, 2) kest = (1 dhe +i a3) w= fhe), a4) where ce, ir, and yp ate consumption, investment, and output at petiod t, ke is the capital stock at the beginning of the period, which depreciates at rate 0 < 6 <1, and ky,1 is the capital stock eartied into the next period. Equation (1.2) can either be interpreted as the national income identity (ic total output is composed of consumption goods and investment goods) or the aggregate resource constraint (ic. income is divided between consumption and savings, s¢ — ye — cr, where s, can only be used to buy investment goods. Equation (1.3) is the law of motion for capital. In period t, a fraction of the capital stock depreciates, dk. ‘Thus, the capital stock available in the next period is equal to the fraction that did not depreciate, (1 ~ é)kz, plus any investment made in period ¢ Equation (1.4) is the production function. Output is produced using the capital stock available at the beginning of the period. An increase in capital increases output (f"(-) > 0) but at a decreasing rate (£0) <0). Also Jim s'()=00 and im £7) which are known as Inada conditions. These equations state that at the origin, there are infinite output gains to inereasing capital, but the gains decline and eventually approach 0. We can combineA.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY (1.2.1.4) to obtain cot hes = S(he) + (1 ake. (sy ‘The goal of the planner is to make consumption-savings decisions in every period that maximize (LA) subject to (1.5). 1.1.1 Euler Equation ‘We can substitute for ¢ in (1.1) using (1.5) to reduce the problem to an unconstrained maximization problem. In this case, the planner’s problem is to choose {h141} 9 to maximize DY Btu Fe) + (1 — 8)he — hes), which is equivalent to differentiating + Btu fe) + (1 ~ 8)ky — Bega) + BM Bul fe) + (1 8)he4 — heya) + with respect to ky+1. Thus, we obtain the following first order condition —Btul (er) + Bw (eesa)LF (ey) +1 — 4] = 0, jon, becomes w (ce) = Bu'(cesr)[F"(Rev1) +1 — 4] (1.6) This equation is known as an Euler equation, which is the fundamental dynamic equation in in- tertemporal optimization problems that include dynamic constraints. It relates the marginal utility ‘of consumption at time ¢ to the discounted marginal utility benefit of postponing consumption for cone period, More specifically, it states that marginal cost of forgoing consumption today must equal to discounted marginal benefit of investing in capital for one period. ‘The optimal consumption-savings decision must also satisfy «yy which, after simplifi jim Btul(ex) which is known as the rransversality condition. To understand the role of this condition in intertem- poral optimization, consider the implication of having a finite capital stock at time T'. If consumed, this would yield discounted utility equal to 3? w/(cr)kp. If the time horizon was also T, then it ‘would not be optimal to have any capital left in period T,, since it should have been consumed in- stead. Hence, as ¢ + 00, the transversality condition provides as extra optimality condition for intertemporal infinite-horizon problems. 1.1.2. Dynamie Programming ‘Although we stated the problem in section 1.1 as choosing infinite sequences for consumption and capital, the problem the planner faces at time t = 0 can be viewed more simply as choosing today's consumption and tomorrow's beginning of period capital! The value function, (kg). is given by V(bo) =, max > s'u(er) (1.8) fener} Oy "The presentation in this section is kept simple, with the hope of communicating the main ideas quickly and enabling the reader to use these techniques to solve problems. Fora more thorough presentation see Stoky eal (1989),A.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY It represents the maximum value of the objective function, given an initial level of capital at time + =0, ko. V(k1) is the maximum value of utility that is possible when capital at time t = 1 is ki, and 3V (ky) discounts this value to time ¢ = 0. Thus, we can rewrite (1.8) as cok feokert V(ko) = max fu(eo) + max _ S>5tu(er) a = max [u(eo) + 3V(n)], which is known as Bellman’s functional equation, The study of dynamic optimization problems through the analysis of such functional equations is called dynamic programming. When we look at the problem in this recursive way, the time subscripts are unnecessary, since the date is imelevant for the optimal solution. After substituting for cy using (1.5), Bellman’s equation, conditional on the state at time t is Vike) mal f(r) + (1 ~ 8)ke — kesa) + BV (Kus) as) ‘The first order condition is given by —ul(ea) + BVia (ben) = 0. (1.10) ‘To illustrate the Envelope Condition, postulate a law of motion for capital given by key1 = h(Ke), ‘which intuitively asserts that tomorrow's capital stock is a function of today’s stock. Next, substitute jptimal investment plan” into the initial problem, (1.9), to obtain V (he) = w( fe) + (1 5)kr = hUe)) + BV (( Re) Optimizing with respect to the state variable, ke, yields V"(Ki) = w'(ea) fee) + 1 — 8 — bh )] + BV’ esa) (he) > V"(he) = w (ea) Lf (he) +1 = 8) + (=u! (ee) + BV" (esr )]R' (ie) ‘The expression multiplying I(x) is zero according to (1.10). Thus, the expression simplifies to V' (he) = wl (ee)LP"(Ke) +1 — 5) aap In short, the Envelope Condition allows us to differentiate (1.9) directly with respect to the state variable ky, ignoring its effect on ky, to get the exact same result If we advance equation (1.11) forward one period, we can use the result to substitute for V"(ke41) im (1.10) to obtain we) = Bu'(cerr)IF (key) + 1 ~ 5] (12) Which is the same as the Euler equation given in (1.6). Note that we could have alternatively used 1.5) to substitute for ke in (1.8) and applied similar steps to obtain the same Euler equation, 1.1.3. Lagrange Method We could also solve the constrained maximization problem using the Lagrange method. The La- sgrangian is given by Le = So 9'uler) + dalf (ha) + (=) = het = altsA.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY where the multiplier on the constraint is *2,. There are two choice variables, c; and kry1. The first-order conditions with respect to these variables are (13) (14) Bu’ (ce) — Ad] = 0, Met 8 ea LF" (heya) +19) ‘The Lagrage multiplier is easily obtained from (1.13). Substituting for dy and Ay, 1 in (1.14) yields we) = Bu'(cerr)LF (key) +1 ~ 5] Which, once again, is the same Euler equation given in (1.6) 1.1.4 Intuitive Derivation of the Euler Equation If we reduce c; by a small amount, dey, how much larger must ¢41 be to fully compensate, leaving, utility across the two periods unchanged? Define total utility in any two consecutive periods as (ce) + Bulcrs). Keeping total utility constant, the total derivative is uw (er)dee + Bul (era )ders (1.15) ‘The loss in utility is u/(cy)der. In order for Vj to remain constant, this loss must be compensated by the discounted gain in utility 3u!(crs1)dex41 ‘The resource constraint, (1.5), must also be satisfied in every period. Totally differentiating the resource constraints in periods t and ¢ + 1 implies dey + dks = f'(ka)dhy + (1 = 8) dh ders + dkese = f'(hesa)dkegs + (1 — S)dkey Since hy is given and beyond period ¢ + 1 we are constraining the capital stock to be unchanged, dk, = dhiya = 0. Hence dey + digs = egy = f'(kesi)dhesa + (1 d)dkes, which can be combined by eliminating dy. to obtain dear = —[f"(hev1) +1 — Sldee (1.16) This is an indifference curve that trades consumption tomorrow for consumption today. The output ‘no longer consumed in period t is invested and increases output in period ¢+1 by —J"(Kr41)der. This amount, in addition to the undepreciated increase in the capital stock —(1 — 6)dey = (1 —8)dkes, aves the total increase in consumption in period f + 1. Plugging this value in for dey, in (1.15) implies ul(er)des = Bu'(cesa) Lf" (ess) + 1 — dlder Cancelling out de; yields the same Euler equation given in (1.6).A.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY Vewe,)*Bute,..) Figure 1.1: Graphical solution to the basic dynamic general equilibrium model 1.1.5 Graphical Solution ‘The production possibility frontier is associated with a production function with more than one type of output. It measures the maximum combination of each type of output that can be produced with a fixed amount of factors. The intertemporal production possibility frontier (IPPF) is associated with ‘output at different points in time and is derived from the resource constraint, (1.5). Combine the constraints at periods ¢ +1 and ¢ + 2 to eliminate ke, and obtain the IPPF, given by, cost = Shes1) — hea + (1 Shea = F(F(he) — ec + (1 ~ 8)he) — hese + (1 — 8)LF (Re) — ce + (1 — 5)Ke ayy which is a concave relation between cy and 141. The slope of the IPPF is Be = Thea) +13), which is concave given that e41/%e¢ = f"(ke41) <0. ‘The solution to the two period problem is represented in figure 1.1. The upper curved line is the indifference curve, given in (1.16). The lower curved line isthe IPPF, which touches the indifference curve atthe point of tangency with the budget constraint. Hence, the solution must satisfy dees dey Bees er lpr S's) +1 =ltre=— Vom where the net marginal product, f‘(Key1) — 6 = rea represents the implied real rate of return on capital, An increase in r1 makes the resource constraint steeper, which increases Vi, cy, and ¢r41. 1.1.6 Stabi A useful graphical too! for studying two-dimensional nonlinear dynamic systems is a phase diagram, ‘To construct the phase diagram presented in figure 1.2, consider the two equations that describe the optimal solution at each point in time—the Euler equation and the resource constraint, which are ty and Saddlepath DynamicsA.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY reproduced below: cet he. (he) + (1 8) he (1.18) ul(e) = Bul(eess)[F (ker) +1 - 4} (1.19) We first consider the long-run equilibrium properties. The long-run equilibrium is a static solution, implying that in the absence of shocks to the macroeconomic system, consumption and the capital stock will be constant through time. Thus, c¢ = c*, ky = k*, Ace = 0, and Aky = 0 for all. In static equilibrium the Euler equation can be written as f(t) =B 45-1 5+0, where @ = 3~! — 1.This condition shows that the steady state level of capital is independent of consumption. We depict this on the phase diagram in (k,c)-space with a vertical line. To see what happens to consumption as k & k*, note that the Euler equation, (1.19), implies con 2 = Bul(cey1) < Buller) we) a = Fheayt tas $0) = 15 AU (hes1) + 1-4) > 640 f(b) ha si Thus, whenever k S k*, Ae 2 0, which is represented by vertical arrows. From the budget constraint, it is easy to see that Ak = 0 implies. c= S(k) — bk, which we can depict on the phase diagram with a hump-shaped curve that peaks at k= 6 > k*. To see what happens to capital above and blow this line, note that the budget constraint, (1.18), implies Kap > he > filly) + (1-8) — 4 > be = 4S She) — dhe ‘Thus, whenever ¢ S f(k) ~ 6k, Ak 2 0, which is represented by horizontal arrows. Figure 1.2 shows that there is a unique level of capital where the two lines intersect (point B). Thus, a steady state (", c) that satisfies the equilibrium conditions must exist and is unique. Note that the origin (0,0) is also a steady state, since an economy that begins with zero capital remains at (0,0). However, this steady state violates the Euler equation, since limy-so J’(K) = 20. Thus, trajectories that converge to the vertical axis are not equilibria. Likewise, trajectories that converge wo the imersection of te Ak = 0 schedule and the horizontal axis do not reach an eyul the transversality condition, given in (1.7) is violated. To see this note that (1.19) implies. whew) L “asl we) | BP) Bt ‘The inequality follows from that fact that at this point, kt > K, which implies f"(k) < 6. In other ‘words, the rate of growth of u/(c,) is larger than the rate of decline of the discount factor, 1/3 — 1. Since iis constant, this implies the transversality condition is violated.A.W. Richter LI. BASIC ROBINSON CRUSOE ECONOMY Ako e & z Figure 1.2: Phase diagram (o the basic dynamic general equilibrium model ‘The SS tine through point B is known as the saddlepath. Only trajectories that converge to the intersection of (k*,c*) satisfy the equilibrium conditions. Given the direction of the arrows, it seems likely that the steady state, given by (k*,c*), is reachable through trajectories that follow ‘SS. In the northeast quadrant, consumption is excessive and the capital stock is so large that the ‘marginal product of capital is less than 6 +0. This is not sustainable and therefore consumption and the capital stock must decrease. The opposite is true in the southwest region. The other two regions, as discussed above, are not stable. Equations (1.18) and (1.19) represent a first-order nonlinear difference equation in the vector (heey). To better understand the stability properties of the model, level linearize the system of equations by taking a first-order Taylor series approximation around the steady state to obtain we ae Bul (e*)(F(A*) + 1 — 8)(cop — €*) + Bulle) F(R) esa — BP) = F(R) — BY) + (1 — 8) — BY (hep — B). Given that percent changes are a good approximation for log deviations, we can represent this system in log-linear form as 6 = bays — BoTT G(R) keg hy — keys Ine —Inat* % («ee — 2°) x" and. = —u'"(c)e/u'(c) > Os the coefficient of relative isk aversion. In derivaing these equations, we made use of the fact that in stationary equilibrium 1 = BLf"(k*) + 1 ~ 4]. The system of equations can be rewritten in matrix form as Bom f"(k)k* 1] key) _ [0 1 ke 1 0} Léa] = [a-! -e' a] La. BI"(h)R* Jo > 0, we obtain (ee) = [ow ee] =x8 1+ xct/e] Le —SS ” Defining y =A.W. Richter 1.2. EXTENSIONS TO THE BASIC ROBINSON CRUSOE ECONOMY The eigenvalues of the matrix, A, are the solutions to the characteristic equation, given by, (A) =? = (14. 91 + ye*/R)A +B ‘Thus, the trace, T, and determinant, D, of the matrix, A, are given by TH1lt Ss +ye/k>148%>2 and D=s>1 Hence, the discriminant, T? — 4D > (1+ 9-1)? 48-1 = (1 — 3-)? > 0, and the characteristic polynominal has two positive real roots, whose sum exceeds 2 and whose product exceeds 1. Since p(1) = —xe"/k* < 0, the two real eigenvalues lie on either side of unity, which means that O< Ar <1 < Azan (K*,c*) is a saddle. If we define a = [ke G]", then our difference equation is z¢ = Azi-1. Let D = diag(A1, 2) and P be the corresponding projection matrix. Then if we define Z = P~'z, we obtain Z, = D2. Hence Zi. = A Zig—1 and Zo = d2Zo4—1. The general solutions are Zi = a1} and Za, = azX,for some constants a; and ay. To recover the solutions for fy and é,. apply the projection matrix to Z; to obtain tay = PuZig + ProZ2¢= Prd} + Prars, 2.4 = Poi Zig + PooZo4 = Povard + Proards, where P,; (i,j € {1,2}) are the elements of the projection matrix. Since ky is given, 23,9 = ko (hy — k*)/k* is also given. Hence fy = Prrar + Proar (1.20) We also know that the optimal paths for cy and ke converge to the steady state. Thus, lim ac 21¢ = lims-s20 %2¢ = 0. This implies that a = 0, since Ay > 1. From (1.20), a1 = ko/Pi1, and the solutions for ky and é are hy = fo é = PakoM/Pu = Paki/Puy Which is a the unique stable solution that converges to the stationary equilibrium, 1.2. Extensions to the Basic Robinson Crusoe Economy 1.2.1 Endogenous Labor Supply In the basic model, labor is inelastically supplied. That is, since agents did not derive utility from leisure, the optimal choice of labor was to spend all available time working. To extend the basic ‘model, we assume leisure provides utility, so that the planner makes an endogenous labor supply decision about how much time to spend working and consuming leisure. We assume that the total available time is 1. Thus, nc-+ £ = 1, where ny is hours worked and f; is leisure. The instantaneous utility function is u(ce,f1), where ue > 0, up > 0, lee <0, and ug <0. This says that there is positive, but diminishing marginal utility to both consumption and leisure, For convenience, we assume ue = 0, which rules out substitution between consumption and leisure. We also assume labor is a second factor of production. Thus, the production function becomes J (ce, 1), with Jig > 0, fn > 0, Sik $0, fan $0, fin > 0, and lity sac fie = Os Littin sae fn = 0, Himaso fic = 90, and limin-s0 Jnr = 0, which are the Inada conditions.A.W. Richter 1.2. EXTENSIONS TO THE BASIC ROBINSON CRUSOE ECONOMY ‘The problem is as follows. A representative planner chooses sequences {ce, m1, kr+1}?29 0 ‘maximize lifetime utility given by Y Btuler, tm) = subject to the resource constraint, c+ est ~ (1 6)he = Seon) where we have substituted the time constraint, ny + é¢ = 1, into the instantaneous utility function, ‘The Lagrangian is given by Le = YB {ulee, 1 = me) + Ae(F(Bes me) — Resa + (1 = Be = e4)}, \where Apis the Lagrange multiplier on the resource constraint. The first order conditions are given by ce: Weg =e me: wee = Arfnt Kerri de = BrceiLfigers + 1 ~ 9] ‘The first order conditions for ¢; and ky 1 yield the same Euler equation as the model where labor is inelastically supplied, and is given by, tot = Buotsilfiter + 1-4). ‘The first order condition for labor implies We, = Weabnt ‘This equation says that the loss in utility from giving up diy = —dny < 0 units of leisure is compensated by an increase in utility due to producing extra output equal 10 —uefn,ille > 0. To summarize, we have found that when we allow the planner to choose how much to work, the solutions for consumption and capital are virtually unchanged from those of the basic model 1.2.2 Investment Adjustment Costs ‘The basic model includes investment, but thus far we have assumed that there are no costs to in- stalling new capital. Suppose there is a permanent change in the long-run equilibrium level of capital, Although capital takes time to adjust to its new steady state level, investment in the basic ‘model adjusts instantaneously to the level that is optimal each period. In practice, however, itis usually optimal to adjust investment more slowly, due to installation costs ‘Toillustrate, suppose that new investment imposes an additional resource cost equal to dir /(2hr) for each unit of investment, where ¢ > 0. In this case, the cost of a unit of investment depends on how large it is in relation to the size of the existing capital stock. The planner’s decisions are now constrained by - é > Sle) = 04 (1+$2)i 2p Kuga = (1 = Oke + it. (1.22)A.W. Richter 1.3. COMPETITIVE ECONOMY ‘The Lagrangian is given by L Lot {ue +a (100 -- SE a) + mi = 8)h vo) where 2, is the Lagrange multiplier on (1.21) and ji; is the Lagrange multiplier on (1.22). The first onder conditions are given by ce: Wer =e + Bpesa(l — 8). ‘The first order condition for investment implies GH = 1+ O(ie/he) > ie = (ae — Dhl, where the ratio of the Lagrange multipliers 4s = ju/'e > 1 is known as Tobin’s q. There exists positive investment if q > 1. \y is the marginal utility value, in terms of net output, of an additional unit of k. jy measures the marginal utility value of 1 unit of investment. Hence, q; represents the benefit from investment per unit of benefit from capital in terms of units of output. Or, qi is the ‘market value of 1 unit of investment relative to its replacement costs. 1.3 Competitive Economy In the Robinson Crusoe economy, one person made consumption and production decisions for the whole economy. In a competitive economy, consumers rent capital to firms and sell labor. In this section, we assume there is a continuum of identical agents of a unit mass and that all agents can provide up to one unit of labor to the market. All agents are the same so that we can take the behavior of one agent as that of the whole economy since we simply integrate from 0 to I over identical agents. 1.3.1 Consumer's Problem Individual i chooses sequences {c},nj, hj, } to maximize lifetime utility, given by, fu(cd, ni) subject to cet Ries vwyni + rekd + (1 — dR, (1.23) where w is the wage rate and r is the rental rate on capital. ‘The Lagrangian is given by Le = YB {ulefs nj) + Arwen} +r + (1-8)Ki =~ Kia) 10A.W. Richter 1.3. COMPETITIVE ECONOMY ‘here 2 is the Lagrange multiplier on (1.23). The first order conditions are given by hs Wea = Ae et gg = we Fast = BAealrest +14) ‘After combining these results, we obtain eg = tee Uo = Buegalresa +1 ~ 5] ‘The first equation says that the loss in utility from giving up dl; = dry < 0 units of leisure is compensated by an increase in utility due to earning i. The second equation says that the marginal cost of foregoing consumption today in favor of investing in capital is equal to the discounted utility value of that investment tomorrow. The net return on investment equals 1+ resa~ 6 1.3.2. Firm’s Problem ‘The firm maximizes profit each period by choosing {fe, nr} to maximize J (asm) — wore = ree ‘The first order conditions are felkesme) = 14 alkene ‘These equations show that this is a competitive firm as each of the factor prices is equal to its ‘marginal product. 1.3.3 Competitive Equilibrium ‘The competitive equilibrium system is composed of the consumer's and firm's optimality condi- tions, et = tet Int uot = Buetrilfieri +1 — 4), the budget constraint, (1.23), and the aggregation rules, m= fn and ky = fii hb lb ‘When all the unit mass of individuals are identical, as in the nj and ky = ky ‘The production function is homogeneous of degree one (constant returns to scale) and under conditions of perfect competition with free entry, firms to do not make any profits. Hence the aggregation rules simplify 10 Fatma + Fieehe = (heme) uA.W. Richter 1.4. SOLUTION METHODS. and the budget constraint can be rewritten as cet hey = f(kesna) + (1 = 8k where aggregate consumption is c = ff cidi. Notice that the conditions for the equilibrium in the competitive economy turn out to be an aggregate version of the same conditions for the Robinson Crusoe economy. ‘The equilibrium that ‘we found to the Robinson Crusoe economy is Pareto optimal. It is the result of the social planner finding consumption and production points that maximize the utility of the single individual in the economy, given his technology constraints. ‘The first fundamental welfare theorem tells us that any competitive equilibrium is necessarily Pareto optimal, so that the equilibrium found using a decentralized economy with factor and goods markets is also Pareto optimal. The second welfare theorem tells us that, since the production technologies and preferences are the same in the two economies, then with the right initial wealth conditions, the competitive equilibrium can achieve an equilibrium that is identical to the social planner economy. Itis the second fundamental welfare theorem that permits us to use Robinson Crusoe economy to mimic a competitive economy. Since the second fundamental theorem is carefully worded, it should be clear thatthe solution to the social planner’s problem will not always give the appropriate resulls, If the economy is not perfectly competitive, if part of the economy has some monopoly power or if there are some internal or external restrictions that prevent some agents from behaving perfectly competitive, then the economy found in the decentralized economy will not necessarily be achievable in the Robinson Crusoe economy. 1.4 Solution Methods ‘There are several different ways to solve nonlinear dynamic models: 1. Method of undetermined coefficients (guess and verify method): Guess a functional form for the value function and then verify that this functional form satisfies Bellman’s functional equation 2. Value function iteration (Bellman Iteration) 3. Euler equation iteration 4, Howard’s Improvement Algorithm: guess a functional form for the policy function and plug it into the Euler equation. ‘Some solution methods will be easier to apply to certain models. Below is set of examples that are designed to help you develop the basic techniques involved in each of the above solution methods. Note, however, that these are very stylistic examples. Most models can not be solved analytically and will be require numerical techniques. 1.4.1 Method of Undetermined Coefficients Example 1: Cake-Eating Problem We begin with a very simple dynamic optimization problem. Suppose that you have a cake of size twp. At each point in time, ¢ = 1,2,3,... you eat some of the cake and save the remainder. The 12