MacroeconomicsII : HW1 Solution

KONISHI Teruki,
KOGANEMAEU Ryohei*

October 22, 2024

 Q1. Detrending and Balanced Growth 

Find quarterly data on U.S. real GDP per capita from 1950 Q1 to 2023 Q4. Also find the 10-year real interest
rate (of any frequency) and average annual hours worked per worker for as long as available.
(a) Generate three time-series plots for log real GDP per capita.
• Quarterly data with a linear trend,
• Quarterly data with a HP filtered trend,
• HP detrended cyclical components.
(b) Do the same exercise for real interest rate and average weekly hours worked per worker as in the
question (i). Explain in words how the time series pattern of hours differ from those of GDP per
capita.
 

(a) Answer.

Figure 1: GDP per capita

(b) Answer. The average annual hours worked per worker decreases in time, while GDP per capita increases.
* Both are 2nd-year master students at the Graduate School of Economics, The University of Tokyo. Contact : ut-
macro2024a@gmail.com

1
 Figure 2: cyclical components of GDP per capita

Figure 3: average annual hours worked per worker

Figure 4: cyclical components of average annual hours worked per worker

2
 Figure 5: 10-year real interest rate

Figure 6: cyclical components of 10-year real interest rate

3
 Listing 1: trend-cycle decomposition
%%%%%%%%%%%%%%% hours %%%%%%%%%%%%%%%%%%%%%%%%
%find data of average annualhours worked per worker at OECD
%I call such data UShoursworkedperworker19502023

year=1950:2023;

p=polyfit(year,UShoursworkedperworker19502023,1);

trend=polyval(p,year);
cycle=UShoursworkedperworker19502023-trend;

[t,c]=hpfilter(UShoursworkedperworker19502023,100);
%annual, lambda=100

[length]=size(UShoursworkedperworker19502023);
length=length(1,2);
stationary=zeros(length);

%trend
hold off
hold on
plot(year,t) %hp
plot(year,trend) %linear
plot(year,UShoursworkedperworker19502023) %data
legend('HP','linear','data')
xlabel('year')
label('average annual hours per worker')
hold off

%cycle
plot(year,c) %hp
hold on
plot(year,cycle) %linear
plot(year, stationary) %data
xlabel('year')
ylabel('average annual hours per worker')
legend('HP','linear','data')
hold off

%%%%%%%%%%%% GDP %%%%%%%%%%%%%%%%

%find data of real GDP per capita at FRED
quarter=1:length(GDPpercapita);
p=polyfit(quarter,GDPpercapita,1);
trend=polyval(p,quarter);
cycle=GDPpercapita-trend';
[t,c]=hpfilter(GDPpercapita,1600);
%quarter lambda=1600
stationary=zeros(length(GDPpercapita),1);

4
hold off
hold on
plot(quarter,t) %hp
plot(1950:0.25:2023.75,t)
plot(1950:0.25:2023.75,trend) %linear
plot(1950:0.25:2023.75,GDPpercapita) %data
legend('HP','linear','data')
xlabel('time')
ylabel('GDP per capita')
hold off

hold on
plot(1950:0.25:2023.75,c) %hp
plot(1950:0.25:2023.75,cycle) %linear
plot(1950:0.25:2023.75, stationary) %data
xlabel('time')
ylabel('GDP per capita')
legend('HP','linear','data')
hold off

%%%%%%%%%%%% interest rate %%%%%%%%%%%%%%%

%find data of real interest rate at FRED
monthly=1:length(REAINTRATREARAT10Y);
%REAINTRATREARAT10Y(:,2)
p=polyfit(monthly,REAINTRATREARAT10Y(:,2),1);
trend=polyval(p,monthly);
cycle=REAINTRATREARAT10Y(:,2)-trend';
[t,c]=hpfilter(REAINTRATREARAT10Y(:,2),14400);
%monthly lambda=14400
stationary=zeros(length(REAINTRATREARAT10Y),1);

hold off
hold on
plot((1982+1/12):1/12:(2024+10/12),t)
plot((1982+1/12):1/12:(2024+10/12),trend) %linear
plot((1982+1/12):1/12:(2024+10/12),REAINTRATREARAT10Y(:,2)) %data
legend('HP','linear','data')
xlabel('time')
ylabel('real interest rate')
hold off

hold on
plot((1982+1/12):1/12:(2024+10/12),c) %hp
plot((1982+1/12):1/12:(2024+10/12),cycle) %linear
plot((1982+1/12):1/12:(2024+10/12),stationary) %data
xlabel('time')
ylabel('real interest rate')
legend('HP','linear','data')
hold off

5
 Q2. Computational Exercise : Labor Supply 

Consider the following general equilibrium problem. Suppose that the representative household solves

( 1
)
c 1− σ − 1 h 1+ θ
max −ψ
c,h 1−σ 1 + 1θ
s.t. c = wh, 0⩽h⩽1

where c and h denote consumption and work hours that the household chooses optimally and σ and θ are
parameters. w denotes the market wage.
The representative firm maximizes its profit π

π = max { Al α − wl }
l

where A and α are parameters.a

The household and the firm take the market wage rate as given. The equilibrium wage is determined in
the labor market so that the market clearing condition, l = h, holds in equilibrium.
(a) Write down the first-order conditions of the household and the firm.
(b) Follow the steps below to compute the equilibrium wage and labor.
i. Assume that the parameters are σ = 0.7, ψ = 1.5, θ = 0.5, A = 2.0, α = 0.6.
ii. Generate 20 grids of wages in the range between 1.0 and 5.0. You can use the built-in functions
in your programming language.
iii. Plot the labor supply and demand curves on the grids generated above.
iv. Find the equilibrium wage using the bisection and Golden-section search methods without
using the built-in functions.

(c) Assuming the parameter values given in (b), compute an equilibrium by solving the system of two
equations using a built-in function (e.g. fsolve in Matlab) and verify your solutions obtained in (ii).
(d) What would happen if you changed σ to 1.3? Discuss the economic intuition behind this.b
(e) Compute the equilibrium assuming σ = 1.3 and A = 4.0 instead of A = 2.0.
aAssume that π is “thrown into the ocean” (or sent to a foreign owner of the firm) and plays no other role in the model.
b Hint: Remember the Slutsky’s decomposition.
 

(a) Answer. To begin with, let us derive the household’ condition. By substituting c in the budget into the
utility function and taking the differential, we obtain the first-order condition
( 1
) 1
!
(wh)1−σ − 1 h 1+ θ ∂ (wh)1−σ − 1 h 1+ θ 1
max −ψ =⇒ −ψ = w1−σ h−σ − ψh θ = 0
h 1−σ 1 + 1θ ∂h 1−σ 1 + 1θ

6
 Next, let us solve the firm’s problem. The first-order condition of the firm is simply given by

∂π
= Aαl α−1 − w = 0
∂l

Although you are not supposed to solve the problem for each agent for now, you can obtain the labor
supply and demand functions by solving the first-order conditions. 1

(w/ψ̃) σ1+−1/θ
σ
if (0 ⩽)h ⩽ 1 1
(Labor Supply) h(w) = , where ψ̃ = ψ 1−σ
1 if 1>h

1
(Labor Demand) l (w) = (w/Aα) α−1

(b) Answer. The following code is an example of how to generate grid points and plot labor demand and sup-
ply curves in MATLAB. Note that the labor demand curve is downward sloping and labor supply curve
is upward-sloping with this parameters setting. DO NOT forget to set h = 1 if h exceeds 1.

(a) Q2.(b)-(iii) : Labor Supply and Demand Curves (b) Q2.(b)-(iv) : Objective Functions

Figure 7

Listing 2: Generate Plot of Labor Market

% Initial Parameters
global psi theta alpha
sigma = 0.7; psi = 1.5; theta = 0.5; alpha = 0.6; A = 2.0;
% Generate Grids
w = linspace(1.0,5.0,20);
% Evaluate labor supply and demand at grid points
h_s = (w.^(1-sigma)/psi).^(1/(1/theta+sigma));
h_s(h_s>1.0) = 1.0;
l_d = (w/A/alpha).^(1/(alpha-1));

1 Since you are asked to just write down the first-order conditions, you don’t need to consider the corner solution issues. We are

really sorry if you are confused whether you should consider the corner solution problem or not.

7
The following code is an example of how to solve the equilibrium wage using the bisection and GSS
methods in MATLAB. First of all, you have to define the excess labor demand (or supply) function as the
objective function of the root-finding problem. Here is the m.file for the excess labor demand function.
I defined it as a function of w, σ and A because we have to change the parameter values of σ and A in
the later questions. Other parameters are defined globally.
Listing 3: excess_labor_demand.m
function [exdem] = excess_labor_demand(w,sigma,A)
global psi theta alpha
% Labor supply
hs = (w.^(1-sigma)/psi).^(1/(1/theta+sigma));
hs(hs>1.0) = 1.0;
% Labor demand
ld = (w/A/alpha).^(1/(alpha-1));
% Excess labor demand
exdem = ld - hs;
end

After preparing the m.file, we can start to write the main algorithm of bisection and GSS methods.
These are the codes for them written in the main script. The initial range of wages, ( a0 , b0 ), must be
defined for the bisection method so that excess_labor_demand( a0) × excess_labor_demand( a0) < 0.
When you successfully implement the algorithm, you obtain w∗ = 1.2612, h∗ = l ∗ = 0.8830 from both
algorithm.
Listing 4: Bisection Method
% Bisection method to find the equilibrium
maxiter = 100;iter = 0; dif = 1; tol = 1e-8;
a0 = 1; b0 = 5;
while dif > tol && iter < maxiter
c0 = (a0 + b0)/2;
fc = excess_labor_demand(c0,sigma,A);
fa = excess_labor_demand(a0,sigma,A);

if fa*fc ≤ 0
b0 = c0;
else
a0 = c0;
end
iter = iter + 1;
dif = abs(a0-b0);
end
w0_bisec = a0
h0_bisec = (w0_bisec^(1-sigma)/psi)^(1/(1/theta+sigma))
l0_bisec = (w0_bisec/A/alpha)^(1/(alpha-1))

8
 Note that although the GSS method is widely used to find minimum/maximum value of a unimodal
function, you can utilize it for discover the root of the function by taking the absolute value or the square
of the function. (See Figure 7.(b))
Listing 5: Golden-Section Search Method
% GSS method to find the equilibrium
a0 = 1; b0 = 5;
phi = (1+sqrt(5))/2; resphi = 2-phi;
c0 = a0 + resphi*(b0 - a0);
d0 = b0 - resphi*(b0 - a0);
fc = abs(excess_labor_demand(c0,sigma,A));
fd = abs(excess_labor_demand(d0,sigma,A));
iter = 0; diff = 1;
while diff > tol && iter < maxiter
if fc ≤ fd
b0 = d0;
d0 = c0;
fd = fc;
c0 = a0 + resphi*(b0-a0);
fc = abs(excess_labor_demand(c0,sigma,A));
else
a0 = c0;
c0 = d0;
fc = fd;
d0 = b0 - resphi*(b0-a0);
fd = abs(excess_labor_demand(d0,sigma,A));
end
iter = iter + 1;
diff = abs(b0-a0);
end
w0_GSS = a0
h0_GSS = (w0_GSS^(1-sigma)/psi)^(1/(1/theta+sigma))
l0_GSS = (w0_GSS/A/alpha)^(1/(alpha-1))

(c) Answer. You can make use of fzero function as follows. Note that you may not reach the solution
without specifying the searching region. If so, use a constraint like fzero(exdem, [0.01 10]).
Listing 6: Root-Finding using fzero
% fzero
exdem = @(w) excess_labor_demand(w,sigma,A);
w0_fzero = fzero(exdem,[0.01 10])
h0_fzero = (w0_fzero^(1-sigma)/psi)^(1/(1/theta+sigma))

This code also generates w∗ = 1.2612, h∗ = l ∗ = 0.8830.

9
(d) Answer. As you can see from the labor demand and supply functions, a change in σ only affects the
labor supply behavior. By plotting the labor supply curves, you realize that the labor supply curve with
σ = 1.3 becomes downward-sloping. Therefore, parameter σ is an important determinant of the response
of labor supply to TFP increase as you will see in Q2.(e).
An intuition behind this is as follows: A wage increase leads to a positive substitution effect on labor
supply, encouraging more work. However, the income effect has the opposite impact, potentially reduc-
ing labor supply in total. The relative strength of the income effect is controlled by the parameter σ: when
σ = 0.7, the substitution effect exceeds the income effect and we obtained a upward-sloping labor supply
curve; when σ = 1.3, the income effect outweighs the substitution effect and then a downward-sloping
labor supply curve appears. (In addition, show a plot like Figure.8)

Figure 8: New Labor Market with σ = 1.3

Comment. You can understand this as follows: the effect of a wage change on the labor supply can be
decomposed into the substitution effect and income effect as you learned in a basic micro course. To see
this decomposition, let’s write the first-order condition differently.
 θ
w/ψ
h(w, c) =
cσ

Put it casually, the substitution effect refers to how an individual’s labor supply choices change when the
wage changes, holding their income constant2 . Hence, the strength of the substitution effect is controlled
by

∂ ln h(w, c)
θ= >0
∂ ln w

which is called the Frisch elasticity of labor supply. The labor income also changes as a wage changes
2 The mathematical definition is different from this explanation. Check your micro textbook for rigorous definition.

10
 and overall impact of the wage change is affected by this income change. Using c = wh, we can compute
θ  (1−σ)/(σ+1/θ )
w1−σ /ψ

w
h= =⇒ h =
hσ ψ̃

Hence, the overall elasticity of the labor supply is calculated as

dh(w) 1−σ
ε= = ⋛ 0 ⇐⇒ 1 − σ ⋛ 0
dw σ + 1/θ

In summary, a wage increase leads to a positive substitution effect on labor supply, encouraging more
work. However, the income effect has the opposite impact, potentially reducing labor supply in total.
The relative strength of the income effect depends on whether σ is greater than one. If σ exceeds one, the
negative income effect outweighs the positive substitution effect.
∗
(e) Answer. Using the same methods as in (b) or (c), you obtain wnew = 2.6104 and h∗new = 0.8105. Here
is the MATLAB code. You can see that decreased hours worked after the TFP increase. Note that this
question asks you to find the equilibrium, so you have to report the equilibrium hours worked as well as
the equlibrium wage.
Listing 7: The Effect of TFP Increase
% New parameters
A_new = 4.0;
% Solve New Eq
exdem1 = @(w) excess_labor_demand(w,sigma_new,A_new);
w1 = fzero(exdem1,[0.1,5])
h1 = (w1^(1-sigma_new)/psi)^(1/(1/theta+sigma_new))

Please read the following supplemental information regarding the intent of the questions.

11
Q2. Supplement: Balanced Growth and Labor Supply
The theme of Q1.(b) and Q2. is to give you implications for the concept of balanced growth and historical
labor supply data. Declining trends of labor supply measured as hours worked per worker have been widely
observed in many advanced countries. Along the usual BGP, however, the hours worked per worker must be
constant over time. So, what class of utility functions does reconcile the tension between the BGP and declining
hours? Homothetic utility as taught in lecture 1? I would answer to the two questions: (a) How to harmonize
the BGP and declining labor supply, (b) How to achieve the BGP with labor supply decision.
(a) You found the downward trend in hours worked per worker in the U.S. by solving Q1.(b), but you also
realized that the real GDP has been evolving along the BGP. This problem can be resolved based on
the idea you saw in Q2.(e), that is, labor supply decreases along the income growth if households have
a strong income effect on their labor supply. This becomes clear when you see the Figure 10.(b): As
TFP, A, increases, labor demand shifts rightward, and then the household works less hours. In fact, this
model predicts that h∗new = 0.8105 < 0.8653 = h∗old with σ = 1.3.

(a) Weekly Hours Worked per Worker (b) TFP Growth and Labor Supply

Figure 9: With σ > 1, we can replicate the BGP with declining hours workerd

(b) Regarding the optional question (f), BGP asks us to find the utility function with which the hours worked
is constant over time.3 By plotting the labor supply function with σ = 1.0 as asked in (f), you will find
the constant labor supply regardless of the wage levels. Therefore, this is an example of a utility function
consistent with the BGP. (See red -x- line in Figure 10.(b))

3 This depends on the definition of the BGP. Usually, a balanced growth path is defined as the part of an equilibrium path where

growing variables (such as consumption, GDP, capital stock, population (total labor supply), wage rate, etc...) grow at constant rates
whereas other variables (such as hours worked per worker, interest rate, R&D worker ratio, etc...) are constant over time. The variables
defined as a ratio are typically classified as a non-growing variable. Now, the hours worked per worker is a ratio-type variable as it
means how you divide your 24 hours into working and leisure. On the other hand, the total hours worked by Japanese people is a
growing variable.

12
 Q3. Intertemporal Elasticity of Substitution 

Consider the following production function with capital K and labor L as two inputs of production.
  ρ−ρ 1
ρ −1 ρ −1
Y = F (K, L) = α · K ρ + (1 − α ) · L ρ

(a) Show that this production function exhibits a constant returns to scale.
(b) Derive the elasticity of substitution between capital and labor.
(c) Show that this production function nests the Cobb-Douglas production function when ρ → 1.a
a Hint: You could apply L’Hopital’s rule and use the lemma (without a formal proof):

lim ln f ( x; θ ) = ln g( x ) ⇒ lim f ( x; θ ) = g( x )
θ →α θ →α

for all x ∈ X, where f : X × Θ → R++ , g : X → R++ and α ∈ Θ where X ⊂ R N , Θ ⊂ R, and N ∈ N. (Note that α can be a cluster
point of Θ and the lemma is applicable for the case of ρ = 1 above.)
 

Q3

Q3-(a)
Take ∀c > 0.
h ρ −1 ρ −1 i ρ−ρ 1
F (cK, cL) = α(cK ) + (1 − α)(cL) ρ ρ

h ρ −1  i ρ
ρ −1 ρ −1 ρ −1
= c ρ αK ρ + (1 − α ) L ρ

=cF (K, L)

Therefore, the production function is homogeneous of degree one.

Q3-(b)
The marginal products of capital is

ρ −1 ρ −1 i ρ−1 1
− ρ1
h
αK αK ρ + (1 − α ) L ρ

The marginal products of labor is

h ρ −1 1
1 ρ −1 i ρ −1
(1 − α) L− ρ αK ρ + (1 − α) L ρ

13
 The marginal rate of transformation between capital and labor is

∂F (K, L)/∂L
MRTKL =
∂F (K, L)/∂K
 − 1ρ
(1 − α ) L
=
α K

The elasticity of substitution between capital and labor is

∂( L/K )/( L/K ) MRTKL /( L/K )

− =−
∂MRTKL /MRTKL ∂MRTKL /( L/K )
 − ρ1 −1
1− α L
α K
=−   − 1ρ −1
1− α 1 L
α − ρ K

=ρ

Q3-(c)
Take limit of log production function as ρ → 1:

(  ρ−ρ 1 )
ρ −1 ρ −1
lim ln F (K, L; ρ) = lim ln α·K ρ + (1 − α ) · L ρ
ρ →1 ρ →1
 
ρ −1 ρ −1
ln α · K ρ + (1 − α ) · L ρ

= lim ρ −1
ρ →1
ρ
 
ρ −1 ρ −1
d
dρ ln α · K ρ + (1 − α ) · L ρ

= lim   (L’Hôpital’s Rule)

ρ →1 d ρ −1
dρ ρ
ρ −1 ρ −1
1 α·K ρ ln K +(1−α)· L ρ ln L
ρ2 ρ −1 ρ −1
α·K ρ +(1−α)· L ρ
= lim 1
ρ →1
ρ2

=α ln K + (1 − α) ln L
 
= ln K α L1−α .

Using the lemma in the footnote, we obtain

lim F (K, L; ρ) = K α L1−α .

ρ →1

14
 Q4. Arrow-Debreu Equilibrium 

Consider an exchange economy with two infinitely-lived individuals, indexed by i = 1, 2. Individuals

have the utility function
∞
∑ βt log cit
t =0

where β ∈ (0, 1) is the discount factor. The endowment streams of the two individuals are given as

(e01 , e11 , e21 , e31 , ...) = (4, 1, 4, 1, 4, ...)

(e02 , e12 , e32 , e32 , ...) = (1, 4, 1, 4, 1, ...)

Goods are nonstorable.

(a) Define an Arrow-Debreu equilibrium (ADE) for this economy. Carefully specify all elements of the
equilibrium.
(b) Compute the ADE allocations and prices. Make sure to derive optimality conditions and show the
steps to derive your solutions.
(c) Does the access to the Arrow-Debreu market make the two agents better off compared to them living
in an autarky? Compute the welfare gains or loss for each agent, by living in the Arrow-Debreu
economy compared to the autarky in terms of consumption equivalence.
(d) Suppose that a financial trader starts to sell a perpetual bond at time 0, which delivers 1.0 units of
consumption each period forever. What would be the price of this bond? How would the access to
the bond affect welfare of the individuals of this economy? Explain why.
 

15
Q4

Q4-(a)
Definition of an Arrow-Debreu equilibrium 

The Arrow-Debreu equilibrium in this economy consists of a set of allocations {c1t , c2t }∞
t=0 and a set of
prices { pt }∞
t=0 such that

(a) Given { pt }∞ i ∞
t=0 , for each i ∈ {1, 2}, { ct }t=0 solves

∞
max{ci }∞
t t =0
∑ βt log cit
t =0
∞ ∞
s.t. ∑ pt cit ≤ ∑ pt eit
t =0 t =0

∀t ≥ 0, cit ≥ 0

(b) The market clearing condition is

∀t ≥ 0, c1t + c2t = e1t + e2t (= 5)

 

Q4-(b)
We set up the Lagrnagean for agent i ∈ {1, 2} as

∞  ∞ ∞ 
Li = ∑ βt log cit + λi ∑ pt eit − ∑ pt cit
t =0 t =0 t =0

The FOCs are

βt
∀t ≥ 0, = p t λi
cit
∞ ∞
∑ pt eit = ∑ pt cit
t =0 t =0

The Euler Equation is

1 1
∀t ≥ 0, i
=β
pt ct pt+1 cit+1
Then, we have
∀t ≥ 0, pt+1 cit+1 = βpt cit

By summing up the equations over i, we obtain

   
∀t ≥ 0, pt+1 c1t+1 + c2t+1 = βpt c1t + c2t

16
The market clearing condition implies
   
∀t ≥ 0, pt+1 e1t+1 + e2t+1 = βpt e1t + e2t

Since the total endowment is invariant over time, we have

∀t ≥ 0, pt+1 = βpt
p t = β t p0

We normalize p0 to 1, and we get pt = βt , which means that for each i ∈ {1, 2}, the Euler Equation becomes

∀t ≥ 0, cit+1 = cit

Since the consumption is time-invariant, we let for each i ∈ {1, 2}, there exits c̄i ≥ 0 such that

∀t ≥ 0, cit = c̄i

In the utility maximization problem, the equality of the lifetime budget constraint is satisfied. Therefore,
for each i ∈ {1, 2},

∞ ∞
∑ βt c̄i = ∑ βt eit
t =0 t =0

For agent 1, we have

c̄1 4 β
= 2
+
1−β 1−β 1 − β2
4+β
c̄1 =
1+β

For agent 2, we get

c̄2 1 4β
= 2
+
1−β 1−β 1 − β2
1 + 4β
c̄2 =
1+β
   
4+ β 1+4β
The ADE allocations are ∀t ≥ 0, c1t , c2t = 1+ β , 1+ β and the ADE prices are ∀t ≥ 0, pt = βt .

Q4-(c)
Yes.
Both agents are better off in the Arrow-Debreu market than in the autarky.

17
 In the autarky, the welfare of agent 1 is

∞
1
∑ βt log e1t = 1 − β2 log 4
t =0

The welfare of agent 2 in the autarky is

∞
β
∑ βt log e2t = 1 − β2 log 4
t =0

The welfare of agent 1 in the Arrow-Debreu market is

∞
1 4+β
∑ βt log c1t = 1 − β log 1 + β
t =0

The welfare of agent 2 in this economy is

∞
1 1 + 4β
∑ βt log c2t = 1 − β log 1+β
t =0

Let µi denote a fraction of period consumption that agent i would require in the autarky so that they achieve
the same utility as in the Arrow-Debreu securities market.
For agent1,

1   β   1 4+β
2
log 4 ( 1 + µ 1 ) + 2
log 1 + µ 1 = log .
1−β 1−β 1−β 1+β

This generates
4+β 1
µ1 = − 1.
1 + β 4 1+1 β
For agent2,

1   β   1 1 + 4β
2
log 1 + µ 2 + 2
log 4 ( 1 + µ 2 ) = log .
1−β 1−β 1−β 1+β

Then, we obtain
1 + 4β 1
µ2 = − 1.
1 + β 4 1+β β

µ1 > 0 and µ2 > 0 show the weifare gains of both agents in terms of consumption equivalence.

18
Q4-(d)
The price of consumption good at time t is βt . At period 0, by arbitrage, the price of the perpetual bond can
be computed by
∞
1
∑ βt 1 = 1 − β .
t =0

Indeed, access to this derivative does not affect the individuals’ welfare because both the ADE and SME are
complete markets.

19
 Q5. Negishi’s Algorithm 

Consider an exchange economy populated by two individuals indexed by i = 1, 2, who live forever. Each
individual values sequences of consumption {cit }∞
t=0 according to preferences

∞
∑ βti log cit
t =0

with 0 < β 1 < β 2 . The endowment sequence is given by eit = 1 for all t ≥ 0 and i = 1, 2.

(a) Define a Pareto-efficient allocation for this economy. State the first welfare theorem for this economy.

(b) Define and solve the social planner’s problem for this economy. Denote by α = (α1 , α2 ), with
∑2i=1 αi = 1, Pareto weights that the planner assigns to the two agents. Also denote by µt the
Lagrange multiplier on the resource constraint in period t.

(c) Find the transfer payments

∞
ti ( α ) = ∑ µt (cit (α) − eit )
t =0

necessary for agents to afford the Pareto efficient allocation indexed by α. Express them as a function
of α, { β i }2i=1 and {eit }2i=1 .

(d) Demonstrate that the functions {ti (α)}2i=1 are homogeneous of degree 1 and that they sum to zero.

(e) By solving for the Pareto weights (α1 , α2 ), find the Arrow-Debreu equilibrium allocations and prices.
What is the consumption of the two individuals in the long-run, as t → ∞?
 

Q5

Q5-(a)
Definition of a Pareto efficient allocation 
n o∞
An allocation (cit )i∈{1,2} is Pareto-efficient if it is feasible and there is no other feasible allocations
n o∞ t =0
(c̃it )i∈{1,2} such that
t =0

∞ ∞
∀i ∈ {1, 2}, ∑ βt log c̃it ≥ ∑ βt log cit
t =0 t =0
∞ ∞
∃i ∈ {1, 2}, ∑ βt log c̃it > ∑ βt log cit
t =0 t =0
 

20
The First Welfare Theorem 
n o∞
Let (c∗t i )i∈{1,2} denote the competitive equilibrium allocation. Then, the allocation is Pareto-efficient.
t =0
 

Q5-(b)
Definition of the social planner’s problem 

The social planner’s problem with Pareto-weights α is

∞
max{(ci )i∈{1,2} }∞
t t =0
∑ ∑ αi βti log cit
t=0 i ∈{1,2}

s.t. ∀t ≥ 0, ∑ cit = ∑ eit (= 2)

i ∈{1,2} i ∈{1,2}

∀i ∈ {1, 2}, ∀t ≥ 0, cit ≥ 0

 
We set up the Lagrangian for the social planner as

∞   
L= ∑ ∑ αi βti log cit + µt ∑ eit − ∑ cit
t =0 i ∈{1,2} i ∈{1,2} i ∈{1,2}

The FOCs and the resource constraint are

αi βti
∀i ∈ {1, 2}, ∀t ≥ 0, = µt ,
cit
∑ cit = ∑ eit = 2.
i ∈{1,2} i ∈{1,2}

Then, ∀t ≥ 0,
α1 βt1 α2 βt2
=
c1t c2t
∀t ≥ 0,
t
α2

β2
c2t = 1 c1t
α β1
The resource constraint implies that
t
α2

β2
∀t ≥ 0, c1t + 1 c1t = 2
α β1

Generally, we have
2αi βti
∀i ∈ {1, 2}, ∀t ≥ 0, cit =
∑i∈{1,2} αi βti

∑i∈{1,2} αi βti
∀t ≥ 0, µt =
2

21
Q5-(c)

∞   ∞
∑i∈{1,2} αi βti

2αi βti

∀i ∈ {1, 2}, ti (α) = ∑ µt cit (α) − eit =∑
2 ∑i∈{1,2} αi βti
− eit
t =0 t =0

Finally, we obtain the transfer payments.

∞  ∑i∈{1,2} αi βti i 
∀i ∈ {1, 2}, ti (α) = ∑ αi βti −
2
et
t =0

Q5-(d)
Take ∀k > 0. Then, we have

∞  ∑i∈{1,2} kαi βti i 

ti (kα) = ∑ kαi βti − et
t =0 2
∞  ∑i∈{1,2} αi βti i 
=k ∑ αi βti − et
t =0 2
=kti (α).

Moreover, we obtain

∞ 
∑i∈{1,2} αi βti

∑ ti ( α ) = ∑ ∑ αi βti −
2
i ∈{1,2} i ∈{1,2} t=0
∞ ∞
= ∑ ∑ αi βti − ∑ ∑ αi βti
i ∈{1,2} t=0 i ∈{1,2} t=0

=0

Thus, the transfer functions are homogenous of degree one, and the sum is zero.

Q5-(e)
Using Negishi’s algorithm, we find α such that ∀i ∈ {1, 2}, ti (α) = 0. If the conditions are satisfied, the
Pareto efficient allocations and prices are equivalent to the Arrow-Debreu equilibrium prices and allocations,
respectively.
t1 (α) = 0 gives

∑i∈{1,2}
αi
α1 1− β i
= .
1 − β1 2

22
Similarly, t2 (α) = 0 implies
∑i∈{1,2}
αi
α2 1− β i
= .
1 − β2 2
By combining these two equations, we get

α1 α2
= ,
1 − β1 1 − β2
α1 (1 − β 2 )
α2 = .
1 − β1

Since ∑i∈{1,2} αi = α1 + α2 = 1, we have

1 − β1
α1 = ,
2 − β1 − β2

1 − β2
α2 = .
2 − β1 − β2
2αi βti
Substituting α1 and α2 into cit = ∑i∈{1,2} αi βti
yields Arrow-Debreu equilibrium consumption ∀t ≥ 0,

2 2
c1t = 1− β 2 β 2 t
, c2t = 1− β 1 β 1 2
.
1+ 1− β 1 ( β 1 ) 1− β 2 ( β 2 ) +1

The Arrow-Debreu equilibrium prices are ∀t ≥ 0,

∑i∈{1,2} αi βti
µt =
2
∑i∈{1,2} (1 − β i ) βti
=
2(2 − ∑i∈{1,2} β i )

By the assumption β 2 > β 1 , as t approaches ∞,

 t
β2
→∞
β1
 t
β1
→0
β2

Therefore, as t → ∞, the consumption of the three individuals becomes

(c1t , c2t ) → (0, 2)

23
 Q6. Arrow-Debreu and Sequential Market Equilibria 

Consider a two-period exchange economy populated by two types of individuals indexed by i = 1, 2.

Type 1 individual has an endowment sequence of (e1 , 0) and type 2 has an endowment of (0, e2 ). They
have preferences over a life-time consumption {c1i , c2i } given as follows.

u(c1i , c2i ) = ln c1i + β ln c2i

where β > 0.

(a) Define a competitive Arrow-Debreu equilibrium (ADE) of this economy.

(b) Now allow for a bond market, in which individuals can borrow and lend at interest rate r. Define a
sequential market equilibrium (SME) for this economy.
(c) Show that allocations are identical in the ADE and SME.
(d) Suppose e1 = e2 = 1 and β = 0.9. Compute the competitive equilibrium allocations and prices in the
ADE. Who consumes more in each period? Provide the intuition. What if β = 1.0?
 

Q6

Q6-(a)
Definition of Arrow Debreu Equilibrium 

A competitive Arrow-Debreu equilibrium in this economy consists of a set of prices ( p1 , p2 ) and allocations
(c11 , c12 , c21 , c22 ) such that

(a) Given ( p1 , p2 ), for i = 1, 2, (c1i , c2i ) solves

max {ln c1i + β ln c2i }

c1i ,c2i

s.t. p1 c1i + p2 c2i ≤ p1 e1i + p2 e2i ,

c1i , c2i ≥ 0.

(b) c1t + c2t = e1t + e2t , t = 1, 2.

 

24
Q6-(b)
Definition of Sequential Market Equilibrium 

A sequential market equilibrium in this economy consists of an interest rate r and allocations
(c11 , c12 , a12 , c21 , c22 , a22 ) such that

(a) Given r, for i = 1, 2, (c1i , c2i , a2i ) solves

max {ln c1i + β ln c2i }

c1i ,c2i ,a2i

a2i
s.t. c1i + ≤ e1i ,
1+r
c2i ≤ e2i + a2i ,
c1i , c2i ≥ 0.

(b) c1t + c2t = e1t + e2t , t = 1, 2, and a12 + a22 = 0.

(Assume initial bond holdings are zero, a11 = a21 = 0.)

 

Q6-(c)
We first compute the allocations (c11 , c12 , c21 , c22 ) in the ADE. Set up Lagrangian for the utility maximization
problem of agent i ∈ {1, 2}:

Li = ln c1i + β ln c2i + λi [ p1 e1i + p2 e2i − ( p1 c1i + p2 c2i )].

The first-order necessary conditions are therefore

1
= p1 λ i ,
c1i

β
= p2 λ i ,
c2i

p1 c1i + p2 c2i = p1 e1i + p2 e2i budget constraint.

Using optimality conditions, we get the Euler equation

p2 c2i = βp1 c1i .

Summing this over i, we get

p2 (c12 + c22 ) = βp1 (c11 + c21 ).

25
Then, the goods market clearing gives

p2 (e21 + e22 ) = βp1 (e11 + e12 ).

Since e11 + e12 = e1 and e21 + e22 = e2 , we have

p2 e2 = βp1 e1 .

βe1
Normalizing p1 = 1, we obtain p2 = e2
. The Euler equation becomes

e1 i
c = c1i ,
e2 2

Thus, the budget constraint becomes

e1 i βe1 i i βe1 i
c 2 + c 2 = e + e .
e2 e2 1
e2 2

which gives the allocations in the ADE as

e1
c1,ADE
1 = ,
1+β
e2
c1,ADE
2 = ,
1+β
βe1
c2,ADE
1 = ,
1+β
βe2
c2,ADE
2 = .
1+β
Now we compute the allocations (c11 , c12 , c21 , c22 ) in the SME. The constraints

a2i
c1i + ≤ e1i
1+r

and
c2i ≤ e2i + a2i

give the lifetime budget constraint

c2i ei
c1i + ≤ e1i + 2 .
1+r 1+r
Set up Lagrangian for the utility maximization problem of agent i ∈ {1, 2}:

e2i c2i
  
i
L = ln c1i + β ln c2i +λ i
e1i + i
− c1 + .
1+r 1+r

26
The first-order necessary conditions are therefore

1
= λi ,
c1i

β λi
= ,
c2i 1+r

c2i ei
c1i + = e1i + 2 budget constraint.
1+r 1+r
Eliminating λi yields the Euler equation
c2i = β(1 + r )c1i .

Summing this over i, we get

c12 + c22 = β(1 + r )(c11 + c21 ).

The goods market clearing condition implies

e2 = β (1 + r ) e1 ,

that is,
e2
1+r = .
βe1
Plugging this into the Euler equation, we obtain

e1 i
c = c1i .
e2 2

Then, the lifetime budget constraint becomes

e1 i βe1 i e1 i i βe1 i
c 2 + c 2 = ( 1 + β ) c 2 = e + e .
e2 e2 e2 1
e2 2
βe1 i e2 βe1 i βe2
For i = 1, e1i + e
e2 2
= e1 , so c12 = 1+ β . For i = 2, e1i + e
e2 2
= βe1 , so c22 = 1+ β . These results and the Euler
equation generate
e1 e2 βe1 2,SME βe2
c1,SME
1 = , c1,SME
2 = , c2,SME
1 = , c2 = .
1+β 1+β 1+β 1+β

These are exactly the same as c1,ADE

1 , c1,ADE
2 , c2,ADE
1 , c2,ADE
2 , so the allocations in the ADE are the same as in
SME.

Q6-(d)
Plugging e1 = e2 = 1 and β = 0.9 into c1,ADE
1 , c1,ADE
2 , c2,ADE
1 , c2,ADE
2 , we obtain

1 10
c1,ADE
1 = = ,
1.9 19

27
 1 10
c1,ADE
2 = = ,
1.9 19
0.9 9
c2,ADE
1 = = ,
1.9 19
0.9 9
c2,ADE
2 = = .
1.9 19
βe1
Then, we have p2 = e2 = 0.9, while p1 is normalized as 1.
As we can see, agent 1 consumes more than agent 2 in both periods. The intuition is as follows: β is the
factor agents used to discount future utilities (coming from consuming goods), and β < 1 reflects the fact
that the agents are impatient, that is, by consuming the same amount of goods they obtain more utility today
than in the future. This equivalently means that, the same amount of resource today is more valuable than
in the future, so that agent 1’s “life-time" endowment has a higher value than that of agent 2, so that agent
1 consumes more than agent 2 in both periods. Consumption is the same among agents if β = 1 because
endowment at period 2 is as valuable as that at period 1, and the amount of endowment each agent has is
identical e1 = e2 .

28