Bewley Models

Prof. Lutz Hendricks

Econ720

November 22, 2019

1 / 19
Bewley Models

For many applications we need models with heterogeneous

agents.
In Bewley models, agents are ex ante identical.
They are ex post heterogeneous because they are hit by
idiosyncratic shocks.
Incomplete markets prevents sharing these risks.

In this section, we study a simple endowment economy.

The goal is to get the mechanics down in a simple setting.

2 / 19
Endowment Economy
An Endowment Economy

I Demographics:
I There is a unit measure of households.
I Each lives forever.
I Preferences:
∞
E ∑ β t u (ct ) (1)
t=0

I Technology:
I Households receive random endowments yt ∈ Y (finite).
I Transition matrix: π (y0 |y).

4 / 19
No aggregate uncertainty

I Assume a "law of large numbers."

I Let Π (y) be the stationary distribution of y.
I Assume that the fraction of households with endowment y is
Π (y).
I The aggregate endowment ȳ is constant over time.
I With complete markets, households would not face any
uncertainty.

5 / 19
Household problem

I Flow budget constraint:

a0 = y + (1 + r) a − c (2)

I Borrowing constraint:
a0 ≥ −b (3)

6 / 19
Household problem

I Focus on a stationary equilibrium.

I Meaning: Aggregates & prices are constant over time.
I State vector: (a, y).
I Given: r.
I Bellman equation:

v (a, y) = max u (c) + β ∑ π y0 |y v a0 , y0

(4)
y0

s.t. budget constraint and borrowing constraint.

7 / 19
Household problem I
The borrowing constraint may bind. We need Kuhn-Tucker.
Bellman equation

v (a, y) = max u y + (1 + r) a − a0


(5)
+β ∑ π y0 |y v a0 , y0 + µ(a0 + b)



(6)
y0

First-order conditions:

∂ v (a0 , y0 )
u0 (c) = β ∑ π y0 |y +µ (7)
∂ a0
∂ v/∂ a = u0 (c) (1 + r) (8)
0
µ(a + b) = 0 (9)

8 / 19
Household problem

I Euler equation
u0 (c) ≥ Eβ (1 + r) u0 c0


(10)
with equality if a0 > −b.
I Solution: v (a, y) , c (a, y) , a0 (a, y) that satisfy the usual
conditions.

9 / 19
How does the household behave?

Assume (for the moment) that y ∼ iid

Then (obviously?), consumption and saving depend on "cash on
hand"
x = y + (1 + r) a (11)
Also (obviously?), a0 is increasing in x
If x is sufficiently high: Choose a0 > −b and satisfy standard Euler
equation.
If x is below a cutoff, set a0 = −b and "violate" the Euler equation.
The borrowing constraint depresses current consumption.

10 / 19
Stationary Recursive Competitive Equilibrium

I Aggregate state: The joint distribution of assets and

endowments: Φ (a, y).
I This is needed to compute aggregates.
I Objects:
I Household: v (a, y) , c (a, y) , a0 (a, y).
I Φ (a, y).
I Price function: r (Φ).
I Equilibrium conditions:
I Household: see above.
I Market clearing.
I Time invariance of Φ.

11 / 19
Stationary Recursive Competitive Equilibrium
Market clearing

I Goods:
Z Z Z
C= c (a, y) Φ (da, dy) = yΠ (dy) = ȳ (12)

I Bonds: Z Z
a0 (a, y) Φ (da, dy) = 0 (13)

12 / 19
Time invariance of the distribution

I Informally, household choices determine tomorrow’s

distribution Φ0 .
I The policy function a0 (a, y) implies a law of motion for Φ.
I In stationary equilibrium, the law of motion must imply Φ0 = Φ.

13 / 19
Law of motion for the distribution

I Define a transition function Q ((a, y) , (A, Y)).

I Its value is the probability (mass) of households in state (a, y)
today that end up in (a0 , y0 ) ∈ (A, Y) tomorrow.
(
∑y0 ∈Y π (y0 |y) if a0 (a, y) ∈ A
Q ((a, y) , (A, Y)) =
0 otherwise

I This is because a0 is deterministic.

14 / 19
Law of motion for the distribution

I Law of motion
Z Z
Φ0 (A, Y) = Q ((a, y) , (A, Y)) Φ (da, dy) (14)

I In words:
I Φ0 (A, Y): What is the mass of households in the set of states
(A, Y) tomorrow?
I For any (a, y), this is given by Q ((a, y) , (A, Y)).
I Sum over all (a, y) to get the total mass.
I Stationarity then means: Φ0 (A, Y) = Φ (A, Y) for all (A, Y).

15 / 19
Non-stationary Equilibrium

I The equilibrium concept generalizes easily to economies where

Φ evolves over time.
I Household:
I Add the aggregate state Φ to the household’s state: v (a, y, Φ)
and a0 (a, y, Φ).
I The household takes prices as functions of the aggregate state:
r (Φ).
I The household knows the law of motion for Φ: Φ0 = H (Φ).
I Equilibrium:
I Drop stationarity of Φ.

16 / 19
Where is this useful?

I Models of the wealth distribution:

I Krusell and Smith (1998)
I Models of business cycles with heterogeneous agents:
I Castaneda et al. (1998)

17 / 19
Reading

I Acemoglu (2009), ch. 17.4.

I Krueger, "Macroeconomic Theory," ch. 10.

18 / 19
References I

Acemoglu, D. (2009): Introduction to modern economic growth,

MIT Press.
Castaneda, A., J. Dıaz-Giménez, and J.-V. Rıos-Rull (1998):
“Exploring the income distribution business cycle dynamics,”
Journal of Monetary economics, 42, 93–130.
Krusell, P. and J. Smith, Anthony A. (1998): “Income and Wealth
Heterogeneity in the Macroeconomy,” The Journal of Political
Economy, 106, 867–896.

19 / 19