11.

Fiscal Policy in the Growth Model

Yvan Becard
PUC-Rio

Macroeconomics I, 2023
 Introduction

▶ Today we return to the deterministic neoclassical growth model

▶ We introduce fiscal policy in the form of government spending and taxes
▶ The goal is to understand how government policies affect the behavior of
private agents – households and firms – and aggregate economic outcomes

2
 In a Nutshell

▶ In the basic neoclassical growth model, the resource constraint is

Yt = Ct + It

▶ We now add an important component of GDP

Yt = Gt + Ct + It

3
Government Spending Share of Output

4
Public Debt in the United States

5
Public Debt in Brazil

6
 Lecture Outline

1. Model Setup 6. Inelastic Labor Supply

2. Sequential Budget Constraint 7. Steady State
3. Term Structure of Interest Rates 8. Equilibrium Path
4. Competitive Equilibrium 9. Fiscal Policy Experiments
5. Ricardian Equivalence 10. Exercises

Main Reference: Ljungqvist and Sargent, 2018, Recursive Macroeconomic

Theory, Fourth Edition, Chapter 11

7
1. Model Setup
 Model Diagram

Government

Tax capital income, labor income, consumption

Purchase goods, issue debt

Supply capital,
Markets Demand capital,
Households supply labor,
demand goods goods, labor, capital
demand labor,
supply goods
Firms

9
 Preferences

▶ There is no uncertainty, agents have perfect foresight

▶ Time is normalized to unity, ℓt + nt = 1
▶ A representative household enjoys consumption and leisure
∞
X
β t U (ct , 1 − nt )
t=0

▶ U is strictly increasing in consumption ct and leisure 1 − nt , is twice

continuously differentiable, strictly concave, satisfies the Inada conditions

10
 Resource Constraint

▶ The aggregate resource constraint is

gt + ct + it ≤ F (kt , nt )

▶ All produced goods in the economy are either consumed by the government
gt , consumed by households ct , or saved by households it
▶ The production function F (k, n) is linearly homogeneous, has positive and
decreasing marginal products, and satisfies the Inada conditions

11
 Capital Accumulation

▶ Capital evolves according to

kt+1 = (1 − δ)kt + it

▶ Plug that into the resource constraint

gt + ct + kt+1 ≤ F (kt , nt ) + (1 − δ)kt

12
 Time Zero Trading

▶ Let’s consider a competitive equilibrium with all trades occurring at time 0

▶ The representative household owns the capital stock, makes investment
decisions, rents out capital and labor to firms, and buys goods from firms
▶ The representative firm finances the rental of capital and labor with sales of
output goods it produces using the production function F (kt , nt )
▶ Even though there is no uncertainty, the household and the firm meet in
time 0 and trade securities for every period from t = 0 to the infinite future

13
 Price System

▶ A price system is a triple of sequences {qt , rt , wt }∞

t=0
▶ qt is the time 0 pretax price of one unit of good at time t
▶ rt is the pretax rental rate of capital at time t
▶ wt is the pretax wage at time t
▶ qt is expressed in terms of time 0 goods, wt and rt are expressed in terms of
time t goods

14
 Government

▶ Government behavior is exogenous

▶ The government finances a sequence of goods purchases {gt }∞ t=0 from firms
with a sequence of taxes {τtc , τtk , τtn , τt }∞
t=0 on households
▶ τtc is tax rate on consumption
▶ τtk is tax rate on capital income
▶ τtn is tax rate on labor income
▶ τt is a lump-sum tax

15
 Lump-Sum Tax

▶ A lump-sum tax is a tax based on a fixed amount; eg a flat fee for all workers
in the township (a head tax); or a flat fee to register a vote (a poll tax)
▶ It is independent of the taxpayers’ actions who pay it no matter what
▶ Since agents cannot affect the amount of the lump-sum tax by changing
their behavior, the tax implies no distortion in choice
▶ Lump-sum taxes play a special role in macroeconomic theory

16
 Government Budget Constraint

▶ The government faces the following time 0 budget constraint

∞
X ∞
X h i
qt gt ≤ qt τtc ct + τtk (rt − δ)kt + τtn wt nt + τt (1)
t=0 t=0

▶ The lifetime market value of purchases cannot exceed that of tax revenues
▶ A government expenditure and tax plan is budget feasible if it satisfies the
government budget constraint

17
 Household Budget Constraint

▶ The household faces the time 0 budget constraint

∞
X ∞
X n o
qt {(1 + τtc )ct + [kt+1 − (1 − δ)kt ]} ≤ k n
qt rt kt − τt (rt − δ)kt + (1 − τt )wt nt − τt
t=0 t=0

▶ Notice the positive sign of τtc and negative signs of τtk τtn , and τt
▶ The government gives a depreciation allowance δkt from the gross rentals
on capital rt kt and so collects τtk (rt − δ)kt on rentals from capital

18
2. Sequential Version of Government Budget Constraint
 From Time Zero to Sequential Trading

▶ We have used the time 0 trading abstraction seen in lecture 6

▶ Sequential trading of one-period risk-free debt can also support the
equilibrium allocations we will study today
▶ Let’s describe the sequence of one-period government debt that is implicit in
the equilibrium tax policies here

20
 Government Budget Constraint

▶ Suppose the government enters period 0 with no government debt

▶ Define total tax collections as

Tt ≡ τtc ct + τtk (rt − δ)kt + τtn wt nt + τt

▶ The government budget constraint (1) becomes

∞
X
qt (gt − Tt ) = 0
t=0

21
 Deficit Today Is Surplus Tomorrow

▶ The previous equation can be written as

∞
X
q0 (g0 − T0 ) + qt (gt − Tt ) = 0
t=1
∞
X qt
or g0 − T0 = (Tt − gt )
q0
t=1

▶ The government deficit at time 0, g0 − T0 , equals the present value of future

government surpluses, ∞ qt
P
t=1 q0 (Tt − gt )

22
 Introducing Debt
▶ In the previous equation, B0 ≡ ∞ t=1 q0 (Tt − gt ) is the value of government
qt
P
debt issued at time 0, denominated in units of time 0 goods
∞
X qt
g0 − T0 = B0 where B0 ≡ (Tt − gt )
q0
t=1

▶ Rewrite the value of debt at time 0

∞
q1 X qt
B0 = (T1 − g1 ) + (Tt − gt )
q0 q0
t=2

▶ Multiply by q0 /q1 on both sides

∞
q0 X qt
B0 = T1 − g1 + (Tt − gt )
q1 q1
t=2

23
 Real Interest Rate

▶ Define R0,1 ≡ q0
q1 as the gross one-period real interest rate
∞
X qt
B0 R0,1 = T1 − g1 + B1 where B1 = (Tt − gt )
q1
t=2

▶ B1 is the value of one-period government debt issued at time 1 and repaid

at time 2, expressed in units of time 1 goods

24
 Flow Budget Constraint

▶ Express the last equation in terms of t and t − 1 periods to get a sequence of

period-by-period government budget constraints

gt + Rt−1,t Bt−1 = Tt + Bt (2)

▶ One for each t ≥ 1, where Bt is issued in t and repaid in t + 1

∞
qt−1 X qs
Rt−1,t = and Bt = (Ts − gs )
qt qt
s=t+1

▶ The left side of (2) is time t government expenditures including interest and
principal payments; the right side is total revenues including new debt

25
 Today’s Deficit Is Tomorrow’s Surplus

▶ Thus, embedded in the time 0 government budget constraint (1) is a

sequence of one-period government debts satisfying Bt ≡ ∞ qs
P
s=t+1 qt (Ts − gs )
▶ The value of government debt at t is equal to the present value of
government surpluses from date t + 1 onward
▶ Government debt at time t signals future government budget surpluses

26
3. The Term Structure of Interest Rates
 A Digression

▶ The price system {qt }∞t=0 embeds within it a term structure of interest rates
▶ To see this, write qt as
q1 q2 qt
qt = q0 ...
q0 q1 qt−1
qt+1
= q0 m0,1 m1,2 . . . mt−1,t where mt,t+1 ≡
qt
▶ mt,t+1 is the one-period (nonstochastic) discount factor

28
 Gross and Net

▶ Represent the discount factor as

1 1
mt,t+1 = = ≈ exp(−r̄t,t+1 )
Rt,t+1 1 + r̄t,t+1

▶ Rt,t+1 is the one-period gross interest rate between t and t + 1

▶ r̄t,t+1 is the one-period net interest rate between t and t + 1

29
 Short and Long

▶ Notice qt can also be expressed as

qt = q0 exp(−r̄0,1 ) exp(−r̄1,2 ) · · · exp(−r̄t−1,t )

= q0 exp[−(r̄0,1 + r̄1,2 + · · · + r̄t−1,t )]
= q0 exp(−tr̄0,t )

where r̄0,t is the net t-period interest rate between 0 and t

r̄0,t = t−1 (r̄0,1 + r̄1,2 + · · · + r̄t−1,t )

30
 Zero Coupon Bond

▶ qt is the time 0 price of one unit of time t consumption

▶ So r̄0,t is the yield to maturity on a zero coupon bond that matures at t
▶ A zero coupon bond promises no coupon before maturity and pays only the
principal due at the date of maturity t

31
 Yield Curve

▶ r̄0,t = t−1 (r̄0,1 + r̄1,2 + · · · + r̄t−1,t ) expresses the expectations theory of the
term structure of interest rates
▶ The theory states that interest rates on long (t-period) loans are averages of
rates on short (one-period) loans expected to prevail over the long horizon
▶ More generally, the s-period long rate at time t is

1
r̄t,t+s = (r̄t,t+1 + r̄t+1,t+2 + · · · + r̄t+s−1,t+s )
s
▶ A graph of r̄t,t+s against s is called the yield curve at t

32
 Yield Curve

Yield
7%
6%
5%
4%
3%
2%
1%
0% Years
0 5 10 15 20 25 30
Time to maturity
33
4. Competitive Equilibrium
 Household Budget Constraint

▶ Back to the model: collect capital terms in the household budget constraint
∞
X n o
qt rt kt − τtk (rt − δ)kt − [kt+1 − (1 − δ)kt ]
t=0
∞
X n o
= qt [(1 − τtk )(rt − δ) + 1]kt − kt+1
t=0

= q0 [(1 − τ0k )(r0 − δ) + 1]k0 − q0 k1 + q1 [(1 − τ1k )(r1 − δ) + 1]k1 − q1 k2 + . . .

+ · · · + qT [(1 − τTk )(rT − δ) + 1]kT − qT kT +1
∞ n
X o
= [(1 − τ0k )(r0 − δ) + 1]q0 k0 + [(1 − τtk )(rt − δ) + 1]qt − qt−1 kt − lim qT kT +1
T →∞
t=1

35
 Household Budget Constraint

▶ The household budget constraint thus writes

∞
X ∞
X ∞
X ∞ n
X o
qt [(1 + τtc )ct ] ≤ qt (1 − τtn )wt nt − qt τt + [(1 − τtk )(rt − δ) + 1]qt − qt−1 kt
t=0 t=0 t=0 t=1
+ [(1 − τ0k )(r0 − δ) + 1]q0 k0 − lim qT kT +1
T →∞

▶ All other things being equal, the household wants to increase resources
(right side) to buy more consumption goods (left side)

36
 Unbounded Profit

▶ The household assembles capital in t − 1 at cost qt−1 kt ; it then rents it out in

t and sells the undepreciated stock to make a return [(1 − τtk )(rt − δ) + 1]qt kt
▶ If [(1 − τtk )(rt − δ) + 1]qt − qt−1 > 0, the agent makes profit and wants kt → ∞
▶ If [(1 − τtk )(rt − δ) + 1]qt − qt−1 < 0, the agent short sells capital, ie it sells
“synthetic” units in t − 1 and buys these units back at a lower price in t
▶ That is, the agent makes profit and wants kt → −∞

37
 No Arbitrage

▶ In equilibrium, capital is nonnegative and bounded, therefore it must be

that the terms multiplying k equal zero
qt k
= [(1 − τt+1 )(rt+1 − δ) + 1] for all t ≥ 1
qt+1
▶ This is the zero-profit, or no-arbitrage, condition

38
 Transversality Condition

▶ A similar argument holds for the limiting term − limT →∞ qT kT +1

▶ The household wants this term to be as small as possible, by short selling
capital in the limit, in order to buy unlimited amounts of consumption goods
▶ But the market would stop the household from undertaking such a short
sale since there would be no party on the other side of the transaction
▶ Thus, as a condition of optimality, we have the transversality condition

− lim qT kT +1 = 0
T →∞

39
 Household Problem

▶ To solve the household problem, we write a Lagrangian

∞
X ∞
X
L= β t U (ct , 1 − nt ) + µ qt {(1 − τtn )wt nt − τt − (1 + τtc )ct }
t=0 t=0

▶ Note the budget constraint is not discounted: implicitly, µ is a present-value

multiplier, which embeds discounting

40
 First-Order Conditions

▶ The first-order conditions are, for each t

ct : β t Uc (ct , 1 − nt ) = µqt (1 + τtc )

nt : β t Un (ct , 1 − nt ) = µqt wt (1 − τtn )

▶ We verify indeed that µ depends on β t

41
 Firm Problem

▶ The firm solves

∞
X
max qt [F (kt , nt ) − wt nt − rt kt ]
kt ,nt
t=0
▶ The first-order conditions are, for each t

kt : rt = Fk (kt , nt )
nt : wt = Fn (kt , nt )

42
 Competitive Equilibrium

A competitive equilibrium with distorting taxes is a price system {qt , rt , wt }, a

budget-feasible government policy {gt , τtc , τtn , τtk , τt }, and an allocation
{ct , nt , kt+1 } that solve the system of difference equations consisting of the
resource constraint (or feasible allocation), the zero-profit (or no-arbitrage)
condition, the household’s first-order conditions, and the firm’s first-order
conditions, subject to the initial condition on k0 and the terminal condition

43
 Summary of Equilibrium Conditions

▶ We end up with a system of six equations for six endogenous variables

t=0 , given the exogenous variables {gt , τt , τt , τt , τt }t=0
{ct , kt , nt , qt , rt , wt }∞ c k n ∞

gt + ct + kt+1 = F (kt , nt ) + (1 − δ)kt

k
qt = [(1 − τt+1 )(rt+1 − δ) + 1]qt+1
β t Uc (ct , 1 − nt ) = µqt (1 + τtc )
β t Un (ct , 1 − nt ) = µqt wt (1 − τtn )
rt = Fk (kt , nt )
wt = Fn (kt , nt )

44
 Exogenous But Constrained

▶ On top of the previous six equations, the budget constraint of the

government must hold
∞
X ∞
X h i
qt gt ≤ qt τtc ct + τtk (rt − δ)kt + τtn wt nt + τt
t=0 t=0

▶ Thus, one of the five exogenous variables {gt , τtc , τtk , τtn , τt }∞
t=0 must adjust to
satisfy this equation

45
5. Ricardian Equivalence
 No Lump-Sum Tax in Equilibrium

▶ Notice that the lump-sum tax τt does not appear in any of the system’s six
equilibrium conditions
▶ But its present value ∞t=0 qt τt does appear in the time 0 government
P
budget constraint

47
 No Lump-Sum Tax in Equilibrium

▶ We conclude two things

1. The path or timing of lump-sum taxes has no effect on the model’s dynamics
2. Only the present value of current and future taxes matters
▶ Let’s explore these results in detail

48
 Two Modifications

▶ We consider the same model

▶ But to simplify we set all distorting taxes to zero, τtk = τtc = τtn = 0
▶ Also, we introduce government debt

49
 Government Budget Constraint

▶ The government issues debt bt at price qt

▶ bt is one-period debt due at t, denominated in t goods
▶ The government’s period t budget constraint writes

qt gt + qt bt = qt τt + qt+1 bt+1
| {z } | {z }
expenses resources

▶ Solve for τt
qt+1
τt = gt + bt − bt+1
qt

50
 Household Budget Constraint

▶ The household’s period t budget constraint is

qt ct + qt [kt+1 − (1 − δ)kt ] + qt+1 at+1 = qt wt nt + qt rt kt + qt at − qt τt

▶ at is financial wealth excluding capital holdings

▶ Divide by qt
qt+1
ct + kt+1 − (1 − δ)kt + at+1 = wt nt + rt kt + at − τt
qt

51
 Merging the Two Budget Constraints

▶ Plug the government budget constraint into the household budget

constraint to substitute out for τt
qt+1 qt+1
ct + kt+1 − (1 − δ)kt + at+1 = wt nt + rt kt + at − gt − bt + bt+1
qt qt
▶ In equilibrium, demand of bonds equals supply, at = bt , and hence we obtain

ct + kt+1 − (1 − δ)kt = wt nt + rt kt − gt

▶ Taxes τt and debt bt no longer appear in the household budget constraint

▶ The household only cares about government spending gt

52
 More Government Spending Makes Me Poorer

▶ From the previous equation, we see that an increase in government

spending gt directly reduces the household’s resources
▶ Why? The reason is that the household is rational and infinitely-lived and
therefore internalizes the budget constraint of the government
▶ The household knows that an increase in government expenditure gt must
be financed either by an increase in taxes τt or by an increase in debt bt

53
 Irrelevant Tax-Debt Mix

▶ But to repay debt in the future, taxes will have to go up

▶ Higher debt is simply future higher taxes
▶ Thus the choice is between higher taxes today or higher taxes tomorrow
▶ For a rational and infinitely-lived household, that is equivalent
▶ Thus it does not matter how the government finances its spending: the
tax-debt mix is irrelevant

54
 Ricardian Equivalence

▶ The way the government finances its expenditures – through lump-sum

taxes or debt – has no impact on the dynamics of the economy
▶ This is the Ricardian equivalence proposition, discussed first by Ricardo
(1820) and shown formally by Barro (1974) in a seminal paper
▶ An increase in government spending makes households feel poorer today
▶ Again, this is because they realize that taxes will have to go up, either today
or in the future, to finance the increase in spending

55
 Alternative Way

▶ We can show Ricardian equivalence in another way

▶ Solve for bt in the government’s period t budget constraint
qt+1
bt = τt − gt + bt+1
qt
qt+1 qt+2
▶ Solve this equation forward: bt = τt + gt + qt (τt+1 + gt+1 ) + qt bt+2 or
∞
X qt+s qt+T +1
bt = (τt+s − gt+s ) + lim bt+T +1
qt T →∞ qT
s=0 | {z }
=0
▶ The transversality condition rules out a government-induced Ponzi scheme

56
 Household Budget Constraint

▶ Do the same for assets at in the household’s period t budget constraint

∞
X qt+s
at = (ct+s + kt+1+s − (1 − δ)kt+s + τt+s − wt+s nt+s − rt+s kt+s )
qt
s=0

▶ Since ct + kt+1 − (1 − δ)kt = F (kt , nt ) − gt and since zero profit implies

F (kt , nt ) − wt nt − rt kt = 0, we have
∞
X qt+s
at = (τt+s − gt+s )
qt
s=0

57
 Spending Is Taxing

▶ We conclude for all t

∞
X qt+s
at = (τt+s − gt+s ) = bt
qt
s=0

▶ Suppose that in period t, government debt is zero, at = bt = 0

∞ ∞
X qt+s X qt+s
gt+s = τt+s
qt qt
t=0 s=0

▶ The present value of government spending equals the present value of taxes

58
 Conditions for Ricardian Equivalence

▶ The conditions for Ricardian equivalence to hold are threefold

1. Rational and infinitely-lived households
2. Complete markets
3. Lump-sum taxes
▶ With uncertainty, Ricardian equivalence holds
▶ But with distorting taxes, Ricardian equivalence breaks down

59
6. Inelastic Labor Supply
 Inelastic Labor Supply

▶ Let’s consider the simplifying case

U (c, 1 − n) = u(c)

▶ No utility from leisure, so constant labor supply n = 1

▶ Also, define f (k) = F (k, 1), so Fk (k, 1) = f ′ (k) and Fn (k, 1) = f (k) − f ′ (k)k

61
 Inelastic Labor Supply

▶ Rewrite the previous system with nt = n = 1

gt + ct + kt+1 = f (kt ) + (1 − δ)kt (3)

qt = [(1 − τt+1k
)(rt+1 − δ) + 1]qt+1 (4)
t ′
β u (ct ) = c
µqt (1 + τt ) (5)
rt = f ′ (kt ) (6)
wt = f (kt ) − f ′ (kt )kt (7)

▶ The household’s FOC for labor drops out

62
 Non-Distorting Labor Income Tax

▶ The labor income tax τtn no longer appears in the equilibrium conditions
▶ Thus when the labor supply is inelastic, ie constant, the labor income tax is
non distortionary
▶ Intuitively, since workers have no marginal disutility of work, taxing their
labor income will not affect their willingness to work

63
 Reducing the System

▶ Combine equations (4), (5), and (6) of the system

1 + τtc h i
u′ (ct ) = βu′ (ct+1 ) c (1 − τ k
t+1 )[f ′
(kt+1 ) − δ] + 1
1 + τt+1

▶ The resource constraint (3) is

ct = f (kt ) + (1 − δ)kt − kt+1 − gt

▶ We are down to a system of two difference equations for two endogenous

variables, ct and kt , given exogenous variables gt , τtk , and τtc

64
 Non-Distorting Constant Consumption Tax

▶ Suppose the consumption tax is constant, τtc = τt+1 c = τ̄ c

▶ With an inelastic labor supply, a constant consumption tax is non
distortionary
h i
′ ′ k ′
u (ct ) = βu (ct+1 ) (1 − τt+1 )[f (kt+1 ) − δ] + 1

65
7. Steady State
 Constant Fiscal Policy

▶ We assume that fiscal policy is eventually constant

lim gt = ḡ lim τtk = τ̄ k lim τtc = τ̄ c

t→∞ t→∞ t→∞

▶ Given this assumption and those on the production and utility functions, we
know that the system converges to a unique steady state

67
 Steady State

▶ The system in steady state is

h i
1 = β (1 − τ̄ k )(f ′ (k̄) − δ) + 1
c̄ = f (k̄) − δ k̄ − ḡ

▶ In a steady state with inelastic labor supply, only the capital income tax is
distorting, ie changes the behavior of agents

68
8. Equilibrium Path
 System

▶ Our system is

1 + τtc h i
u′ (ct ) = βu′ (ct+1 ) c (1 − τ k
t+1 )[f ′
(kt+1 ) − δ] + 1 (8)
1 + τt+1
ct = f (kt ) + (1 − δ)kt − kt+1 − gt (9)

70
 Optimal Path

▶ We know the initial condition k0

▶ We have the terminal steady state k̄ and c̄
▶ We know the path of exogenous variables gt , τtc , τtk
▶ We want to compute the equilibrium path that starts from the initial
condition and ends at the terminal value

71
 Shooting Algorithm

▶ We are going to use a shooting algorithm, a numerical method for solving a

two-point boundary value problem
▶ We “shoot” out trajectories in different directions until we find a trajectory
that has the desired boundary value
▶ We search for an initial c0 that makes the two equations (8) and (9) imply
kS ≈ k̄, where S is a finite but large time index

72
 Steps

1. Solve the steady state

2. Select a large time index S and guess an initial consumption c0 ; compute
u′ (c0 ) and solve (9) for k1
3. Given ct and kt+1 , use (8) and (9) to compute ct+1 and kt+2
4. Iterate on step 3 to compute candidate values k̂t , t = 1, . . . , S
5. Compute k̂S − k̄; if k̂S > k̄, raise c0 and compute a new k̂t ; if k̂S < k̄, lower c0
and compute a new k̂t
6. In this way search for a c0 that makes k̂S ≈ k̄

73
 After Convergence

▶ If S is not big enough the algorithm might not converge

▶ We can implement the shooting algorithm in Matlab
▶ Dynare implements the shooting algorithm
▶ Once we solve for the equilibrium {kt } sequence, we recover all variables
using the system’s other equations

74
 Balanced Government Budget

▶ How do we ensure the time 0 government budget constraint is satisfied?

▶ Two cases
1. The government can impose lump-sum taxes
2. The government cannot impose lump-sum taxes

75
 With Lump-Sum Taxes

▶ With lump-sum taxes, we assume a path for {gt , τtk , τtc }∞

t=0
▶ We solve for all equilibrium variables with the algorithm
▶ We find a value for ∞ t=0 qt τt that balances the budget
P

▶ In other words, lump-sum taxes are the residual variable that satisfies the
government budget constraint

76
 No Lump-Sum Taxes

▶ If lump-sum taxes are not available, we need an extra loop

▶ Compute a candidate equilibrium for an arbitrary tax mix
▶ Change some elements of the expenditure or tax sequence until the
government budget constraint is satisfied
▶ There is a huge number of possibilities

77
9. Fiscal Policy Experiments
 Experiments Using Models

“One of the functions of theoretical economics is to provide fully articulated,

artificial economic systems that can serve as laboratories in which policies that
would be prohibitively expensive to experiment with in actual economies can be
tested out at much lower cost.”
Robert Lucas, 1980, Journal of Money, Credit, and Banking

79
 Fiscal Policy Experiments

▶ We are going to run some fiscal policy experiments

▶ At time t = 0, the economy is in steady state
▶ At time t = 1, the government announces a policy change that will be
implemented at time t = 10
▶ Our goal is to analyze how the economy responds to these changes

80
 Functional Forms

▶ We use the following utility function

c1−σ
t
u(ct ) =
1−σ
▶ We use the following production function

f (kt ) = ktα

81
 Calibration

▶ We calibrate the model at annual frequency, one period is a year

Parameter Description Value

α Capital share in production 0.33
β Discount factor 0.95
δ Depreciation rate of capital 0.2
σ Inverse elasticity of substitution 2.0
ḡ Steady-state government spending 0.2
τ̄ c Steady-state consumption tax 0
τ̄ k Steady-state capital income tax 0
τ̄ n Steady-state labor income tax 0

82
 Permanent Increase in Government Spending

▶ The first experiment is an increase in government spending

▶ In period t = 1, the government announces it will double gt in t = 10
▶ Financed only with the lump-sum tax, no distorting tax
▶ This is a once-and-for-all change: unique and permanent
▶ The change is perfectly foreseen, ie no uncertainty

83
Permanent Increase in Government Spending

84
 Permanent Drop in Income

▶ Consumers are forward-looking: they react immediately to the policy shock

even though it takes effect only in ten periods
▶ Consumers are Ricardian: they know that an increase in government
expenditure means an increase in taxes, now or later
▶ The permanent increase in gt means a permanent negative wealth effect
▶ Households consume less and save more to compensate (hours are fixed)
▶ This leads to a gradual build-up of capital until t = 10

85
 Crowding Out of Investment

▶ In period t = 10, the policy change takes place

▶ Taxes increase, households become poorer, they dissave, and capital
gradually falls back to its steady-state value
▶ Notice how the increase in gt is matched by a decrease in investment: this is
the crowding out effect of public spending on private investment
▶ Variation in capital allows households to smooth consumption; changes in
the return to capital rt , inversely related to kt , help accomplish this

86
 Permanent Drop in Consumption

▶ The permanent increase in government spending gt has no permanent

effect on capital kt : remember that in steady state, ḡ does not affect k̄
▶ But ḡ does affect c̄ in steady state: c̄ = f (k̄) − δ k̄ − ḡ
▶ Thus the permanent increase in government spending makes consumers
permanently poorer and, as a result, they consume less forever

87
 Permanent Income Hypothesis

▶ Consumption smoothing illustrates the permanent income hypothesis put

forward by Milton Friedman (1957)
▶ Changes in permanent income (assets, wage), not temporary income, are
what drive changes in a person’s consumption patterns
▶ The key determinant is the individual’s lifetime income; here the present
value of disposable income, ie capital and labor income net of taxes

88
 Dynamics

▶ Before t = 10, the response of each variable in the economy is entirely due
to expectations about future policy changes
▶ After t = 10, there is a purely transient response to a new stationary level of
exogenous variables, the forcing function
▶ Before t = 10 the forcing function changes, after t = 10 the policy vector is
constant and the sources of dynamics are transient

89
 Temporary Increase in Government Spending

▶ Let’s now simulate a temporary increase in gt

▶ In period t = 0 the government announces it will double gt but only in
t = 10, after that gt returns to its steady state
▶ Again the change is foreseen, ie anticipated

90
Temporary Increase in Government Spending

91
 Temporary Increase in Government Spending

▶ Same mechanism, a negative wealth effect

▶ Consumption drops immediately and falls further in anticipation of the
increase in taxes
▶ At time t = 10, capital and investment jump downward because the
government consumes mores and raises taxes, making households poorer
▶ Consumption slowly returns to its steady-state value

92
 Permanent Increase in Consumption Tax

▶ We now simulate an increase in the consumption tax

▶ In period t = 0, the government announces it will raise τtc from zero to 20
percent in period t = 10 for ever, ie a permanent and anticipated shock
▶ The increase in τtc is accompanied by a reduction in the present-value of
lump-sum taxes that leaves the government budget balanced

93
Permanent Increase in Consumption Tax

94
 Permanent Increase in Consumption Tax

▶ Consumers know they are not taxed today but will be taxed later:
consumption jumps on impact and a consumption binges ensues
▶ Consumers finance that binge by saving less, capital de-accumulates
▶ At time t = 10 the party is over, consumption plunges and households turn
to saving instead
▶ Capital returns slowly to steady state at the cost of austerity

95
 Permanent Increase in Capital Income Tax

▶ We now simulate an increase in the capital income tax

▶ In period t = 0, the government announces it will raise τtk from zero to 20
percent in period t = 10 for ever, ie a permanent and anticipated shock
▶ Again, the value of the lump-sum tax is reduced

96
Permanent Increase in Capital Income Tax

97
 Permanent Increase in Capital Income Tax

▶ Capital is taxed tomorrow but loses value already today

▶ Households save less and consume more
▶ But eventually less capital produces less output and consumption falls
▶ The new steady state is a permanently smaller economy with lower capital,
lower output, lower consumption, and a higher return to capital

98
 Temporary Increase in Capital Income Tax

▶ We now simulate a temporary increase in the capital income tax

▶ In period t = 0, the government announces it will raise τtk from zero to 20
percent in period t = 10 only

99
Temporary Increase in Capital Income Tax

100
 Permanent Increase in Capital Income Tax

▶ Same mecanism, capital loses value today due to the future tax
▶ Households dissave until the day of the policy change arrives
▶ This sustains a temporary consumption binge from periods 1 to 9
▶ After the shock, all variables gradually return to steady state

101
 Taking Stock

▶ In this simple model, we can study the effects of temporary and permanent
policy changes
▶ We can also study anticipated and unanticipated policy changes
▶ We can extend the model to account for more realistic features of the
economy and the tax system

102
 Conclusion

▶ We have introduced fiscal policy in the standard growth model

▶ We have done that by adopting a positive approach, ie we have described
the mechanisms at play without judgment
▶ Next class we will adopt a normative approach
▶ We will ask the question: what is the optimal tax policy?

103
10. Exercises
 Exercise 1 – Tax Reform 1

Consider the nonstochastic growth model studied above. The government only
levies a consumption tax τtc and a lump-sum tax τt . Utility depends only on
consumption.
1. Define a competitive equilibrium.
2. Suppose the government has unlimited access to the lump-sum tax and sets
gt = ḡ and τtc = 0. This situation is expected to go on forever. Tell how to
find the steady-state capital-labor ratio for this economy.
3. Prove that the timing of taxes is irrelevant.

105
 Exercise 1 – Continued

4. Suddenly lump-sum taxes are declared illegal. Starting at time t = 0 the

government must finance expenditures with τtc . Policy advisor 1 proposes
the following tax policy: find a constant τ̄ c that satisfies the time 0
government budget constraint, and impose it from time 0 onward. Compute
the new steady-state value of kt under this policy. Analyze the transition
path from the old to the new steady state.
5. Policy advisor 2 proposes an alternative policy. Instead of imposing the
increase suddenly, he proposes to ease the pain by postponing the increase
for 10 years: τtc = 0 for t = 0, . . . , 9 and then τtc = τ̄ c for t ≥ 10. Compute the
steady-state level of capital under this policy. Can you say anything about
the transition path to the new steady state?
6. Which policy is better, the first or the second one?

106
 Exercise 2 – Tax Reform 2

Consider the nonstochastic growth model studied above. The government only
levies a consumption tax τtc and a capital income tax τtk . Utility depends only
on consumption.
1. Define a competitive equilibrium.
2. Assume an initial condition in which the government finances constant
expenditures ḡ entirely with a constant τ̄ k and zero τ c . Tell how to find the
steady-state levels of capital, consumption, and the rate of return on capital.

107
 Exercise 2 – Continued

3. Suddenly the capital tax is repealed, τ k = 0 for ever. Now the government
must use τ c to finance ḡ. Tell what happens to the new steady-state values
of capital, consumption, and the return on capital.
4. Compare the two alternative policies of (1) relying completely on the
taxation of capital or (2) relying completely on the consumption tax, by
looking at the discounted utilities of consumption in steady state, ie 1−β
1
u(c̄)
in the two equilibria. Is this a good way to measure the costs or gains of one
policy vis-a-vis the other?

108
 Exercise 3 – Shooting Algorithm

1. Write a Matlab code to simulate the nonstochastic growth model without

fiscal policy using a shooting algorithm.
2. Write another code to simulate the nonstochastic growth model with
exogenous spending and lump-sum taxes.
3. Compare the effect of a temporary increase in technology in the two models
using impulse response functions.
4. Repeat the procedure using Dynare.

109