Macroeconomics 2

Master in Economics
Lecture 4: The Determination of the Price Level

Jean-Baptiste Michau
Ecole Polytechnique

January 2025

1 / 43
Introducing Money
In the Ramsey model, there is no money ⇒ Only relative prices are
determined ⇒ We could normalize the price of the final good to
one.

In order to be able to say anything about the aggregate price level,

we need to introduce money into the model.

Question: Why do people hold money even though bonds yield a

higher interest rate than money?

Answer: Money is necessary to perform transactions.

Three key models of the transaction demand for money:

▶ Cash-in-advance models
▶ Shopping time models
▶ Transaction cost models
2 / 43
Cash-in-Advance Constraint
Following Clower (1967), assume that, at the beginning of any
time period, households must have enough money to pay for their
consumption throughout the time period.
∞
max ∑ βt u (ct )
{ct ,Mt ,Bt }t∞=0 t =0

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt − Pt ct

Pt ct ≤ Mt
Bt +1 + Mt +1
lim t ≥0
t →∞ ∏
s =0 (1 + is )
M0 + B0 given
where Wt is the nominal labor income of the agent at t, ct is his
real consumption, Pt the aggregate price level, it the nominal
interest rate from the end of t − 1 to the end of t, Bt is the
agent’s holdings of bonds from the end of time t − 1 to the end of
t and Mt +1 is his holding of money at the end of time t.
3 / 43
Cash-in-Advance Constraint
The timing of the cash in advance model is as follows:
1. Throughout time t, the agent:
▶ Buys a quantity ct of consumption goods;
▶ Pays for his consumption with his money holdings Mt saved at
the end of time t − 1.
2. At the end of time t, the agent:
▶ Receives income from labor Wt and from bonds savings
(1 + it )Bt ;
▶ Decides how many bonds Bt +1 to buy and how much money
Mt +1 to keep for transactions next period.

The opportunity cost of holding money is the forgone nominal

interest rate it ⇒ Whenever it > 0 for t ≥ 0, the cash in advance
constraint is binding:
Mt = Pt ct .

4 / 43
Cash-in-Advance Constraint

The household’s problem of the cash-in-advance model could be

written as:
∞   
Mt
max ∑ β u min ct , Pt
{ct ,Mt ,Bt }t∞=0 t =0
t

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt − Pt ct

Bt +1 + Mt +1
lim ≥0
t → ∞ ∏t
s =0 (1 + is )
M0 + B0 given

Instead of solving the cash-in-advance model, we shall later solve a

closely related money-in-the-utility-function model.

5 / 43
Shopping Time

Each period, an agent has one unit of time that can either be
allocated to leisure lt or shopping st :

1 = lt + st ,

where, for simplicity, we assume that the amount of time spent

working is fixed and exogenous.

Shopping time is increasing in consumption and decreasing in real

money balances:  
Mt +1
st = H c t , ,
Pt
with Hc (., .) ≥ 0 and HM/P (., .) ≤ 0.

6 / 43
Shopping Time

The household’s problem is:

∞
max ∑ βt u (ct , lt )
{ct ,Mt +1 ,Bt +1 }t∞=0 t =0

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt − Pt ct

 
Mt + 1
1 = lt + H ct ,
Pt
Bt +1 + Mt +1
lim t ≥0
t →∞ ∏
s =0 (1 + is )
B0 and M0 given

where ul (., .) > 0 and ull (., .) < 0.

7 / 43
Shopping Time

The household’s problem of the shopping time model could be

written as:
∞   
Mt +1
max ∑ β u c t , 1 − H c t , Pt
{ct ,Mt +1 ,Bt +1 }t∞=0 t =0
t

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt − Pt ct

Bt +1 + Mt +1
lim ≥0
t → ∞ ∏t
s =0 (1 + is )
B0 and M0 given

The shopping time model could be seen as a special case of a more

general money-in-the-utility-function model.

8 / 43
Shopping Time: A Baumol-Tobin Specification
A natural specification of the shopping time function is:
 
Mt +1 1 ct
H ct , = ϵ,
Pt 2 Mt +1 /Pt

where ct /(2Mt +1 /Pt ) is the number of trips to the bank and ϵ is

the time cost of each trip to the bank.

This specification is based on Baumol (1952) and Tobin (1956)

who were the first to emphasize the trade-off between:
▶ The size of the real money balances
▶ The number of trips to the banks

Indeed, small real money balances imply small forgone returns on

savings but a large number of trips to the bank (and vice versa).

Note that, under this interpretation, Mt +1 /Pt is the average real

money holdings in period t.
9 / 43
Transaction Cost
Consumers must incur a real transaction cost:
 
Mt +1
ϕ , ct ,
Pt
where ϕM/P (., .) < 0 and ϕc (., .) > 0.

The household’s problem is:

∞
max ∑ βt u (ct )
{ct ,Mt +1 ,Bt +1 }t∞=0 t =0

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt

  
Mt + 1
−Pt ct + ϕ , ct
Pt
Bt +1 + Mt +1
lim ≥0
t → ∞ ∏t
s =0 (1 + is )
B0 and M0 given
10 / 43
Transaction Cost
Let:  
▶ ct + ϕ MPt +1 , ct = c̃t ;
t
 
▶ u (ct ) = ũ c̃t , MPt +1 .
t

The household’s problem with a transaction cost could be written

as:
∞  
Mt +1
max ∑ β ũ c̃t , Pt
{c̃t ,Mt +1 ,Bt +1 }t∞=0 t =0
t

subject to Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Wt − Pt c̃t

Bt +1 + Mt +1
lim t ≥0
t →∞ ∏
s =0 (1 + is )
B0 and M0 given

The transaction cost model could be seen as a special case of a

more general money-in-the-utility-function model.
11 / 43
Money in the Utility Function
Following Sidrauski (1967), the simplest way to generate a demand
for money is to assume that utility is directly increasing in real
money balances.
▶ This is known as the money in the utility function approach.

The objective of the household is to maximize:

∞  
Mt +1
∑ β u c t , Pt
t
t =0
where uM/P (., .) ≥ 0, uM/P,M/P (., .) < 0 and
limM/P →0 uM/P (., M/P ) = ∞.

The timing convention implies that agents value real money

holdings at the end of time t (whereas in a cash-in-advance model,
agents value their real money holdings at the beginning of time t).

We shall now characterize the demand for money and solve for the
aggregate price level in a decentralized Ramsey economy.
12 / 43
Money in the Utility Function: Producer
At any time t, a representative firm maximizes its profits:

max Pt F (Kt , Lt ) − Rt Kt − Wt Lt ,
{Kt ,Lt }

where Rt is the nominal rental cost of capital and Wt is the

nominal wage rate.

Kt is the amount of capital rented by the firm from the end of

time t − 1 to the end of t.

In equilibrium, we must have:

▶ Rt = Pt FK (Kt , Lt ) = Pt f ′ (kt )
▶ Wt = Pt FL (Kt , Lt ) = Pt [f (kt ) − kt f ′ (kt )]
where f (kt ) = F (Kt , Lt )/Lt and kt = Kt /Lt .

Note: Constant returns to scale imply that, at the optimum,

profits are equal to 0.
13 / 43
Money in the Utility Function: Government
In this Ramsey economy, the government earns some revenue by
issuing money.

We must therefore introduce the government budget constraint:

Bt +1 = (1 + it )Bt − (Mt +1 − Mt ) + Tt
where:
▶ Bt +1 is the amount of government debt at the end of time t.
▶ Mt +1 − Mt is revenue that the government obtains at the end
of t by increasing the money supply. This revenue is known as
seignorage.
▶ Tt is the value of net transfers to households at the end of t.

Note that this government budget constraint is a consolidation of

the budget constraints of:
▶ The central bank
▶ The ministry of finance
14 / 43
Money in the Utility Function: Government
Bonds issued by the treasury must have either been purchased by
households or by the central bank:
BtT = Bt + BtCB
where:
▶ BtT : Bonds issued by the ministry of finance (the Treasury)
▶ BtCB : Bonds purchased by the Central Bank
▶ Bt : Bonds purchased by households

We have:
BtT+1 = (1 + it )BtT + Tt − St
BtCB CB
+1 = (1 + it )Bt + (Mt +1 − Mt ) − St

where St denotes a transfer from the central bank to the treasury.

Combining these two constraints yields the government budget

constraint (of the previous slide).
15 / 43
Money in the Utility Function: Household
At the end of time t − 1, a household can buy three assets:
▶ Money Mt
▶ Money does not yield any interest but is directly valued by the
household (as it facilitates transactions).
▶ Government bonds Bt
▶ Government bonds yield a nominal interest rate it at the end
of t (which is determined by the price of bonds at t − 1).
▶ Claims on units of physical capital At
▶ Pt is the price of a unit of physical capital at t (as output can
either be transformed into consumption or investment goods).
▶ A unit of physical capital yields the rent Rt at the end of t and
the value of undepreciated capital at the end of the rental
period Pt (1 − δ).
The return on a unit of physical capital from t − 1 to t is:
Pt (1 − δ + Rt /Pt ) = Pt (1 + rt ),
where the real interest rate rt is defined as the real rental cost of
capital net of depreciation, i.e. rt = Rt /Pt − δ = f ′ (kt ) − δ.
16 / 43
Money in the Utility Function: Household

The household’s problem is:

∞
!
Md
max ∞ ∑βu
t
ct , t +1
Pt
{ct ,Mtd+1 ,Btd+1 ,At +1 }t =0 t =0

subject to Mtd + (1 + it )Btd + (1 + rt )Pt At

+Wt + Tt − Pt ct = Mtd+1 + Btd+1 + Pt At +1
M0 , B0 and A0 given

where Mtd and Btd denote the demand for money and government
bonds, respectively, at the end of t − 1.

Note that labor supply is exogenous and equal to 1.

17 / 43
Money in the Utility Function: Household
At any time t:
▶ The state variables are Mtd , Btd and At .
▶ The control variables are ct , Mtd+1 , Btd+1 and At +1 .

The household’s problem could be written recursively:

Vt (Mtd , Btd , At ) =
!
Md
max u ct , t + 1 + βVt +1 (Mtd+1 , Btd+1 , At +1 )
Mtd+1 ,Btd+1 ,At +1 Pt

where:
Mtd Bd Wt Tt Mtd+1 Btd+1
ct = + (1 + it ) t + (1 + rt )At + + − − − At +1
Pt Pt Pt Pt Pt Pt

Note: As it , rt , Pt , Wt and Tt are not constant over time, the

problem is not stationary. The value function must therefore be
indexed by time.
18 / 43
Money in the Utility Function: Household
Envelope conditions with respect to Mtd , Btd and At :
uc (ct , Mtd+1 /Pt )
Vt,M (Mtd , Btd , At ) =
Pt
uc (ct , Mtd+1 /Pt )
Vt,B (Mtd , Btd , At ) = (1 + it )
Pt
Vt,A (Mt , Bt , At ) = (1 + rt )uc (ct , Mtd+1 /Pt )
d d

First-order conditions with respect to Mtd+1 , Btd+1 and At +1 :

−uc (ct , Mtd+1 /Pt ) + uM/P (ct , Mtd+1 /Pt )

Pt
+ βVt +1,M (Mtd+1 , Btd+1 , At +1 ) = 0

uc (ct , Mtd+1 /Pt )

− + βVt +1,B (Mtd+1 , Btd+1 , At +1 ) = 0
Pt
−uc (ct , Mtd+1 /Pt ) + βVt +1,A (Mtd+1 , Btd+1 , At +1 ) = 0
19 / 43
Money in the Utility Function: Household

Combining the envelope conditions with the first-order conditions

yields the following optimality conditions:
" #
uc (ct , Mtd+1 /Pt ) − uM/P (ct , Mtd+1 /Pt ) uc (ct +1 , Mtd+2 /Pt +1 )
=β
Pt Pt +1
" #
uc (ct , Mtd+1 /Pt ) uc (ct +1 , Mtd+2 /Pt +1 )
= β (1 + it +1 )
Pt Pt + 1
h i
uc (ct , Mtd+1 /Pt ) = β (1 + rt +1 )uc (ct +1 , Mtd+2 /Pt +1 )

20 / 43
Money in the Utility Function: Household
Combining the first and second equations yields the demand for
money equation:
it +1
uM/P (ct , Mtd+1 /Pt ) = uc (ct , Mtd+1 /Pt )
1 + it +1

The third equation is the consumption Euler equation:

uc (ct , Mtd+1 /Pt ) = β(1 + rt +1 )uc (ct +1 , Mtd+2 /Pt +1 )

Combining the second and third equations yields the Fisher

relationship:
Pt 1 + it +1
1 + rt +1 = (1 + it +1 ) =
Pt + 1 1 + π t +1
where πt +1 = (Pt +1 − Pt )/Pt is the inflation rate from time t to
t + 1. The Fisher relationship is usually approximated as
rt ≃ it − πt .
21 / 43
Decentralized Economy: Competitive Equilibrium
Definition

A
 competitive equilibrium consists of
∞
ct , Lt , At , Kt , Btd , Bt , Mtd , Mt , Tt , Pt , Wt , Rt , rt , it t =0 such that:
▶ rt = Rt /Pt − δ and 1 + it = (1 + rt )(1 + πt ).
▶ ct , At , Btd , Mtd t∞=0 solves the household’s problem given M0 ,


B0 and A0 and {Pt , Wt , rt , it }t∞=0 .

▶ {Kt , Lt }t∞=0 solves the producer’s problem given
{Pt , Wt , Rt }t∞=0 .
▶ {Bt , Mt , Tt , it }t∞=0 satisfies the government budget constraint.
▶ At {Pt , Wt , Rt , rt , it }t∞=0 all markets clear:
▶ Lt = 1
▶ Kt = At
▶ Btd = Bt
▶ Mtd = Mt
▶ ct + It = Yt
22 / 43
Money in the Utility Function: General Equilibrium
The general equilibrium of the economy is therefore jointly
characterized by the Fisher relationship, the solutions to the firm’s
and to the household’s problems, by the government and the
household’s budget constraints and by market clearing conditions.

Substituting the government budget constraint with Btd = Bt and

Mtd = Mt into the household’s constraint yields:

At +1 = (1 + rt )At + Wt /Pt − ct .

Imposing the equilibrium conditions in the market for capital,

At = Kt , and in the market for labor, Lt = 1, implies that
At = Kt = kt Lt = kt . We therefore have:

kt +1 = (1 + rt )kt + Wt /Pt − ct
= (1 − δ + f ′ (kt ))kt + f (kt ) − kt f ′ (kt ) − ct
 

= (1 − δ)kt + f (kt ) − ct
23 / 43
Money in the Utility Function: General Equilibrium
The general equilibrium of this economy is jointly determined by:
▶ The resource constraint:
kt +1 = (1 − δ)kt + f (kt ) − ct
▶ The consumption Euler equation:
uc (ct , Mt +1 /Pt ) = β(1 − δ + f ′ (kt +1 ))uc (ct +1 , Mt +2 /Pt +1 )
▶ The Fisher relationship:
1 + it = (1 + πt )(1 − δ + f ′ (kt ))
▶ The demand for money equation:
uM/P (ct , Mt +1 /Pt ) = [it +1 / (1 + it +1 )] uc (ct , Mt +1 /Pt )

Note: To fully characterize the equilibrium, we assume that

{Mt }t∞=0 is exogenous and that {Tt }t∞=0 adjusts such as to exactly
satisfy the intertemporal government budget constraint.
24 / 43
Money in the Utility Function: Steady State Equilibrium
Let gt +1 be the growth rate of money supply from time t to t + 1:
Mt +1 − Mt
gt + 1 = .
Mt

In steady state, where each variable grows at a constant rate:

▶ Consumption, the capital stock and the nominal interest rate
are constant, i.e. ct = c ∗ , kt = k ∗ and it = i ∗ .
▶ The money supply and the price level both grow at rate g , i.e.
gt = g and πt = g .
▶ This implies that the level of real money balances is constant,
i.e. Mt +1 /Pt = m∗ .

It follows that, in steady state, consumption c ∗ and the capital

stock k ∗ are jointly determined by:
c ∗ = f (k ∗ ) − δk ∗
β(1 − δ + f ′ (k ∗ )) = 1
as in the cashless Ramsey economy.
25 / 43
Money in the Utility Function: Neutrality of Money
If a real variable is independent of:
▶ The level Mt of the money supply ⇒ Money is neutral.
▶ The growth rate g of the money supply ⇒ Money is
superneutral.

In steady state, money is both neutral and superneutral.

If utility is additively separable between consumption and real

money balances, then the marginal utility of consumption
uc (ct , Mt +1 /Pt ) is independent of real money balances Mt +1 /Pt
and, hence, money is always neutral and superneutral (even out of
steady state).

Note: Even if utility is not additively separable, money is unlikely

to have large real effects in this family of models.

26 / 43
Money in the Utility Function: Optimal Monetary Policy

Optimal monetary policy:

▶ In steady state, money only affects welfare through its direct
effect on utility.
▶ Thus, the optimal monetary policy consists in setting
uM/P (c ∗ , m∗ ) = 0.
▶ This requires i ∗ = 0 and, hence, π ∗ = −r ∗ /(1 + r ∗ ) < 0,
where r ∗ = f ′ (k ∗ ) − δ = 1/β − 1 > 0.
▶ This is known as the Friedman rule (Friedman 1969).

27 / 43
Money in the Utility Function: Assessment

These models emphasize the transaction demand for money. They

are useful to characterize:
▶ The demand for money
▶ The aggregate price level (and hyperinflation)
▶ The interaction between seignorage and fiscal policy

However, these models abstract from key channels through which

money has real effects:
▶ Money illusions (as induced by adaptive expectations in the
IS-LM AD-AS model)
▶ Nominal rigidities (which will be central to the new-Keynesian
framework of Lecture 6)

28 / 43
Money in the Utility Function: Assessment
In practice, in the short-run, an increase in the money supply leads
to a fall in the nominal interest rate.
▶ This is known as the liquidity effect.

In the short-run:
▶ Models of the transaction demand for money are inconsistent
with the liquidity effect:
▶ An increase in the money supply raises inflation while leaving
the real rate nearly unaffected.
▶ Models of money illusions or nominal rigidities typically
generate a liquidity effect:
▶ Prices are slow to adjust ⇒ The increase in the money supply
raises real money balances ⇒ The price of assets increases ⇒
The nominal interest rate falls ⇒ The real interest rate falls
⇒ Aggregate demand increases ⇒ Output booms.

29 / 43
The Price Level

Let us now determine the aggregate price level and the inflation
rate.

For simplicity, we focus on the steady state and assume that utility
is additively separable between consumption and real money
balances:    
Mt + 1 Mt +1
u ct , = v (ct ) + w .
Pt Pt

Thus, money is completely neutral and consumption is equal to c ∗ ,

the capital stock to k ∗ and the real interest rate to r ∗ .

30 / 43
The Price Level: Steady State
In this economy, the price level is fundamentally determined by the
equilibrium in the money market.

For any t ≥ 0, the demand for money equation is:

[(1 + πt +1 )(1 + r ∗ ) − 1] ′ ∗
 
′ Mt + 1
w = v (c ),
Pt (1 + πt +1 )(1 + r ∗ )
which could be written as:
Mt + 1
= h ( π t +1 ),
Pt
where h′ (.) < 0.

Assume that money supply grows at rate g , i.e. Mt = (1 + g )t M0 .

▶ In steady state, prices must also grow at rate g , i.e. πt = g .
▶ This implies:
Mt +1
Pt = .
h (g )
31 / 43
The Price Level: Indeterminacy
Does the price level necessarily converge to a steady state? No!

Even if the growth rate of the money supply is constant, prices are
in fact indeterminate.
▶ The money market is in equilibrium for any sequence {Pt }t∞=0
that satisfies:
Pt + 1 − Pt
 
Mt +1
=h for all t ≥ 0.
Pt Pt
▶ Out of steady state, this either generates:
▶ A strictly decreasing sequence of prices with the price level
tending to zero, i.e. speculative deflation.
▶ A strictly increasing sequence of prices with the price level
tending to infinity, i.e. speculative hyperinflation.
▶ Obstfeld Rogoff (1983) showed that:
▶ Speculative deflation can easily be ruled out (as infinitesimally
small prices violate the household’s transversality condition).
▶ Speculative hyperinflation cannot, in general, be ruled out.
32 / 43
The Price Level: Quantity Theory
The quantity theory of money is the cornerstone of monetarism.
▶ Let Vt = Pt Yt /Mt +1 be defined as the velocity of money
(where Yt is output at t).
▶ The quantity theory of money postulates that the velocity Vt
is determined by technological factors (e.g. ATMs) and is
therefore exogenous.
▶ Hence, prices are proportional to the money supply.
▶ Milton Friedman claimed: ”Inflation is always and everywhere
a monetary phenomenon.”

In our model, the quantity theory is

▶ Inconsistent with speculative hyperinflation, which implies
that velocity has an upward trend;
▶ Consistent with the steady state, which is monetarist.

The quantity theory could be seen as an equilibrium selection

device that justifies focusing on the steady state.
33 / 43
The Price Level: Timing Assumptions
Following the cash-in-advance model, we could have assumed that
agents value real money holdings at the beginning of time t:
∞  
Mt
∑ β u ct , Pt
t
t =0

With additively separable utility, the resulting money demand

equation can be written as:
Pt − Pt − 1
 
Mt
= h̃ ,
Pt Pt − 1
where h̃′ (.) < 0.
 
▶ Question: Why do economists often use MPt +1 = h Pt +P1 −Pt ?
t t
▶ Answer: To have a linear relationship between Mt +1 and Pt
▶ In steady state, if at the end of time t the central bank
doubles Mt +1 while committing to keep the subsequent
growth rate of the money supply unchanged, then Pt doubles.
34 / 43
The Price Level: A Fiscal-Monetary Theory of Inflation
So far, we have ignored the government budget constraint.
▶ We have assumed that the ministry of finance takes {Mt }t∞=0
as given and adjusts {Tt }t∞=0 such as to balance the budget.
▶ Let us now assume that the central bank takes {Tt }t∞=0 as
given and adjusts {Mt }t∞=0 such as to balance the budget.

Iterating on the government budget constraint from time 0 to T

yields:
T
BT +1 + MT +1 Tt
∏T
= (1 + i0 )B0 + ∑ ∏t
k =1 (1 + ik ) t =0 k =1 (1 + ik )
T
Mt +1 − Mt M
−∑ t + T T +1
t =0 ∏k =1 (1 + ik ) ∏k =1 (1 + ik )

The no-Ponzi condition is:

lim (Bt +1 + Mt +1 ) / ∏k =1 (1 + ik ) ≤ 0.
t
t →∞

35 / 43
The Price Level: A Fiscal-Monetary Theory of Inflation
The intertemporal government budget constraint at time 0 is:
∞ ∞
Tt Mt + 1 − Mt
(1 + i0 )B0 + ∑ ∏t ≤ ∑ ∏t
t =0 k =1 (1 + ik ) t =0 k =1 (1 + ik )
Mt +1
− lim ,
t → ∞ ∏t
k =1 (1 + ik )

where the right hand side corresponds to the present value of

seignorage.

Using the fact that 1 + it = (1 + πt ) (1 + r ∗ ) = (1 + r ∗ )Pt /Pt −1 ,

this constraint simplifies to:
∞ ∞
B0 Tt /Pt (Mt +1 − Mt ) /Pt Mt +1 /Pt
(1 + i0 ) +∑ ∗
≤ ∑ ∗
− lim .
P0 t =0 (1 + r ) t (1 + r ) t t → ∞ (1 + r ∗ )t
t =0

36 / 43
The Price Level: A Fiscal-Monetary Theory of Inflation
Let us assume:
▶ Real government expenditures are constant over time, i.e.
Tt /Pt = τ for any t ≥ 0.
▶ At time 0, the central bank performs an open market
operation in order to decrease M1 .
▶ From time 0 onwards, the economy is in steady state with
money supply growing at a rate g , i.e. (Mt +1 − Mt )/Mt = g
for all t ≥ 1. This implies that πt = g for all t ≥ 1.

In steady state, the (binding) government budget constraint could

be written as:
B0 1 + r ∗ M1 − M0
(1 + i0 ) + ∗
τ=
P0 r P0
∞
1 Mt +1 − Mt Mt Pt −1 Mt +1 /Pt
+∑ ∗
− lim .
t t → ∞ (1 + r ∗ )t
t =1 (1 + r ) Mt Pt −1 Pt
37 / 43
The Price Level: A Fiscal-Monetary Theory of Inflation
The government budget constraint simplifies to:
B0 1 + r ∗ M1 − M0 g h (g )
(1 + i0 ) + τ= + .
P0 r∗ P0 1 + g r∗

Seignorage must be used to balanced the budget from time 1

onwards ⇒ g is determined by this government budget constraint.
The price level at time 0 is determined by:
M1 /P0 = h (g ).

In that context, it is possible that a decrease in the money supply

M1 leads to a rise in inflation g .
▶ The decline in M1 causes a reduction in seignorage at time 0,
which must be compensated by higher seignorage in the
future.

Sargent and Wallace (1981) call this ”unpleasant monetarist

arithmetic”.
38 / 43
The Price Level

It is important to distinguish two types of policy regimes:

▶ Monetary dominance: The ministry of finance takes
seigniorage as given and adjusts taxes and expenditures such
as to balance the government budget constraint.
▶ Quantity theory of money
▶ Fiscal dominance: The central bank takes the fiscal policy as
given and relies on seigniorage to balance the government
budget constraint.
▶ Unpleasant monetarist arithmetic

39 / 43
The Price Level: Fiscal Theory of the Price Level
A final possibility is that neither the ministry of finance nor the
central bank is willing to make the necessary adjustment to
balance the government budget constraint.

Let us assume:
▶ Real government expenditures are constant over time, i.e.
Tt /Pt = τ for any t ≥ 0.
▶ The money supply growing at a fixed rate g , i.e.
(Mt +1 − Mt )/Mt = g for all t ≥ 0.

Recall that the money market is in equilibrium for any sequence

{Pt }t∞=0 that satisfies:
Pt +1 − Pt
 
Mt + 1
=h ,
Pt Pt
with Mt +1 = (1 + g )t +1 M0 .

Hence, there are several feasible values of P0 .

40 / 43
The Price Level: Fiscal Theory of the Price Level
The iteration on the government budget constraint from 0 to T
could be written as:
T
BT +1 + MT +1 Tt
∏T
= (1 + i0 )B0 + M0 + ∑ ∏t
k =1 (1 + ik ) t =0 k =1 (1 + ik )
T
it Mt
−∑ t
t =1 ∏k =1 (1 + ik )

Taking the limit as T tends to infinity, imposing the no-Ponzi

condition and using the Fisher relationship, i.e. 1 + it =
(1 + r ∗ )Pt /Pt −1 , the intertemporal government budget constraint
can be written as:

∞ ∞
(1 + i0 ) [B0 + M0 ] Tt /Pt it Mt /Pt −1
+∑ ∗ t
≤ ∑ ∗ t
.
P0 t =0 (1 + r ) t =1 Pt /Pt −1 (1 + r )

41 / 43
The Price Level: Fiscal Theory of the Price Level
The (binding) government budget constraint could be further
simplified to:
∞
(1 + i0 ) [B0 + M0 ] 1 + r ∗ (1 + πt )(1 + r ∗ ) − 1 h(πt )
P0
+
r∗
τ = ∑ 1 + πt (1 + r ∗ )t
t =1

The initial price level P0 adjusts such as to balance the

government budget.

This is known as the fiscal theory of the price level (Sims 1994,
Woodford 1995).
▶ This theory is a selection device that solves the price level
indeterminacy.
▶ However, it generically selects a speculative path for the price
level which is inconsistent with steady state.
▶ The quantity theory of money seems to be a more sensible
selection device (Kocherlakota Phelan 1999).
42 / 43
The Price Level: Fiscal Theory of the Price Level
The government issues debt in its own currency.

Hence, according to Cochrane (2005), there is no government

budget constraint, but only a government debt valuation equation.
▶ Government debt is a claim on future surpluses;
▶ Low future surpluses ⇒ Low real value of government debt ⇒
High price level.

In practice, as the government cares about the price level, neither

monetary nor fiscal policy are exogenous.
▶ Hence, we cannot test empirically the fiscal theory of the price
level.

Cochrane (2023) specifies the stochastic process for surpluses such

as to be able to reinterprets recent macroeconomic outcomes
through the lens of the fiscal theory.
43 / 43