doi: 10.1111/j.1742-7363.2007.00051.

Land tax and economic growth under credit

market imperfection
Masaya Sakuragawa∗ and Yukie Sakuragawa†

We study the general equilibrium effects of land taxation on economic growth by

extending the model developed by Kiyotaki and Moore (1997) to an endogenous
growth model, where land is used not only as an input of production but also
as collateral. Land taxation tends to hamper economic growth through the credit-
contraction effect, but the overall direction on economic growth depends on the
redistribution scheme of the tax revenue. Surprisingly, we show that if the tax revenue
is fully refunded to entrepreneurs, the economy grows faster than a no-taxation
economy. We calibrate our model and show that if taxation on land is raised by
1 percent, the land price initially falls by approximately 9.09 percent, while the
economy grows faster by 0.6%.

Key words endogenous growth, land taxation, collateral

JEL classification H20, O40

Accepted 20 February 2006

1 Introduction

The general equilibrium impact of land taxation on the aggregate economy is a contro-
versial topic. It depends on the choice of economic models. In the representative-agent
growth model with fixed supply of land used as a production factor, land taxation does
not influence the allocation of resource and economic growth. Using Diamond’s (1965)
overlapping-generations model, in contrast, Feldstein (1977) demonstrates that the im-
position of a tax on land rent promotes capital accumulation and might appreciate land.
Hashimoto and Sakuragawa (1998) extend Feldstein’s idea to an endogenous growth model
and demonstrate that a tax on land rent always promotes economic growth.
The topics on land taxation have received considerable attention in Japan because the
Japanese economy is greatly influenced by the behavior of land prices.1 This comes mainly
from the practice of the Japanese banking sector in which land has been widely used as

∗ Faculty of Economics, Keio University, Tokyo, Japan. Email: masaya@econ.keio.ac.jp

† Faculty of Management, Atomi University, Saitama, Japan.

We are greatly thankful to an anonymous referee, Hidehiko Ishihara and Makoto Yano for valuable comments.
1 From 1958 to 1998, the annual growth rates of land prices and per-capita GDP show a positive correlation.
The correlation coefficient is 0.72.

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 131
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

collateral for loans. Several recent works support the significant role of land as collateral for
corporate investment (see Ogawa et al. 1996; Ogawa and Kitasaka 1998; Ogawa and Suzuki
1998).2
In the present paper we examine the effect of land tax on economic growth in an
economy where land is effectively used as collateral for raising loans. Kiyotaki and Moore
(1997) develop an economy under credit market imperfection in which land is not only
a factor of production but also serves as collateral for loans and demonstrate that the
fluctuation of the land price induces persistent fluctuations in aggregate investment and
output.
We extend the model developed by Kiyotaki and Moore to an endogenous growth
model in which externalities of the Romer (1986) type are the driving force of growth. We
characterize a balanced growth path of an economy where land is used not only as an input
of production but also as collateral. On a balanced growth path, output, capital and land
price grow at the same rate.
Our model is designed to investigate the general equilibrium effect of land tax on an
economy. Because of the incentive problem associated with raising loans, there emerges a
channel through which land tax hampers economic growth. The land price falls promptly
in response to taxation, and consequent depreciation of the collateral value of land makes
entrepreneurs face severe credit constraints. For this reason, land tax tends to hamper
capital accumulation and, hence, economic growth.
The overall effect of land tax on economic growth depends on how the tax revenue is
distributed. Surprisingly, when tax revenue is fully refunded to entrepreneurs, our economy
grows faster than in a no-taxation economy. We show that if land tax is raised by 1 percent,
the land price initially falls by approximately 9.09 percent, whereas the economy grows
faster than for the no-taxation economy, at 0.6 percent.
This paper is organized as follows. In Section 2 we construct a theoretical model. In
Section 3 we examine the general equilibrium effect of land taxation. In Section 4 we
calibrate the model. Finally, Section 5 concludes this paper.

2 The model

Consider an economy consisting of two types of agents: entrepreneurs and investors. Each
of them lives infinitely. The population size of entrepreneurs is normalized to be unity
and that of investors is N . A significant difference between our model and that in Kiyotaki
and Moore (1997) is the production technology that each entrepreneur has access to. We
describe the production function as
1−α
Yt + 1 = AK tα L t1 − α K t , (1)

where K t and Lt are capital stock and land that are used as inputs for production at period
t, and Yt + 1 is the output of production at period t + 1.K t represents the aggregate capital
2 Additionally, Sakuragawa and Sakuragawa (2002) study the relationship between economic growth and

appreciation in land prices in postwar Japan, and find causation between land prices and economic growth
in all sub-sample periods examined.

132 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

stock at period t that captures the Romer (1986) type’s technological externality.
∞ t The
preference of an entrepreneur is described by a risk neutral utility function: t = 0 β C tE ,
where C tE is the consumption of the entrepreneur at period t. The preference of an investor


is also described by a risk neutral utility function: ∞t = 0 β̂ C t , where Ct is the consumption
t

of the investor at period t. Investors do not have access to any production technology and,
hence, earn income only by lending their own funds to others.
The discount factors, β and β̂, both lie in (0,1). We assume that investors are more
“patient” than entrepreneurs; that is:
Assumption 1 β < β̂.
The critical assumption in the present study is that there is no enforcement mechanism to
fulfill financial contracts between debtors and creditors. In the economy, lenders cannot
force borrowers to repay their debt unless the debts are secured. In particular, we assume
that creditors cannot seize output or capital of their debtors. Creditors can seize only land
that the debtor holds. In other words, the debtors can consume their output or run away
with their capital after they exchange capital for consumer goods, but they cannot run away
with their land.
We now assume that holding land is subject to taxation. Assume that landholders are
imposed a proportional tax on land at rate τ ∈ [0, 1]. The timing of taxation influences the
outcome. We explain this problem in detail in later sections. The tax revenue is transferred
to either entrepreneurs or investors in a lump-sum fashion. We assume that the government
budget is balanced period by period.
The maximization problem of an entrepreneur is formulated as:
∞

max β t C tE (2)
{C tE , Bt , K t , L t }∞
t =0 t =0

subject to
1−α
θ Tt + AK tα− 1 L t1 −− 1α K t − 1 + Bt − Rt Bt − 1 ≥ K t − K t − 1 + Q t {(1 + τ )L t − L t − 1 } + C tE
(3)

(1 − s )K t − 1 ≤ C tE , (4)

Rt Bt ≤ γ Q et,t + 1 L t , (5)

where Bt is the amount borrowed, Qt is the land price, Q et,t + 1 is the expected land price
formed at period t, and Rt is the gross borrowing interest rate that is determined at period t
and has to be paid at period t + 1. A fraction θ (0 ≤ θ ≤ 1) of the tax revenue Tt ( = τ Qt Lt ,
under the balanced budget) is refunded to entrepreneurs, whereas the rest (1 − θ )Tt is
transferred to investors.
Equation (3) is the entrepreneur’s budget constraint. Equation (4) shows that con-
sumption has to exceed a certain ratio of the capital carried over from the preceding
period. One interpretation behind (4) is that capital depreciates at rate s and, therefore,

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 133
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

entrepreneurs find it more beneficial to consume than to depreciate. Equation (5) repre-
sents the borrowing constraint that implies that entrepreneurs can borrow so long as the
debt repayment, Rt Bt , never exceeds the liquidation value of the land, γ Q et,t + 1 Lt , where
the constant γ ∈ (0, 1] reflects the fact that a fraction (1− γ ) of the market value is spent
or bankruptcy procedures. Hereafter, we call γ the “collateral rate”. Note that creditors may
evaluate the liquidation value net of the cost of taxation by perceiving that part of the value
is spent for taxation in marketing the pledged collateral. Then, the inequality (5) can be
replaced by

Rt Bt ≤ γ Q et,t + 1 L t − τ γ Q et,t + 1 L t . (5 )

Marketing collateral is typically an activity involving collecting loan losses, with the revenue
being tax-exempt. For this reason, we proceed to the analysis by using not (5 ) but (5).3
The maximization problem of an investor is the following:
∞

max
∞
β̂ t C t
{C t }t = 0
t =0

subject to W t + 1 = Rt W t − Ct , given W 0 > 0, where W t is the asset that the investor holds
at the beginning of period t.
Assume at this stage that, when Assumption 1 is satisfied, investors become creditors and
entrepreneurs become debtors at the neighborhood of the steady state. Assume additionally
that the rate of interest always equals investors’ constant rate of time preference in the credit
market; that is, Rt ≡ R = 1/β̂. Because then βR < 1, entrepreneurs prefer borrowing to
lending at this interest rate and, hence, the constraint (5) binds with equality at least in
the neighborhood of the steady-state equilibrium. If, instead, R = 1/β̂ holds, investors
are indifferent as to the allocation of their assets between consumption and saving at any
period. Therefore, investors provide the fund passively to entrepreneurs who would demand
borrowing, the amount of which is determined by the constraint (5), and consume the
remaining. The world faced by entrepreneurs may be interpreted as the small-open economy
in which the gross interest rate is exogenously fixed at R = 1/β̂.
Solving for the maximization problem of the entrepreneur, we formulate the following
Lagrangean function as
∞

  
 C te , K t , Bt , L t = β t C te
t =0
 1−α
+ µt θ Tt + AK tα− 1 L t1 −− 1α K t − 1 + Bt − Rt Bt − 1 − (K t − K t − 1 )
  
− Q t {(1 + τ )L t − L t − 1 } + C te + ∈t C te − (1 − s )K t − 1
 
+ λt γ Q et,t + 1 L t − Rt + 1 Bt , (6)

where µ t , ∈ t , and λ t are the Lagrange multipliers. Optimal conditions are given by four first-
order conditions, ∂/∂C te = 0, ∂/∂ K t = 0, ∂/∂ Bt = 0 and ∂/∂ L t = 0, three com-
3 We tried the analysis for the case when the borrowing constraint (5 ) replaces (5), but could not find any

further insights at the sacrifice of troublesome calculation.

134 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

plementary slacknesses, µt ≥ 0, ∂/∂µt ≥ 0 or µt × ∂/∂µt = 0, ∈t ≥ 0, ∂/∂ ∈t ≥ 0

or ∈t × ∂/∂ ∈t = 0, and λt ≥ 0, ∂/∂λt ≥ 0 or λt × ∂/∂λt = 0, and the transversality
condition, lim β T µT K T = 0.
T →∞
Four first-order conditions are described as

∈t = µt − 1, (7)

β(1 + α A)µt + 1 − β(1 − s ) ∈t + 1 − µt = 0, (8)

λt = Rt−+11 µt − βµt + 1 (9)

and

β(1 − α)Aµt + 1 K t + βµt + 1 Q t + 1 − (1 + τ )µt Q t + γ λt Q et,t + 1 = 0, (10)

where we use K t = K̄ t and Lt = 1 to derive (8) and (10) on the grounds that entrepreneurs
are homogeneous and that only the entrepreneurs demand land. It follows from (7) and
(8) that
1 1−s
µt + 1 = µt − . (11)
β(α A + s ) αA+s
We impose restrictions on parameters in the following way:
Assumption 2 β(s + α A) < 1 < β(1 + α A).
The first inequality of Assumption 2 implies that when entrepreneurs invest as much as
possible utility does not explode. The second inequality implies that entrepreneurs prefer
investment to consumption, where (1 + α A)is the gross marginal product of capital.4 The
following holds.
Lemma µ t > 0, ∈ t > 0, and λ t > 0 for all t ≥ 0.

PROOF: Under Assumption 2, except for the steady state µ t + 1 = µ t , (11) implies that µ t
explodes to either positive or negative infinity. If µ t explodes to positive infinity, it violates

4 To see that this is the case, we compare the utility when an entrepreneur consumes an additional unit with the

utility when he or she invests it at period t. It is obvious that the utility when he or she consumes the additional unit
is unity. However, when he or she invests it, the investment produces the amount with interest added, (1 + α A),
at period t + 1. Given that the new purchase of land is zero, as will be shown below, all the investment is used for
purchasing capital. The amount with interest added, (1 + α A), is divided between consumption, (1 − s ), and rein-
vestment, s + α A, at period t + 1. The investment produces the amount with interest added, (s + α A)(1 + α A)
at period t + 2, which, in turn, is divided between consumption, (1 − s )(s + α A), and reinvestment, (s + α A)2 ,
at period t + 3, and so on. This process continues infinitely, so the discounted present value of the increase in
utility from investment is
β(1 − s )
β(1 − s ) + β 2 (1 − s )(s + α A) + β 3 (1 − s )(s + α A)2 + · · · = .
1 − β(s + α A)
When Assumption 2 is satisfied, this utility gain is greater than the utility when the entrepreneur consumes the
additional unit.

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 135
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

the transversality condition, which contradicts with optimality. If µ t explodes to negative

infinity, it violates the constraint µ t ≥ 0 at some finite period, which contradicts with
optimality. This implies that only the steady state is consistent with optimality. Therefore,
(11) implies
β(1 − s )
µt + 1 = µt = µ = > 0. (12)
1 − β(α A + s )

It follows from (7), (12) and Assumption 2 that ∈t = 1 −β(1 −s)

β(α A + s )
−1=
β(1 + α A) − 1
> 0.
1 − β(α A + s )
Finally, it follows from (9), (12), Rt = 1/β̂ and
Assumption 2 that
 −1  β(1 − s )
λt = Rt+1 − β µ = (β̂ − β) > 0. (13)
1 − β(α A + s )
Q.E.D.
This lemma implies that all of the three inequalities (3), (4) and (5) are satisfied with
equality. Using (12) and (13), (10) is rewritten as

β(1 − α)AK t + β Q t + 1 − (1 + τ )Q t + γ (R − 1 − β)Q et,t + 1 = 0. (14)

There is no aggregate uncertainly in the model. Therefore, under rational expectations,

agents have perfect foresight for future land prices. By inserting Q et,t + 1 = Qt + 1 into (14),
and rearranging, the no-arbitrage condition for land becomes

Q t + 1 1 + τ β A(1 − α) K t
= ∗ − , (15)
Qt β β∗ Qt

where β ∗ ≡ β + γ λ = (1 − γ )β + γ β̂(<1) is the effective discount rate faced by en-

trepreneurs. Because β < β̂ (Assumption 1), the greater the value of γ , the greater the
value of β ∗ . A greater γ might reflect better condition of land as collateral.
We next study the collateral channel through which the growth rate of the land price
influences economic growth. If the budget constraint (3) is satisfied with equality, by using
Lt = Lt− 1 = 1, (4) with equality, and the government’s budget constraint Tt = τQt Lt , (3)
can be rewritten as

AK t − 1 + Bt − Rt−1 Bt−1 = K t − s K t − 1 + (1 − θ )τ Q t . (16)

By inserting the borrowing constraint (5) with equality into (16) and rearranging, we finally
obtain
Kt Qt Qt + 1
= (s + A) + γ R −1 − {γ + (1 − θ )τ } . (17)
Kt − 1 Kt − 1 Qt

Equation (17) shows that the current growth rate of capital tends to depend positively on the
expected growth rate of the land price Q t + 1 /Q t . This suggests that a possible appreciation
in the liquidation value of land accelerates capital accumulation. Finally, we obtain the

136 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

following relation from the definition.

K t /K t − 1 K t /Q t
= . (18)
Q t /Q t − 1 K t − 1 /Q t − 1

The equilibrium of this economy is described by the sequence of three variables

{K t /K t − 1 , Q t /Q t − 1 , K t /Q t }∞
t = 1 , which satisfy (15), (17) and (18), simultaneously.
We analyze the dynamic properties of this equilibrium. Rearranging (15), (17) and
(18), the dynamic system is described by the first-order difference equation of the growth
rate of the land price defined by g tQ+ 1 ≡ Q t + 1 /Q t ;

(1 + τ ) + β ∗ (s + A) + {γ + (1 − θ )τ }β A(1 − α) (s + A)(1 + τ ) g Qt
g Qt+ 1 = −
β ∗ + γβ A(1 − α)/R β ∗ + γβ A(1 − α)/R
 
≡ F g Qt , τ, θ , (19)

where β ∗ ≡ β + λγ = (1 − γ )β + γ β̂ < 1. It is straightforward to obtain that F (g tQ , τ , θ )

is increasing in g Qt , lim F (g tQ , τ, θ ) = −∞, and
g Qt →0

  (1 + τ ) + β ∗ (s + A) + {γ + (1 − θ )τ }β A(1 − α)
lim F g Qt , τ, θ < .
Q
g t →0 β ∗ + γβ A(1 − α)/R

Given β̂ > β ∗ , if β̂(s + A) < 1 holds, we obtain

(1 + τ − β ∗ R){1 − (s + A)/R} + (1 − θ )τβ A(1 − α)

F (R, τ, θ ) − R = >0
β ∗ + γβ A(1 − α)/R

for any τ ∈ [0, 1] and any θ ∈ [0, 1]. As Figure 1 illustrates, F (g tQ , τ , θ ) intersects the
45-degree line twice when s + A < R(R = 1/β̂) holds. The growth rate of the land price
realized at point F, however, exceeds the interest rate R, which violates the transversality
condition. Hence, we focus on only the neighborhood of point E. Because point E is
dynamically unstable, all the exploding paths violate the transversality condition. Therefore,
the balanced growth path realized at point E is the only equilibrium.
Finally, we prove that the credit market clears only at the interest rate Rt = 1/β̂. If
Rt < 1/β̂, not only entrepreneurs but also investors want to borrow, and excess demand
is created in the credit market, which contradicts the market clearing. If Rt > 1/β̂, any
investor would be willing to lend to entrepreneurs unboundedly, whereas entrepreneurs
can borrow the amount only to the extent that the borrowing constraint binds, so the excess
supply would prevail in the credit market. The credit market clears only at Rt = 1/β̂. In
equilibrium, entrepreneurs are borrowers and investors are lenders. Therefore, we obtain
the following:

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 137
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

Figure 1 The existence of the balanced growth path.

Proposition 1 Assume that Assumptions 1, 2 and β̂(s + A) < 1 are satisfied. Then there
exists a unique balanced growth path.
Output, capital and the land price grow at the same rate at the balanced growth path.5 In
the next section we study how a change in the tax rate, τ , will influence economic growth.
5 From (15), the land price is represented by
 2  t + T
β∗ β∗ β∗
Q t = β ∗ A(1 − α) K t + K te + 1 + K te + 2 + · · · + lim Q et + T .
1+τ 1+τ T →∞ 1+τ
∗
We impose the transversality condition, which excludes land price bubbles; that is, lim ( 1β+ τ )t + T Q et + T = 0.
T →∞
Let G t + i + 1 ≡ K te + i + 1 /K te + i (i = 0, 1, · · · ) denote the (expected) growth rate of capital. Then the growth rate
of the land price is represented by
∗  ∗ 2
Qt + 1 1 + 1β+ τ G et + 2 + 1β+ τ G et + 3 G et + 2 + · · ·
= G et + 1 ∗  ∗ 2 .
Qt 1 + 1β+ τ G et + 1 + 1β+ τ G et + 1 G et + 2 + · · ·
The growth rate of the land price depends on the sequence of future growth rates of capital. In particular, if people
expect the growth rates of capital to be the same throughout future periods, the above equation can reduce to the
simple relation given by Q t + 1 /Q t = G et + 1 .

138 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

3 Land tax and economic growth

In this section we consider the effect of land tax on economic growth. Assume that the
government announces taxation on land from period t on wards after the period t − 1
production begins but before the period-t debt repayment is made. Figure 2 illustrates the
timing of events.
Because it is hard to investigate the general case, we examine two polar cases. In one
case, the tax revenue is fully transferred to investors, whereas in the other case it is fully
refunded to entrepreneurs. First, we examine the case when investors receive all the transfer.
The following holds.

Proposition 2 Assume that land tax is fully transferred to investors (θ = 0). The tax on land
reduces the growth rate of the economy at the balanced growth path.

PROOF: We prove this by comparing two functions, F (g tQ , 0, 0) and F (g tQ , τ , 0). First,

we obtain
     2
∂ F g Qt , 0, 0 ∂ F g Qt , τ, 0 − τ (s + A)/ g Qt
− = ∗ <0 (20)
∂ g Qt ∂ g Qt β + γβ A(1 − α)/R

for any g tQ > 0 and any τ > 0. Equation (20) implies that the slope of F (g tQ , 0, 0) is always
smaller than that of F (g tQ , τ , 0) for any g tQ > 0. As Figure 3 illustrates, both functions
F (g tQ , 0, 0) and F (g tQ , τ , 0) are monotone increasing and intersect only once. Let M
denote the intersection.
Second, we show that point M lies below the 45 degree line. Letting ḡ (θ ) denote
+A
the growth rate at the intersection, we obtain ḡ (θ ) = 1 + (1 −sθ)β A(1 − α)
, where ḡ (θ ) satisfies
∗
F (ḡ (θ ), 0, θ ) = F (ḡ (θ ), τ, θ ) = β (s + A) − (1 − γ − θ)β A(1 − α)
β ∗ + γβ A(1 − α)/R
≡ F (ḡ (θ )). When θ = 0, point
M is given by the following pair:

(s + A) β ∗ (s + A) − (1 − γ )β A(1 − α)
{ḡ (0), F (ḡ (0))} = , .
1 + β A(1 − α) β ∗ + γβ A(1 − α)/R

Figure 2 The timing of events.

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 139
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

Figure 3 The relation of balanced growth paths with and without texation (the case for θ = 0).

Under Assumption 2 and β ∗ = (1 − γ )β + γ β̂, we obtain

β A(1 − α)(1 − γ )[1 − β(α A + s )]

F (ḡ (0)) − ḡ (0) = − < 0, (21)
[β ∗ + γβ A(1 − α)/R][1 + β A(1 − α)]
which implies that point M necessarily lies below the 45 degree line.
Finally, we show that ḡ (0) is always smaller than R. We obtain R − ḡ (0) =
β A(1 − α) + 1 − β̂(s + A)
β̂[1 + β A(1 − α)]
> 0. Let E0θ= 0 be the smaller value of the intersection of F (g Q
t , 0, 0) with

the 45 degree line and let Eτθ= 0 be that of the intersection of F (g tQ , τ , 0) with the 45 degree
line. As Figure 3 shows, F (g tQ , τ , 0) should lie above F (g tQ , 0, 0) for g Qt ∈ (ḡ (0), +∞). It
is easy to see that point Eτθ= 0 always lies in the southwest of E0θ= 0 . Therefore, the growth
rate under land tax is smaller. Q.E.D.
The land price falls promptly in response to the tax and the consequent depreciation of the
collateral value of land makes entrepreneurs face more severe credit constraints. Land tax
hampers capital accumulation and economic growth.

140 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

Next, we examine the other polar case when tax revenue is refunded to entrepreneurs.
The economy is expected to grow faster to the extent that the tax revenue is refunded to
entrepreneurs because the increased net worth of entrepreneurs will mitigate their credit
constraints (e.g. Bernanke and Gertler 1989). The following holds.
Proposition 3 Assume that land tax is fully refunded to entrepreneurs (θ = 1). The tax on
land increases the growth rate of the economy on a balanced growth path.

PROOF: We prove this by comparing F (g tQ , 0, 1) and F (g tQ , τ , 1). The procedure of proof

is analogous to that of Proposition 2. First, we obtain
     2
∂ F g Qt , 0, 1 ∂ F g Qt , τ, 1 τ (s + A)/ g Qt
− =− ∗ <0 (22)
∂ g Qt ∂ g Qt β + γβ A(1 − α)R − 1

for any g tQ > 0 and τ > 0. Equation (22) implies that the slope of F (g tQ , 0, 1) is always
smaller than that of F (g tQ , τ , 1) for any g tQ > 0. As Figure 4 illustrates, both F (g tQ , 0, 0)
and F (g tQ , τ , 0) are monotone increasing and intersect only once. Let N be the intersection.

Figure 4 The relation of balanced growth paths with and without texation (the case for θ = 1).

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 141
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

Second, we show that point N lies above the 45 degree line. Point N is given by
∗
the following pair: {ḡ (1), F (ḡ (1)} = {s + A, ββ(s∗ ++γβ
A) + γβ A(1 − α)
A(1 − α)R − 1
}. Under β̂(s + A) < 1 and
β ∗ = β + γ (R − 1 − β) = (1 − γ )β + γ β̂, we obtain
γβ A(1 − α){1 − (s + A)R − 1 }
F (ḡ (1)) − ḡ (1) = > 0, (23)
{β ∗ + γβ A(1 − α)R − 1 }
which implies that point N necessarily lies above the 45 degree line.
Finally, we show that ḡ (1) is always smaller than R. Under β̂(s + A) < 1, we obtain
R − ḡ (1) = 1/β̂ − (s + A) = 1 − β̂(sβ̂ + A) > 0. Let E0θ= 1 be the smaller value of the intersec-
tion of F (g tQ , 0, 1) with the 45 degree line and let Eτθ= 1 be that of the intersection of F (g tQ ,
τ , 1) with the 45 degree line. As Figure 4 shows, because F (g tQ , τ , 1) is steeper than F (g tQ ,
0, 1), point Eτθ= 1 always lies in the northeast of E0θ= 1 . Therefore, the growth rate under land
tax is greater. Q.E.D.
Proposition 3 implies that when the tax revenue is fully refunded to entrepreneurs, the
credit-mitigation effect dominates the credit-contraction effect. This result is somewhat
surprising because economic growth is enhanced although entrepreneurs are refunded only
the same amount as that paid for taxation.
Finally, we briefly comment on the effects of land taxation on the economy without
any constraint on borrowing. In a perfect credit market, the growth rate of the balanced
growth path is independent of the borrower’s capital structure. Therefore, the change in
the land price that will be induced by the taxation never affects economic growth.

4 Calibration

In this section, we calibrate how land taxation affects the behavior of the aggregate economy.
As a benchmark, we calculate the growth rate of the balanced-growth path in the no-
taxation economy. Next, we calculate the growth rate when land is taxed by 1 percent in
each of the two transfer cases. Finally, we trace out the behavior of the land price and
capital.
We describe parameters used for calibration. The selected time period is 1 year. The
discount factor of investor β̂ is 0.96, and the discount factor of entrepreneur β is 0.931.
Note that β̂ equals the inverse of the gross interest rate, R. Next, we consider technological
parameters, α and A. It is difficult to find α because research estimating the technological
parameters of the production function including both capital and land is not found.
We take an approach of approximating α and A by using the benchmark model under
perfect information. Two equations, KKt +t 1 = β(1 + α A) and β{A(1 − α) QK tt + QQt +t 1 } = 1,
can constitute the benchmark economy. We choose annual growth rates of capital and
land prices to be 2 percent, which was the average growth rate in the developed countries
in the 1990s (also used by Hayashi and Prescott (2002)). We set KKt +t 1 = QQt +t 1 = 1.02. We
construct QK tt by using the Japanese data because the data on land is not available for other
developed countries. We set QK tt = 1.36 by averaging the capital stock over land, both of
which were held by non-financial corporations in Japan, over the period from 1994 to 2003

142 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

Table 1 Selected parameters and calibration results

Selected Parameters
α (the technical parameter of capital) 0.706
A (the level of technology) 0.135
β̂ (discount factor of investors) (β̂ = R −1 ) 0.96
β (discount facotr of entreprenuers) 0.931
s (saving rate of entreprenuers) 0.9
γ (collateral rate) 0.7
Calibration Results
The growth rate of balanced-growth path (τ = 0) 1.56%
The growth rates of balanced-growth paths (τ = 0.01)
The case for θ = 1† 2.17%
The case for θ = 0‡ 1.13%
The change rates of the initial land price (τ = 0 → 0.01)
The case for θ = 1 9.09% down
The case for θ = 0 28.5% down
†
θ = 1 represents the transfer scheme when the tax revenue is fully refunded to entrepreneurs. ‡ θ = 0
represents the transfer scheme when the tax revenue is fully transferred to investors.

(Economic and Social Research Institute 2005). If follows from this procedure that α = 0.706
and A = 0.135. Following the model, we interpret the entrepreneur’s consumption as the
depreciation of capital. We set the saving rate of entrepreneurs to be 0.9 by averaging 1
minus the consumption of fixed capital over the fixed capital over the period from 1994
to 2003.6 Finally, we set the collateral rate, γ , to be 0.7.7 We calculate the growth rate of
the balanced-growth path of the benchmark economy as 1.56 percent. Table 1 lists chosen
parameter values and calibration results.
The overall effect of land taxation on economic growth depends on how tax revenue is
distributed. We again calibrate the two polar cases. The growth rate for the case of θ = 0 is
1.13 percent, which is 0.43 percent than the no-taxation economy, whereas the growth rate
for the case of θ = 1 is 2.17 percent, which is 0.61 percent greater.
Next, we consider the behavior of the levels of land price and capital. In doing so, we have
to investigate carefully the process of debt repayment because investors might experience
unanticipated declines in the liquidation value of land as a result of policy changes and
might be forced to renegotiate initial debt contracts.
Assume that the land price that was expected to prevail at period t, Q̂ t , varies to Q̃ t in
response to land taxation, where the variable with a hat “∧ ” denotes those before the policy
change and the variable with a tilde “∼” denotes those after it. If the initially contracted
6 In the actual counterpart of the economy, the entrepreneurs’ consumption in the model might be interpreted

as dividends paid to shareholders. Therefore, we might be able to interpret 1 − s as the sum of the dividends paid
from corporations and the consumption of fixed capital over the fixed capital. The ratio of dividends over the
fixed capital is no more than 1 percent over this interval.
7 To our small knowledge about the banking practice, the collateral rate ranges from 0.5 to 0.8. We also tried the

cases for γ = 0.6 and 0.8, but the results are insensitive to the variation in the collateral rate.

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 143
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

repayment, R B̂t − 1 , exceeds the realized liquidation value of land, borrowers would have
an incentive to repudiate the initial contract, γ Q̂ t , and to negotiate the debt down equal to
the “new” collateral value, γ Q̃ t .
The borrower’s budget constraint at period t, the counterpart of (17), is written as

K̂ t − 1 K̃ t
(s + A) = + {γ + (1 − θ )τ } − γ R − 1 g Q (τ, θ ), (24)
Q̃ t Q̃ t

where g Q (τ, θ ) = Q̃ t + 1 / Q̃ t is the growth rate of the land price of the new balanced-growth
path. The new no-arbitrage condition, the counterpart of (15), requires

K̃ t
β ∗ g Q (τ, θ ) + β A(1 − α) = 1 + τ. (25)
Q̃ t
With this preparation, we first calculate the land price at period t when land begins to be
taxed. The land price falls by 28.5 percent for the case of θ = 0, and 9.09 percent for the
case of θ = 1. In either transfer scheme, the land price falls immediately at the timing of
taxation. Figure 5 illustrates the behavior of the land price. Importantly, the land price falls
more sharply when the tax revenue is transferred to investors. Entrepreneurs face a more
severe agency problem and, therefore, demand less land. The market clearing forces the
land price to fall sharply, which in turn leads to severe credit constraints.
Finally, we illustrate the behavior of the stock of capital in Figure 6. We set the initial
level of the stock of capital before land is taxed at
567847.6bn, which was the value at the

Figure 5 Behavior of land price.

144 International Journal of Economic Theory 3 (2007) 131–146 

C IAET
Masaya Sakuragawa and Yukie Sakuragawa Land tax and economic growth

Figure 6 Behavior of stock of capital.

end of the 2003 calendar year in Japan.8 Because entrepreneurs can optimally renegotiate
the initial debt contract in the face of a decline in the liquidation value of land, capital can
grow at the same rate as the new balanced growth path immediately after land is taxed.
For the case of θ = 0 when the tax revenue is fully transferred to investors, capital grows
at 1.13 percent, which is slower than for the no-taxation economy, whereas for the case of
θ = 1 when the tax revenue is fully refunded to entrepreneurs, capital grows faster at 2.17
percent.

5 Conclusion

We have studied how land taxation influences economic growth in an economy under
credit market imperfection, where land is used not only as an input of production but also
as collateral for borrowing. The effect of land tax on economic growth depends on how tax
revenue is distributed.
The theoretical findings of the present paper suggest that faster economic growth is
associated with initial depreciation in land price, which is consistent with the published
literature on overlapping-generations economies, including Feldstein (1977), Drazen and
Eckstein (1988) and Deaton and Laroque (2001), all of which argue that the fund released
from purchasing land can contribute to capital accumulation. It is particularly reminiscent
8 This value is available from fixed assets for non-financial corporation. We use data from the Economic and

Social Research Institute (2005).

International Journal of Economic Theory 3 (2007) 131–146 

C IAET 145
Land tax and economic growth Masaya Sakuragawa and Yukie Sakuragawa

of Feldstein (1977) in that a tax on land finally leads to the appreciation in the land price.
The distinguishable feature of our finding from Feldstein is that we derive this result in a
model of infinitely-lived agents.
This theoretical finding bears an important policy implication. An appropriately com-
bined land-tax-subsidy scheme, for example, such as a reduction of the corporate tax rate
combined with a rise in the tax rate of land, can lead to faster economic growth followed by
appreciations in land prices in an economy where land plays an important role as collateral.

References

Bernanke, B., and M. Gertler (1989), “Agency costs, net worth, and business fluctuations,” American Economic
Review 79, 14–31.
Deaton, A., and G. Laroque (2001), “Housing, land prices, and growth,” Journal of Economic Growth 6, 87–105.
Diamond, P. A. (1965), “National debt in a neoclassical growth model,” American Economic Review 55, 1126–50.
Drazen, A., and Z. Eckstein (1988), “On the organization of rural markets and the process of economic develop-
ment,” American Economic Review 78, 431–43.
Economic and Social Research Institute (2005), Annual Report on National Accounts 2005, Cabinet Office, Gov-
ernment of Japan, Tokyo, Japan.
Feldstein, M. (1977), “The surprising incidence of a tax on pure rent: A new answer to an old question,” Journal
of Political Economy 85, 349–60.
Hashimoto, H., and M. Sakuragawa (1998), “Land-tax, transfer and growth in an endogenously growing economy
with overlapping generations,” The Japanese Economic Review 49, 412–25.
Kiyotaki, N., and J. Moore (1997), “Credit cycles,” Journal of Political Economy 105, 211–48.
Ogawa, K., and S. Kitasaka (1998), Sisan shijyou to keiki hendou (in Japanese), Nippon Keizai Shinbun Sha, Japan.
Ogawa, K., S. Kitasaka, H. Yamaoka, and Y. Iwata (1996), “Borrowing constraints and the role of land asset in
Japanese corporate investment decision,” Journal of the Japanese and International Economics 10, 122–49.
Ogawa, K., and K. Suzuki (1998), “Land value and corporate investment: Evidence from Japanese panel data,”
Journal of the Japanese and International Economics 12, 232–49.
Romer, P. M. (1986), “Increasing returns and long-run growth,” Journal of Political Economy 94, 1002–1037.
Sakuragawa, M., and Y. Sakuragawa (2002), “The long-run interaction between land price and economic growth
in a credit-constrained economy: Theory and evidence in Japan,” Discussion Paper No. 317, Nagoya City
University, Nagoya, Japan.

146 International Journal of Economic Theory 3 (2007) 131–146 

C IAET