Evaluating the Economic Impacts of a Disaster:

A CGE Application to the Tokai Region of Japan

Hiroyuki SHIBUSAWA
*


Abstract
Natural disasters have a negative effect on people and the regional economy.
The central and regional governments have made natural disaster reduction a high
priority. In this paper, we develop a dynamic spatial CGE model for evaluating the
economic impact of an earthquake on the Tokai region of J apan. Our model is
characterized as a decentralized economy with utility-maximizing consumers and
value-maximizing firms in a dynamic context. This model embodies both the spatial
interactions among regions and the dynamics of regional investments. A simulation
model is constructed of an inter-regional inter-sectoral economy in which J apan is
subdivided into 47 regions. All the regions are connected by transportation networks.
The model is calibrated for the regional economy in J apan. By numerical simulation,
we examine the economic impacts of an earthquake on the Tokai region using our
analytical scenarios.

1. Introduction
In this paper, we develop a dynamic spatial computable general equilibrium
(DSCGE) model to evaluate the economic impacts of an earthquake on the Tokai
region of J apan. Our model is characterized as a decentralized economy with
utility-maximizing consumers and value-maximizing firms in a dynamic context. This
model embodies both the spatial interactions among regions and the dynamics of
regional investments.
A numerical simulation model is developed of an inter-regional inter-sectoral
economy in which J apan is subdivided into 47 regions. All the regions are connected
by road, rail, sea, and air transportation networks. The model is calibrated for the
regional economy in J apan. The dynamic impacts of an earthquake on the Tokai region
are analyzed by numerical simulation. In our analytical scenario, a reduction in the
capital stock is the primary physical damage caused by an earthquake.
CGE analysis is a major artery in economics, regional science, and
engineering. It is also widely recognized as a policy evaluation method. There is a vast
literature reporting applications of static CGE models, but few studies have employed

*
Toyohashi University of Technology, 1-1 Tempaku Toyohashi, Aichi 441-8580, J apan
dynamic and spatial frameworks. Recent trials of dynamic and spatial CGE modeling
have been undertaken by Giesecke (2002,2003), McGregor, Swales and Yin (1995),
and McKibbin and Wilcoxen (1992). These previous studies mostly rely on a
quasi-dynamic framework, which is characterized by an evolutionary approach and a
sequential procedure. Economic impacts of disasters in a spatial context have been
analyzed by Okuyama, Chang (2004) and Shon, Kim, Hewings, Lee and J ang (2003).
Koike and Ueda (2005) examined the economic damage caused by a catastrophe in
J apan by using a static spatial CGE model.
In this study, the CGE model is developed based on dynamic macroeconomic
theory with a multi-regional and multi-sectoral specification. The investment is
endogenously determined by the behavior of value-maximizing firms, which face
capital adjustment costs. The dynamic impact of an earthquake in J apan is evaluated
by using the spatial CGE model. In this paper, we show a modified version of the
model which was developed by Shibusawa, Yamaguchi and Miyata (2009).

2. Basic Assumptions
The world is subdivided by region. Throughout the world there are general
industries, transportation industries and households. The economy is endowed with the
primary factors of labor and capital. Labor is mobile across industries but not regions
and capital is immobile across industries and regions. Goods and factor prices are
determined in perfectly competitive regional markets. Commodity trade between
regions in the country generates demand for transportation services and unit
transportation costs are endogenous. Commodities are perfect substitutes, i.e., the
trades are determined by the trade coefficients. The interaction of commodities among
regions is enabled by road, rail, sea and air transportation networks. The modal share is
also given. The model is solved for rational expectation equilibrium under the
assumptions of perfect competition and foresight. However, we assume that firms
place priority on the investment-savings balance. Then the level of investment is
determined by the firms optimization behavior.
The model is finitely set up in discrete time. {1,2, , }
F
T t denotes a
planning period index and
F
t is the final planning period. The world is divided into a
home country and foreign country. These are subdivided by region. R denotes a
regional index in the home country. There are three kinds of industries, i.e. general,
transportation and distribution industries. The general industry involves domestic and
foreign trade between regions. I denotes a sector index for the general industry. M
is a sector index of the transportation industry. All the regions interact with each other
via the transportation networks. A transportation network is defined by nodes and links.
A transport path connecting two regions is fixed and the transport link distance is
exogenously given.

3. The Model
The model is based on dynamic macroeconomic theory with a multi-region
and multi-sector specification. Each region has production and household sectors.
Commodity trade flows are determined by the trade and modal share coefficients. We
characterize the problems related to the maximization of the production and household
sectors in this economy.

3.1 Production Sectors
Each sector of the general and transportation industries maximize their present
cash flow values in each period
r
jt
NC and the asset value of their industrial capital in
the final period ( )
r
j
u . The sector operates with constant returns to scale technology.
The sector chooses the optimal investment and labor employment strategies. The
behavior of the production sector j I M e in region r R e is given as
1 , 1
{ , , , }
( )
max
F F
r r r r
jt jt jt jt
r r r
t jt t j j t
K L
t T
NC K
+ +
e
+ u

X Z
,
subject to
, 1
(1 ) ( )
r r r r
j t j jt jt jt
K K K o
+
= + A ,
where ( , , ) ( )
r Or r r r r r r Dr r Dr r r
jt jt j jt jt jt t jt it ijt it i ijt
i I M i I M
NC p Y K L wL p X p G Z
e e


X .
1
1/(1 )
t
t


+ represents the discount factor and is the positive discount rate.
( )
r
j
Y is a production function of capital
r
jt
K , labor
r
jt
L , and a vector of intermediate
input
1
{ , , }
r r r
jt jt Ijt
X X = X . The value added production function for labor and capital
has a Cobb-Douglas form, while the intensities of intermediate goods are fixed. The
asset value for the final period ( )
r
j
u is a linear function of the capital stock for the
final period. The capital stock
r
jt
K is accumulated by an investment function ( )
r
jt
K A
with constant returns to scale. It is a function of a vector of intermediate inputs for the
investment
1
{ , , }
r r r
jt jt Ijt
Z Z = Z , and a Leontief type technology is assumed.
j
o is the
depreciation rate. It is assumed that the cost function of intermediate goods for
investment ( )
r
i
G has increasing returns to scale. It can be interpreted that the
function reflects both the costs of intermediate goods and the costs of adjusting their
capital inputs.
In these sectors, there are two kinds of prices in each region. One is the
producers price
Or
jt
p and the other is the purchasers price
Dr
jt
p in region r . If a
commodity j is tradable between regions o and d, then the producers price in
region o is represented by
Oo
jt
p and the purchasers price is represented by
Dd
j
p
( ) j I e . In the transportation sector,
Or
jt
p ( ) j M e means the unit price of the
transportation services in region r .
r
t
w is the wage rate.
After having paid wages to households, the sector has to decide how to
distribute profit and finance investment. In this model, the net investment is financed
by new bonds. Let
r
jt
B be the number of bonds in period t and
r
jt
r be the interest
rate. The bonds are traded in each region. The initial number of bonds is normalized by
1 1
r r
j j
B K = . In this case, the profit dividend is calculated as
( , , )
IO
r Or r r r r r r r r Dz z r r
jt jt j jt jt jt jt jt t jt i ij Bjt j jt
i I
p Y K L r B wL p X p K t o
e
=

X .
If the net investment is financed by issuing new bonds, it holds that
( )
r r Dr r r r r
Bjt jt it i ijt Bjt j jt
i I M
p B p G Z p K o
e
A =

,
where
r
jt
B A is the number of new bonds issued by sector j in region r for period
t.
r
Bjt
p is the price of the new bond. Therefore the outstanding bond is given by
, 1
r r r
j t jt jt
B B B
+
= + A with
1 1
r r
j j
B B = . It is assumed that the price of the new bond is given
by
r r
Bjt jt
p q = where
r
jt
q is the costate variable of the current-value Hamiltonian
function [ ( ) ]
r r r r r r
jt jt jt jt jt j jt
H NC q K Z K o = + A .
In this model, we assume that tradable goods are perfect substitute. The profit
of the distribution sector is given by ( ) j M e
( )
r Dr or or Oo Tor or or
Djt jt jmt jt jt jmt jmt jt
o R m M o R m M
p F p p F t
e e e e
= +

where
Tor Or or
jmt jmt mt jmt
p p D k .
The commodity flow
or
jt
F is calculates as
( )
or or r r
jt j jit j
i I M
F X C t
e
= +

where
or
j
t is the given trade coefficient.
or
jmt
is the given modal share.
Tor
jmt
p is the
transportation cost of mode m from region o to region r .
or
mt
D is the distance
between origin and destination along with a path. The path is exogenously given by the
shortest path rule.
mjt
k is a given unit transportation services of mode m for goods
i . From the zero profit condition, the purchasers price is given by
( )
Dr Oo Tor or or
jt j jm jm j
o R m M
p p p F
e e
= +

.

3.2 Household Sector
A representative household maximizes the utility level subject to income
constraints. The full income consists of wages and interest on bond holdings. The
behavior of a household in region r R e is given as
{ }
( )
max
r
r r
t t
t T
U
e

C
C ,
subject to 0
r r r r r Dr r r r
t it it t t it it Bit it
i I M i I M i I M
w r A d FA p C p A
e e e
+ + + A >

.
( )
r
U is a Cobb-Douglas utility function for period t and it is a function of
consumptions
1
{ , , }
r r r
t t It
C C = C .
r
it
A is the number of bond holdings per household.
r
it
A A represents new bonds issued for industrial investments. The household can
receive the interest income but must pay to obtain a new bond.
r
t
FA is the income
transfer that provides a balance against a surplus or deficit in foreign and regional
trades.
r
t
d is the profit dividend that is given as /
r r r
t it t
i I M
d N t
e

since the utility

function is not identical among regions. If the production function is of the constant
returns to scale, then the profit of the firm becomes zero.
In this model, we assume that the level of investment is determined by a
firms optimization behavior. Firms place priority on the investment-savings balance.
Therefore, the level of household savings is adjusted to coincide with the level of
investment. In this case, the new bonds and the bond holdings per household are
calculated as
/
r r r
it it it
A B N A = A ( ) i I M e and
, 1 1
( ) /
r r r r r
i t it it t t
A A A N N
+ +
= A + ( ) i I M e .

3.3 Equilibrium Conditions
To obtain an equilibrium solution, the following market clearing conditions
should be satisfied in each region( ) r R e .
(1) Goods and Services Markets
General Goods
( , , )
r r r r r r r rd or r r
j jt jt jt jit jt t j j jt jt
i I M d R o R
Y K L X C N T T E M
e e e
= + + +

X ( ) i I e
where
r
jt
E is a given export from region r and
r
jt
M is a given import to region r .
Transportation Services
( , , )
I
r r r r r or or or
m mt mt mt imt im imt imt
i I o R
Y K L D F k
e e
=

X ( ) m M e
(2) Labor
r r
t it
i I M
N L
e
=


r
t
N is the total labor force (population ) in each region and it is exogenously given.
(3) Capital
r r r
it t it
A N B A = A ( )
I M
i I

e
r r r r
it t it it
A N B K = = with
1 1 1 1
r r r r
i i i
A N B K = = ( ) i I M e
1
r
i
A is the initial number of bond holdings of a household.
(The simulation results will be shown in the meeting.)

References
Abel, A.B. and Blanchard, O. J ., An Inter temporal Model of Saving and Investment,
Econometrica, Vol.51, No.3, 1983, pp.675-692.
Ciesecke, J ., Explaining Regional Economic Performance: An Historical Application
of a Dynamic Multi-Regional CGE Model, Papers in Regional Science, Vol.81,
2002, pp.247-247.
Ciesecke, J ., Targeting Regional Output with State Government Fiscal Instruments: A
Dynamic Multi-Regional CGE Analysis, Australian Economic Papers, Vol.42,
2003, pp.214-223.
Kehoe, T.J ., Srinivasan, T.N. and Whalley, J ., Frontiers in Applied General
EquilibriumModeling, Cambridge University Press, 2005.
Koike, A. and Ueda, T., Economic Damage Assessment of Catastrophe by using
Spatial General Computable General Equilibrium Analysis, Proceeding of the
19
th
Pacific Regional Science Conference, 2005.
McGregor, P.G., Swales, K., and Yin, Y.P., Migration Equilibria in Regional
Economies: A Multi-Period CGE Analysis of an Improvement in Local Amenities,
in Recent Advances in Spatial EquilibriumModeling, eds. J eroen C.J .M. Van Den
Bergh, P. Nijkamp, and P.Rietveld, New York: Springer, 1995.
McKibbin ,W. and Wilcoxen, P., GCUBED: A Dynamic Multi-Sector General
Equilibrium Growth Model of the Global Economy, Brookings Discussion Papers
in International Economics, No.97, 1992.
Okuyama, Y. and Chang, S.E., Modeling Spatial and Economic Impacts of Disasters,
Springer, 2004.
Shibusawa, H., Yamaguchi, M. and Miyata, Y., Evaluating the Impacts of a Disaster in
the Tokai Region of J apan: A Dynamic Spatial CGE Model Approach, Studies in
Regional Science, Vol.39, No.3, 2009, pp.539-553.
Sohn, J ., Kim, T.J ., Hewings, J .D., Lee, J .S. and J ang, S-G, Retrofit Priority of
Transport Network Links under an Earthquake, J ournal of Urban Planning &
Development, 2003, pp.195-210.
Taniguchi, H., Development of an Estimation Method for Direct Economic Damage
Loss caused by Earthquake, Bulletin of the Graduate School of Social and
Cultural Studies, Kyushu University, Vol.4, 1998, pp.19-26.