KEIO UNIVERSITY MARKET QUALITY RESEARCH PROJECT (A 21 st Century Center of Excellence Project) KUMQRP DISCUSSION PAPER SERIES DP2007-003 Deciphering Japanese Economy Using Social Accounting Matrix (SAM) Multiplier Approach Hyun Suk * Abstract This study aims to clarify the causes of Japan’s long recession by inter-industry linkage analysis incorporating the production sector and the simplified loanable funds market within a social accounting matrix (SAM) framework from multi-sectoral perspectives. SAM macroeconomic multiplier could be decomposed further into savings-investment balance multiplier (M2) which could be construed as financial intermediary effect on the industry. This paper additionally attempts to perform the policy simulation in order to test the efficient re-allocation effects on the macroeconomy considering transmission linkage and multiplier effects in input-output table. The results of policy simulations indicate that it is not necessarily better to reallocate the resources of a low TFP industry to a high TFP industry especially when considering multi-sectoral linkages through intermediary goods and investment multiplier effects within SAM framework. *Bond Market Specialist, International Finance Department, Japan Bank for International Cooperation Graduate School of Economics and Graduate School of Business and Commerce, Keio University 2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan
31
Embed
(A 21st Century Center of Excellence Project)ies.keio.ac.jp/old_project/old/gcoe-econbus/pdf/dp/DP2007-003.pdf · (A 21st Century Center of Excellence Project) KUMQRP DISCUSSION PAPER
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
KEIO UNIVERSITY MARKET QUALITY RESEARCH PROJECT (A 21st Century Center of Excellence Project)
KUMQRP DISCUSSION PAPER SERIES
DP2007-003
Deciphering Japanese Economy Using Social Accounting Matrix (SAM) Multiplier Approach
Hyun Suk *
Abstract
This study aims to clarify the causes of Japan’s long recession by inter-industry linkage analysis incorporating the production sector and the simplified loanable funds market within a social accounting matrix (SAM) framework from multi-sectoral perspectives. SAM macroeconomic multiplier could be decomposed further into savings-investment balance multiplier (M2) which could be construed as financial intermediary effect on the industry.
This paper additionally attempts to perform the policy simulation in order to test the efficient re-allocation effects on the macroeconomy considering transmission linkage and multiplier effects in input-output table. The results of policy simulations indicate that it is not necessarily better to reallocate the resources of a low TFP industry to a high TFP industry especially when considering multi-sectoral linkages through intermediary goods and investment multiplier effects within SAM framework. *Bond Market Specialist, International Finance Department, Japan Bank for International Cooperation
Graduate School of Economics and Graduate School of Business and Commerce, Keio University
2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan
1
Deciphering Japanese Economy Using Social Accounting Matrix (SAM) Multiplier Approach†
HYUN SUK*
Abstract
This study aims to clarify the causes of Japan’s long recession by inter-industry linkage
analysis incorporating the production sector and the simplified loanable funds market within a
social accounting matrix (SAM) framework. SAM macroeconomic multiplier could be
decomposed further into savings-investment balance multiplier (M2), which could be
construed as financial intermediary effect on the industry.
This research additionally attempts to perform policy simulation in order to test the
efficient reallocation effects on the macroeconomy, considering transmission linkage and
multiplier effects in input-output table. The results of policy simulations indicate that it is not
necessarily better to reallocate the resources of a low TFP industry to a high TFP industry
especially when considering multi-sectoral linkages through intermediary goods and
investment multiplier effects within SAM framework.
†The author is very grateful to Colin McKenzie, Kazuhito Ikeo, Naoyuki Yoshino, Naosumi Atoda and Nagendra Shrestha for their helpful suggestions and valuable comments. Any remaining errors are solely mine. I also acknowledge that this research was supported partly from Research Grant of Fuji Xerox Setsutaro Kobayashi Memorial Fund. *E-mail: [email protected], International Finance Department I, Japan Bank for International Cooperation, 4-1 Ohtemachi 1-Chome, Chiyoda-ku, Tokyo 100-8144, Japan
2
Introduction
There have been lots of attempts to explain and clarify the causes of Japan’s long
recession, the so called “lost decade” and what happened in Japan during the 1990s.
Clarifying the causes of the “lost decade” is very important in revitalizing the Japanese
economy as well as understanding the long recessions in the 1990s. These issues can be
explored mainly from the perspectives of the demand-side and the supply-side. From the
demand-side, it is usually explained that the GDP gap between real GDP and potential GDP
was lost and that fiscal and monetary policy could stimulate the economy. From the
supply-side, total factor productivity (TFP) growth and GDP growth led by low TFP growth
slowed during the 1990s.
Fewer papers investigate supply-side factors compared with demand-side factors. Using
a stochastic growth model, Hayashi and Prescott (2003) point out that the lost decade of the
Japanese economy in the 1990s resulted from the following: 1) a fall in TFP growth, and 2) a
reduction in the length of the work week. Since then, low TFP growth has been pointed out as
a major and stylized factor of Japan’s long recession after the bubble burst.
It has been argued that this low TFP growth results from the inefficient allocation of
bank credit, so called “Oikashi (over-loan, ever-greening)” and there are several empirical
researches on this relation between economic productivity and the misallocation of credit. By
using sectoral loan data from 1993 to 1997, a seminal paper of Hoshi (2000) shows that bank
loans to manufacturing industries with high profits decreased while loans to real estate
industries with low profits increased in the form of additional loans or extensions and interest
exemptions, and loans to real estate consequently crowded out loans to manufacturing.
Sakurakawa (2002) also shows that the share of real estate loans was positively related to land
price that is a proxy variable for profitability of real estate industry up to 1991 but since 1992
there has been no such significant relation by estimating loan share function of real estate
industry using banks’ panel data. Peek and Rosengren (2003) estimate the probability function
3
of bank loan with panel logit model and their empirical result shows that firms with low ROA
and liquidity have an increased probability of securing bank loans. They also show that the
stronger relationship the main bank has with borrowing firms or affiliate firms, the more loans
are made. Caballero, Hoshi and Kashyap (2004) identify the zombie firms and analyze the
impact of their existence on the macroeconomy. They indicate that the ratio of zombie firms
increased from 1993 especially in the industry of construction and real estate and the more
zombie firms exist, the wider the gap in productivity between zombie firms and non-zombie
firms becomes. These studies show that the “Oikashi” phenomenon is found in the real estate
and construction industries which are burdened with excessive bank loans by investigating
empirically the balance sheet variables of banks and firms.
However, under the circumstances of declining TFP in the 1990s, it is easy to guess that
those firms heavily dependent on bank loans with an optimistic view about the rising TFP
during the bubble period must have encountered difficulties in paying back their loans when
high TFP growth was not realized because of financial disintermediation. Also the
malfunction of financial intermediation might have aggravated the negative effects coming
out of lower TFP growth. The deterioration of financial market quality in the 1990s might be
partially responsible for the long recession of the Japanese economy because the efficient
allocation of capital is positively correlated with good-function and development of loanable
funds market1. Japanese recession could be deteriorated and prolonged by the falling TFP of
each industry, which resulted from declining liquidity and flexibility of the bank loan markets
that channel savings into investments opportunities.
This paper attempts to construct social accounting matrix (SAM)2 and calculated
SAM-based multipliers in order to perform multi-sectoral linkage analysis especially focusing
1 See more details in Wurgler (2000) 2 The social accounting matrix (SAM) is one of the useful tools of economic research, policy analysis and economic planning. This matrix is usually used as the database for computable general equilibrium (CGE) models. On the construction of Japanese SAM, refer to Hosoe (2004) for more details.
4
on simplified loanable funds market under savings-investment balance. SAM multipliers
calculated in each industry could be further decomposed to construe M2 multiplier of
investments on the real sector by dividing macroeconomic SAM into real sector and financial
sector. This decomposed multiplier (M2) captures production-savings-investment linkages
within this SAM framework. And then policy simulation tests the effects of efficient
re-allocation from low TFP industries to high TFP industries on the macroeconomy,
considering transmission linkages of industries because it is pointed out that economic
resources such as bank loans were allocated inefficiently to industries performing badly (low
TFP growth) during the bubble period. The results of policy simulation indicate that it is not
necessarily better to reallocate the resources from a low TFP industry to a high TFP industry
especially when considering the transmission mechanism through intermediary goods in the
macroeconomy.
1. Social Accounting Matrix (SAM)3
1.1 Social Accounting Matrix
The Social Accounting Matrix (SAM) is a statistical “snapshot” of the circular flow of
the economy that provides the basic information on transactions in an economy. It may be
described as a coherent square matrix system4 that collects data on the production and
income generation in Input-Output (IO) table and Systems of National Accounts (SNA) on the
one hand, and data on the incomes as received by different institutional groups, and the
expenditures of these incomes on the other hand.
Each cell shows the payment from the account of its column to the account of its row –
the incomes of an account appear along its row, its expenditures along its column. The
underlying principle of double-entry accounting requires that, for each account in the SAM, 3 SAM was proposed by Stone and elaborated on by Pyatt and Thorbecke (1976) and Pyatt and Round (1985). 4 For the data used a SAM, various kinds of consistency checks have to be carried out, in order to assure full consistency between the different data sources within the framework.
5
total revenue (row total) should equal total expenditure (column total).
SAM-multiplier analysis can be performed by partitioning the accounts into
endogenous and exogenous accounts. This multiplier analysis takes into account all the
interactions within each step of the process of linkages among incomes, expenditures, and
production. There is not a single multiplier, but a matrix of multipliers representing the
potential effects of the additional one cell of the original SAM on the rest of the cells. This
SAM relationship may trace the complicated interactions inherent in the circular process.
1.2 Macroeconomic Structure in SAM
Table 1 basically consists of a set of rows and columns with similar headings, but
different meanings, that is, production activities, institutions, capital, government and foreign
sector, are all considered to be interrelated directly or indirectly within the SAM framework.
In this model, production, households and capital are treated as endogenous and the
government and the rest of world (ROW) are partitioned as exogenous.
Table 1. Basic Macroeconomic SAM
Expenditures
Incomes 1 Expenditure 2 consumption 3 Investment4 Government Expenditure
5 ROW (Export)
1 Production A C INV G EX 2 Household
Income Y 0 0 0 0
3 Savings 0 hSAV 0 gSAV F 4 Government
Revenue 1T 2T 0 0 0
5 ROW (Import)
IM 0 0 0 0
Based on the above SAM, the basic macroeconomic model consists of equations (1)-(5).
These five equations satisfy the properties of the SAM in Table 1, namely, the totals of each
6
column and row must be same.
1TIMEXGINVCY −−+++= -----1)
2TSAVCY h ++= -----2)
FSAVSAVINV gh ++= -----3)
GTTTSAV g −+≡= )( 21 -----4)
EXIMF −= -----5)
Where Y denotes GNP, C denotes Consumption, INV denotes Investment, G denotes Government
Expenditure, EX denotes Exports, IM denotes Imports, hSAV denotes Household Savings, F(=IM-EX)
denotes the Balance of Trade ,.T denotes Total Taxes, 1T denotes Production Tax + Tariff, 2T denotes
Income Tax, A denotes Intermediate goods. It is assumed that cYC = and YsavSAV ii = , so that c is
the marginal propensity to consume, and savi: is the marginal propensity to save of sector i, where I denotes
either the government or households.
1.3 SAM-based Multipliers
The fundamental approach to the SAM-based multiplier models is to compute column
shares divided by the total amount of each column S , column coefficients matrix from a SAM
represents the structure of the economy which is analogous to Leontief inverse matrix
calculation in the IO table analysis. This SAM multiplier matrix captures the interdependence
of the endogenous variables given exogenous shocks. To perform SAM multiplier analysis, a
SAM first should be partitioned into two accounts, endogenous accounts and exogenous
accounts.
Following the approach of IO analysis, a linear multipliers model is constructed based
on the assumption that all the expenditure (column) coefficients in the SAM are constant. As
in most applications, it has been customary to regard the government and the rest of the world
as exogenous because the government accounts are essentially policy-determined, and the
external account is out of domestic control, while sectoral production, factor returns, and
7
household incomes are treated as being endogenous.
In this model, the capital account is also endogenized to capture the role of the
savings-investment balance in the determination of national income following Robinson et al
(1988) and endogenous accounts are further grouped into two. The first includes the factor
and production markets in the real side of the economy, and consists of supplier (a) and
household accounts (c, y). The second group incorporates financial flows consisting of
capital accounts that summarize the loanable funds market circulating savings into
investment.
XYNSXYNs
yinvca
YNh
+⋅=+⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛=
0000 ,
YNNS = -----6)
XMXSIYN ⋅=−= −1)( -----7)
where YN , N, X and M denote total SAM, the endogenous accounts, the exogenous
accounts, and the SAM multiplier matrix respectively. I is the identity matrix, and S is the
share coefficient matrix. This coefficient matrix S is decomposed into two groups in order
to capture the interaction between real side and financial side of the economy. The right upper
in 2S stands for the demand for investment and the left lower in 2S stands for savings.
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛+⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛=+=
00000
00
000000
21hs
invy
caSSS ------8)
This grouped SAM multiplier matrix can be further decomposed into three
multiplicative components following Pyatt and Round (1979) because this multiplicative
decomposition elucidates the routine of complex transmission linkages in the macroeconomy.
XYNSI =− )( ------9)
8
XMMMXSIDIDIMXYN 1231
112 ))(()( =−+−== −− ---10)5
1M stands for the ‘within accounts (intra-group or transfer) effect’ that is the multipliers of an
exogenous injection into one set of accounts will have on that same set of accounts. 2M
represents the ‘cross (open-loop or extra-group) effect’, an injection of an exogenous shock
into one set of accounts has on the other set of accounts with no reversal effects. This analysis
focuses this open-loop multiplier effect of investment on the production operating through the
savings-investment balance. 3M captures the multiplier effect of the full circular flow, so
called ‘between-accounts (inter-group or closed-loop) effect’ which measures the full circular
effects, resulting from an exogenous injection into the system, after returning to the account
where the injection originate in.
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
−−+−−−−+−
= −−−
−−−−
IcaIyIcyaIy
caIyIcaIcyaIM
000])([)]([0])([)()]([
111
1111
1
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
+−+−
= −
−
IsinvcyaIyI
invcyaIIM
h0)]([0
)]([01
1
2
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
+−−⋅+−−
⋅+−−⋅+−=
−−
−−
−−−
11
11
111
3
])]([[000])]([[00])]([[)]([
invcyaIysIsinvcyaIyI
sinvcyaIyIsinvcyaIIM
h
h
hh
5 To derive multiplicative SAM multipliers
xySI =− )(
S is partitioned into 1S and 2S then xySSI =−− )( 21
xSIySSIy 112
11 )()( −− −=−−
xSIySSII 112
11 )())(( −− −=−−
xSISSIIy 11
12
11 )())(( −−− −−−= xSIDIy 1
11 )()( −− −−= where 2
11)( SSID −−=
)()()( 121 DIDIDI +−=− −−
9
2. SAM Multiplier Analysis
There are few multi-sectoral analyses using a SAM which provides a framework
including both national income in SNA and the production in IO table, while there are many
input-output (IO) analyses focusing on intermediate flows and on the sectoral production and
demand. Therefore, this study attempts to build Japanese SAM and calculate the SAM
multipliers and decompose further the total multipliers into M2 multiplier especially focusing
on savings-investment balance in order to analyze the effects of investments on the real sector.
Figure 1 shows the total multiplier effect (M) of investment given a unitary exogenous
shock to government expenditure in each industry between 1980 and 2000. For example, the
investment multiplier in the construction sector (sector number 17) is the largest of all the
multipliers and peaked in the bubble period which was affected by public construction in the
1990s through the stimulating economic policy packages (the values of the multiplier for this
industry were: 1.1998 1.3601 1.0974 1.198 1.0945). Furthermore, commerce,
electrical machinery, food, personal services, and real estate have a large multiplier given an
exogenous shock, while the public service sector such as other public services, water supply
and waste management services, and public administration have a small multiplier effect as
usually expected. Interestingly R&D and service industries in 2000 such as business services,
personal services sector, communication and broadcasting, and education and research have
higher multiplier effects than during the economic bubble in 1985 while the total multipliers
of construction, iron and steel, foods, etc., have declined.
Specifically TFP85 scenario increases government expenditure by 1 million yen
respectively in high TFP industries such as commerce and finance & insurance and
simultaneously decreases government expenditure by 1 million yen respectively in low TFP
industries such as construction and real estate and then the same scenario is iterated over 1990,
1995 and 2000. Resource reallocation of this scenario increases total industry investment
especially in 1990 by 1.15 million yen. The interesting finding of this simulation is that
investment in the construction industry is increased even though government expenditure is
decreased by 1 million yen. In contrast, decrease in government expenditure downsizes the
investment in the real estate industry almost three-fold. This simulation indicates that
inter-industry linkages and industrial structure affect the multipliers of investment between
6 It is not feasible to reallocate resources directly in SAM framework therefore an increase or a decrease in government expenditure stands for resource re-allocation in this simulation.
15
industries.
M2 scenario in simulation I increases government expenditures by 1 million yen
respectively in large M2 multiplier industries such as construction and commerce, and
simultaneously decreases government expenditure by the same amount respectively in
relatively small investment M2 multiplier industries of finance & insurance and real estate.
Like the result of TFP85 simulation, total industry investment in 1990 is bigger than those in
other periods. This reallocation from large M2 industries to small M2 industries has a
negative impact on investments in service sectors in 2000 and consequently total industry
Activities not elsewhere classified 0.006042 0.004825 -0.00388 0.00034
Total 0.165175 1.150433 -0.11369 0.139633
18
4. Conclusion and Summary
Many works have been done in order to elucidate the Japanese long recession since the
bubble burst from the supply side or demand side or sometimes from both sides. However
there are few multi-sectoral linkage analyses to explain the cause of the lost decade within
SAM framework focusing on the interaction between real sectors and the financial sector.
Therefore this research attempts to construct five Japanese SAMs from 1980 to 2000 and to
calculate SAM-based multipliers in order to analyze the effects of investments on each
industry.
Both total investment multipliers and M2 multipliers of investment that peaked in the
bubble period of 1985 have declined and these declining M2 multipliers could implicitly
indicate that the quality of the loanable funds market also was worsened under severe
financial environments such as non-performing loans and BIS capital adequacy ratio. One of
the interesting empirical findings shows that the real estate industry has an increasing M2
multiplier of investment which could consequently contribute to the growth of that industry,
while the construction industry a decreasing M2 multiplier and the share of that industry
decreased. Both the construction industry and the real estate industry have been pointed out as
industries that induced economic bubble and misallocation of bank credit (Oikashi).
However, when considering the inter-dependence and transmission mechanism of industries,
this decomposed M2 multiplier analysis dilutes the possibility of “Oikashi” in the real estate
industry while this supports the possibility of “Oikashi” in the construction industry.
Simulation results shows that the efficient re-allocation of resources from low TFP
growth industries to high TFP growth industries is not necessarily better for the economy
when considering inter-industry linkages and investment multipliers effects. By using SAM,
this multi-sectoral analysis could be extended to the computable general equilibrium model
which incorporates macroeconomic variables and the behaviors and interactions of each
economic agent.
19
References Jorge Alarcon et al (1991), The Social Accounting Framework for Development: Concepts,
construction and applications, Institute of Social Studies, Avebury. Caballero, R. J., T. Hoshi and A. K. Kashyap (2004) “Zombie Lending and Depressed
Restructuring inJapan,” mimeo Fukao et al (2005), “Sectoral Productivity and Economic Growth in Japan, 1970-98: An
Empirical Analysis Based on the JIP Database”, Hi-Stat Discussion Paper Series D03-19, Institute of Economic Research, Hitotsubashi University.
Fumio Hayashi and Edward Prescott (2002). “The 1990s in Japan: A Lost Decade,” Review of Economic Dynamics, 5: pp 206-235.
Hoshi,T.(2000).” Naze Nihon wa Ryudosei no Wana kara Nogarerarenainoka? (Why is the Japanese Economy Unable to Get Out of a Liquidity Trap?)” in Fukao,M. and H. Yoshikawa (eds.) Zero Kinri to Nihon Keizai (Zero Interest Rate and the Japanese Economy), Nihon Keizai Shimbunsha, pp.233-266.
Nobuhiro Hosose (2004), Textbook of Applied General Equilibrium Modeling, Tokyo University Press
Masahiro Kawai (2005), “Reform of the Japanese Banking System”, Hi-Stat Discussion Paper Series No.102, June 2005.
Kikuo Iwata and Tsutomu Miyakawa eds (2004), What is the true cause of the lost decade?, Toyo Keizai Shinposha
Yasushi Nakamura (2005), “Multiplier Analysis of Social Accounting Matrix”, Economia Vol 51 pp11-25, Yokohama National University.
Humio Otake and Noriyuki Ryukawa eds (2004), The Issue of Heisei Depression: The Verification of the Lost Decade, Toyo Keizai Shinposya
Peek and Rosengren (2003), “Unnatural Selection; Perverse Incentive and the Misallocation of Credit in Japan” NBER Working Paper Series 9643
Pyatt and Round (1985). Social accounting matrices: a basis for planning, World Bank Verbruggen Pyatt, G. & Thorbecke, E. (1976) Planning Techniques for a Better Future: a
summary of a research project on planning for growth, redistribution, and employment (Geneva, International Labour Office).
Robinson and Roland-Holst (1988), Macroeconomic Structure and Computable General Equilibrium Models, Journal of Policy Modeling, 10 (3), pp353-375
Sherman Robinson (2003), “Macro Models and Multipliers: Leontief, Stone, Keynes and CGE Models”, International Food Policy Research Institute (IFPRI) Working Paper.
Masaya Sakuragawa (2002), “Kin-yu Kiki no Keizai Bunseki ” Economic Analysis of Financial Crisis, University of Tokyo Press
Jeffrey Wurgler (2000), “Financial markets and the allocation of capital”, Journal of Financial Economics (2000), pp. 187–214
20
[Websites] Input-Output Table, Statistics Bureau, Ministry of Internal Affairs and Communications http://www.stat.go.jp/data/io/ Systems of National Accounts (SNA), Cabinet Office Economic and Social Research Institute, http://www.esri.cao.go.jp/jp/sna/menu.html Japan Industry Productivity (JIP) Database 2003, ESRI http://www.esri.go.jp/en/archive/bun/abstract/bun170index-e.html Japan Industry Productivity (JIP) Database 2006, METI http://www.rieti.go.jp/jp/database/d04.html