The Economic Growth of Korea after 1990: Identifying Contributing Factors from Demand and Supply Sides

This study is purposed to identify major factors that explains the growth path of the Korean economy in the past decades and evaluate their relative contributions. To that end, we present four economic models: Two of them contrast the recent changes in the determination of foreign exchange rate as well as the monetary policy rule Korean economy underwent right after the East Asian Currency Crisis in 1998, while the others are Blanchard and Quah (1989)’s original 2variable model and its 3variable extension. Converted properly into the corresponding SVAR systems with longrun restrictions, their estimation results confirm that the decreased rate of economic growth of Korea since 2000 seems attributable to the decrease in Korea’s potential growth rate.

The Economic Growth of Korea after 1990: Identifying Contributing Factors from Demand and Supply Sides

Seok-Kyun Hur Korea Development Institute

Abstract

This study is purposed to identify major factors that explains the growth path of the Korean economy in the past decades and evaluate their relative contributions To that end, we present four economic models: Two of them contrast the recent changes in the determination of foreign exchange rate as well as the monetary policy rule Korean economy underwent right after the East Asian Currency Crisis in 1998, while the others are Blanchard and Quah (1989)’s original 2- variable model and its 3-variable extension Converted properly into the corresponding SVAR systems with long-run restrictions, their estimation results confirm that the decreased rate of economic growth of Korea since 2000 seems attributable to the decrease in Korea’s potential growth rate

JEL classifications: E32, E60

Key Words: Structural Vector Auto Regression (SVAR), long-run restrictions, fixed VS flexible exchange rate system, monetary aggregate targeting Vs inflation targeting

I Introduction

Our study stems from a question, "How should we understand the growth pattern of the Korean economy after the 1990s?" Among various quantitative methods applicable, this study chooses a Structural Vector Auto Regression (SVAR) with long-run restrictions, identifies diverse impacts from either demand or supply sides that gave rise to the current status of the Korean economy, and differentiates relative contributions of those impacts

Following Blanchard and Quah(1989)'s tradition, we distinguish permanent supply shocks from transient demand ones by levying various identification restrictions In the first half of this paper, we replicate Blanchard and Quah’s original 2-variable model with the Korean macro data Then, we extend the models to a three variable one, which consist of demand, supply and price shocks We demonstrate the estimation results of these two models here because these types of models are quite popular and could be used as benchmarks for other

may reflect Korea specific features and historical experiences

As of year 2007, looking back to 1990s, the East Asian Currency Crisis in 1997 marks a break point for the Korean economy Especially, in the foreign exchange market and in the money market, a flexible exchange rate system and an inflation targeting rule of monetary policy were introduced respectively Needless to say, all the reform measures taken since the financial crisis exerted huge impact on the whole economy Among them, however, the transition from a fixed exchange rate system to a floating one as well as the transition from a monetary aggregate targeting to an inflation targeting regime was very crucial

In this context, we introduce the next two linear stochastic differential systems of equations, both of which contrast such drastic institutional changes while adhering to the same backbone in other aspects By solving the models we represent key macro variables as linear functions of exogenous shocks coming possibly from various

1 As Blanchard and Quah(1989) admit, their model is a mere example showing how a SVAR with long run restrictions is estimated

sources and derive long-run identification restrictions following Blanchard and Quah (1989) Then, we levy the identifying restrictions to VAR systems consisting of the key variables and estimate them with the Korean data We demonstrate estimation results in terms of Impulse Responses (IR) and Forecasting Error Variance Decomposition(FEVD) and interpret them in terms of economic growth Eventually, it is our destination to discern what portion of the economic growth of Korea is influenced by either the impact of productivity growth through technological progress, or changes in the aggregate demand induced by fluctuating consumption and investment, or exogenous shocks like ones from oil price

The contents of this paper are construed as follows: The second section observes the recent trend of the economic growth of Korea and reviews a few relevant domestic literatures, which might help in clearly defining the scope and analytic methodology of this study The third section provides a quantitative tool to be used in this study, which is Structural VAR (SVAR) as mentioned above Accordingly, variables used, estimation equations, and identification conditions of impacts are also explained here The fourth section reports estimation results derived by the previously introduced models, and the fifth section concludes

II The Economic Growth of Korea: A Phenomenon and Discussions

In this section, we exhibit the economic growth of Korea in the past decades and summarize the relevant domestic literature Despite the abundant existing literature on the issue, we restrict our attention to the ones using the methodology of SVAR

1 1 The Economic Growth of Korea

<Table 1> Averages and Standard Deviations of Real GDP Growth Rate

<Table 1>, the average real GDP growth has been falling from marvelous 8.1% during 1970s to 5.0% during 2000s

In the meantime, the volatility of the GDP growth(measured by standard deviation) rose from 3.4% during 1970s to 4.8% during 1990s and fell again to 2.1% during 2000s Reminded that the East Asian Currency Crisis broke out in the fourth quarter of 1997, it seems that the severity of business cycle fluctuations stayed more or less at the same level up to late 1990s and it was subdued in 2000s (see [Figure 1]) Such dampening business fluctuation seems to be related with the global prevalence of low interest rate and the emergence of China as a new world economic power

Putting all these into consideration, it would be pivotal to identify

post crisis changes in various market institutions of Korea and the global environments surrounding them for understanding the growth path of Korean economy (at least) after 1990s To take a few but most remarkable

and the floating exchange rate system were introduced, while financial institutions were restructured

[Figure 1] Trends of Real GDP Growth Rate (with or without Treatment of Seasonality)

ln(real GDP) ln(real GDP(S.A))

(Year-on-Year % Change, Quarter-on-Quarter % Change)

As is widely believed, financial restructuring led to changes of behaviors in both demand and supply sides of domestic capital market Banks moved their concentration from business finance to consumer loans in order to reduce risk exposure while enhancing profitability Accordingly, households could enjoy the benefit of consumption smoothing from the alleviation of liquidity constraints (Hur and Sung[2003]) In the meantime, most large firms, enforced to lower the debt/equity ratio, began to accommodate required capital by IPOs or internal reserves rather than by debt financing By that much, the banks could hold excess capacity to lend, which in turn directed towards consumer credit In addition, the global phenomenon of low interest rates and mild inflation, which sustained stable growth for the last

2 Some of these changes were caused by the Crisis but others happened to be taken after (or around) the outbreak of the Crisis Here, however, it is not our immediate concern to verify their causalities with the Crisis Instead we focus on evaluating the macroeconomic consequences of these post-crisis changes on the Korean economy

decade, contributed to settling the newly adopted inflation targeting rule (in 1997) and the flexible exchange rate system (in 1998) in post-crisis Korea

may result in lowering the business cycle amplitude, still it remains a puzzle to explain the slowed pace of economic growth in Korea Hence,

it would be crucial to devise analytical frameworks beyond merely introducing major institutional changes so as to include various sources

of shocks and their transmission channels, which may hinder the economic growth

2 2 Literature

In this section we introduce three papers, all of which, in common with others, explore the economic growth and/or the business cycle of Korea since 1990s using SVAR Though, they differ in the time span of data set used and the pool of variables chosen Hence, direct comparison

of their results may be not much of consequence Instead we highlight their methodological differences

First, Shim (2001) decomposes post-crisis business cycle by factors based on B-Q(1989) A linear system of sectoral equations, intended for deriving long-run restrictions, is arranged so that its Structural Moving Average Representation (SMAR) or a long-run impulse response matrix

not consider post-crisis changes in the monetary policy rule and the

Second, Kim (2005) concentrates on analyzing the impact of foreign shocks on the domestic business cycle Hence, Kim uses foreign variables, such as oil price and exchange rates, jointly with domestic ones including interest rate, CPI, and the growth rate Kim (2005)'s model, in

3 In other words, Shim(2001)'s system of equations is simply reduced to a VAR with Cholesky ordering

4 Monetary aggregate targeting and inflation targeting diverge from each other in the treatment

of a money supply schedule Under the monetary aggregate targeting, LM curve is derived from money demand=money supply whereas money supply is replaced by an interest rate setting rule, such as a Taylor rule under the inflation targeting Furthermore, autonomy of monetary policy

is not guaranteed in a fixed exchange rate system because the domestic interest rate should be always equal to the foreign interest rate Otherwise, the exchange rate would fluctuate

common with Shim(2001)'s, does not derive the relationship among shocks from an economic model Instead he orders them in a Cholesky way a priori

Third, Oh (2007) considers an open economy version of the B-Q model Matched with three key variables of world import volume, GDP, and CPI, he introduces three shocks of domestic supply, demand, and world supply In terms of shock identification, Oh (2007) also assigns

Keeping distance from the predecessors, our paper is based on economic models, which allow the presence of shocks from various sources and introduces institutional changes in the monetary policy regime and the foreign exchange market Then, we derive long-run restrictions by solving the models and use them in estimating corresponding SVAR models

III Models

In a neoclassical framework, it is somewhat inevitable that an economy experiences slowdown of growth in the long run In reality, however, it is very intricate to distinguish the long-run trend of slowdown from the short-run business downturns There are various statistical methods for decomposing the path of economic growth into the long run trend and the short-run fluctuations With the long series of macro variables available, these statistical methods are relatively easy to implement but it is not rare to see the results hard to interpret under conventional economic reasoning

In contrast, there have been many trials of identifying transmission channels of shocks with perception that the change of economic growth

is accumulated responses of various sectors to external shocks Analysis

5 Oh (2007) assumes that domestic demand shocks have only temporary effect while domestic supply shocks persist in the long-run He also assumes that world supply shocks increase world production and have permanent effect on world demand for imports whereas domestic supply shocks have only temporary effect on the world demand for imports Under such presumptions, the estimation results report that world supply shocks have larger impact in contrast with the shrunken influence of domestic supply shocks after the financial crisis It is also revealed that the impact of domestic demand shocks has been magnified in the short-run

in this category, based an economic model (however simple it may be), allows economic intuition to work but it is not satisfying in that a slight change in the architecture of the model may lead to a different result In this context, additional robustness check would be required.

This study encompasses the first approach from the standpoint of the second one In details, it derives long-run restrictions of an SVAR representation from a simple macro economic model In the meantime a number of shocks are introduced in the model Some of them affect a sector while others have influence on the economy beyond a sector Roughly speaking, those shocks are categorized into two groups- demand and supply shocks, which are, in turn, believed to match with business cycle and long-run trend of growth respectively Behind such a logic lies a general notion that demand shocks are transient whereas supply ones are permanent As admitted by B-Q(1989), however, transient supply shocks or persistent demand ones may exist in reality Thus, it would be absurd to associate demand shocks with volatile business cycle and supply ones with the changing growth trend In this context, our paper keeps itself apart from trials to decompose the economic growth into the long-run trend and the short run fluctuations

1 Sources of Shocks and Transmission Channels

In the following models all the shocks are classified into demand driven and supply driven ones, which are, in turn, grouped into domestic and foreign ones In addition, we consider the possibility that the changes of economic environments (internal or external) Korean economy has experienced since 1990 may have altered transmission channels while providing new sources of shocks

To begin with, noticeable internal changes of the economic environments have been made in restructuring the financial markets and adopting the inflation targeting rule and the floating exchange rate system, most of which seem to contribute to altering transmission channels On the while, the global phenomenon of low interest rates, housing price hike and emergence of China as a world economic power are major external changes surrounding Korean economy

Next, understanding demand and supply shocks within a framework

of AD-AS, we define internal demand shocks to be idiosyncracies in consumption, investment, government budget, and the markets for domestic and foreign currencies, and represent external demand shocks

to be rooted in the terms of trade and the world economic growth On the other hand, we comprehend that supply shocks are caused internally

by changes in factor and total factor productivities and externally by price fluctuations of raw materials (such as oil and iron ore) and technology spill-over

A notable point here is that such a way of sorting shocks (and discerning changes in transmission channels from those in the magnitudes of shocks) is rather conceptual and does not provide a reliable yardstick to apply in the reality For example, alleviation of household credit constraints induced by the restructuring of the financial sector accompanies consumption growth Also, it is not fully convincing

to define TFP growth to be a sole domestic supply shock TFP growth may results from international competition or TFP growth may interact with the increased demand for investment In this regard, our way of introducing shocks has limitations As a remedy, we present the four incomplete models and compare their results instead of searching indefinitely for just one complete model

2 Equations of Estimation and Identifying Restrictions

In this study, we estimate the following four SVARs with long run restrictions The first two of them, based on Blanchard and Quah(1989), extend the original version of 2-variables and 2-shocks to a variant of 3-variables and 3-shocks, which is to reflect the reality that the Korean economy is exposed to foreign risks more heavily than other economies

unemployment rate (U ) and match them with price shock t 6 (εtπ), supply

Watson(2002), include the real GDP growth(∆y t) and inflation (πt) in common

A A Three Variable Extension of Blanchard and Quah(1989)

Blanchard and Quah (1989) understand a VAR system in an equivalent MAR(Moving Average Representation) and levy additional long-run

based on Fisher (1979) and provides its solution in the form of long-run restrictions on demand and supply shocks Instead of repeating the already famous B-Q setup, we explain how it is extended to a model with three variables and three shocks

As for shocks, we decompose the supply shock into the two parts- the

t

e ) and the productivity shock ( s

t

t

t

determination, while indirectly working against aggregate demand and

(1) Aggregate Demand (a combinations of IS and LM urves)

6 Price shocks represent situations like sudden hikes of oil price and foreign exchange rate By

construction, they denote all the shocks from foreign economies

7 B-Q(1989) mention that the model in the paper is an example intended for explaining the use of SVAR with long-run restrictions

8 SVAR with short run restrictions differs from B-Q(10989) in that it constrains only the

relationships between contemporaneous disturbances

Y t =M t −P t +aθ1,t

(2) Aggregate Supply (Assuming a CRS production technology)

t t

(3) Price Equation

t t t

P = −θ1, −θ2,(4) Wage Equation

t t t

θθ

t t t

d t t

M = −1+ , 1, = 1,−1+ , 2, = 2,−1+

Manipulating (1)~(5) properly, we derive the following matrix, in

π

1 1 1

000

11

111

1

101

11)1(

t

d t

s t

t

d t

s t

t t t

e e

e a

a e

e

e a

a U

Y

9 Most of variables are logarized for scaling In additional all the other variables except the

unemployment rate are used in differences for stationarity.

10 Addition of nominal wage growth ( ∆W t) to these three variables, however, would make the SVMAR(Structural Vector Moving Average Representation) under-identified Hence, we

discard nominal wage growth ( ∆W t) from the SVMAR

By simply ignoring the subscripts t and summing the right hand side of the SVMAR, we obtain the long run restriction matrix C

11

01)1(

0010

00

11

111

1

101

11)1(

a

a a

a a

NA NA

NA

00

With these long-run restrictions, all the shocks are exactly identified Their effects on the three key variables, both long-run and short-run, are summarized

t

real GDP whereas it has permanent influence on price level and

t

t

e ), such as oil price

hike, affects the real GDP and price level temporarily but has permanent effect on the unemployment rate

B An Economy under Inflation Targeting Rule and Flexible Exchange Rate

System

Next, we introduce an open economy New Keynesian model, substantial part of which is borrowed from Stock and Watson(2002) In the economy, the government carries out a monetary policy based on

inflation targeting11 In details, it adjusts the short-term interest rate in response to the anticipated inflation and GDP gap following so called a Taylor rule

For the features of openness, we define the trades of commodities, services and currencies across borders Hence, we describe an equilibrium condition for the foreign exchange market and include the

)(

,][1

,,

1

1

f t t t t k

j t t j t

t

d t

d t

d t

d t t t t

p e p q E

k R r

q r y

(2) New Keynesian Phillips curve(Aggregate supply curve)

e t

d t

p i t t

(3) A forward looking Taylor rule

i y h t t

r

1

][

f t

12

f t P t E t P t Q f t P t E t P t Q f

13 IS curve is influenced by the accumulation of demand shocks( d

t

ε ), d t

θ It should be noted, however, that demand shock doesn’t necessarily have permanent effect on y t because y t is determined not only

by the IS curve but also by its interactions with other equilibrium conditions

Equations (1)~(4) consist of orthogonal demand and exchange rate

p t

1

(

j

s j t j

s t t

s t

ρ

εε

ρ

Next, the equilibrium condition of the foreign exchange market is

j t j

−+

−

=

++

−

=

0 1

1 2 2 2 1

1

1 1 1

)(

)(

)(

)(

j

e j t

e t

e t j

f j t j t

e t

e t t

f t t

f t t

e t t

f t t t

R R

e R R R

R

e R R e

εθ

θ

εεε

t

p ) and inflation (π ) of a foreign country are exogenously given f

15 The author's pretest using the quarterly data of Korea(1979~2001) shows that the GDP

gap(≡GDP minus(-) Hodrick-Prescott filtered GDP) follows a stationary AR(1) process ( 0 < ρ < 1 )

t t t p

y = + ρ −1+ ε , we could say that this economy is a part

of the Keynesian world in a sense that the potential GDP ( p

t

y ) moves one-to-one with the aggregate demand or real GDP (y t ) However, change iny t will not have any impact on p

j t

y+ Once the productivity shock is defined to have permanent effect on the potential GDP, s

t

ε in (5) will be a most conceivable candidate for the productivity shock

Now, we represent the four variables of real GDP(y ), inflation ( t πt),

t

d t

+

−

=

++

−

=++

=

+++

=

0 1

1

0 0

11

11

1

)10

(,1

1

][

j

s j t j h

h

h t

i h

p i t t

e t

d t j

s j t j

e t

d t t e

t

d t

i i

e t

d t i

x i t i t t

i t

X X

y y E

X X

X E

ερδρ

χρρ

ρρ

ρρρ

δρε

εερδρχ

εεδρχεερδχ

εεερ

δχπ

Second, plugging the above equations back to (3), we could represent

t

h y

h

t

h y

t h

t

h y

e h t

d h t h t t

h

p j t t y h t t t

X

X X

X c

X E

y y E E

−

=

−

−+

−

=

−

−+

=

+ +

χρβ

ρ

ρρβδρ

χρβ

ρ

ρρβε

εδρχβ

βπ

β

π π π π

1

11

1

11

1

11

][

]

[

1

we derive the following

≡

=+

=

0 1

1

)(,

)(

1

11

11

11

,1

j

e j t j e

t

e t j

f j t j t t

s h

y h

t s

j t t j t

t

R AX e

s A

AX E

s r R

εφθ

θ

ρ

ρδρ

χρρ

ρρβδρ

χρβ

−

=

ρ

ρρβδρ

χρβκ

θθ

κ

11

,

h y

h

d t t t

d t t t t

B

q BX q

r y

t

t

d t

s

t ε ε

their non-stationarity Accordingly, the first order differences of the real

−+

−

∆

=

−++

)(

)1(

)1(

)1()1(

,1

)(

j

e j t j

e j t j

f j t j t

t t t

j

s j t j

s t

t

s t

t t

t

f t t

e t t t

d t

f t t t t t

d t

d t t t t t

R X

A

e e e

X

X X

L X

e X

X B

e e X

B

q q X B y

εε

ερρε

ρε

ρρ

πε

δρχ

εππ

θθ

Properly rearranged, the vector of (∆y t, π , t ∆e t) is represented as a SVMAR(Structural Moving Average Representation) system of

t

d t

j j

A A

NA NA

NA A

A L

A e

y

t

e t

s t j d

t

e t

s t

t t

t

0

0

0,

)

(

0 εεεε

εεπ

properties of the model economy, which is also characterized by

t

ε ) has permanent impact on the real GDP, price level, and the exchange rate Second, the

ε ) has permanent effect on the exchange

17 Instead of taking the first order differences of I(1) variables (such as real GDP(y t) and the

exchange rate of US $ to Korean Won(e t)), we could adopt a Structural Vector Error Correction Model (SVECM) with cointegrating equations, as suggested by King, Plosser, Stock, and

Watson(1991) In contrast, following B-Q(1989), we form a system of SVAR with long-run

identifying restrictions, which consists of I(0) variables and first-order differences of (1)

variables A reason for not following King et al.(1989) is to avoid disentangling permanent

shocks from transient shocks apriori

the monetary authority controls a monetary aggregate instead of adjusting the short-term interest rate based on so called Taylor rule To rephrase, it is assumed to closely watch the monetary aggregate (neither the anticipated inflation rate nor the GDP gap) and use the money

Such a monetary aggregate targeting rule describes best the policy regime of Bank of Korea before 1997

Second, the exchange rate is assumed to be fixed at a certain level by the government In order to balance the capital account or clear the currency exchange market at the prespecified exchange rate, interest rate differential with other countries is not allowed and the monetary aggregate is constantly controlled for maintaining zero interest rate differential In this context, we could say that the central bank loses

in the case of the monetary aggregate targeting, such a fixed or managed exchange rate system existed until the financial crisis in 1997

(1) IS curve

[ ]

f t f f t p f t

f t

f t f

f t

f t

f t

f t t t t k

j t t j t

t

IS t

IS t

IS t

IS t t t t

X y

y X

X S p p

p e p q E

k R r

q r y

1 ,

, 1 1

1

,)(

,1

,,

−

=

∑

ρπ

π

εθθθκ

π

(2) LM curve

t t t

d t

IS t d t

d t t k

j t t j t

t t

b

a b

a

q E

k p m b

πκ

κ

+

=+

≡

++

1

,)

1(

][)

(

(3) New Keynesian Phillips curve(Aggregate supply curve)18

d t

p i t t

i

κδ

χ

0

++

(5) Money Supply

t f t t

t

ε ) and four macro

t

ε , which is in turn defined to be the productivity shock

1(

10

,

j

s j t j

s t t

s t t

s t t t

p t t

L X

L X

X y y X

ερρ

εε

ρ

ρε

ρ

t

s

t ε

following procedure First, inserting the AR(1) representation of the