productivity and employment in a developing country: evidence from republic of korea

8/22/2019 Productivity and Employment in a Developing Country: Evidence from Republic of Korea

1/31

Economics and REsEaRch dEpaRtmEnt

pruvy Elye

develg cury:

Evee fr Reubl f Kre

Sangho Kim, Hyunjoon Lim, and Donghyun Park

June 2008

RD WoRking PaPER SERiES no. 116


2/31


3/31

ERD Wrin Paper N. 116

Productivityand EmPloymEntina dEvEloPing country:

EvidEncEfrom rEPublicof KorEa

Sangho Kim, hyunjoon lim, and donghyun ParK

junE 2008

Sangho Kim is professor at the College of Business, Honam University, Republic of Korea; Hyunjoon Lim is a Ph.D.

student at the Department of Economics, University of Rochester, New York; and Donghyun Park is senior economist

in the Macroeconomics and Finance Research Division, Asian Development Bank, Manila.


4/31

Asian Development Bank6 ADB Avenue, Mandaluyong City

1550 Metro Manila, Philippineswww.adb.org/economics

2008 by Asian Development BankJune 2008

ISSN 1655-5252

The views expressed in this paperare those o the author(s) and do notnecessarily reect the views or policies

o the Asian Development Bank.


5/31

FoREWoRD

The ERD Working Paper Series is a orum or ongoing and recently completedresearch and policy studies undertaken in the Asian Development Bank or onits behal. The Series is a quick-disseminating, inormal publication meant tostimulate discussion and elicit eedback. Papers published under this Series

could subsequently be revised or publication as articles in proessional journalsor chapters in books.


6/31


7/31

CoNtENts

Abstract vii

I. IntroductionI. Introduction 1

II. Basic Empirical rameworkII. Basic Empirical ramework 2

A. Data ConstructionA. Data Construction 2 B. Bivariate Structural VAR Model 4

III. Empirical ResultsIII. Empirical Results 5

A. Results o the Bivariate Structural VAR ModelA. Results o the Bivariate Structural VAR Model 5 B. Trivariate Structural VAR Model and Its Results 10

IV. Concluding Remarks 1IV. Concluding Remarks 14

Appendix 1Appendix 16

Reerences 1Reerences 19


8/31


9/31

AbstRACt

The paper empirically investigates the relationship between productivityand employment in Republic o Korea using structural vector autoregression

(VAR) models. Productivity-enhancing technology shocks signifcantly increasehours worked, which lends support to the real business cycle theory. The resultsshow that technology shocks can explain most elements o a business cycle bothin the short and long run. On the other hand, demand shocks can only explain

price uctuations. The evidence thus suggests that Korean policymakers should

give higher priority to supply-side policies that promote technological progressand innovation.


10/31


11/31

I. INtRoDuCtIoNTraditional Keynesian theory emphasizes the central role o demand-side actors such as

monetary, fscal, and investment shocks in macroeconomic uctuations. In contrast, real businesscycle (RBC) theory puts orth technology shocks as the main drivers o business cycles. A majorprediction o the RBC theory is a high positive correlation between productivity and employment.The underlying idea is that a positive technology shock increases both productivity and demand or

labor, which, in turn, increases employment. Unortunately or RBC theorists, a well-known stylizedact rom United States (US) data that shows no correlation and indeed oten negative correlationbetween productivity and employment has led many economists to question the relevance o theirtheory. A substantial literature has recently emerged to empirically examine more rigorously the

relationship between productivity and employment.

The pioneering paper by Gali (1999) fnds that productivity-enhancing technology shocksreduced hours worked in the US as well as in all other G-7 economies except Japan. The substantial

body o research that confrms and supports Galis milestone fndings include Basu et al. (2006),rancis and Ramey (2005), rancis et al. (2003), Gali (2004), Gali and Rabanal (2004), Shea (1999),and Kiley (1998). A number o studies have challenged the robustness o such evidence, primarily

on methodological grounds. These include Christiano et al. (2003), Uhlig (2004), Dedola and Neri(2004), Peersman and Straub (2004), Chang and Hong (2006), and Chang et al. (2004). In any case,a negative eect o positive-enhancing technology shocks on employment cannot be reconciledwith the RBC theory and is more consistent with the sticky prices o Keynesian models. The basic

idea is that price rigidity prevents demand rom changing in the ace o lower marginal costs dueto productivity gains, leading frms to produce the same output with less labor.

The central objective o the paper is to empirically investigate the eect o technological shockson productivity and employment in Republic o Korea (henceorth Korea). The paper reexaminesthe relationship between productivity-enhancing technology shocks and employment using Koreandata. The paper hopes to make a signifcant contribution by using data rom a developing country

to investigate this relationship and the broader issue o empirical validity o the RBC theory. Thevast majority o the existing empirical literature on the relationship between productivity andemployment looks at data rom the US and other developed countries. This is also true or thebroader literature testing the validity o the RBC theory. The limited number o studies on RBCs

in developing countries includes Mendoza and Smith (2006), Carmichael et al. (1999), and Chyi(1998). Studies on RBC in Korea are similarly limited and includes Yoon (2006), Park (2000), andMasih and Masih (1995).

However, neither set o studies looks at the technologyemployment relationship nor seeksto otherwise test the RBC theory. That is, those studies look at issues other than how technologyshocks aect productivity and employment or, more generally, how such shocks drive the businesscycle in developing countries. Yet there is no theoretical reason why technology shocks should


12/31

have a smaller impact on productivity and employment or on the business cycle in developingcountries than in developed countries. On the contrary, technological shocks are likely to have abigger impact on developing countries, due to their relative technological backwardness and hencegreater scope or technological progress. This is especially true or Korea, where imported oreign

technology played a central role in the countrys rapid industrialization and economic growth. Whilethe role o technology in long-term economic development and growth has long been recognizedand studied, there has been very little research on the role o technological shocks on the business

cycle o developing countries, as noted above. This paper seeks to examine the eect o productivity-enhancing technology shocks on Korean employment as a way o testing the RBC theory, to helpshed light on the sources o business cycles in developing countries, and thus contribute to thelimited empirical literature on the topic.

The rest o the paper is organized as ollows. In Section II, we describe the basic rameworkor our empirical analysis. More specifcally, we describe the construction o our data, along withthe bivariate structural vector autoregression (VAR) model o productivity and hours worked, which

is used to identiy the technology shocks. In Section III, we report and discuss the results o ourestimation o the bivariate model. We also introduce the variable prices, to build a trivariate structuralVAR model, estimate it, and report and discuss the results. In Section IV, we draw conclusions andpolicy implications rom our main empirical fndings.

II. bAsIC EmPIRICAl FRAmEWoRk

In this section, we explain why we choose total actor productivity (TP) as our measure oproductivity as well as how we construct our TP data. We also describe how we plan to identiytechnology shocks using the bivariate structural VAR model.

A. Daa Cnrcin

Many empirical studies on the employmentproductivity relationship use labor productivity asthe measure o productivity, but this partial measure ails to take into account actor substitutionbetween capital and labor. Such substitution is especially important or the Korean economy, whichhas continuously experienced capital deepening and adoption o new production technologies. Labor

productivity generally depends on capital deepening as well as technological progress and structuralefciency changes. urthermore, it is oten argued that Korean economic growth has been drivenmostly by actor accumulation rather than by productivity growth. In light o these acts, we useTP, which incorporates the eects o both structural and technological changes, as well as labor

productivity, as our productivity measures.1

The data or labor productivity, which is defned as the ratio o gross domestic product (GDP)

to total hours worked, is compiled by the Bank o Korea. We constructed our TP data rom various

sources in the Bank o Korea database and used the data to estimate Solow residuals or theperiod 1980Q12002Q4. The capital stock is the real amount o tangible fxed assets, adjusted orthe capital utilization rate. Our measure o employment is hours worked, which we derived as the

product o total number o employed workers and average hours worked per week (hi). All variablesare converted into constant 2000 prices.

1 Chang and Hong (2006) and Chang et al. (2004) used TP to investigate the dynamic relationship between technologyshocks and employment in the US manuacturing sector.

June2008

Productivityand emPloymentina develoPing country: evidencefrom rePublicof Korea

Sangho Kim, hyunJoon lim, and donghyun ParK


13/31

Section ii

baSicemPirical frameworK

The measured Solow residual is an imperect measure o productivity growth in the absenceo perect competition, constant returns to scale technology, and ull employment o labor andcapital. This implies that the measured Solow residual may be aected by demand-side variables.2In the case o Korea, Kim and Lim (2004) fnd that the Solow residual is not a strictly exogenous

variable but instead co-moves with demand shocks. I measured productivities are inuenced by thebusiness cycle, a correlation between productivity and employment may be spurious and due to acorrelation between employment and the business cycle. or this reason, it is desirable to control

or cyclical bias in the productivity measure.

To address this problem, we ollow the method suggested by Basu and Kimball (1997) andBall and Moftt (2001). This frst step in this method is to regress the log dierence o measured

labor productivity and Solow residual on the log dierence o the capital utilization rate, whichis a proxy or business cycles. The next step is to adjust the average o the regression error termso that it equals the original productivity measure when the productivity measure is adjusted orcyclical actors. Appendix igure A.1 shows the growth rates o measured Solow residual and TP

estimates we obtained ater eliminating the cyclical eects rom the residual. TP grew rapidly aterthe mid-1980s but slowed somewhat in the 1990s, and collapsed during the crisis o 19971998.TP recovered rom the crisis shortly thereater but then ell again ater 2000.

Table 1 below presents the lagged correlation coefcients between the variables used inour empirical analysis. Real GDP has coefcient values o 0.54 and 0.48 with hours worked andemployment, respectively. TP is highly correlated with hours worked and the GDP, leading the two

variables by a quarter or showing simultaneous correlation. The correlation coefcients support thepredictions o the RBC theory in the sense that productivity growth is closely related to macroeconomicuctuations. Based on this observation, in the next section, we apply a structural VAR model toKorean data to investigate the dynamic relation between technology shocks and employment.

tablE 1lag corrElation coEfficiEntSbEtwEEn KEy variablES

in KorEa, 1980Q12003Q4

4Q 3Q Q 1Q 0Q +1Q +Q +3Q +4Q

Hours worked, GDP 0.08 0.10 0.10 0.15 0.54 0.14 0.14 0.06 -0.06

Employment, GDP 0.06 0.12 0.26 0.28 0.48 0.20 0.13 0.07 -0.10

TP, GDP 0.18 0.05 0.18 0.45 0.71 0.63 0.42 0.22 0.07

TP, hours worked 0.13 0.03 0.14 0.31 0.43 0.43 0.30 0.14 0.02

TP = total actor productivity; GDP = gross domestic product.Note: Lag correlation coefcients measure correlation between the ormer variable in the present period and the latter variable in the

t iperiod.

2 See Hall (1989) and Mankiw (1989) or more comprehensive discussions.

erd worKingPaPer SerieSno. 116 3


14/31

Prior to our empirical analysis, we carried out augmented Dickey-uller (AD); Phillips-Peron(PP); and Kwiatkowski, Phillips, Schmidt, and Shin (KPSS) unit root tests to examine whether thetime-series o the variables ollow stochastic trends. Appendix Table A.1 reports the test results orboth levels and frst dierences. The tests unambiguously suggest the existence o one unit root

or every variable, indicating that the time-series are integrated o order 1, I(1). We perormedJohansens cointegration test on various sets o variables to check or the existence o a long-runrelationship among the variables. The results, which we report in Appendix Table A.2, indicate

that there is no cointegration vector and thus no long-run time-series relationship among thevariables.

b. bivariae srcra VAR mde

In view o the absence o a cointegrating relationship among the variables, we speciy abivariate structural VAR model o TP and hours worked to identiy technology shocks in the Korean

economy. While Shea (1999) used the number o patents and research and development expendituresas proxies or technology shocks, Gali (1999) used the long-run restriction that only technology

shocks can aect productivity permanently in a structural VAR model. Although Sheas method maybe able to solve some measurement problems, such as those associated with procyclical movements

o productivity, it cannot replace Galis identifcation method due to the low explanatory power othe proxies.3

Letztbe a vector o TP growth (xt) and total hours worked growth (ht),z x ht t t= [ , ] , and et

be a vector o log o technology shocks ( etx) and log o nontechnology shocks ( et

h ), e e et tx

t

h= [ , ] .Then, the k-lag VAR o TP growth and hours growth can be written as

( )L zt t= e (1)

where

(L) is a kth-order matrix polynomial in the lag operator. The VAR can be rewritten in itsmoving average (MA) representation:

z C Lt t= ( )e (2)

where C(L) is an infnite polynominal matrix in the lag operator ( ) ( )L C L= 1 . We can rewriteequation (2) as

zC L C L

C L C Lt

t

x

t

h=

11 12

21 22

( ) ( )

( ) ( )

e

e (3)

3 To minimize the misspecifcation error, Peersman and Straub (2004) used sign restrictions, frst suggested by aust(1998), to identiy structural shocks in VAR.

4 June2008




15/31

Section iii

emPirical reSultS

Each o the matrices is a polynominal in the lag operator. Two disturbances o technologyand nontechnology shocks cause uctuations in TP and hours worked, and are assumed to beorthogonal to each other.4 To identiy the technology shock, et

x, we assume that both technologyand nontechnology shocks can have a permanent eect on hours worked, but only technology shocks

can have a permanent eect on TP. This assumption imposes the long-run restriction that thenontechnology shocks long-run impact on productivity be equal to zero and implies that C12(1)=0,which restricts the unit root in TP to originate solely rom technology shocks.

III. EmPIRICAl REsults

We report and discuss the results o our estimation o the bivariate VAR model outlined in thepreceding section. We also introduce an additional variable, prices, to speciy a trivariate structuralVAR model, estimate it, and report and discuss the results.

A. Re f he bivariae srcra VAR mde

In this subsection, we report the results o our estimation o the bivariate model o productivityand hours worked. We chose the lag length o our to minimize the Akaike Inormation Criterion,Schwarz Criterion, or Bayesian Inormation Criterion. However, changing the lag length does notaect our results.5 We frst defne productivity as labor productivity, as in Gali (1999), and reportthe coefcient estimates in equation (4) below.

lp

h

t

t

=

0 0083

0 00160

0 0025

0 0024

0 0085

0 0014

.

( . )

.

( . )

.

( . )

e

e

t

t

t

nt

(4)

where lpt

and ht

denote labor productivity growth and hours worked growth, respectively, and

et

tand et

ntdenote technology and nontechnology shocks, respectively. The standard errors are in

parentheses, and zero is the long-run restriction used to identiy technology shocks. Our resultsshow that the impact o positive technology shocks on hours worked is negative but statisticallyinsignifcant.

igure 1 below shows the cumulative impulse response o labor productivity and hours workedto technology and nontechnology shocks in the bivariate model. The responses are defned in terms

o the natural logs o the levels rather than growth rates o the endogenous variables. 6 Laborproductivity rose permanently to higher levels ater initial adjustments in response to a one-standard-deviation positive technology shock. Hours worked initially rose, then uctuated and settled intoa new steady-state level ater six quarters, in response to the same shock. The response o hours

worked to technology shock was negative but insignifcant.

4 Some studies, including McGrattan (2004) and Holzl and Reinstaller (2004) interpreted the two shocks in the structural

VAR as technology shocks and demand shocks. However, many supply shocks other than technology shocks, such asshocks arising rom uctuations in production costs or labor supply, have no long-run impact on productivity. Thereore,

it seems more appropriate to classiy shocks as technology shocks and nontechnology shocks rather than as technologyshocks and demand shocks.

5 The Ljung-Box Q statistics cannot reject the nonexistence o cross- and auto-correlations between the estimated error

terms, implying the nonexistence o autocorrelation and heteroskedasticity.6 We computed the standard errors by bootstrapping 1,000 random draws.

erd worKingPaPer SerieSno. 116


16/31

2.4

2.0

1.6

1.2

0.8

0.4

0.0

0.4

0.2

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

2

4

2

0

-2

-4

-6

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

0 64 8 1210 1614 2018

20 64 8 1210 1614 2018 20 64 8 1210 1614 2018

20 64 8 1210 1614 2018

FIGURE 1

CUMULATIVE IMPULSE RESPONSES OF LABOR PRODUCTIVITY AND HOURS WORKED

TO TECHNOLOGY AND NONTECHNOLOGY SHOCKS IN KOREA, 19852002

Response of labor productivity to

technology shock

Response of labor productivity to

nontechnology shock

Response of hours worked to

technology shock


nontechnology shock

Our fnding o negative but insignifcant eect o productivity-enhancing technology shocks onhours worked is qualitatively very similar to Gali (1999). As noted earlier, it is possible to interpretsuch evidence as casting doubt on the validity o the RBC theory.

However, the negative correlation between hours worked and technology shock disappearedwhen we replaced labor productivity with TP. The identifed coefcients o the structural VAR oTP and hours worked are:

6 June2008




17/31

Section iii

emPirical reSultS

S

h

t

t

=

0 0225

0 00630

0 0073

0 0031

0 0077

0 0019

.

( . )

.

( . )

.

( . )

e

e

t

s

t

h

(5)

where Stand htdenote TP growth and hours worked growth, respectively, andet

s

and eth

denote

technology and nontechnology shocks, respectively. The standard errors are in parentheses, andzero is the restriction used to identiy technology shocks.

The statistically signifcant and positive coefcients suggest that productivity and hours worked

permanently increases under avorable technology shocks. The results also imply that hours workedincreases in the long run under nontechnology shocks. Our fnding o technology shocks permanentlyraising hours worked lends empirical support to the RBC argument that technology shocks play amajor role in short-run business uctuations by raising both output and employment.

igure 2 shows the impulse response o the bivariate model ater we replaced labor productivity

with TP as our measure o productivity. Hours worked did not show a positive and signifcantresponse to a positive technology shock until the second quarter, but increased steadily thereater,

peaking in the fth quarter and settling into its new steady-state level ater six quarters. The eecto positive nontechnology shock on TP was statistically insignifcant, even in the short run. Thisresult is consistent with the assumption that TP is statistically orthogonal with nontechnology

shocks such as demand or mark-up shocks, even in the short run. All other results, including thepermanent increase in TP to higher levels, are qualitatively similar to the results we obtainedusing labor productivity instead, and consistent with economic theory.7

7 Technology shocks can have a permanent eect on productivity because the level o the TP is an unstable time serieswith a unit root.



18/31

3.2

2.6

2.4

2.0

1.6

1.2

0.8

0.4

0.02

4

3

2

1

0

-1

-2

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

0 64 8 1210 1614 2018 20 64 8 1210 1614 2018

20 64 8 1210 1614 201820 64 8 1210 1614 2018

1.4

1.0

0.6

0.2

-0.2

-0.6

FIGURE 2

CUMULATIVE IMPULSE RESPONSES OF TOTAL FACTOR PRODUCTIVITY

AND HOURS WORKED TO TECHNOLOGY AND NONTECHNOLOGY SHOCKS IN KOREA, 19852002

Response of TFP to technology shockResponse of TFP to technology shock


technology shock


nontechnology shock

The most striking dierence between the two sets o results based on dierent productivitymeasures has to do with the response o hours worked to technology shocks. The response was

positive and signifcant when we used TP but negative and insignifcant when we used laborproductivity. It is possible to interpret the positive and signifcant response as empirical evidencesupportive o real business cycles in Korea, since according to the RBC theory, technology shocksdrive the business cycle through their positive impact on both output and employment.

June2008




19/31

Section iii

emPirical reSultS

As pointed out in the previous section, the estimated Solow residual may be an imperectmeasure o TP in the presence o cyclical eects. To eliminate the cyclical eects, we adjustedthe Solow residual by using a composite index o business cycles and demand-related instrumental

variables. We replaced the original Solow residual (TP) with the adjusted Solow residual (TFP),

and reestimated the structural VAR. The identifed coefcients are:

T

h

t

t

=

0 0146

0 00350

0 0054

0 0024

0 0076

0 0019

.

( . )

.

( . )

.

( . )

e

e

t

T

t

H

(6)

where Ttand htdenote TFP growth and hours worked growth, respectively, andet

T

and etHdenote

technology and nontechnology shocks, respectively.

Broadly speaking, the results or the adjusted TFP are qualitatively similar with the earlierresults or the unadjusted TP. Most signifcantly, the identifed coefcients still imply a permanent

rise in productivity and hours worked under positive technology shocks, and thus continue to provideevidence o a real business cycle. The results also suggest that hours worked increase permanentlyunder a nontechnology shock.

igure 3 shows the impulse responses o the bivariate model when we used the adjusted TFP

as our measure o productivity. They are generally similar to the responses we obtained earlier whenwe used the unadjusted TP as our productivity measure. Hours worked began to show a positiveand signifcant response to positive technology shocks in the second quarter, and settled into itsnew steady-state level ater ten quarters. We may interpret our results, which show the co-movement

o output and employment, as lending support to the empirical validity o the RBC theory in thecontext o the Korean economy.8

8 To check or the robustness o our structural VAR results, we also used impulse responses rom the standard VAR with

Cholesky actorization to construct the innovations. The estimation results are very similar to those rom our structuralVAR models.



20/31

0.34

0.33

0.32

0.31

0

0.006

0.004

0.002

0

-0.002

-0.004

0.018

0.016

0.012

0.008

0.004

0

0 5 10 15 20

0 5 10 15 2050 10 15 20

0.10

0.01

0

-0.01

FIGURE 3

CUMULATIVE IMPULSE RESPONSES OF ADJUSTED TOTAL FACTOR PRODUCTIVITY

AND HOURS WORKED TO TECHNOLOGY AND NONTECHNOLOGY SHOCKS IN KOREA, 19852002

Response of TFP to technology shockResponse of TFP to technology shock


technology shock


nontechnology shock

0 5 10 15 20

b. trivariae srcra VAR mde and I Re

So ar we have used the bivariate structural VAR model o productivity and hours worked to

investigate the impact o technology shock on employment in Korea. Our estimation results indicatethat positive technology shocks raise both output and employment, and are thus consistent withthe predictions o the RBC theory. The bivariate model lumped together all shocks other than

technology shocks as nontechnology shocks. These include demand shocks such as monetary policy,fscal policy or shits in business confdence, mark-up shocks associated with changes in oil prices,other input prices or terms o trade, and labor supply shocks. Since it is unlikely that any o these

10 June2008




21/31

Section iii

emPirical reSultS

diverse shocks aect productivity in the long run, the long-run restriction we use in our modelremains appropriate.

Nevertheless, decomposing nontechnology shocks may be helpul or a more in-depth analysis.

or example, dividing nontechnology shocks into labor supply shocks and price shocks allows us

to analyze their eects on employment, output, and prices. Price shocks generally reect demandshocks. We now expand our bivariate model into the ollowing trivariate model:

T

h

p

C C C

C C C

C

t

t

t

=11 12 13

21 22 23

3

1 1 1

1 1 1

( ) ( ) ( )

( ) ( ) ( )

11 32 331 1 1( ) ( ) ( )C C

t

T

t

H

t

P

e

e

e

(7)

where Tt, ht, and pt denote adjustedTFP growth, hours worked growth, and GDP deator

growth, respectively, and etT , et

H , and etP denote technology shock, labor supply shock, and

price shock, respectively. C(1) represents the long-run multipliers o the shocks on the endogenousvariables.

Identiying the trivariate structural VAR model requires three additional restrictions, along

with symmetry and normalization conditions o the covariance matrix o error terms, = Var t( ).eWe adopted the long-run restrictions o Blanchard and Quah (1989). irst, we retain our earlierrestriction that only technology shock can aect productivity in the long run. This means that

labor supply shock and price shock have no impact on long-run productivity, i.e., C12(1)=C13(1)=0.Second, demand or price shocks do not aect hours worked in the long run, implying the long-runrestriction oC23(1)=0.

igure 4 below shows the impulse responses in the trivariate structural VAR model. In responseto a positive technology shock, hours worked slightly ell at frst, then started to rise within aquarter, peaking in the sixth quarter, and settling into its new steady-state level ater ten quarters.

In response to a positive labor supply shock, hours worked rose by about 1.5% at frst, then ell

beore reaching its new equilibrium, about 0.8% higher than initially, ater ten quarters. In responseto a positive demand or price shock, which was assumed to have no long-run eect on hours workedand productivity, hours worked rose slightly at frst but returned to its initial level ater 2 years.



22/31

1.4

1.0

0.6

0.4

-0.2

-0.4

1.4

1.0

0.6

0.4

-0.2

-0.4

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

6

4

2

0

-2

-4

4.2

3.6

3.2

2.8

2.4

2.0

1.6

1.2

0.8

0.4

0.0

4

3

2

1

0

-1

-2

7

6

5

4

3

2

1

0

-1

FIGURE 4CUMULATIVE IMPULSE RESPONSES OF ADJUSTED TOTAL FACTOR PRODUCTIVITY (TFP),

HOURS WORKED, AND PRICES TO TECHNOLOGY AND NONTECHNOLOGY SHOCKS IN KOREA, 19852002

technology shock labor shock price shock

Response of TFP to:


Responses of hours worked to:


Response of GDP deflator to:

0 2 4 6 8 10 12 14 16 1820 0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10 12 14 16 1820 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10 12 14 16 18 20

0 2 4 6 8 10 12 14 16 18 200 2 4 6 8 10 12 14 16 18 200 2 4 6 8 10 12 14 16 18 20

GDP = gross domestic product.

1 June2008




23/31

Section iii

emPirical reSultS

In response to a positive demand shock, the GDP deator increased rapidly or 12 quartersand kept increasing modestly thereater. In contrast, the deator ell in response to positivetechnology and labor supply shocks. Those who are skeptical about the empirical validity o realbusiness cycles, including Gali (1999), have argued that technology and labor supply shock cannot

aect the business cycle because nominal rigidity limits their impact on demand. However, ouranalysis o Korean data strongly suggests that technology and labor supply shocks aect prices,which supports the empirical relevance o the RBC theory or Korea.

Table 2 below reports the decomposition results o orecast error variances o the trivariatestructural VAR model o adjusted TP, hours worked and GDP deator. or hours worked, labor supplyshock explained about 40% o the orecast error variance in the early periods, but its impact gradually

ell aterward, until it explained less than 10% o the variance. The proportion o variance in hoursworked explained by technology shock was initially around 30%, but rose to more than 80% twoyears ater the shock. Demand or price shock explained about 25% o the orecast error varianceuntil one year ater the shock, but the proportion rapidly ell thereater, and almost disappeared

ater 5 years. Labor supply shock had a bigger impact than the price shock, even in the short run.or the GDP deator, the price shock explained about 5060% o the orecast error variance until5 years ater the shock. The labor supply and technology shocks jointly explain about 40% o thevariance throughout.

tablE 2dEcomPoSitionof forEcaSt Error variancEof adjuStEd total factor

Productivity (TFP), hourS worKEd, and gdP dEflatorfor KorEa, 19852002

HoRIzoN(QuARtERs)

tECHNology sHoCk lAboR suPPlysHoCk

PRICE sHoCk

Totalhours worked

0 24.2 50.1 25.7

1 33.4 42.8 23.82 42.7 35.6 21.7

4 60.7 22.7 16.78 86.4 7.4 6.2

12 92.6 6.2 1.1

20 92.4 6.6 1.0

TFP

0 31.0 43.5 25.5

1 47.1 39.6 13.32 63.2 26.7 10.1

4 66.9 23.4 9.78 78.5 12.9 8.6

12 83.9 8.8 7.2

20 86.3 8.3 5.4

GDP deator

0 13.1 24.9 62.1

1 14.2 25.8 60.02 15.4 26.7 57.8

4 18.5 27.7 53.88 25.1 25.3 49.6

12 28.0 20.6 51.4

GDP = gross domestic product.



24/31

All in all, our results indicate that labor supply and technology shocks can explain mucho the Korean business cycle, both in the short run and the long run. This is because our resultsindicate that the two supply-side shocks jointly account or a large part o both output and priceuctuations in Korea. On the other hand, our evidence indicates that demand shocks aect prices

but have only a limited impact on output.

IV. CoNCluDINg REmARks

According to the real business cycle theory, the business cycle is driven largely by technologyshocks rather than the traditional Keynesian demand shocks associated with macroeconomic policy

or business confdence. A major empirically testable prediction o the RBC theory is a positiverelationship between productivity and employment. A substantial empirical literature initiated byGali (1999) fnds that productivity-enhancing technology shocks reduced employment in the USand other developed countries. Although a number o studies challenge the robustness o this

literature, the balance o evidence seems more supportive o a negative relationship than a positiverelationship. This has cast serious doubt on the empirical validity o the RBC theory among many

economists.

In this paper, we reexamined the relationship between productivity-enhancing technology shocksand employment using quarterly Korean data. More specifcally, we used a bivariate structural VARmodel o productivity and hours worked with two types o shocks, technology and nontechnology,along with the long-run restriction that nontechnology shocks cannot permanently aect productivity.

Our empirical results show a negative but insignifcant eect o positive technology shocks onhours worked when we used labor productivity as the measure o productivity. However, whenwe used TP instead o labor productivity as our productivity measure, we ound that technology

shocks had a signifcant positive eect on hours worked, which lends support to the presence oa real business cycle. We were able to replicate this fnding when we adjusted our measure o TP,the Solow residual, to control or cyclical eects. Signifcantly, in a study o the US manuacturing

sector, which produced qualitatively similar results as our study, Chang and Hong (2006) also useTP rather than labor productivity and fnd that positive TP shocks increase employment.

We then added another variable, overall price level, to expand our bivariate model to atrivariate model o productivity, hours worked, and GDP deator. We divide nontechnology shocks

into labor supply shocks and demand or price shocks. Our empirical results reconfrm a positiveeect o productivity-enhancing technology shocks on hours worked. They also suggest that thetwo supply-side shocks (technology shocks and labor supply shocks) jointly account or a large part

o the uctuations in both output and prices, both in the short run and long run, and thus help toexplain much o the Korean business cycle. On the other hand, while demand shocks seem to havea big impact on prices, their impact on output is very limited. All in all, in the case o Korea, ourfndings lend support to the RBC notion that supply-side shocks drive the business cycle.

A signifcant contribution o our study to the literature is the use o data rom a developingcountry to look at the relationship between productivity-enhancing technology shocks and employment,and the broader issue o the empirical validity o the RBC theory. The overwhelming majority o

the literature on these issues is based on data rom the US and other developed countries. Thisis perectly understandable in light o the act that developing countries typically accord a muchhigher priority to achieving higher long-run growth rather than smoothening out the business cycle.

14 June2008




25/31

Section iv

concluding remarKS

However, in the wake o the Asian crisis, understanding the sources o macroeconomic uctuationshas become more relevant or developing countries. urthermore, there is no theoretical reason whytechnology shocks should have a smaller impact on the business cycle in developing countries thanin developed countries. On the contrary, technology shocks may be relatively more important or

developing countries due to their technological backwardness. The short-run and long-run impacto the inormation technology revolution on the Indian economy is a well-known case in point.

We hope that the analysis in this paper will contribute meaningully to the very limited literature

on the empirical validity o the RBC theory in developing countries, and inspire researchers to pursuethe same topic with data rom other developing countries in the uture. At a broader level, suchstudies will help developing-country policymakers better understand the orces behind the business

cycles o their respective countries and thus provide useul policy guidance. or example, in thecase o Korea, our results imply that policymakers should give higher priority to institutional andstructural supply-side policies that promote technological progress and innovation. urther, ourfndings, along with those o Chang and Hong (2006), suggest that uture studies that look at the

impact o productivity-enhancing technology shocks on employment should use TP as well as laborproductivity or a more comprehensive and robust empirical analysis o the productivityemploymentrelationship.



26/31

APPENDIx

aPPEndix tablE a.1unitroottEStSof KEy variablES

ADF PHIllIPs-PERRoN kPss

I(0) I(1) I(0) I(1) I(0) I(1)

ln(LPD) -1.72 6.35*** -1.26 -13.63*** 1.29*** 0.19

ln(TP) -2.59* -5.74*** -2.28 -3.15** 1.25*** 0.43*

ln(hour) -1.45 -14.03*** -1.01 -22.64*** 1.24*** 0.20

ln(GDP) -1.53 -9.58*** -1.46 -9.62*** 1.29*** 0.30

ln(DE) -1.59 -7.28*** -1.76 -7.21*** 1.30*** 0.10

***, **, and * denote rejection o the null hypothesis at the 1%, 5%, and 10% signifcance level, respectively.AD = augmented Dickey-uller; KPSS = Kwiatkowski, Phillips, Schmidt, and Shin.Note: Test regressions contain a constant and a linear time trend, and lags o the dependent variable are chosen by Akaike Inormation

Criterion, Schwarz Criterion, and Bayesian Inormation Criterion. The null hypothesis is the existence o unit root or AD andPhillips-Perron tests, and the nonexistence o unit root or KPSS test.

16 June2008




27/31

aPPendix

aPPEndix tablE a.2johanSEnS log liKElihood tEStfor cointEgration

Labor productivity, hours worked

NumbER oF CE(s) EIgENVAluE tRACE st. % CRItICAl mAx-EIgEN st. % CRItICAl

None 0.149 19.72 25.32 14.18 18.96

At most 1 0.061 5.55 12.25 5.546 12.25

Total actor productivity, hours worked


None 0.116 17.19 25.32 11.93 18.96

At most 1 0.063 5.96 12.25 5.96 12.25

Total actor productivity, hours worked, GDP deator


None 0.227 36.70 42.44 23.39 25.54

At most 1 0.091 13.31 25.32 8.71 18.96

At most 2 0.049 4.60 12.25 4.60 12.25

Note: Test regression includes a constant and a linear deterministic trend in the data. The test indicates zero cointegrating equation atthe 5% signifcance level or each set o variables.



28/31

APPENDIX FIGURE A.1GROWTH OF THE SOLOW RESIDUAL AND CYCLICALLY ADJUSTED TFP (PERCENT)

Modified TFP

Solow Residuals

4

3

2

1

0

-1

-2

-3

-4

1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

TP = total actor productivity.Note: To eliminate cyclical eects rom the measured Solow residual, we regressed thelog dierences o the measured Solow residual on the log dierences o the compositeindex (CI) or business cycles. We then adjusted the averages o the regression errorterms to equal the original productivity measures, ater controlling or cyclical eects. Toaddress the endogeneity problem, we used the generalized method o moments with 1-and 2-period-lagged CI as well as M2 and 1-period-lagged M2 growth as instruments.

1 June2008




29/31

referenceS

REFERENCEs

Ball, L., and R. Moftt. 2001. Productivity Growth and the Phillips Curve. NBER Working Paper No. 8421,National Bureau o Economic Research, Massachusetts.

Basu, S., and M. Kimball. 1997. Cyclical Productivity with Unobserved Input Variation. NBER Working Paper

No. 5915, National Bureau o Economic Research, Massachusetts.Basu, S., J. ernald, and M. Kimball. 2006. Are Technology Improvements Contractionary?American Economic

Review96:141848.Blanchard, O., and D. Quah. 1989. The Dynamic Eects o Aggregate Demand and Supply Disturbances.

American Economic Review79(4):65573.Carmichael, B., S. Keita, and L. Samson. 1999. Liquidity Constraints and Business Cycles in Developing

Economies. Review of Economic Dynamics 2(2):370402.

Chang, Y., and J. Hong. 2006. Do Technological Improvements in the Manuacturing Sector Raise or LowerEmployment? American Economic Review96(1):35268.

Chang, Y., A. Hornstein, and P. Sarte. 2006. Understanding How Employment Responds to Productivity Shocksin a Model with Inventories. ederal Reserve Bank o Richmond Working Paper 06-06, Richmond, VA.

Christiano, L., M. Eichenbaum, and R. Vigusson. 2003. What Happens Ater a Technology Shock? NBER WorkingPaper No. 9819, National Bureau o Economic Research, Massachusetts.

Chyi, Y. 1998. Business Cycles Between Developed and Developing Economies. Review of InternationalEconomics 6(1):90104.

Dedola, L., and S. Neri. 2004. What Does A Technology Shock Do? A VAR Analysis with Model-based Sign

Restrictions. CEPR Discussion Paper No. 4537, Centre or Economic Policy Research, London.aust, J. 1998. The Robustness o Identifed VAR Conclusions about Money. International inance Discussion

Paper 610, Board o Governors o the ederal Reserve System, Washington, DC.rancis, N., and V. A. Ramey. 2005. Is the Technology-Driven Real Business Cycle Hypothesis Dead? Shocks

and Aggregate luctuations Revisited.Journal of Monetary Economics 52:137999.

rancis, N., M. Owyang, and A. Theodorou. 2003. The Use o Long-run Restrictions or the Identifcation oTechnology Shocks. Federal Reserve Bank of St. Louis Review85:5366.

Gali, J. 1999. Technology, Employment and the Business Cycle: Do Technology Shocks Explain Aggregateluctuations?American Economic Review89:24971.

. 2004. On the Role o Technology Shocks as a Source o Business Cycles: Some New Evidence.Journal of the European Economic Association 2:37280.Gali, J., and P. Rabanal. 2004. Technology Shocks and Aggregate luctuations: How Well Does the RBC

Model it Postwar U.S. Data? NBER Working Paper No. 10636, National Bureau o Economic Research,Massachusetts.

Hall, R. 1989. Invariance Properties o Solows Productivity Residual. NBER Working Paper No. 3034, NationalBureau o Economic Research, Massachusetts.

Holzl, W., and A. Reinstaller. 2004. The Impact o Technology and Demand Shocks on Structural Dynamics:Evidence rom Austrian Manuacturing. MERIT-Inonomics Research Memorandum Series 2004-015, TheNetherlands.

Kiley, M. 1998. Labor Productivity in U.S. Manuacturing: Does Sectoral Comovement Reect TechnologyShocks? Board o Governors o the US ederal Reserve System, Washington, DC. Unpublished.

Kim, S., and H. Lim. 2004. Does Solow Residual or Korea Reect Pure Technology Shocks? Paper presented

at the ar Eastern Meeting o the Econometric Society, 1 July, Seoul.Mankiw, N. 1989. Real Business Cycles: A New Keynesian Perspective. Journal of Economic Perspectives

3:7990.Masih, A., and R. Masih. 1995. Temporal Causality and the Dynamic Interactions among Macroeconomic Activity

within a Multivariate Cointegrated System: Evidence rom Singapore and Korea. Weltwirtschaftliches

Archiv131(2):26585.



30/31

McGrattan, E. 2004. Comments on Gali and Rabanals Technology Shocks and Aggregate luctuations: How Well Doesthe RBC Model it Postwar US Data? ederal Reserve Bank o Minneapolis Sta Report 338, Minnesota.

Mendoza, E., and K. Smith. 2006. Quantitative Implications o a Debt-Deation Theory o Sudden Stops and

Asset Prices.Journal of International Economics 70(1):82114.Park, H. 2000. Trade and Real Business Cycles in a Small Open Economy: The Case o Korea. Bank of Korea

Economic Papers 3(2):13563.Peersman, G., and R. Straub. 2004. Technology Shocks and Robust Sign Restrictions in a Euro Area Structural

VAR. European Central Bank Working Paper No. 373, rankurt.

Shea, J. 1999. What Do Technology Shocks Do? In NBER Macroeconomics Annual 1998. Cambridge: MITPress.

Uhlig, H. 2004. Do Technology Shocks Lead to a all in Total Hours Worked? Journal of the EuropeanEconomic Association 2(23):36171.

Yoon, J. 2006. International Real Business Cycle Model with Human Capital and Skill Acquisition Activities.Journal of the Korean Economy7(2):25376.

0 June2008




31/31

Printed in the Philippines

abu e per

Sangho Kim, Hyunjoon Lim, and Donghyun Park find that productivity-enhancingtechnology shocks have a significant positve effect on employment in the Republicof Korea. Their evidence also indicates that technology shocks can explainfluctuations in output and prices, both in the short run and long run, but demandshocks can only explain price fluctuations. This suggests an expanded role forsupply-side policies that promote technological progress and innovation.

Asian Development Bank6 ADB Avenue, Mandaluyong City1550 Metro Manila, Philippineswww.adb.org/economicsISSN: 1655-5252Publication Stock No

abu e a devele Bk

ADBs vision is an Asia and Pacific region free of poverty. Its mission is to help itsdeveloping member countries substantially reduce poverty and improve the qualityof life of their people. Despite the regions many successes, it remains home to twothirds of the worlds poor. Nearly 1.7 billion people in the region live on $2 or lessa day. ADB is committed to reducing poverty through inclusive economic growth,environmentally sustainable growth, and regional integration.

Based in Manila, ADB is owned by 67 members, including 48 from the region.Its main instruments for helping its developing member countries are policydialogue, loans, equity investments, guarantees, grants, and technical assistance.In 2007, it approved $10.1 billion of loans, $673 million of grant projects, andtechnical assistance amounting to $243 million.

productivity and employment in a developing country: evidence from republic of korea

Documents