1

2

34

56789

101112

1314

15

16171819202122232425262728293031323334353637383940414243444546

World Development Vol. xx, No. x, pp. xxx–xxx, 2007� 2007 Elsevier Ltd. All rights reserved.

0305-750X/$ - see front matter

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

doi:10.1016/j.worlddev.2006.11.013www.elsevier.com/locate/worlddev

Openness, Growth and Poverty: The Case of Taiwan

PAN-LONG TSAI and CHAO-HSI HUANG *

National Tsing Hua University, Hsinchu, Taiwan

F

Pleaseerty: T

OO

Summary. — Using time series data for Taiwan during 1964–2003, this paper examines how eco-nomic growth, openness, and the role of government contribute to poverty alleviation. Sustainedeconomic growth is found to be the major driving force for poverty reduction in Taiwan, and open-ness to foreign trade helps the poor through a direct distribution effect as well as indirect growtheffect, in both the long term and short term. While inward FDI has no significant impact on themean income of the poor, outward FDI from Taiwan in the past two decades seems to havehad an adverse effect on the poorest 20% of the population.� 2007 Elsevier Ltd. All rights reserved.

Key words — trade, foreign direct investment, growth, poverty, Asia, Taiwan

R

E47484950515253545556575859606162636465666768

* We are most grateful to Professor Oliver T. Coomes

and four anonymous referees for very constructive co-

mments and suggestions. Thanks also go to Mr. Peng-

Yuan Su for his excellent research assistance. Financial

support to the first author from the National Science

Council under 92-2415-H-007-005 is highly appreciated.

Ofcourse, we are exclusively responsible for remaining

errors or oversights. Final revision accepted: November27, 2006.

UN

CO

RR

1. INTRODUCTION

Although the world economy has grown wellduring the 1990s, many people in developingcountries continue to live in extreme poverty.As of 2001 more than half of the populationin the developing world, 2.7 billion, live on lessthan US$2 a day, and a fifth of them, 1.2 bil-lion, live on less than US$1 a day (Chen & Rav-allion, 2004). For decades, a considerableamount of resources and efforts have beenmobilized to fight against poverty. However,poverty reduction remains the central challengeto many developing countries as well as inter-national development institutions such as theWorld Bank and International MonetaryFund.

While there is little dispute that povertyreduction is a final goal of all economic poli-cies, the problem of how to achieve that goalis still hotly debated (Dollar & Kraay, 2002;The Economist, 2000; Lopez, 2004; WorldBank, 2000a, 2000b). The central question is,would the poor benefit from economic growthirrespective of the quality of growth? Manyvoices, notably from NGOs and street protes-tors from rich countries, argue that the wrongkind of growth fails to help the poor, damagesthe environment, and benefits only the middleand upper classes. Others believe in the so-called ‘‘Bhagwati hypothesis,’’ namely, growthis the principal driver of poverty reductionwhich will lead to poverty decline no matter

1

cite this article in press as: Tsai, P. -L., & Huhe Case of Taiwan, World Development (200

CTED

Pwhat the reasons are (Bhagwati, 2005; Bhagw-ati & Srinivasan, 2002; Dollar & Kraay, 2002,2004; The Economist, 2000). However, alongwith the strong empirical theories that opennessto international trade and investment are effi-cient means to achieving rapid growth, moreand more empirical evidence suggests a closerelationship between rapid growth and povertyreduction. These lead to a growing consensusthat integration into the global economic sys-tem might be essential for poverty alleviation,though there is certainly no simple relationshipbetween openness and poverty (Agenor, 2004;Bhagwati, 2005; Sharma, 2003; Srinivasan,2001; Winter, McCulloch, & McKay, 2004).

In this respect Taiwan distinguishes itself as aparticularly interesting case. Like many devel-oping countries, poverty was widespread inTaiwan during the early postwar years. Afterthe government decisively reoriented its devel-opment strategy from import substitution to-ward export promotion at the end of the

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

69707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124

125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156

157

158159160161162163164165166167168169170171172173174175176177

2 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

1950s, the exceptional economic growth has notonly brought with it the well-known record ofincome distribution, but has also resulted in ra-pid poverty reduction. What Taiwan has expe-rienced in the past four decades suggests thatthere is a close link between openness, eco-nomic growth and poverty reduction, and thusconstitutes an ideal case for a country-specificstudy as strongly recommended by Srinivasan(2001) and Ravallion (2001). Admittedly, it isunlikely to provide a definite answer to thequestion posted above simply based on the Tai-wanese experience. Nevertheless, studying thesuccessful Taiwanese lessons could shed usefullight on the ways of attacking poverty in otherdeveloping economies.

The aim of this paper is to examine how open-ness and growth in Taiwan have contributed toits poverty reduction since the 1960 s. Asidefrom being open to foreign trade, we defineopenness to include open to inward and out-ward foreign direct investment (FDI) as a liber-al inward foreign direct investment regime isamong the essential ingredients of Taiwan’soverall export-oriented development strategy.There are two reasons for us to pay specialattention to the role of FDI. First, the relation-ship between trade and poverty has been the fo-cus of virtually all the recent studies asexemplified in Dollar and Kraay (2002, 2004),Bhagwati and Srinivasan (2002), Sharma(2003), and the review article of Winter et al.(2004), whereas research on that of FDI andpoverty is surprisingly scant. To the authors’knowledge, Klein, Aaron, and Hadjimichael(2001), Jalilian and Weiss (2002), Majid(2003), Agenor (2004), and Santarelli and Figini(2004) are the few who directly address thispotentially important aspect in the fight againstpoverty. Second, Klein et al. (2001) only dis-cusses the likely impacts of inward FDI on thepoverty of the host country, which is under-standable given that poverty is more a problemof capital-receiving developing countries. Like-wise, the other three papers treat countries un-der investigation who are importers of foreigncapital. Using net inward FDI in their regres-sion analysis, however, they implicitly assumeinward and outward FDI to have equal butopposite effects on poverty. Since the nature ofinward and outward FDI of a particular coun-try in a particular year is generally quite differ-ent, there is no a priori rationale to make thatad hoc assumption. In Taiwan, the dominanceof inward FDI has given way to outward FDIsince 1988. Consequently, Taiwan provides a

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

TED

PR

OO

F

very good example of the difference between in-ward and outward FDI with respect to their im-pacts on poverty.

To sum up, the major contributions of ourpaper include (1) it empirically studies the linkbetween openness and poverty reduction usingtime series data from a specific country. Notonly could this further our understanding ofthe openness–growth–poverty nexus, but it alsofills the gap in the current empirical literaturewhich depends primarily on the cross-countryregression analysis; 1 (2) it provides informa-tion concerning the other important, yet sofar neglected aspect of openness, namely, towhat extent inward FDI could possibly help re-duce poverty in the host country; (3) this is thefirst paper to investigate the likely impact ofoutward FDI on the poverty of the home coun-try. The last result has crucial policy implica-tions for a middle-income developing countrythat starts to invest abroad.

The rest of the paper is organized as follows.Section 2 gives an overview of the postwar pov-erty problem in Taiwan. Section 3 provides the-oretical underpinnings of how trade and FDIcan contribute to growth and poverty allevia-tion, and sketches the evolution of Taiwan’strade and foreign investment policies in thepostwar period. An empirical model is set upin Section 4 to perform a quantitative assess-ment. The final section summarizes our majorfindings and gives concluding remarks.

C 2. POVERTY IN TAIWAN

In spite of the remarkable economic achieve-ments, Taiwan is no exception in that manydeveloping countries had to face the povertyproblem, especially in the early postwar years.However, the problem was completely ne-glected by economists in their studies of thedevelopment experience in Taiwan. When look-ing over important works on Taiwan’s eco-nomic development like Lin (1973), Ho(1978), Galenson (1979), Fei, Ranis, and Kuo(1979), Wade (1990), and Li (1995), one is sur-prised to notice that the word ‘‘poverty’’ is vir-tually absent from all the indexes. The neglectcould be attributed to the fact that the govern-ment’s top priority before the 1960s was simply‘‘security,’’ ‘‘survival’’ and ‘‘legitimacy’’ of theNationalist regime. Moreover, poverty was tra-ditionally treated as an individual’s rather thana society’s problem for which the governmentor the society is responsible. The economy took

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208

209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 3

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

off in the mid-1960s, yet the success unexpect-edly disguised or even distorted the fact that aconsiderable number of Taiwanese were strug-gling in hardship.

There was no consistent definition of povertyin Taiwan in the early postwar years. As a mat-ter of fact, social assistance to the poor wassporadic, mainly for the so-called ‘‘winter sea-son relief’’ as well as relief for natural disastersand catastrophes. The ‘‘Means Test Act Gov-erning the Relief of Poverty in the TaiwanProvidence’’ of 1963 for the first time set a uni-form standard for poverty and made the meanstest a regular work of local governments. Thedefinition of poverty has been revised timeand again since then. At present, there are threecategories of low-income families:

(i) First-type low-income family: a familywith no real estate, no one in the family ableto work, no sensible revenues, and the fam-ily cannot survive without assistance.(ii) Second-type low-income family: a fam-ily with less than a third of the total familymembers able to work, and with per capitamonthly revenue less than two thirds of theper capita monthly minimum costs of living(MCL).(iii) Third-type low-income family: a familywith per capita monthly revenue less thanthe per capita monthly MCL.

UN

CO

RR

E

1965

10

9

8

7

6

5

4

3

2

1

01967 1969 1971 1973 1975 1977 1979 1981 1983

Y

%

Figure 1. Official headcount index of poverty in Taiw

www.moi.gov

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

ED

PR

OO

F

The MCL is currently defined as 60% of theaverage local per capita expenditure in the mostrecent year published by the central govern-ment.

Figure 1 depicts the official headcount indexof poverty since 1965. Caution should be exer-cised in interpreting this measure. First of all,the definition of poverty has been changed fromtime to time. For instance, in 1978 the conceptof poverty was changed from absolute to rela-tive poverty. As a result, the number of individ-uals considered to be poor jumped threefold,from 33,170 in 1977 to 108,667 in 1978. Strictlyspeaking, therefore, the measure is not compa-rable over time. Second, the measure could beaffected by the way the MCL is defined. Forexample, with the MCL fixed at NT$200(New Taiwan Dollar) during 1963–75, the realpurchasing power declined because the inflationrate reached 120% during that period. Conse-quently, the dramatic decline in the official pov-erty index in 1963–75 was somewhatmisleading; it reflected more the effect of infla-tion than the effect of the reduction in poverty.Finally, the government sometimes manipu-lated the poverty incidence for political pur-poses. A notable example is the ‘‘Well-To-DoProgram (WTDP)’’. 2 Just one year after imple-menting the program in 1972, the number ofindividuals considered to be poor was cut by

CT

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

ear

an (Source: Monthly Bulletin of Interior Statistics,

.tw/stat/).

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

239240

241242

4 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

half (from 391,463 to 196,362). This is becauselocal governments greatly tightened the eligibil-

UN

CO

RR

E

243244245246247248249250251252253254255256257258259260261262

Table 1. Selected social indicators

Year A B C D

1964 7.71 5.33 29.63 25.501965 7.81 – 30.20 23.651966 7.90 5.25 28.31 22.071967 7.87 – 26.03 21.111968 7.84 5.28 27.86 21.311969 8.14 – 27.45 19.431970 8.44 4.58 26.21 17.381971 8.52 – 25.68 15.931972 8.60 4.49 23.78 16.381973 8.72 – 22.16 16.181974 8.84 4.37 21.92 14.051975 8.88 – 19.52 13.691976 8.91 4.18 14.96 12.901977 8.96 4.21 14.21 12.371978 8.89 4.18 13.49 11.271979 8.64 4.34 12.84 10.951980 8.82 4.17 12.26 10.991981 8.80 4.21 11.62 10.031982 8.69 4.29 11.07 8.981983 8.61 4.36 10.58 8.311984 8.49 4.40 10.09 7.521985 8.37 4.50 9.62 7.361986 8.30 4.60 9.18 6.621987 8.11 4.69 8.76 5.611988 7.89 4.85 8.34 6.231989 7.70 4.94 7.98 6.081990 7.45 5.18 7.59 5.891991 7.76 4.97 7.15 5.371992 7.37 5.24 6.84 5.601993 7.13 5.42 6.59 5.311994 7.28 5.38 6.26 5.071995 7.30 5.34 5.99 6.431996 7.23 5.38 5.68 6.661997 7.24 5.41 5.34 6.351998 7.12 5.51 5.08 6.571999 7.13 5.50 4.72 6.072000 7.07 5.55 4.45 5.862001 6.43 6.39 4.21 5.992002 6.67 6.16 3.97 5.352003 6.72 6.07 3.03 4.87

Notes: A: share of income of the lowest quintile (%); B:ratio of the highest to the lowest quintile; C: illiteracyrate (%); D: infant mortality rate (a tenth of %).Sources:

1. Report on the Survey of Family Income and

Expenditure in Taiwan Area, DGBAS, various

years.

2. Taiwan–Fukien Demographic Fact Book, Minis-

try of Interior, 2001.

3. Taiwan Statistical Data Book, CEPD, various

years.

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

RO

OF

ity of the poor to minimize the heavy financialburdens under the WTDP. As a result, many ofthe poor were ‘‘disqualified from being poor’’simply by administrative manipulation.

Because of these defects it is not appropriateto use the official measure of poverty for anyquantitative study. Nevertheless, as shown inTable 1, the share of the bottom quintile in na-tional income exceeded 8% for the first time in1969 and reached its highest value, 8.96%, in1977; the illiteracy rate dropped below 20% in1975, while the infant mortality rate has de-creased to less than 15 per 1000 infants since1974. Consequently, a drastic decrease in theincidence of poverty in Taiwan after the mid-1970s seems indisputable when these socialindicators are taken into account. The factsthat the apparent alleviation of poverty oc-curred only after Taiwan veered toward a moreopen regime and after the export-led economicgrowth had continued for more than a decadehave important policy implications.

P

263264

265

266267268269270271272273274275276277278279280281282283284285286287288289290291292

CTED3. ANALYTICAL CONTEXT, AND TRADE

AND FDI POLICIES IN TAIWAN

(a) Analytical context

Trade could affect poverty either througheconomic growth or through income distribu-tion. As mentioned above, countries open tointernational trade tend to have a higher rateof economic growth. This is because an opentrade regime facilitates efficient transmissionof price signals from the international marketto the national economy, enhances diffusionof production and management knowledge,and improves domestic efficiency as a result ofintensive international competition. Undis-torted price signals allow more efficient re-source allocation in accordance with acountry’s comparative advantage, which inturn leads to more rapid economic growth.Through the ‘‘trickle down’’ effect, trade thuscontributes to poverty reduction. Moreover,higher economic growth means more tax reve-nue, enabling the government to invest in infra-structure such as education, transportation,disease control and a social safety net, whichare of crucial importance for poverty allevia-tion (Dollar & Kraay, 2004; Hoekman, Micha-lopoulos, Schiff, & Tarr, 2001).

Whether trade-led economic growth indeedalleviates poverty depends also on the accom-panying distribution effect, or, on where a

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347 Q1

348

349350351352353354355356357358359360361362363364365366367368

369

370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 5

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

country’s comparative advantage lies. Giventhe fact that the absolute majority of theworld’s poor live in developing countries andthat their comparative advantage lies in the la-bor-intensive sector, adoption of freer tradepolicy in developing countries should favorthe poor as labor-intensive exports increase.Expansion of the export sector would result ineither a higher real wage rate for labor a lathe celebrated Stolper–Samuelson theorem ininternational trade theory, or more jobs forthe unemployed who are usually poor almostby definition. It should be cautioned that, atleast in the short term, rigidities in factor mar-kets, and in the labor market in particular, maydefy the Stolper–Samuelson theorem and leadto more unemployment and poverty (Agenor,2004). 3

Similar to the case of trade, most of the re-cent literature treats openness to inward FDIas part of the integration into the global econ-omy. Being a collection of external capital, pro-duction technology and management skills,inward FDI could affect the host country’s levelof poverty directly through employment crea-tion, technical training and higher wage rates,or indirectly through economic growth as wellas backward and forward linkages. Under suchcircumstances, we would expect inward FDI toaffect the average income of the bottom quintilepositively. Ofcourse, it would be naı̈ve to con-clude that inward FDI contributes automati-cally to poverty alleviation. It is perfectlylikely that low-skilled labor-intensive produc-tion activities for the investing countries arein fact high-skilled labor-intensive for the hostdeveloping countries. As a result, the benefitsof inward FDI bypass the low-skilled low-in-come workers and exacerbate income inequal-ity. In that case the envisaged positiverelationship between FDI-led growth and pov-erty reduction would fail to stand up. On bal-ance, however, researchers and policy makersbend to accept that the growth-enhancing effectof inward FDI is stronger than the income dis-tribution effect. Klein et al. (2001) have pro-posed various reasons to argue that FDI-ledgrowth is very likely to be pro-poor. 4, 5

The impact on poverty of outward FDI issubtler than that of inward FDI. If the conten-tion concerning the beneficial effect of inwardFDI on the host country is correct, an interest-ing question would be whether this beneficialeffect is also extended to the home country ofoutward FDI? Or, are the impacts on the hostand home countries in fact asymmetric?

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

Namely, the poverty problem will worsen withthe outflows of domestic capital, as implicitlyassumed in Jalilian and Weiss (2002), Majid(2003), Agenor (2004), and Santarelli andFigini (2004). If globalization is indeed, asmany of its proponents maintain, a process ofefficiently reallocating world resources, thenoutward FDI might not harm the poor, at leastin the long term. In the short-term transitionperiod, however, investing abroad definitely re-sults in rising unemployment or puts downwardpressure on the domestic wage rate. This wouldbe especially harmful to the poor if outwardFDI is made mainly by the firms intensivelyusing low-skilled labor. The detrimental effectcould persist even in the long term if outwardFDI leads to de-industrialization and economicstagnation. In fact, these have been the voicesheard time and again in Taiwan since the late1980s.

(b) Trade and FDI policies in Taiwan

(i) Trade policiesSince trade policies are the core of the overall

development strategy, their formulation andimplementation naturally are in sync with thestages of economic development in Taiwan.Roughly speaking, economic development inTaiwan could be divided into three phases: pri-mary import substitution (1950–58), transitionand export promotion (1958–80), and acceler-ated liberalization (1980–present) (Li, 1995).

Like many developing countries in the 1950s,Taiwan adopted a primary import substitutionstrategy. A battery of interventions, includingforeign exchange controls, protective tariffs,and import restrictions, were used to save for-eign exchange and to encourage domestic pro-duction of substitutes for imported goodssuch as textiles, cement and fertilizers (Ho,1978; Lin, 1973). Along with measures like def-icit financing and selective credit allocation,these policies had significant impacts on Tai-wan’s early economic development. However,as is usually the case, the small protecteddomestic market was quickly saturated, andprimary import substitution lost momentum.Real GNP growth rates fell every year during1952–56; the growth rate of industrial outputdeclined and that of employment dropped inthe second half of 1958. Hardly any foreign ex-change was saved; in 1958, Taiwan still ran atrade deficit of some US$70 million. Theseproblems led to a wide-ranging reorientationof development strategy.

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451

452453454455456457

458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513

6 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

To promote exports, two types of export pro-motion policies were implemented. The firsttype aimed at removing or neutralizing distor-tions resulting from protectionist policies en-acted during the import substitution phase. Itincluded, inter alia, liberalization of the foreignexchange allocation system, substitution of tar-iff for nontariff protection, and rebates of im-port duties. The other type involved the offerof new incentives for exports, to which theestablishment of export processing zones andtax incentives for exports belong. Despite sev-eral rounds of import liberalization and tariffreduction in the 1970s, it is noteworthy that,as argued by Wade (1990) and others, trade lib-eralization was generally limited to the exportsector, with the domestic market remainingheavily protected. In other words, by 1980 Tai-wan had a dual-track trade regime attemptingto encourage exports while sheltering thedomestic market from foreign competition atthe same time.

Ironically, the rapid accumulation of foreignexchange made possible by the successful dual-track trade regime turned out to become themajor driving force for accelerated trade liber-alization after 1980 (Chen, 1999). A shift intrade policies came in 1984 when trade surplusjumped to US$8.5 billion, a 76% increase fromits 1983 level, which put tremendous pressureon domestic money supply as well as the ex-change rate of the New Taiwan Dollar in addi-tion to worsening trade frictions with theUnited States. Under the guidance of ‘‘liberal-ization, internationalization,’’ both tariff andnontariff barriers were drastically reduced.Many export incentives, such as special exportloan programs and the duties rebate system,were either lifted or greatly scaled down bythe early 1990s. Whether driven by judiciousdecisions or simply responses to impendingproblems, the description above reveals thatthere has been a general, clear trend toward lib-eralization with respect to trade policies in Tai-wan. If trade liberalization and the ensuingincrease in foreign trade do have merits in alle-viating poverty as many contend, then whathappened in Taiwan during the past half a cen-tury warrants a careful examination.

(ii) FDI policiesTaiwan has been welcoming inward FDI

since the early 1950s, first for the purpose ofobtaining badly needed foreign exchange, thenfor implementing export-oriented industrializa-tion, and finally for transferring advanced tech-

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

nology. Apart from providing a favorable legalenvironment and streamlining the bureaucraticsystem, measures such as offering fiscal andfinancial incentives, granting 100% foreignownership, and notably, establishing exportprocessing zones and science-based industrialparks have been taken to facilitate inwardinvestment (Hoesel, 1996). Nevertheless, in-flows of FDI proved to be disappointing duringthe early years. Only after the first export pro-cessing zone was formally opened in 1966 didinward FDI start picking up. In 1967, the ac-tual inflows of FDI exceeded US$20 millionfor the first time, and it has grown steadily overthe past four decades. It is worth noting, how-ever, that inward FDI is generally not very sig-nificant in terms of value. For example, in1991–96 the average annual inward FDI as apercentage of gross fixed capital formation inTaiwan is 2.4%, which is not only far less thanthat of East and Southeast Asia (7.4%) but alsolower than that of the developing countries as awhole (6.5%) (UNCTAD, 2003).

While Taiwan has maintained a favorableattitude toward inward FDI, it was not the casewith respect to outward investment prior to themid-1980s. The government clearly leaned to-ward prohibition due to worry about capitalflight and for the purpose of saving preciousforeign exchange. Only in the mid-1980s whenforeign reserves quickly soared and trade fric-tions with the United States loomed larger didthe government’s attitude change. The deregu-lation of capital outflows in June 1987 was awatershed; it permitted a business or an indi-vidual to send annually up to US$5 millionabroad without the government’s approval.As a result, outward FDI surged. In 1987–88both the flows and stocks of Taiwan’s out-bound FDI surpassed those of inbound FDI,and Taiwan has since then become a net capitalexporter.

Aside from snowballing foreign reserves andliberalization of capital control, a couple ofother factors also contributed to driving Tai-wanese firms abroad. The more important onesare (1) the increase in wage rates and loomingshortages of unskilled labor, thanks to decadesof remarkable economic growth which made itpossible for the younger generations to seekhigher education and to shy away from the un-skilled low-paid jobs, (2) the rapid appreciationof NT$ after 1985, (3) a deteriorating domesticinvestment climate due to a drastic increase inrental and land prices, stricter environmentalregulations, and the promulgation of the Labor

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

514515516517518519520521522523524525526527528529530531532

533534

535

536537538539540541542543544545546547548549550551552553554

556556

557558559560561562

563564565566567568569570571572573574575576577578579580581

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 7

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

Standard Law in 1984, and (4) rising protec-tionism in the US market and the loss of GSPstatus in 1989. Since a substantial proportionof Taiwanese exports were labor-intensive upuntil the early 1980s, labor shortages and in-creases in wage rates were particularly detri-mental. The situation was exacerbated by theabrupt appreciation of the NT$ after 1985.Compared to 1980, the hourly rate of laborearnings rose more than 180% in 1988, andmore than 400% in all years 1992–96 in termsof US$! To most small and medium-sized enter-prises (SMEs) specializing in labor-intensiveproducts, this was indeed devastating. Beingunable to import a sufficient number of foreignworkers legally, and unable to upgrade produc-tion structure in a timely manner, many SMEswere forced to relocate production to SoutheastAsian countries and China. 6

E582583584585586587588589590591592593594595596597598599600601602603604605

607607

608609610611612

UN

CO

RR

4. OPENNESS, GROWTH AND POVERTY:A QUANTITATIVE ASSESSMENT

(a) Empirical models

We have shown that there has been a cleartrend toward liberalization in both trade andFDI policies in Taiwan during the past half acentury, and after the mid-1980s in particular.We will now examine the Taiwanese experiencequantitatively to see whether it bears out thetheoretical prediction that openness is generallyconductive to poverty alleviation. The empiri-cal exercise is along the lines of Dollar and Kra-ay (2002, 2004) as well as Jalilian and Weiss(2002), though on a much smaller scale usingonly the time series data of Taiwan.

We define the poor, following Dollar andKraay (2002, 2004), as the bottom quintile ofthe population. The mean income of the poor(yr) is calculated as the share of income earnedby the bottom quintile times the mean income(y), divided by 0.2. The basic model is

lnðyrÞt � lnðyrÞt�1

¼ aþ bðlnðyÞt � lnðyÞt�1Þ

þ c0ðlnðxÞt � lnðxÞt�1Þ þ et; ð1Þwhere ln(x)t is a vector of control variables inlogarithmic form at year t, and et is an errorterm with the usual properties. In estimatingEqn. (1), the control variables other than meanincome include a measure of openness to tradedefined as the sum of imports and exports di-

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

vided by GDP (TR/Y), the share of governmentconsumption in GDP (GOV/Y), the share ofgovernment spending on social security in gov-ernment consumption (SS/GOV), and the ra-tios of inward FDI and outward FDI to GDP(FDII/Y and FDIO/Y), respectively. Since thegrowth of the mean income is included, the ef-fect of the other control variables that workindirectly through overall growth is alreadycaptured. The coefficient for each of the othercontrol variables therefore captures its directimpact on the income of the poor, or equiva-lently, on the share of income earned by thepoor (Dollar & Kraay, 2002, 2004).

Government consumption, and governmentspending on social security in particular, is gen-erally regarded to be related to the provision ofpublic goods, social safety nets and redistribu-tion of incomes and assets toward the poor.Both of them should therefore have positive im-pacts on the average income of the poor. As forthe effect of trade on the poor, while still amongthe hottest debates, it is widely expected to beassociated with higher growth and lower pov-erty as noted above. While inward and outwardFDI are just long-term international capitalflows in opposite directions, their impacts onpoverty might be very different in the homeand host countries. Given the generally acceptedpositive role of inward FDI in Taiwan’s devel-opment process, we would expect it to be of noharm to the poor, even if it did not help them.In contrast, the abrupt massive outward invest-ment by the labor-intensive Taiwanese firmsafter the mid-1980s is likely to harm the bottomquintile of the population via job destruction orwage depression. Consequently, the coefficientof outward FDI is expected to be negative.

Besides the basic model, we have also aug-mented it to include two dummy variables totake into account special characteristics withrespect to trade and FDI during the period.Specifically, we define

Dummy 8703 ¼1 1987–2003;

0 1964–86;

(

Dummy 6474 ¼1 1964–74;

0 1975–2003:

(

While the trend toward overall liberalization inTaiwan became clear in the early 1980s, 1987 isthe critical year in terms of economic, social,and political liberalization. A popular impres-sion in Taiwan is that the SME-based export-

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

EC

TED

PR

OO

F

613614615616617618619620621622623624625626627628629

630

631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666

667668669670671672673674675676677678679680681682

Table 2. Unit root tests

Variables t-Statistics T ðq̂� 1Þ-statistics

Augmented Dickey–Fuller tests

ln(yr) �3.3469* �1.3357ln(y) �2.2825 �1.3610ln(yr) � ln(y) �0.1344 �0.4317ln(TR/Y) �2.3493 �3.3592ln(GOV/Y) �3.0444* �15.5655*

ln(SS/GOV) �1.4354 �5.6369ln(FDII/Y) �4.4386* �29.7878*

ln(FDIO/Y) �1.0086 �1.8426Dln(yr) �3.7464* �21.6254Dln(y) �3.3173* �18.2235*

Dln(TR/Y) �6.9776* �44.8858*

Dln(FDIO/Y) �7.0993* �94.6496*

Dln(SS/GOV) �11.6704* �60.1196*

Phillips–Perron tests

ln(yr) �2.7994 �1.3820ln(y) �2.3317 �1.1186ln(yr) � ln(y) �0.0528 �0.1005ln(TR/Y) �2.4486 �3.2863ln(GOV/Y) �3.0676* �14.8770*

ln(SS/GOV) �3.3178* �18.8526*

ln(FDII/Y) �4.3752* �26.0055*

ln(FDIO/Y) �0.9438 �2.0095Dln(yr) �3.9561* �23.5287*

Dln(y) �3.4265* �18.4945*

Dln(TR/Y) �7.1519* �57.3103*

Dln(FDIO/Y) �5.1097* �18.7553*

Dln(SS/GOV) �13.7427* �55.7891*

Note: (a) For augmented Dickey–Fuller t tests andPhillips–Perron t tests, the critical values are �3.58,�2.93, and �2.60 for 1%, 5%, and 10% significancelevels, respectively. For augmented Dickey–Fuller andPhillips–Perron T ðq̂� 1Þ tests, the critical values are�18.9, �13.3, and �10.7 for 1%, 5%, and 10% signifi-cance levels, respectively.(b) * Denotes the rejection of the null hypothesis of the

process containing a unit root at 5% significance level.

8 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

led equitable growth gradually ran out of steamas a consequence of economic liberalization inthe mid-1980s and the subsequent deteriorationin income distribution. Therefore, we add aninteraction term between trade and Dum-my8703 to test the validity of the impression.The interaction term of inward FDI and Dum-my6474 intends to capture the impact of labor-intensive inward FDI, as compared to otherforms of inward FDI. According to Schive(1978), the kind of inward FDI which relieson cheap labor was apparent only before1974. Since unskilled labor is usually the onlyasset owned by the poor, this type of inwardFDI should have greater potential to helpthem, implying a positive coefficient for thisinteraction term.

(b) Data and preliminary analysis

The models will be estimated by the regres-sion technique, using a data set with 40 observa-tions from 1964 to 2003. 7 Before proceeding, asa preliminary, we investigate time series proper-ties of the variables which will be employed laterin our regression analysis. We focus on examin-ing whether these variables contain unit rootsand whether they share common stochastictrends.

Our unit root test results are reported inTable 2, which contains both t and T ðq̂� 1Þstatistics for augmented Dickey–Fuller testsand Phillips–Perron tests. The results indicatethat for ln(y), ln(TR/Y), and ln(FDIO/Y), thenull hypothesis of the existence of a unit rootin the process cannot be rejected at the 5% sig-nificance level. For ln(yr), most test statistics(with the exception of the augmented Dickey–Fuller t statistic) point toward the existence ofa unit root in the process. For ln(SS/GOV),while both Phillips–Perron t and T ðq̂� 1Þ sta-tistics indicate the rejection of a unit root, theaugmented Dickey–Fuller statistics indicatethe opposite. For ln(GOV/Y) and ln(FDII/Y),all tests point toward the rejection of the unitroot hypothesis at the 5% significance level.Moreover, our results reveal that while ln(yr),ln(y), ln(TR/Y), and ln(FDIO/Y) may all con-tain unit root, the first differences of these seriesare all stationary.

It is noteworthy that, according to our testresults, the null hypothesis that ln(yr) � ln(y)contains a unit root cannot be rejected at the5% level, implying that there exists no longrun equilibrium in the yr/y series. That is, thereis no fixed per-capita income ratio to which the

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

yr/y series tends to revert in the long run. Thelack of mean-reverting in the yr/y series impliesthat during the sample period some long-termfactors might have driven ln(yr) and ln(y) apart.Two possible candidates of those long-term fac-tors are TR/Y and FDIO/Y. This is becauseamong the variables deemed important inaffecting the mean income of poor families,only TR/Y and FDIO/Y may contain unit rootwhile all others appear to be stationary. To ex-plore whether the long-term divergence be-tween ln(yr) and ln(y) could be attributed tothe long-term movements of TR/Y and FDIO/Y, we conduct co-integration tests to determinewhether there exists any long run relation be-tween ln(yr), ln(y), ln(TR/Y), and ln(FDIO/Y).

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723

724725726727728729730

731

732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762

Q4

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 9

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

R

Table 3 reports our co-integration test results.As the table shows, Johansen’s co-integrationtests generally reject the null hypothesis thatln(yr), ln(y), ln(TR/Y), and ln(FDIO/Y) haveno co-integrating relation (that is, the nullhypothesis of r = 0, where r is the number oflinearly independent co-integrating relations)at 10% level. Actually, both kmax and ktrace sta-tistics indicate that there might exist three line-arly independent co-integrating relations sincethe null hypothesis of r = 2 is rejected at the10% level while that of r = 3 cannot be rejectedat the 10% level. To explore the robustness ofour co-integration test results, we also testwhether any variable of the four is redundantin the co-integration relations. The null hypoth-esis for the redundancy of the ith variable in allco-integration relations can be written as H0:bij = 0 with j = 1, . . . , r, where bij is the ith ele-ment of the jth co-integrating vector. Underthe condition that the number of linearly inde-pendent co-integration relations is three (i.e.,r = 3), we find that for all four variables the nullhypothesis cannot be rejected at the 10% signif-icance level. 8 Therefore, our co-integration testresults generally support the existence of co-integrating relations among the four variables.

To further examine how the four variablesare related in the long run, notice that the co-integrating vector for ln(yr), ln(y), ln(TR/Y),and ln(FDIO/Y) corresponding to the largest ei-gen value (i.e., 0.5885) is [1,�1.0639,�0.1202,0.0181]. That is, although ln(yr), ln(y), ln(TR/Y), and ln(FDIO/Y) are all unit root processeswhen looked at individually, a linear combina-tion of the process—ln(yr) � 1.0639 ln(y) �0.1202 ln(TR/Y) + 0.0181 ln(FDIO/Y)—whichis an ‘‘error correction’’ term, is stationary.This co-integrating relation reveals that in thelong run ln(yr) � 1.0639 ln(y) tends to movepositively with ln(TR/Y) and negatively with

UN

CO

R

Table 3. Co-inte

Eigen value hkmax hktrace H0:r # o

0.5885 31.08* 92.36* 00.5677 29.36* 61.28* 10.5042 24.56* 31.92* 20.1897 7.36 7.36 3

Note: (a) kmax and ktrace are statistics of Johansen’s co-integrln(TR/Y), and ln(FDIO/Y) there are r linearly independent4 � r common stochastic trends. kmax,90% and ktrace,90% arsponding to 10% significance level.(b) * Denotes the rejection of the null hypothesis that betwe

integrating relations (with the alternative that there exist r +significance level.

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

ln(FDIO/Y). Even though the detected long-term relation between ln(yr) � 1.0639 ln(y),ln(TR/Y), and ln(FDI O/Y) could not be di-rectly interpreted as any causal relation be-tween these variables, it does provide clues forthe likely causes of the divergence in ln(yr)and ln(y) detected during the sample period.

(c) Regression results

We now turn to the basic regression model ofEqn. (1). Two complications of model estima-tion arise here. First, since the dependent vari-able and the explanatory variables all includecontemporaneous terms and are all likely tobe ‘‘endogenous’’ variables in a system ofsimultaneous equations, the OLS estimates ofthe coefficients are biased and inconsistent. Toresolve the simultaneous-equation bias prob-lem, we shall employ instrumental-variable esti-mation. Second, since in Eqn. (1) bothdependent and explanatory variables are all intheir first differences, it only captures theshort-term relations between the variables whileit fails to take into account the long-term rela-tion inherent in the co-integration relationmentioned in the previous section. To incorpo-rate the long-term relation of the variables inour regression analysis, we use not only thelagged dependent variable and explanatoryvariables (all in first differences) but also thelagged ‘‘error correction’’ term as instrumentalvariables. In addition, the lagged error correc-tion term is also included as an explanatoryvariable to account for its possible direct effecton the dependent variable.

Starting with the basic model with mean in-come, openness to trade and the share ofgovernment consumption in GDP as theexplanatory variables, we try several differentcombinations of the relevant control variables

gration tests

f unit roots: (4 � r) hkmax,90%, hktrace,90%

4 19.88 58.963 16.13 39.082 12.39 22.951 10.56 10.56

ation tests for the null hypothesis that between ln yr, ln y,co-integrating relations, that is, the four variables sharee critical values for kmax and ktrace, respectively, corre-

en the four variables there are r linearly independent co-1 linearly independent co-integrating relations) at 10%

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

763764765766767768769770771772773774775776777778779780781

782783784785786787788789790791792793794795796797798799800

10 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

to see how these variables affect the mean in-come of the bottom quintile. In all regressionestimates we employ the same set of instrumen-tal variables: one- and two-period laggeddependent variables, one- and two-periodlagged explanatory variables, and a one-periodlagged error correction term.

Our major results are reported as columns(1)–(5) in Table 4. In each case the Durbin–Watson statistic indicates the absence of first-order autocorrelation. There is no serious mul-ticollinearity problem as implied by the low va-lue of the condition number. 9 Some salientfeatures stand out in these results. First, thereis essentially no difference in explanatory poweramong different specifications; all of them ex-plain around 95% of the variation in the depen-dent variable. Second, in all cases thecoefficients of mean income, trade openness,

UN

CO

RR

ETable 4. Openness, growth and income of t

Independentvariable

Dependent va

(1) (2) (3

Intercept �0.012 0.003 0.0(�0.12) (0.03) (0.

Mean income 1.022*** 1.006*** 1.00(12.52) (10.92) (10

Trade/GDP 0.175** 0.174** 0.17(2.47) (2.43) (2.

Governmentconsumption/GDP

0.084* 0.086* 0.0

(1.91) (1.92) (1.Expenditure on socialsecurity/governmentconsumption

�0.007 �0.

(�0.41) (�0Inward FDI/GDP �0.0

(�0Outward FDI/GDP

(Trade/GDP) * Dummy8703

(InwardFDI/GDP) * Dummy6474

Error correction �0.0002 0.006 0.0(�0.004) (0.13) (0.

Number of observation 34 34 3DW statistics 2.061 2.068 2.0Condition number 8.223 11.076 11.R2 0.948 0.947 0.9

Notes: (a) All variables are in changes of log value.(b) t-Statistic in parentheses; *, ** and *** indicate significan

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

OO

F

and the share of government consumption inGDP are rather stable and reach the conven-tional levels of significance. Third, in eachequation the estimated coefficient of mean in-come is not statistically different from 1, cor-roborating the findings of Dollar and Kraay(2002, 2004) that there is a one-to-one relation-ship between the growth of the average incomeof the poor and the growth of the mean income.Fourth, in no case does the government’sexpenditure on social security, or inward oroutward FDI have a statistically significant im-pact on the average income of the poor, thoughthere are mild indications that outward FDItends to have a negative effect.

Contrary to the findings of Dollar and Kraay(2002, 2004), but similar to those of Majid(2003) and Santarelli and Figini (2004), the esti-mated coefficient of trade openness is consis-

C

TED

PR

he poor: Instrumental-variable estimates

riable: ln(mean income of the first quintile)

) (4) (5) (6) (7) (8)

03 0.007 0.007 �0.023 �0.023 �0.00403) (0.07) (0.06) (�0.19) (�0.19) (�0.03)6*** 1.009*** 1.010*** 0.998*** 0.998*** 1.030***

.72) (12.20) (11.94) (10.78) (10.56) (10.82)4** 0.169** 0.170** 0.258* 0.259* 0.167**

31) (2.38) (2.27) (1.74) (1.70) (2.10)85* 0.087** 0.087* 0.105* 0.105* 0.087*

78) (1.97) (1.83) (1.92) (1.80) (1.74)007

.40)003 �0.0003 �0.0005 0.002

.02) (�0.02) (�0.04) (0.16)�0.005 �0.005 �0.006 �0.006 �0.007(�0.98) (�0.96) (�0.95) (�0.93) (�1.06)

�0.136 �0.136

(�0.71) (�0.70)�0.027

(�0.63)06 0.008 0.008 �0.007 � 0.007 0.00312) (0.17) (0.16) (�0.12) (�0.12) (0.06)4 34 34 34 34 3479 2.088 2.099 2.102 2.117 1.995760 8.572 9.151 17.764 19.092 11.41445 0.948 0.946 0.937 0.935 0.940

tly different from zero at 10%, 5% and 1%, respectively.

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

801802 Q2

803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856

857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 11

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

tently positive and statistically significant at the5% level in each equation. Our results thus sug-gest that in the case of Taiwan, openness to for-eign trade has contributed to raising the meanincome of the poor not only in the long termas shown in the co-integration analysis, but alsoin the short term through its positive impactson the share of income accruing to the poorestquintile. Specifically, in the short term every 1%increase in the ratio of trade to GDP leads to a0.17% increase in the mean income of the poorin addition to that brought about by economicgrowth. Two likely but not mutually exclusiveexplanations could be advanced. One is thattrade liberalization, along with other export-promoting measures, helps Taiwan exploit itscomparative advantage fully. The other possi-ble explanation is that the export-led economicgrowth in Taiwan is unique in that it is basedon tens of thousands of SMEs. The availabledata show that SMEs have always accountedfor 60–80% employment and 50–70% of manu-facturing exports in Taiwan (Tsai, 2001). Theirisland-wide presence and adoption of simple,unskilled labor-intensive technologies play acrucial role in creating jobs and nonfarm in-come for the poor in the less developed areas.

Columns (1)–(5) also show that the share ofgovernment consumption in GDP tends to beassociated with higher average income for thepoor, whereas government expenditure on so-cial security has no significant impact on theshare of income of the poorest quintile. The for-mer is consistent with the findings of Santarelliand Figini (2004), but is at variance with thoseof Dollar and Kraay (2002, 2004). That govern-ment expenditure on social security does not doany good for the poor is somewhat depressingsince it is widely regarded to be pro-poor andnecessary. The result could be due to ill-targetedspending of the social expenditure, or due to thefact that the simple share of government spend-ing on social security is not a good indicator ofwhether the policy is particularly pro-poor (Dol-lar & Kraay, 2002). Another likely explanation,as the political scientist Varshney recently sug-gested, is that ‘‘direct’’ methods of tackling pov-erty through subsidies and hand outs are lesseffective than ‘‘indirect’’ methods of faster eco-nomic growth (The Economist, 2005).

As for the effects of inward and outwardFDI, our results are not directly comparableto those obtained in earlier studies since theyall use net inward FDI in their analysis. Thelack of significant impacts of inward FDI onpoverty might be due to (1) its relative unim-

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

portance in Taiwan’s capital formation as al-luded to above; (2) the export-oriented, labor-intensive inward FDI was relatively short-livedas explained in the specification of DUM-MY6474; (3) the majority of the early days oflabor-intensive inward FDI was located in theexport processing zones, limiting their linkageeffects. Ofcourse, to the extent that inwardFDI helped overall economic growth is widelyrecognized (Hoesel, 1996; Schive, 1978), itcould still indirectly benefit the poor. The caseof outward FDI is different; our empirical re-sults suggest that the concern over the detri-mental effect of capital outflow is not totallybaseless. On the one hand, the co-integrationanalysis in the last section reveals that there isa long-term negative relation between outwardFDI and the mean income of the poor. On theother hand, the regression results all point to amild detrimental effect of outward FDI on theshare of income accruing to the poorest quintilein the short term. In other words, acceleratinginvestment abroad by Taiwanese firms in thepast two decades might have indeed affectedthose in the bottom end of the population ad-versely. This is most likely to be due to thereduction in employment and wage depressioneffects of outward FDI by firms using low-skilled labor-intensive technology. As a finalpoint, the idea that inward and outward FDIhave completely different empirical implica-tions substantiates our contention that separat-ing the effects of inward and outward FDI isnecessary and important.

The results of the three augmented models re-ported as columns (6)–(8) in Table 4 are notfundamentally different from those obtainedin the basic model. The growth effect does notchange in any sense, showing that Bhagwati’shypothesis is firmly supported by the Taiwan-ese experience. Openness to foreign trade gen-erally directly benefits the poor, with aslightly larger estimated coefficient exceptEqn. (8). However, in contrast to what is ex-pected, accelerated liberalization after 1987has not had a significant adverse impact onthe average income of the poorest quintile,though the estimated coefficients are indeednegatively signed. Consequently, trade liberal-ization has been generally conducive to povertyalleviation in Taiwan. The positive role of gov-ernment consumption is again upheld. Con-trary to theoretical prediction, inward FDI isagain shown to have no direct contribution tothe fate of the poor even when labor intensityis taken into consideration.

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

E

913914

915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965

9669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015

1016

10173

12 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

UN

CO

RR

5. SUMMARY AND CONCLUDINGREMARKS

Poverty reduction remains one of the biggestchallenges faced by the whole world today. Thesuccess stories of some developing countries inthe postwar period are overshadowed by wide-spread poverty in others. Attacking poverty,therefore, will be an imperative for many gov-ernments and international institutions in theforeseeable future. While still under hot debate,a consensus is emerging among researchers andpolicy makers that sustained economic growthis a sine qua non for successful poverty allevia-tion. This consensus, along with the widely ac-cepted empirical regularity that outward-oriented economies indeed grow more rapidly,implies that integration into the global marketis of critical relevance for combating poverty.Being one of the exemplars of successful devel-opment, and with a relatively good record inpoverty reduction, Taiwan is a particularly per-tinent case to the openness–growth–povertynexus. By examining the postwar experienceof Taiwan, this paper hopes to shed some lighton poverty reduction strategies in other coun-tries. Our main findings are as follows:

(1) Growth is the principal driver of povertyreduction, as Bhagwati proposed in the early1960s. Unequivocal poverty reduction inTaiwan came in the mid-1970s, only afterthe country attained sustained economicgrowth for more than a decade. The storymight have been very different had the eco-nomic growth been interrupted. Ofcourse,the Taiwanese experience does not denythe importance of the quality of growth.That economic growth in Taiwan has beendriven by tens of thousands of SMEs anda sound education policy might in itself rep-resent a distinctive pro-poor growth.(2) The increasingly open trade regime notonly brought about remarkable economicgrowth, but also worked to raise the incomeshare of the poorest quintile in Taiwan. Thisresult holds even after the economy was dras-tically liberalized after the mid-1980s andpeople became more and more concernedabout the adverse income distribution effectof liberalization. In contrast to the findingsof most cross-country studies, trade liberal-ization in Taiwan helped alleviate povertythrough both income and distributioneffects, in the long term and short term. Q

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

(3) We could not detect any effect of inwardFDI on poverty, regardless of its labor-intensiveness. On the contrary, acceleratingoutward FDI seems to harm the poor bothin the long term and short term, most likelythrough job destruction and wage depres-sion. While not directly comparable, ourresults are at variance with what was pre-sented in Majid (2003), Agenor (2004), andSantarelli and Figini (2004); all of themfound net inward FDI to be detrimental tothe poor. Our results also differ from thoseof Jalilian and Weiss (2002), who foundnet inward FDI to have a positive contribu-tion to poverty alleviation in the case ofASEAN, though not in others. Most of all,our empirical results clearly point to thenecessity to treat inward and outward FDIseparately when examining their effects onpoverty.(4) Although relatively weak, we havefound the government’s role to be positivelyassociated with higher average income forthe poor as Majid (2003) and Santarelliand Figini (2004) did. This implies that inthe course of fighting against poverty, gov-ernment intervention in providing infra-structure, necessary goods and safety netsshould not be undervalued.

Poverty is an extremely complex multifacetedphenomenon. There is no easy way to under-stand it, not to mention to attack it. Our resultsare tentative and suggestive, and need to beconfirmed in the future using a more compre-hensive data set. To have a more robust conclu-sion, it is imperative to evaluate poverty overtime, using comparable poverty lines, to lookat other countries’ experiences and to examinehow inward/outward FDI in a particularindustry or particular region might affect thepoverty of a particular group of people. Like-wise, insignificance of a regression coefficient,such as the estimated coefficient of expenditureon social security, might be more accuratelyinterpreted to demand more in-depth investiga-tion instead of denying its importance out ofhand. As Ravallion (2001) correctly warned,we have to look beyond average in many cir-cumstances.

6. UNCITED REFERENCE

Dollar (2004).

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013

1018

10191020102110221023

1024102510261027

102810291030103110321033103410351036

103710381039104010411042

104310441045104610471048

1079

108010811082108310841085108610871088108910901091109210931094

OPENNESS, GROWTH AND POVERTY: THE CASE OF TAIWAN 13

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

NOTES

104910501051105210531054

10551056105710581059

106010611062106310641065106610671068106910701071

107210731074

1075107610771078

1. Warr and Wang (1999) also used time series data toexamine the relationship between poverty and economicgrowth in Taiwan. But their methodology differs fromours and they did not go beyond growth to investigatethe roles of openness and other government policies.

2. With the slogan ‘‘Increase Wealth, Eliminate Pov-erty,’’ the Well-To-Do program was launched by Gov-ernor T.M. Shieh of the Taiwan Providence under theinstruction of Premier Chiang Ching-Kuo.

3. Taiwan is quite fortunate to have an ‘‘economically’’rather flexible labor market in its postwar developmentprocess. The working class was indeed repressed andstrikes were prohibited under the martial law (1947–87),but it was aimed mainly at political activities. To theextent that the legitimacy of the Nationalist regime wasnot challenged, a flexible, buoyant labor market wasconsistent with the goal of overall economic develop-ment and was allowed as a result.

4. The reasons are (1) reducing the volatility of capitalflows and income, (2) improving asset and incomedistribution at the time of privatization, (3) helpingimprove social and environmental standards, and (4)helping improve social safety nets and basic services forthe poor (Klein et al., 2001, pp. 7–15).

5. Agenor (2004) has pointed out channels throughwhich financial integration might negatively affect thepoor. These include (1) a high degree of financialopenness may entail significant short-term costs as isevidenced in the 1996–97 Asian crisis, (2) pro-cyclicalworld capital market access, and (3) concentration of

U

RE

Please cite this article in press as: Tsai, P. -L., & HuCase of Taiwan, World Development (2007), doi:10

CTED

PR

OO

F

credits flows toward large firms producing tradables,and reduced access to loans by small and medium-sizedfirms which tend to be more labor-intensive than largerones. It should be noted that these are more relevant toshort-term portfolio investment than to FDI, though heuses FDI as the indicator in the empirical study.

6. Contrary to inward FDI, the ratio of outward FDIto fixed capital formation in Taiwan during 1991–96reached 4.8%, which is higher than those of East andSoutheast Asia (4.2%) and the whole developing world(2.9%) (UNCTAD, 2003).

7. The sample size is determined exclusively by theavailability of the data. For the share of the incomeearned by the poorest quintile, we have data from 1964onward. However, before 1976 it was available onlyevery other year. To avoid losing too many observationsin a small time series data set, we interpolate the 6 yearswith missing values during 1964–76 with the mean of theshares before and after that particular year. All thevariables are available from various years’ TaiwanStatistical Data Book published by the CEPD andReport on the Survey of Family Income and Expendi-ture in Taiwan Area, DGBAS.

8. Under the null hypothesis, the likelihood ratio teststatistic has an asymptotic v2 distribution with r degreeof freedom.

9. No coefficient of the error correction term isstatistically significant. This is not surprising since it isalso used as an instrumental variable which has capturedmost of its effect.

REFERENCES

109510961097109810991100110111021103110411051106110711081109

NC

ORAgenor, P.-R. (2004). Does globalization hurt the poor?.

Journal of International Economics and EconomicPolicy, 1(1), 21–51.

Bhagwati, J. (2005). In defense of globalization: It has ahuman face. The 2005 Angelo Costa Lecture, Rome.

Bhagwati, J., & Srinivasan, T. N. (2002). Trade andpoverty in the poor countries. American EconomicReview Papers and Proceedings, 92(2), 180–183.

Chen, T. J. (1999). Evolution of trade policy. In J. S. Shu(Ed.), Experience of economic development in Taiwansince 1980 (pp. 391–424). Taipei, Taiwan: Chung-Hua Institution for Economic Research (in Chinese).

Chen, S., & Ravallion, M. (2004). How have the World’spoorest fared since early 1980s? The World BankResearch Observer, 19(2), 141–169.

Dollar, D. (2004). Globalization, poverty, and inequalitysince 1980. World Bank Policy Research WorkingPaper 3333.

Dollar, D., & Kraay, A. (2002). Growth is good for thepoor. Journal of Economic Growth, 7(3), 195–225.

Dollar, D., & Kraay, A. (2004). Trade, growth, andpoverty. Economic Journal, 114(493), F22–F49.

Fei, J. C. H., Ranis, G., & Kuo, S. W. Y. (1979). Growthwith equity: The Taiwan case. New York: OxfordUniversity Press.

Galenson, W. (Ed.) (1979). Economic growth and struc-tural change in Taiwan: The postwar experience of theRepublic of China. Ithaca: Cornell University Press.

Ho, S. (1978). Economic development of Taiwan, 1860–1970. New Haven: Yale University Press.

ang, C.-H., Openness, Growth and Poverty: The.1016/j.worlddev.2006.11.013

1110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146

1147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183

1187

14 WORLD DEVELOPMENT

WD 1943 No. of Pages 14

21 August 2007; Disk Used ARTICLE IN PRESS

Hoekman, B., Michalopoulos, C., Schiff, M., & Tarr, D.(2001) Trade policy reform and poverty alleviation.Development Research Group, International TradeTeam, The World Bank.

Hoesel, V. R. (1996). Taiwan: Foreign direct investmentand the transformation of the economy. In J. H.Dunning, & R. Narula (Eds.), Foreign direct invest-ment and governments: Catalysts for economicrestructuring (pp. 280–315). London: Routledge.

Jalilian, H., & Weiss, J. (2002). Foreign direct invest-ment and poverty in the ASEAN region. ASEANEconomic Bulletin, 19(3), 231–253.

Klein, M., Aaron, C., & Hadjimichael, B. (2001).Foreign direct investment and poverty reduction.Development Research Group, International TradeTeam, The World Bank.

Li, K. T. (1995). The evolution of policy behind Taiwan’sdevelopment success (2nd ed.). Singapore: WorldScientific Publishing Co..

Lin, C. Y. (1973). Industrialization in Taiwan, 1946–1972: Trade and import substitution policies fordeveloping countries. New York: Praeger.

Lopez, J. H. (2004). Pro-growth, pro-poor: Is there atradeoff? World Bank Policy Research WorkingPaper 3378.

Majid, N. (2003) Globalization and poverty. Employ-ment Paper 2003/54, Employment Sector, Interna-tional Labor Office, Geneva.

Ravallion, M. (2001). Growth, inequality and poverty:Looking beyond averages. World Development,29(11), 1803–1815.

Santarelli, E., & Figini, P. (2004). Does globalizationreduce poverty? Some empirical evidence for thedeveloping countries. In E. Lee, & M. Vivarelli(Eds.), Understanding globalization, employment andpoverty reduction (pp. 247–306). New York: PalgraveMacMillan.

UN

CO

RE

118411851186

Please cite this article in press as: Tsai, P. -L., & Huerty: The Case of Taiwan, World Development (200

CTED

PR

OO

F

Schive, C. (1978). The impact of foreign investment onthe economy of Taiwan. Taipei: Research, Develop-ment and Evaluation Commission, Executive Yuan(in Chinese).

Sharma, K. (2003). Trade policy, growth and poverty inAsian developing countries. London and New York:Routledge.

Srinivasan, T. N. (2001). Growth and poverty allevia-tion: Lessons from development experience. ADBInstitute Working Paper 17.

The Economist (2000). Quantity and quality. September30th (p. 90).

The Economist (2005). Poverty and the ballot box. May14th–20th (p. 68).

Tsai, P. L. (2001). Poverty reduction in Taiwan: Thepostwar experience. Paper Presented at the confer-ence on poverty, growth and the role of institutions,ADB, Manila, 10–12 October.

UNCTAD (2003). World investment report: FDI poli-cies for development: National and internationalperspectives. New York: United Nations.

Wade, R. (1990). Governing the market: Economic theoryand the role of government in East Asian industrial-ization. Princeton: Princeton University Press.

Warr, P. G., & Wang, W. T. (1999). Poverty, inequalityand economic growth in Taiwan. In G. Ranis, & S.C. Hu (Eds.), The political economy of development inTaiwan: Essays in honor of John C.H. Fei(pp. 133–165). London: Edward Elgar.

Winter, L. A., McCulloch, N., & McKay, A. (2004).Trade liberalization and poverty: The evidence sofar. Journal of Economic Literature, 42(1), 72–115.

World Bank (2000a). World development report 2000/2001: attacking poverty. New York: Oxford Univer-sity Press.

World Bank (2000b). The quality of growth.www.worldbank.org/html/extdr/quality/.

R

ang, C.-H. (2000b), Openness, Growth and Pov-7), doi:10.1016/j.worlddev.2006.11.013