spatial patterns in china's rural industrial growth and prospects for the alleviation of...

23
Spatial Patterns in China’s Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality 1 Denise Hare Reed College, 3203 SE Woodstock Boulevard, Portland, Oregon 97202-8199 E-mail: [email protected] and Loraine A. West International Programs Center, United States Bureau of the Census Received August 3, 1998; revised March 25, 1999 Hare, Denise, and West, Loraine A.—Spatial Patterns in China’s Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality Exploring regional patterns of growth in 26 of China’s rural manufacturing industries between 1986 and 1994, we find that the growth taking place in the inland region may not favor its relatively low-cost labor resources. In addition, the coast appears to have labor productivity advantages that persist across industries. This discovery contradicts the notion that higher returns to labor on the coast arise from increasing dependence on capital-intensive industries in the coastal region relative to the inland. Our results cast doubt on the potential for policy-led rural enterprise investment in inland provinces to alleviate regional income inequality. J. Comp. Econom., September 1999, 27(3), pp. 475– 497. Reed College, 3203 SE Woodstock Boulevard, Portland, Oregon 97202-8199; and International Programs Center, United States Bureau of the Census. © 1999 Academic Press Journal of Economic Literature Classification Numbers: O14, R11, R58. 1 The authors acknowledge constructive comments from John Bonin, Kimberly Clausing, Marc Rubin, Jeffrey Parker, Penelope Prime, and two anonymous referees. Outstanding research assistance was provided by Po-wah Eva Hung, Beverly Mathis, Andrea Miles, and Yi-Ping Tseng. Xin Meng was instrumental in helping the authors obtain some of the data. Journal of Comparative Economics 27, 475– 497 (1999) Article ID jcec.1999.1597, available online at http://www.idealibrary.com on 475 0147-5967/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

Upload: denise-hare

Post on 08-Oct-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

Spatial Patterns in China’s Rural Industrial Growthand Prospects for the Alleviationof Regional Income Inequality1

Denise Hare

Reed College, 3203 SE Woodstock Boulevard, Portland, Oregon 97202-8199

E-mail: [email protected]

and

Loraine A. West

International Programs Center, United States Bureau of the Census

Received August 3, 1998; revised March 25, 1999

Hare, Denise, and West, Loraine A.—Spatial Patterns in China’s Rural IndustrialGrowth and Prospects for the Alleviation of Regional Income Inequality

Exploring regional patterns of growth in 26 of China’s rural manufacturing industriesbetween 1986 and 1994, we find that the growth taking place in the inland region may notfavor its relatively low-cost labor resources. In addition, the coast appears to have laborproductivity advantages that persist across industries. This discovery contradicts the notionthat higher returns to labor on the coast arise from increasing dependence on capital-intensiveindustries in the coastal region relative to the inland. Our results cast doubt on the potential forpolicy-led rural enterprise investment in inland provinces to alleviate regional incomeinequality.J. Comp. Econom.,September 1999,27(3), pp. 475–497. Reed College, 3203 SEWoodstock Boulevard, Portland, Oregon 97202-8199; and International Programs Center,United States Bureau of the Census.© 1999 Academic Press

Journal of Economic LiteratureClassification Numbers: O14, R11, R58.

1 The authors acknowledge constructive comments from John Bonin, Kimberly Clausing, MarcRubin, Jeffrey Parker, Penelope Prime, and two anonymous referees. Outstanding research assistancewas provided by Po-wah Eva Hung, Beverly Mathis, Andrea Miles, and Yi-Ping Tseng. Xin Mengwas instrumental in helping the authors obtain some of the data.

Journal of Comparative Economics27, 475–497 (1999)Article ID jcec.1999.1597, available online at http://www.idealibrary.com on

475 0147-5967/99 $30.00Copyright © 1999 by Academic PressAll rights of reproduction in any form reserved.

Page 2: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

1. INTRODUCTION

The reforms that have led China’s structural transition from an agrarian-basedto an industrially based economy are at the same time driving per capita growthrates to unparalleled heights, averaging 8.1% annually during the period 1978through 1996 (State Statistical Bureau, 1997c). This growth has reduced sub-stantially the number of people living in poverty. The benefits of growth,however, have been unevenly distributed, with China’s Gini coefficient ofincome inequality rising from 28.8 in 1981 to 38.8 in 1995 (World Bank, 1997).The growing regional inequality that China has experienced is well documented(Khan et al., 1992; Lyons, 1991; Rozelle, 1996; Tsui, 1993; and Yang, 1990).The World Bank (1997) estimates that coastal residents earned one-third morethan residents of the inland provinces in 1985 but that by 1995 they were earningtwice as much. Based on evidence from recent trends, Chen and Fleisher (1996)warn that existing regional income gaps may widen. Focusing on the post-Maoera, they apply a Solow growth model to provincial macroeconomic data for theperiod 1978–1993 and conclude that “convergence is occurringwithin thecoastal and noncoastal groups, but notbetweenthem” (p. 154).

Concerns have been expressed about stagnating incomes in China’s interiorregions. Rising inequality poses the threat of slowing growth and fueling socialconflict. Calls for action have frequently focused on the need to make capitalimprovements in the central and western provinces. Veeck (1993, p. 74) echoesa common refrain when he proposes that “The interior provinces . . .must haveinfusions of capital if they are to buy time to adjust their economies. Withoutsuch support, cycles of poverty will only continue.” In recent years, the centralgovernment has attempted to alleviate regional inequality through policies whichencourage investment in the inland provinces. In particular, investment in town-ship and village enterprises is emphasized as this sector has been a key actor inindustrial growth during the reform period (Yang and Wei, 1995).

This paper uses disaggregated data from China’s rural industrial sector toexplore regional patterns of growth in 26 manufacturing industries between 1986and 1994. Results of this exercise shed new light on the sources of regionaldisparities by showing that the industrial growth taking place in China’s inlandregion may not favor this region’s relatively low-cost labor resources. Further-more, the coast appears to have labor productivity advantages; we find evidencethat these advantages exist across industries. This finding contradicts the notionthat higher returns to labor on the coast result from increasing dependence oncapital-intensive industry in the coastal region relative to the inland. Takentogether, these results cast doubt on the potential for policy-led rural enterpriseinvestment in China’s inland provinces to address fully the regional income gaps.

Section 2 describes briefly the main underlying causes of growing inequalitythat have been identified in the literature, including several that are policy-related, and then turns to a discussion of the role of rural enterprises in China’s

HARE AND WEST476

Page 3: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

industrial growth and their contribution to regional inequality. A discussion ofdynamic comparative advantage theory and its predictions for regional industrialgrowth patterns in China is presented in Section 3. Section 4 contains theeconometric results obtained from testing certain predictions of the theory usingprovince-level data from China’s township and village enterprise sector. Section5 includes a discussion of the results and their implications for policy. Section 6concludes the paper.

2. RURAL ENTERPRISES AND REGIONAL IMBALANCES

During Mao’s era, investment decisions fell under the directive of central-planning authorities whose goal of reducing inter-regional disparities and achiev-ing regional self-sufficiency led to the promotion of industry in China’s inlandprovinces. A major portion of investment was channeled through the budget. Incontrast, the economic reforms begun in the 1980s emphasized efficiency overself-sufficiency as a criterion for investment and new patterns of industrializationbegan to emerge. Preferential policies, e.g., designations of special economiczones and open cities, were extended first to the coastal region in order to attractforeign investment and trade. Local authorities in these areas were also givenmore latitude in terms of implementing new reforms. In 1988, the outward-oriented development strategy was implemented in the coastal areas, identifyingthe entire coastal region as an area in which labor-intensive processing industriesgeared toward the international market would be encouraged and promoted(Yang, 1991). Prime (1991, p. 12) argues that the strategy was a deliberateattempt to “concentrate resources where conditions were most suitable” in orderto glean benefits that would not accrue “if resources were spread thinly.” It washoped that the benefits of reform eventually would trickle down to the central andwestern regions.

The inland provinces lagged behind the coastal provinces in terms of humancapital accumulation and physical infrastructure development even prior to theonset of reforms, and the gap has widened over time. Inland provinces havefound it difficult to compete for investment with the more dynamic coastalprovinces because of the latter’s closer proximity to major domestic and inter-national markets, superior infrastructure, and human capital advantages (Veeck,1993; World Bank, 1997).

Compounding the problem of regional inequalities, the central government’sshare of fiscal resources declined dramatically during the reform period, therebylimiting the scope for redistribution. Fiscal decentralization has increased bothlocal government control over fiscal resources and local government expenditureresponsibility. Those regions with plentiful fiscal resources, especially extrabud-getary funds, have been able to use those funds to support social services,including education and health care, and to invest in industry (Tsui, 1991; Wong,1997). On the other hand, regions with weak fiscal capacity have struggled to

CHINA’S RURAL INDUSTRIAL GROWTH 477

Page 4: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

fund basic social services (West and Wong, 1995). Local governments haveresisted recent recentralization efforts by shifting resources from budgetary toextrabudgetary accounts. Local officials in thriving coastal areas have resistedredistribution strongly, arguing that they should retain control over the profitsgenerated in their locale in order to maintain incentives for future profits (Veeck,1993).

Government policy has further contributed to regional income gaps by negat-ing certain comparative advantages of the inland provinces. Initially, the centralgovernment restricted foreign direct investment in strategic sectors that used keyraw materials, harming the inland provinces that typically had a comparativeadvantage in these materials (Huang, 1996). While the role of the market hasbeen expanding during the reform period, the remaining state price controlstended to work against residents of inland provinces by holding down prices ofthe raw materials they supplied to producers in other regions and from whichthey earned their livelihoods.

Concerns about the economic, social, and political ramifications of disparateregional growth rates have led the government to take action. In recent years,policy efforts designed to address the regional imbalances have focused on ruralenterprise development as a means of equalizing incomes. The “Decision onSpeeding Up the Development of Rural Enterprises in Central and WesternRegions” was issued by the State Council in February of 1993. Components ofthe policy include targeted credit and preferential tax treatment for rural enter-prises in the interior region, as well as the promotion of greater economiccooperation between regions. Due to rising labor costs in the coastal regioncoupled with an abundance of labor in the interior, one of the central govern-ment’s objectives is to transfer labor-intensive production inland (Yang and Wei,1996).

The attention paid by policymakers to the rural enterprise sector is welldirected. Economic reforms have radically changed the relative contributions ofthe state and non-state sectors to China’s industrial output. State sector domi-nance gave way as its share of industrial output value fell from 78% in 1978 tojust 37% in 1994. The nonstate sector grew rapidly, in particular township andvillage enterprises, which accounted for 25% of industrial output value by 1994,up from 9% prior to reforms (State Statistical Bureau, 1997c; Naughton, 1995).Since the late 1980’s, rural industrial output has been increasingly targeted tooverseas markets and the sector has become a major player in China’s foreigntrade activity. In 1996, the township and village enterprise sector accounted for34% of national exports (State Statistical Bureau, 1997a; State Statistical Bureau,1997c).

The township and village enterprise sector plays a special role in China’sindustrialization process. Being either collectively or privately owned, theseenterprises typically face a hard budget constraint and are more likely to seekcost-minimizing production methods and to respond to changing market condi-

HARE AND WEST478

Page 5: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

tions than are enterprises in the state sector. The rural economic reforms of theearly 1980s led to the release of large quantities of surplus labor from theagricultural sector and simultaneously eliminated many of the prereform restric-tions on rural commercial activity. Thus, rural locations often provide access toabundant pools of rural workers having few alternative employment opportuni-ties. These changes have facilitated this sector’s growth in real gross output,which has averaged about 20% annually since 1978.

While the emergence of a vibrant township and village enterprise sector hasraised rural incomes, evidence suggests that increased income inequality hasaccompanied the rapid growth of this sector. Tsui (1991) identifies the nonagri-cultural sector as the main source of regional inequality and argues that thepattern of emerging inequality is consistent with rapid growth of the collectivesector, which benefited primarily local governments in the coastal provinces.Rozelle (1994) shows that income differences among counties in Jiangsu prov-ince widened during the 1980s, and that this increased inequality is largely afunction of growth in the rural industrial sector. Using nationwide data on ruralagricultural and non-agricultural output, also at the county level, Howes andHussain (1994, p. 13) concur that “the largest contributor to the increase in theGini coefficient for rural output has been the sharp increase in the share ofnon-agricultural output.” Examining the changing contributions of varioussources of income to inequality between 1985 and 1990, Tsui (1998) argues that,while agricultural income is also an important source of rural income inequality,nonetheless, the contribution of nonagricultural income has increased over time.

The figures in Table 1 illustrate this sector’s spatial concentration in the coastalregion, relative to the inland region.2 Township and village enterprises in thecoastal region accounted for two-thirds of the sector’s employment and more

2 Included in the coastal region are Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang,Fujian, Shandong, and Guangdong. The inland noncoastal region includes Shanxi, Jilin, Hei-longjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Sichuan, Shaanxi, Inner Mongolia, Guangxi,Guizhou, Yunnan, Gansu, Qinghai, Ningxia, and Xinjiang. There are a variety of different regionalclassifications for China (Yang, 1990, footnote 3). Our definition is consistent with that used by theTownship and Village Enterprise Bureau of the Ministry of Agriculture (Yang, 1996, footnote 12).For our purposes, the most important features of the provincial and municipal units in the coastalregion are their relatively higher levels of prosperity and rural enterprise development relative to theinland (see Table 1). Although the province of Guangxi has some coastal access and is sometimesclassified as a coastal province, we consider it included more appropriately among the inlandprovinces because Guangxi remains poor, with a lower level of rural enterprise development, even atthe end of our period of study. While Beijing, Tianjin, and Shanghai more closely resemble cities thanprovinces, all three have rural suburbs with substantial rural enterprise activity, and therefore areincluded in the analysis. Separate observations on Hainan are not available in 1986 because Hainanwas still a part of Guangdong province in that year. Therefore, the 1994 observations on Hainan areadded to the Guangdong series. Finally, comparable data on township and village enterprises in Tibetare not available; therefore, it is excluded from the analysis.

CHINA’S RURAL INDUSTRIAL GROWTH 479

Page 6: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

than three-quarters of its output in 1994.3 In contrast, the number of enterprisesin each region is about equal, suggesting that coastal enterprises are relativelylarge on both an employment and a gross output basis. Finally, the averageproduct of labor in the coastal region is nearly double that of the inland region.The areas in which these enterprises have flourished are characterized by closeproximity to urban centers, well-developed infrastructure, and readily availableresources (Ho, 1995). Because the pattern of development is spatially uneven,residents of those areas in which enterprises are more successful have benefiteddisproportionately from the sector’s growth. The uneven distribution of thetownship and village enterprise sector is consistent with the regional patternsobserved for total output and rural income (Table 1).

China’s drive to increase township and village enterprise investment in the

3 The focus of this paper is the set of enterprises that are engaged in manufacturing. Due to datalimitations, we restrict our analysis to enterprises with township- or village-level ownership, and weexclude the jointly and individually owned rural enterprises, which are sometimes included in abroader definition of township and village enterprises. With respect to the broadly defined townshipand village enterprise sector, our sample accounts for 5% of the total enterprises, 29% of totalemployment, and 47% of total output in 1994. Considering only the enterprises engaged in industrialproduction, including mining activities in addition to manufacturing and including all ownershipcategories, our sample accounts for 12% of all enterprises, 48% of all employment, and 62% of alloutput (China Township and Village Enterprise Yearbook Editorial Committee, 1995).

TABLE 1

Regional and National Statistics for Rural Manufacturing Enterprises, Rural Net Income,and Gross Domestic Product, 1994

Coastal Inland National

TVE manufacturing sectorNumber of enterprises 403,612 409,577 813,189Share of enterprises (percent) 50 50 100Number of employees 21,846,758 11,440,420 33,287,178Share of employees (percent) 66 34 100Gross output value (million yuan) 1,660,569 479,924 2,140,493Share of gross output value (percent) 78 22 100Average product of labor (yuan) 76,010 41,950 64,304

Per capita rural net income (yuan) 1,679 994 1,221Per capita gross domestic product (yuan) 5,689 2,700 3,923

Sources.China Township and Village Enterprise Yearbook Editorial Committee (1995). StateStatistical Bureau (1995a). State Statistical Bureau (1997c).

Note.Our data on the township and village enterprise manufacturing sector are restricted to theenterprises operated at the township or village levels; for a discussion of the sample’s representa-tiveness of the sector as a whole, see footnote 3 in the paper. Adjustments to a per capita basis aremade as follows: rural income figures are adjusted using the rural population, and GDP figures areadjusted using the total population.

HARE AND WEST480

Page 7: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

interior region evolves perhaps from the reasoning that increased inputs ofphysical capital will raise labor productivity and close the income gap. This lineof thinking, however, is contradicted by the results of some recent research. Dongand Putterman (1997) use panel data on 89 rural enterprises to estimate astochastic production frontier. According to their findings, after controlling forinputs of capital and labor as well as industry category and ownership type,location in one of the coastal provinces represented among those in their sampleraises township and village enterprise productivity by 27 to 60% relative tolocation in a central province. Enterprise productivity in Gansu, the only westernprovince in the sample, is 41 to 44% below that in the central provinces.

Fleisher and Chen (1997) use a conventional growth accounting frameworkand provincial-level gross domestic product data to derive provincial estimates oftotal factor productivity. They find that productivity in the coastal region ishigher than that in the non-coastal region by a factor of two. The growth of totalfactor productivity is also higher in the coastal region. Further analysis of theseproductivity differences leads Fleisher and Chen to conclude that much of thegap can be explained by regional differences in human capital investment,measured as the flow of new university graduates, and in foreign direct invest-ment. Citing the differences in factor productivity, they argue that increasedinvestment in physical capital stock in the non-coastal provinces is not sufficientto close the income gap.

3. STRUCTURAL CHANGE AND REGIONAL SPECIALIZATION

Given China’s pattern of regionally divergent industrial growth, to which thetownship and village enterprise manufacturing sector appears to be a majorcontributor, an important question for policy makers to consider is the potentialeffectiveness of a policy designed to address regional income disparities throughthe promotion of rural industry in the inland areas. The theory of dynamiccomparative advantage, based on Vernon’s notion of the product cycle, can beused to shed light on the manner in which location attributes affect manufactur-ers’ production site choices. We draw on this theory to explore the process ofstructural change that evolves as manufacturers respond to both changing pro-duction technologies and differences in factor prices across regions.

The notion that the preferred production location may depend on the life stageof the product was initially proposed by Vernon (1966). His product cycle theorypostulated that new innovation of products would likely take place in regionswith high income and high unit labor costs. In the early stages of production,manufacturers prefer a location with the advantages of flexible access to inputsand good communication. Moreover, small production cost differences areunlikely to be of major significance in picking the production location at thisstage, since a new product is likely to have fewer substitutes and therefore alower price elasticity of demand. However, as the product matures, mass pro-

CHINA’S RURAL INDUSTRIAL GROWTH 481

Page 8: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

duction becomes possible, and the product may be imitated by other manufac-turers. The preferred attributes of the production location are now likely torevolve around unit production costs. Producers of standardized products that canbe transported efficiently are likely to seek out low-wage production sites thatmay be distant from the final goods market.

Using the product cycle notion, the process of structural change, whichinvolves changing comparative advantage, can be given a theoretical foundation(Meier, 1995). Low-wage locations are expected to specialize in the productionof goods that are labor intensive. Of course, new product innovation, followed bystandardization and the emergence of mass production technology, will lead tocontinual turnover in the optimal mix of products produced in any given location.Therefore, even without any accumulation of physical or human capital, the setof goods that is produced in the low-wage location may experience on-goingchange due simply to repeated cycles of innovation and standardization. How-ever, the accumulation of human or physical capital in the labor-abundantlocation can bring about ascension up the ladder of comparative advantage.Movement up entails the transfer out of resource- and unskilled-labor-intensiveproduction and entry into skilled-labor-, capital-, and knowledge-intensive pro-duction. The latter production phases may be thought of as exploiting “created”rather than “natural” comparative advantage (Meier, 1995, p. 458).

The prospect of structural change offers a rather appealing prediction ofeconomic growth and prosperity for inland regions of China if the availability oflow-wage labor serves to attract labor-intensive production activities from thecoast, consistent with the comparative advantage hypothesis. The followingsection analyzes data disaggregated by province and industry in order to betterunderstand the spatial aspects of growth in the township and village enterprisesector. Specifically, we look for evidence of an underlying dynamic comparativeadvantage process in the industrial expansion observed across provinces. Supportwould be provided by the finding that regions were becoming more specializedin industries that were intensive in the respective region’s abundant factor.4 Inparticular, we would expect industrialization patterns to be responsive to differ-ences in the relative price of labor.

The results of our analysis will provide insight into the problem of regionalincome disparities and associated policy solutions. The finding that growth in themanufacturing sector is responsive to regional differences in labor wages wouldbode well for the future resolution of absolute, and perhaps even relative, poverty

4 Prime (1992) explores a related question using industrial output data from China. Contrasting thepost-reform years with the pre-reform era, she finds that in the case of one province, i.e., Jiangsu,industrial production patterns display a mild trend toward greater specialization during the first 10years following reforms from 1978 to 1988, while no such trend is apparent between 1957 and 1978.Her study, however, does not test whether the factor intensities of the industries in which Jiangsubecame more specialized are also those in which the province is relatively abundant.

HARE AND WEST482

Page 9: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

in the inland provinces, particularly if growth is accentuated by a policy ofindustrial promotion in the inland region.

4. TESTING FOR REGIONAL SPECIALIZATION

The empirical analysis makes use of province-level data on the number oftownship- and village-level enterprises in operation, their total employment, andtheir gross output value. We use 1986 and 1994 measures of all variables toconstruct average annual rates of growth over the eight-year interval.5 Attentionis restricted to the manufacturing sector, consisting of 26 industries. Between1986 and 1994, both the number of enterprises in operation and the sector’s totalemployment grew at a faster rate in the inland regions, 2.15 and 4.92% annually,respectively, relative to the coast, 0.48 and 3.02% annually, respectively. Theaverage annual rate of growth in real gross output value, however, is nearly equalacross both regions, 21.62% in the coast and 21.20% in the inland. As a result,growth in the average productivity of labor in the coastal region, 18.61% perannum, exceeded that observed in the inland region, 16.28% per annum.

Assuming that labor costs were relatively lower in the inland region during theperiod under study, these growth statistics may tell a story that is consistent withdynamic comparative advantage theory. If per rural worker annual income is anyguide to rural enterprise employees’ reservation wages, Figure 1 provides evi-dence to suggest that in 1985, just prior to the start of the interval over whichgrowth is measured, workers in the coastal region would have expected higherwages, relative to workers in the inland provinces.6 Thus, the initial evidencesupports the hypothesis of economic development through the process of struc-tural change by suggesting that the coastal provinces may have moved up theladder of comparative advantage into more capital-intensive manufacturing ac-tivities while labor-intensive activities moved westward in order to take advan-tage of lower labor costs.

Further exploration of recent rural industrial growth patterns is carried out bydisaggregating the observed province-level growth rates by industry. We test thehypothesis that industry-level factor intensity contributes to regional patterns ofgrowth in a manner consistent with that predicted by the theory of dynamiccomparative advantage. Multivariate regression equations are estimated usingeach of the four measured growth rates in provincei and industryj as dependent

5 The growth rates in the number of enterprises, the number of employees, and gross output valueare calculated on a per capita basis using a measure of population that excludes urban residents withnon-agricultural registrations. The growth rates of gross output value and average product of labor arecalculated using constant yuan deflated by the overall retail price index.

6 Because the income gap between the two regions has widened over time (see, for instance,Rozelle, 1996), the difference in labor costs almost certainly became wider rather than narrowerbetween 1986 and 1994.

CHINA’S RURAL INDUSTRIAL GROWTH 483

Page 10: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

HARE AND WEST484

Page 11: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

variables and including, as the independent variables, controls for location andindustry capital intensity. The equations estimated are

y1i , j 5 b0 1 b1xi 1 b2zj 1 b3xizj 1 e1i , j (1)

y2i , j 5 g0 1 g1xi 1 g2zj 1 g3xizj 1 e2i , j (2)

y3i , j 5 h0 1 h1xi 1 h2zj 1 h3xizj 1 e3i , j (3)

y4i , j 5 u0 1 u1xi 1 u2zj 1 u3xizj 1 e4i , j, (4)

where the variables denote

y1 i , j the average annual rate of growth in the number of enterprises in provincei and industryj ,

y2 i , j the average annual rate of growth in employment in provincei andindustry j ,

y3 i , j the average annual rate of growth in output in provincei and industryj ,y4 i , j the average annual rate of growth in labor productivity in provincei and

industry j ,xi a dichotomous variable indicating whether provincei is on the coast, andzj the capital–labor ratio for industryj .

The dynamic comparative advantage hypothesis would be supported by thefinding of a statistically significant, positive difference between the coastal regionand the omitted noncoastal region in the effect of industry capital intensity on thegrowth rates of enterprise numbers, employment, and output. This finding wouldindicate that capital-intensive industries are growing more slowly and labor-intensive industries are growing faster in the central and western regions, relativeto the coast. Furthermore, if the comparative advantage story is correct, we wouldexpect that the coastal region control would add little explanatory power to thelabor productivity growth equation after industry capital intensity controls havebeen included. This result would confirm the notion that the observed faster rateof labor productivity growth in the coastal region arises from the growingimportance of capital-intensive industries in this region relative to the inlandarea. In summary, evidence of a dynamic comparative advantage growth processis supported by the finding of positive and significant signs onb3, g3, andh3 aswell as the finding thatu1 is not significantly different from zero.

Unfortunately, data that measure average capital–labor ratios in the townshipand village enterprise sector for each of the 26 industries over the eight-yearinterval under study do not exist. The only existing data that measure nationwide

FIG. 1. China 1985 income per rural laborer. Sources: State Statistical Bureau 1986 and variousprovincial statistical yearbooks.

CHINA’S RURAL INDUSTRIAL GROWTH 485

Page 12: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

capital and labor input in township and village enterprises are from 1995. Thesedata probably overstate capital intensity within the sector during the years of ourstudy, although relative capital intensity across industries might be reasonablyrepresented by these data if capital intensification proceeded at the same paceacross industries. Also available are 1985 and 1995 capital and labor input datafor state-owned enterprises. For the reasons we discuss below, our measure ofcapital intensity is constructed using the 1985 National Industrial Census datacollected from state-owned enterprises. The net value of physical capital peremployee (Table 2) indicates the relative level of capital intensity among the 26manufacturing industries included in our sample.

We believe that capital–labor ratios from state-owned enterprises at the start ofthe interval under study provide a good proxy for relative capital intensity across

TABLE 2

Capital Intensities of State-Owned Manufacturing Industries, 1985

Capital–labor ratio(yuan per employee)

Garments 1,476Furniture manufacture 2,289Leather processing and manufacturing 2,496Education and sports products 2,754Metal products 2,926Wood processing and products 3,576Electrical machinery and equipment manufacture 3,973Printing 4,056Textiles 4,282Nonmetal mineral production 4,394Plastics 4,396Rubber 4,616Precision instrument manufacture 4,764Paper and cardboard products 5,671Food processing and products 5,920Common machinery manufacture 5,927Pharmaceuticals 6,068Beverage products 6,539Electronics and communications equipment manufacture 6,568Transportation equipment manufacture 6,770Tobacco processing 7,557Chemicals 10,910Nonferrous metal refining and reshaping 13,949Ferrous metal refining and reshaping 14,423Oil processing and refining 21,882Synthetic fibers 25,985

Source.State Statistical Bureau (1987).Note.The measure of capital stock is net of depreciation. The labor series measures the number of

employees at year-end.

HARE AND WEST486

Page 13: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

industries in the township and village enterprise sector during the period ofinterest. Support lies in the fact that the township and village enterprise sectorlags behind the state-owned enterprise sector in its production technology.Jefferson and Rawski (1994) report that the state sector is a source of specializedpersonnel and technology to the township and village enterprises; they even goso for as to describe a product cycle in which the state firms are the innovatorsand the rural enterprises are the followers with a cost advantage. In a survey ofindustrial enterprises operating in various industries and with various forms ofownership, conducted by China’s State Statistical Bureau in 1992 and reportedby Jefferson et al. (1999), respondents of all ownership types agreed overwhelm-ingly that state-owned enterprises have the most technically advanced productlines relative to urban cooperatives and township and village enterprises. Fur-thermore, Jefferson et al. find that while township and village enterprises arerelatively more innovative, as measured by shares of new products in grossoutput, their analysis of new product profit rates across enterprises of differentownership types indicates that innovation in township and village enterprisestends to imitate the “frontier innovations made within the state sector” (p. 104).

We recognize, of course, that vast differences exist in average capital intensitybetween the two sectors. One reason for this is that some portion of state enterprisecapital stock is held in nonproductive assets, dedicated to providing housing, edu-cation, health, and other social services (Jefferson, 1999). The effect of thesenonproductive asset holdings on average capital intensity between sectors is tem-pered somewhat by the fact that state enterprises also retain nonproductive labor andare characterized by overstaffing (Zheng et al., 1998). Furthermore, similar to thestate sector, township and village enterprises hold some of their capital stock innonproductive assets, but the share is smaller than that found in the state sector(Jefferson, 1999; Zheng et al., 1998). A second reason for higher levels of capital-ization in the state enterprise sector during the period under study is their greateraccess to subsidized credit coupled with their softer budget constraints. While our1985 state enterprise data may overstate the average capital intensity of township andvillage enterprises during the 1986 to 1994 interval, we believe that all industries aresubject to overstatement, thereby reducing our concerns about the introduction ofbias into our measures of relative capital intensity across industries.7 The sensitivity

7 A possible counter argument might be made that strategically important industries are givenprivileged access to resources (Zheng et al., 1998). If this is true, certain favored industries might becharacterized by excessively high capital–labor ratios in our state sector data. If it is also the case thatstate sector dominance has been more actively preserved among the favored sectors, we might expecttownship and village enterprise growth rates in these industries to be low, relative to those found inother industries. As a result, our estimates of the effect of industry capital intensity on rural enterprisegrowth may be biased downward. China’s five designated pillar industries are machine building,electronics, petrochemicals, automobile manufacture, and construction. Although they are not pillarindustries, certain mining activities and power generation have been identified as sectors also in needof state development. Construction, mining activities, and power generation are not included in our

CHINA’S RURAL INDUSTRIAL GROWTH 487

Page 14: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

of the results to our choice of capital–labor ratio series, relative to the existingalternatives, is discussed later in the paper.

Descriptive statistics for the variables included in the growth regressions areprovided in Table 3. The regression results in Table 4 contradict the notion thatrecent patterns of regional industrialization in China’s township and villageenterprise sector are consistent with the dynamic comparative advantage growthprocess outlined above. Application of both White and Breusch–Pagan testsindicated the existence of heteroskedasticity in our data, so we reportt-ratios thatare calculated using White heteroskedastic-consistent estimates of the standarderrors (Greene, 1997). We find that, with respect to the number of enterprises in

sample of manufacturing industries. In order to test for potential bias resulting from differenttreatment toward favored sectors, we ran a set of regressions that exclude all of the manufacturingactivities in our data sample that fall under the remaining pillar industry categories. These results,along with the results of some other sensitivity tests, are reported in footnote 9.

TABLE 3

Growth Regression Variables

Mean Std. dev. Minimum Maximum

Average annual growth rates, 1986–1994Number of enterprises 0.058 0.091 20.214 0.430Number of employees 0.081 0.095 20.222 0.581Total output value 0.233 0.124 20.316 0.991Average product of labor 0.152 0.077 20.478 0.792

Coastal province dummy variable 0.370 0.483 0 1Non-coastal province dummy variable 0.630 0.483 0 1Capital–labor ratio (Y 10,000 per

employee)1985 SOE KL ratio 0.689 0.565 0.148 2.5991995 SOE KL ratio 4.221 2.858 1.829 13.5631995 TVE KL ratio 2.624 1.283 1.363 7.915

Sources.China Township and Village Enterprise Yearbook Editorial Committee (1995). StateStatistical Bureau (1987). State Statistical Bureau (1995a). State Statistical Bureau (1995b). StateStatistical Bureau (1997b). Township and Village Enterprise Bureau of the Ministry of Agriculture(1987).

Note.The data sample includes 28 provinces and 26 industries. We have a total of 679 observationson each of the four dependent variable growth rates. If all industries were represented in all provinces,we would have a total of 728 observations. Some cells are missing because, in certain industrycategories, a few provinces reported no production activity among township and village enterprisesin one or both of the years that we use to construct the growth rates. Of the 49 missing cells, 17 hadno production activity in either year and 32 had production activity in either 1986 or 1994, but notin both. The growth rate measures are adjusted to reflect per capita and real value changes asdescribed in footnote 5.

HARE AND WEST488

Page 15: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

operation, employment, and gross output, the more capital-intensive industriesare growing at a slower, rather than at a faster, rate in the coastal region. Thesigns of the estimated coefficients on the capital intensity-coastal interaction termare all negative rather than positive, and statistically significant at the 90%confidence level or better using a one-tailed test. Furthermore, controlling forindustry capital intensity, the growth rate in average labor productivity in thecoastal region is 4 percentage points higher, relative to the inland provinces. Infact, labor productivity growth seems to be largely independent of industrycapital intensity.8 Therefore, movement into more capital-intensive industryseems not to account for the coastal region’s higher labor productivity growth

8 Besides the fact that the coefficient estimates are very small, thet-ratios observed on thecapital–labor controls in the labor productivity growth equation suggest that these variables providerelatively little explanatory power. It is worth noting that, when controlling for industry fixed effectsin a model including province fixed effects, the null hypothesis of joint exclusion of the industrycontrols is not rejected at any reasonable confidence level. Therefore, not only is the industrycapital–labor ratio uncorrelated with the observed pattern of industry–province growth rates in laborproductivity, but nothing about the industries appears to be correlated with the productivity growthrates.

TABLE 4

OLS Growth Regression Results Using 1985 SOE Capital–Labor Ratios(679 Industry-Province Observations)

Dependentvariable—averageannual growth in

number ofenterprises,1986–1994

Dependentvariable—averageannual growth in

number ofemployees,1986–1994

Dependentvariable—averageannual growth in

gross outputvalue, 1986–1994

Dependentvariable—averageannual growth in

the averageproduct of labor,

1986–1994

InterceptCoefficient 0.011 0.041 0.173 0.133T-ratio 1.651 5.336 17.729 20.726

CoastalCoefficient 20.000 20.014 0.027 0.041T-ratio 20.041 21.329 2.086 4.762

KL ratioCoefficient 0.077 0.075 0.085 0.010T-ratio 7.972 6.809 7.441 1.231

KL ratio* coastalCoefficient 20.023 20.024 20.034 20.010T-ratio 21.537 21.603 22.289 20.986

R-squared 0.190 0.182 0.113 0.048Adjusted

R-squared 0.187 0.178 0.109 0.044

CHINA’S RURAL INDUSTRIAL GROWTH 489

Page 16: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

rate; on the contrary, rural enterprise growth in capital-intensive industriesappears to be faster in the inland region than in the coast, particularly whenmeasured in terms of gross output growth.9

To further illustrate our findings, we derive the marginal effects on growthrates predicted from a one standard deviation increase from the mean in thecapital–labor ratio. From these derivations, we can observe the magnitude of theeffect of moves to more capital-intensive industries in the coastal region, relativeto the inland for all growth measures. A one standard deviation increase in capitalintensity is associated with a fairly large percentage point increase in gross outputvalue, 2.88 percentage points in the coastal region and 4.80 percentage points inthe inland region. However, changes in capital intensity have little or no effect ongrowth rates in labor productivity as the marginal effect of a one-standarddeviation increase in capital intensity is20.01 percentage points in the coastalregion and 0.57 percentage points in the inland. Furthermore, increases inemployment growth rates at 2.89 percentage points in the coastal region and 4.24percentage points in the inland are nearly as high as those observed for outputgrowth rates. Something other than movement into more capital-intensive indus-

9 A variety of alternative specifications of the regression equations produced results that areremarkably consistent with those reported in Table 3. For example, a set of province fixed-effectregressions produced results with few discernable differences. Because Guangxi is sometimesconsidered a coastal rather than an inland province, we ran a set of regressions in which theobservations from Guangxi province were omitted from the sample. The magnitudes of the coeffi-cients on the capital intensity-coastal interaction term are marginally smaller,20.020,20.022, and20.033, respectively, for the number of enterprises, employment, and output regressions. Meanwhile,the coefficient estimate on the coastal dummy variable in the labor productivity regression increasesto 0.043. Statistical significance levels remain at the 90% confidence level or better for the capitalintensity–coastal interaction estimates and at the 99% confidence level for the coastal dummyestimate. We test for bias generated by the inclusion of pillar industry manufacturing activities byrunning a set of regressions that omit all observations on these industries. Our suspicions that thebaseline results may underestimate the effect of industry capital intensity on rural industrial growthare confirmed; however, our conclusions remain intact. Specifically, the coefficient estimates on thecapital–labor ratio variable and its coastal interaction term all double approximately in size relativeto those reported in Table 3. Smaller standard errors, relative to the coefficients themselves, indicatethat the coefficients are estimated with greater precision as statistical significance reaches the 99%confidence level for all variables of interest. Most notable, the effect of capital intensity on growthremains substantially lower in the coast than in the inland region. It is of interest here that theindustries omitted in this exercise are primarily drawn from among the more capital-intensiveindustries. Therefore, it is possible that the effect of capital intensity on growth is greater amongindustries with relatively lower capital–labor ratios. Support is found in a set of regressions that omitobservations from the two industries with extremely high capital intensity levels, i.e., oil processingand refining and synthetic fibers. The results of these regressions are very similar to those generatedby the sample that excludes pillar industries. Regardless of the absolute size of the effect of industrycapital intensity on growth, all of our results point to the same final conclusions: more capitalintensive industries are growing at a faster rate, as measured by number of enterprises, employment,and output, in the inland relative to the coast and that the coast maintains an advantage acrossindustries with respect to labor productivity growth.

HARE AND WEST490

Page 17: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

tries is driving the labor productivity gains observed in both regions. Hence, thestructural change argument is a particularly poor explanation for the faster ratesof labor productivity growth observed in the coastal region.

Finally, we ran a set of regressions on an alternative capital–labor ratio seriesconstructed using 1995 National Industrial Census data collected from townshipand village enterprises with sales revenues of 1 million yuan or higher.10 Theresults of this analysis are presented in Table 5 and, again, ourt-ratios arecalculated using White heteroskedastic-consistent estimates of the standard er-rors. These results are highly supportive of the key findings presented above. Forinstance, a one standard deviation increase in capital intensity is predicted togenerate an increase in the gross output growth rate of 3.98 percentage points inthe inland and 1.03 percentage points in the coast. Comparable marginal effectson employment growth rates are 3.46 percentage points in the inland and 1.28percentage points in the coast. Coefficient estimates are statistically significant at

10 We also tested the results of using the 1995 state-owned enterprise capital–labor ratio series, butthis series reflected very little correlation with the growth rate measures, indicating that the technol-ogies used by state enterprises in 1995 were relatively poor proxies for the technologies available totownship and village enterprises between 1986 and 1994.

TABLE 5

OLS Growth Regression Results Using 1995 TVE Capital–Labor Ratios(679 Industry-Province Observations)

Dependentvariable—averageannual growth in

number ofenterprises,1986–1994

Dependentvariable—averageannual growth in

number ofemployees,1986–1994

Dependentvariable—averageannual growth in

gross outputvalue, 1986–1994

Dependentvariable—averageannual growth in

the averageproduct of labor,

1986–1994

InterceptCoefficient 0.008 0.022 0.152 0.130T-ratio 0.807 3.868 8.204 11.382

CoastalCoefficient 0.016 0.013 0.063 0.050T-ratio 0.986 0.839 2.835 3.449

KL ratioCoefficient 0.021 0.027 0.031 0.004T-ratio 6.155 6.439 4.318 0.884

KL ratio* coastalCoefficient 20.012 20.017 20.023 20.006T-ratio 22.093 22.964 22.736 21.162

R-squared 0.071 0.112 0.064 0.048Adjusted

R-squared 0.066 0.108 0.060 0.044

CHINA’S RURAL INDUSTRIAL GROWTH 491

Page 18: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

95% confidence levels, or better. While the predicted marginal effects on growthrates are somewhat lower than those reported above, the large gap between theinland and the coastal region remains, with the inland region experiencing higherrates of growth among the more capital-intensive industries. As before, thestructural change hypothesis does not explain growth in labor productivity ineither region, as the marginal effect of an increase in capital intensity is close tozero in both regions. Finally, controlling for industry capital intensity, we findthat the coast has a statistically significant 5 percentage point advantage in laborproductivity growth relative to the inland. The consistency of these results withthose obtained using the 1985 state enterprise capital–labor ratio series, andreported above, alleviates further our concerns about the potential bias arisingfrom the use of state enterprise data.

5. DISCUSSION OF RESULTS AND POLICY IMPLICATIONS

One of the primary results of this study is the finding that industry capital-intensity explains relatively little of the large difference we observe acrossregions in the growth of the average productivity of labor. Labor productivity isgrowing most rapidly in the coastal region, but the rate of growth appears to befairly constant across industry types. As explained above, this result contradictsthe notion that structural change is driving the difference in productivity growth.Although changes in industry mix do not explain the faster labor productivitygrowth observed in the coastal region, it remains possible that a more rapid rateof capital accumulation in the coastal region has contributed to its higher rate oflabor productivity growth.11 That is to say, due to capital deepening, all manu-facturing industries in the coastal region township and village enterprise sectormay have become relatively more capital intensive, thereby raising labor pro-ductivity.

It may be tempting to advocate the use of increased capital investment in theinland area to help to narrow the regional gaps in labor productivity and income.However, the well-documented existence of a substantial productivity gap be-tween the coastal region and the noncoastal region discussed above calls intoquestion any proposal that seeks to address income disparities simply by increas-ing production inputs. The results of this study complement the existing body ofliterature by shedding light on the spatial patterns of industrial expansion andshowing that much of the new enterprise and employment growth in the centraland western regions has taken place in industries with relatively high capitalintensities. A mismatch between technologies, factor endowments, and the sup-porting infrastructure may contribute to the low rates of productivity and pro-

11 The World Bank (1997) notes that in most coastal provinces about half of all nonstate enterpriseinvestment is directed to township and village enterprises while, in the interior provinces, townshipand village enterprises receive less than a quarter of the total.

HARE AND WEST492

Page 19: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

ductivity growth observed in the inland areas. This raises the question of why thespatial aspects of the growth process that we observe are so different from thosepredicted by dynamic comparative advantage theory.

One possible answer is that the process in fact does respond to factor endow-ments, but that factor endowments are endogenous. While physical capital stockaccumulation is believed to favor the coastal region, there has also been a netflow of labor eastward. For instance, in a rural household survey conducted in1993, which included more than 12,000 households nationwide, it was found that70% of rural migrants originating in a coastal province stayed within the coastalregion (Li and Han, 1994). Meanwhile, 72% of the migrants from the centralregion, and 52% of the migrants from the western region travelled to the coastalregion. The fast-growing township and village enterprise sector is a primarysource of attraction to the coastal region for the rural migrants. The steady supplyof low-wage labor from the interior region may reduce the incentive for labor-intensive manufacturing processes to move inland.

Furthermore, the analysis presented here considers only two factors of pro-duction, i.e., capital and labor. The decision to locate a relatively capital-intensive manufacturing process in an inland province may reflect other consid-erations, such as the presence of natural resources or other raw materials. InChina’s mixed system of plan and market, the reliability of input sources is likelyto be an important factor in the location decision. Local processing will alsoincrease value added and thereby enhance the income that can be earned at thesource of production or extraction. This kind of industrial development, however,may not reflect global efficiency. Rather, the establishment of the processingfacilities may create incentives to manufacture other production inputs, leadingthe locality even further away from its natural comparative advantage (Prime,1991).

Local officials, as both facilitators and beneficiaries of rural industry, mayinterfere with market incentives in guiding enterprise development. Examiningthe contribution of rural enterprises to net per capita rural income, Jin and Qian(1998) find that, controlling for the share of non-farm employment in the rurallabor force and the level of per capita community government revenue, the shareof township- and village-level enterprise ownership in the rural enterprise sectoris negatively related to income, suggesting that the collective form of ownershipis less efficient than is private ownership. Jin and Qian use the province as theirunit of analysis. To understand why some enterprises may use resources ineffi-ciently, it is useful to consider the results of studies that focus on the enterpriseas the unit of analysis. Both Dong (1998) and Putterman (1997) find support forthe notion that employment levels in township and village enterprises are setbelow profit-maximizing levels so that the marginal productivity of labor exceedsthe shadow wage rate. This evidence is consistent with an insider–outsider modelsuch as that proposed by Chang and Wang (1994), in which the goal of localincome generation takes priority over employment generation after full employ-

CHINA’S RURAL INDUSTRIAL GROWTH 493

Page 20: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

ment is achieved within the community. Dong and Putterman (1996) propose analternative explanation, which is that enterprises are monopsonists in their locallabor market. Either scenario implies the existence of non-competitive marketsand the accrual of economic rents.

Pei (1996) argues that it is precisely the higher surplus-generating capacity ofrural industry, relative to agriculture, that makes it attractive to village leaders,and that maximization of the rents that accrue to the local constituency isnecessary to protect and promote the leaders’ own interests. This tendency hasbeen reinforced by fiscal decentralization. Increasing dependence on local rev-enues has spurred local governments to develop township and village enterprisesas a primary revenue source. Township and village enterprises are not only amajor source of tax revenue for local governments,12 but they are also called uponto make contributions to self-raised funds that are used to finance major localcapital investments, such as new schools, health clinics, or transportation infra-structure. With very limited alternative sources of revenue, local governmentshave strong incentives to invest in township and village enterprises to establisha revenue base.

Given the existence of resources for capital formation, it is not unreasonableto believe that excess investment might be undertaken in order to maximize theearnings that can be retained by the local government. This, hypothesis, however,begs the question of sources of investment finance. Putterman (1997) argues that,historically, rural collective enterprises benefited from the commune’s ability tocommand local resources during the pre-reform era and from profit retention andfavorable access to bank credit during the immediate post-reform 1980s. In the1990s, Yang and Wei (1996) claim that in central and western regions, certainlocalities have been targeted for enterprise development support. Whether se-lected on the basis of past success with rural industrialization or with theintention of creating planned growth through the establishment of rural industrialzones, enterprises in these communities benefit from preferential treatment andpriority in access to enterprise development funds. If the objective is to maximizeearnings per worker, or more likely, earnings per local resident, and if credit issubsidized, then the overuse of capital and an inefficient choice of industry mixis a predictable outcome.

Finally, China’s market may not be sufficiently integrated to permit theapplication of the dynamic comparative advantage theory. The importance oflocal contacts both in procuring inputs and in marketing outputs, the inadequacies

12 For example, in 1989, taxes paid by township and village enterprises represented 37% of allfiscal revenue at or below the county level (Wong et al., 1995). In areas where township and villageenterprises are highly developed, they easily account for more than half of all fiscal revenue at thetownship and county levels. Jin and Qian (1998) provide empirical evidence showing that thecontributions of rural enterprises to community (and state) governments’ shares of revenues aregreater when the enterprises are owned by the township or village, relative to private ownership.

HARE AND WEST494

Page 21: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

of the communication and transportation system, and institutional barriers tofactor mobility all may combine to make dispersed, smaller-scale productiongeared toward local markets more viable. If continued reforms and growth leadto more complete markets and improved infrastructure, production may begin todisplay greater patterns of regional specialization.

6. CONCLUSION

The key findings presented here indicate that the spatial pattern of growthexhibited by China’s township and village enterprise sector does not displaystrong characteristics of trends toward greater regional specialization as predictedby a theory of comparative advantage. Policy efforts to stimulate enterprisegrowth in China’s interior in order to address regional income imbalances maybe a contributing factor to the outcomes we observe. In particular, preferentialtreatment with respect to credit may lead to the overuse of capital in regions withabundant labor resources. While the patterns of rural industrialization that weobserve may fulfill local objectives of rent maximization, they most likely reflectsome degree of resource misallocation at the nationwide level.

From a nationwide efficiency standpoint, better policies for closing the re-gional income gaps include improving the physical infrastructure, insuring con-tinued support for the formation of human capital in the inland provinces, andfollowing through on commitments to achieve market price levels. As Chen andFleisher (1996) and Fleisher and Chen (1997) point out, these efforts will makethe interior region a more attractive investment environment and will increase theincentives for efficient resource allocation. Reform of the fiscal system thatexpands the tax base beyond enterprises and increases central governmentresources for redistribution to inland provinces is essential in order to finance theinfrastructure and human capital investments. The government has announcedplans to boost infrastructure spending, by as much as $750 billion over threeyears, to counter the economic slowdown caused by the Asian financial crisis.Well-targeted large-scale infrastructure projects could benefit inland provincesand increase the productivity of existing and future township and village enter-prises. We conclude that the fastest way to raise income levels in the poorerregions is to reduce state intervention in pricing and resource allocation decisionsand to increase state support for infrastructure and education services.

REFERENCES

Chang, Chun, and Wang, Yijiang, “The Nature of the Township–Village Enterprise.”J. Comp.Econom.19, 3:434–452, Dec. 1994.

Chen, Jian, and Fleisher, Belton, M., “Regional Income Inequality and Economic Growth in China.”J. Comp. Econom.22, 2:141–164, Apr. 1996.

China Township and Village Enterprise Yearbook Editorial Committee,Zhongguo Xiangzhen QiyeNianjian 1995 (1995 Yearbook of China’s Township and Village Enterprises). Beijing:Agricultural Publishing House, 1995.

CHINA’S RURAL INDUSTRIAL GROWTH 495

Page 22: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

Dong, Xiao-yuan, “Employment and Wage Determination in China’s Rural Industry: InvestigationUsing 1984–1990 Panel Data.”J. Comp. Econom.26, 3:485–501, Sept. 1998.

Dong, Xiao-yuan, and Putterman, Louis, “China’s Rural Industry and Monopsony: An Exploration.”Pacific Econom. Rev.1, 1:59–78, June 1996.

Dong, Xiao-yuan, and Putterman, Louis, “Productivity and Organization in China’s Rural Industries:A Stochastic Frontier Analysis.”J. Comp. Econom.24, 2:181–201, Apr. 1997.

Fleisher, Belton, M., and Chen, Jian, “The Coast–Noncoast Income Gap, Productivity, and RegionalEconomic Policy in China.”J. Comp. Econom.25, 2:220–236, Oct. 1997.

Greene, William,Econometric Analysis,3rd ed. Upper Saddle River, NJ: Prentice–Hall, 1997.Ho, Samuel P. S., “Rural Non-agricultural Development in Post-Reform China: Growth, Develop-

ment, Patterns, and Issues.”Pacific Affairs68, 3:360–391, Fall 1995.Howes, Stephen, and Hussain, Athar, “Regional Growth and Inequality in Rural China.” London

School of Economics Working Paper EF/11, p. 36, Oct. 1994.Huang, Yasheng,Inflation and Investment Controls in China.Cambridge UK: Cambridge Univ.

Press, 1996.Jefferson, Gary H., “Are China’s Rural Enterprises Outperforming State Enterprises? Estimating the

Pure Ownership Effect.” In Gary H. Jefferson and Inderjit Singh, Eds.,Enterprise Reform inChina: Ownership, Transition, and Performance.Oxford/New York: Oxford Univ. Press,1999.

Jefferson, Gary H., and Rawski, Thomas G., “Enterprise Reform in Chinese Industry.”J. Econom.Perspect.8, 2:47–70, Spring 1994.

Jefferson, Gary H., Rawski, Thomas G., and Zheng, Yuxin, “Innovation and Reform in China’sIndustrial Enterprises.” In Gary H. Jefferson and Inderjit Singh, Eds.,Enterprise Reform inChina: Ownership, Transition, and Performance.Oxford/New York: Oxford Univ. Press,1999.

Jin, Hehui, and Qian, Yingyi, “Public versus Private Ownership of Firms: Evidence from RuralChina.” Quart. J. Econom.113,3:773–808, Aug. 1998.

Khan, Azizur Rahman, Griffin, Keith, Riskin, Carl, and Zhao, Renwei, “Household Income and ItsDistribution in China.”China Quarterly,132,1029–1061, Dec. 1992.

Li, Fan, and Han, Xiaoyun, “Age Groups and Educational Composition of Peasants Leaving forNonagricultural Work—Studies of Rural Labor Migration.”Zhongguo Nongcun Jingji(Chi-na’s Rural Economy), no. 8, pp. 10–14, Aug. 20, 1994. [In Chinese, translated in JointPublications Research Service CAR-95-054, pp. 61–69, November 15, 1994].

Lyons, Thomas P., “Interprovincial Disparities in China: Output and Consumption, 1952–1987.”Econom. Develop. Cultural Change39, 3:471–506, Apr. 1991.

Meier, Gerald M., Ed.,Leading Issues in Economic Development,6th ed. New York: Oxford Univ.Press, 1995.

Naughton, Barry,Growing Out of the Plan: Chinese Economic Reform 1978–1993.Cambridge, UK:Cambridge Univ. Press, 1995.

Pei, Xiaolin, “Township–Village Enterprises, Local Governments and Rural Communities: TheChinese Village as a Firm During Economic Transition.”Econom. Transition4, 1:43–66,May 1996.

Prime, Penelope B., “China’s Economic Reforms in Regional Perspective.” In Gregory Veeck, Ed.,The Uneven Landscape: Geographic Studies in Post-Reform China,Geoscience and Man,Vol. 30, pp. 9–27. Baton Rouge, LA, Geoscience Publications, 1991.

Prime, Penelope B., “Industry’s Response to Market Liberalization in China: Evidence from JiangsuProvince.”Econom. Develop. Cultural Change41, 1:27–50, Oct. 1992.

Putterman, Louis, “On the Past and Future of China’s Township and Village-Owned Enterprises.”World Develop.25, 10:1639–1655, Oct. 1997.

Rozelle, Scott, “Rural Industrialization and Increasing Inequality: Emerging Patterns in China’sReforming Economy.”J. Comp. Econom.19, 3:362–391, Dec. 1994.

HARE AND WEST496

Page 23: Spatial Patterns in China's Rural Industrial Growth and Prospects for the Alleviation of Regional Income Inequality

Rozelle, Scott, “Stagnation Without Equity: Patterns of Growth and Inequality in China’s RuralEconomy.”China J.35:63–92, Jan. 1996.

State Statistical Bureau,Zhongguo Tongji Nianjian 1986[1986 China Statistical Yearbook]. Beijing:China Statistical Publishing House, 1986.

State Statistical Bureau,Zhongguo Renmin Gongheguo 1985 Nian Gongye Pucha Ziliao[Data of the1985 Industrial Census of the People’s Republic of China]. Beijing: China Statistical Pub-lishing House, 1987.

State Statistical Bureau,Zhongguo Renkou Tongji Nianjian1995 [1995 China Population StatisticsYearbook]. Beijing: China Statistical Publishing House, 1995a.

State Statistical Bureau,Zhongguo Tongji Nianjian 1995[1995 China Statistical Yearbook]. Beijing:China Statistical Publishing House, 1995b.

State Statistical Bureau,Zhongguo Nongcun Tongji Nianjian 1996[1996 Rural Statistical Yearbookof China]. Beijing: China Statistical Publishing House, 1996.

State Statistical Bureau,Zhongguo Nongcun Tongji Nianjian 1997[1997 Rural Statistical Yearbookof China]. Beijing: China Statistical Publishing House, 1997a.

State Statistical Bureau,Zhongguo Renmin Gongheguo 1995 Nian Di San Ci Quanguo Gongye PuchaZiliao Huibian [The Data of the Third National Industrial Census of the People’s Republic ofChina in 1995]. Beijing: China Statistical Publishing House, 1997b.

State Statistical Bureau,Zhongguo Tongji Nianjian 1997[1997 China Statistical Yearbook]. Beijing:China Statistical Publishing House, 1997c.

Township and Village Enterprise Bureau of the Ministry of Agriculture,Xiangzhen Qiye TongjiZiliao 1986 [1986 Township and Village Enterprise Statistical Data], 1987.

Tsui, Kai Yuen, “China’s Regional Inequality, 1952–1985.”J. Comp. Econom.15, 1:1–21, Mar.1991.

Tsui, Kai Yuen, “Decomposition of China’s Regional Inequalities.”J. Comp. Econom.17, 3:600–627, Sept. 1993.

Tsui, Kai Yuen, “Factor Decomposition of Chinese Rural Income Inequality: New Methodology,Empirical Findings, and Policy Implications.”J. Comp. Econom.26, 3:502–528, Sept. 1998.

Veeck, Gregory, “Development, Regional Equity, and Political Change in China.”Bull. ConcernedAsian Scholars25, 3:73–75, July 1993.

Vernon, Raymond, “International Investment and International Trade in the Product Cycle.”Quart.J. Econom.80, 2:190–207, May 1966.

West, Loraine, A., and Wong, Christine, P. W., “Fiscal Decentralization and Growing RegionalDisparities in Rural China: Some Evidence in the Provision of Social Services.”Oxford Rev.Econom. Policy11, 4:70–84, Winter 1995.

Wong, Christine P. W., “Rural Public Finance.” In Christine P. W. Wong, Ed.,Financing LocalGovernment in the People’s Republic of China,pp. 167–212. Hong Kong: Oxford Univ.Press, 1997.

Wong, Christine P. W., Heady, Christopher, and Woo, Wing T.,Fiscal Management and EconomicReform in the People’s Republic of China.Hong Kong: Oxford Univ. Press, 1995.

World Bank, Sharing Rising Incomes: Disparities in China.Washington, DC: The World Bank,1997.

Yang, Dali L., “Patterns of China’s Regional Development Strategy.”China Quart.122:230–257,June 1990.

Yang, Dali L., “China Adjusts to the World Economy: The Political Economy of China’s CoastalDevelopment Strategy.”Pacific Affairs64, 1:42–64, Spring 1991.

Yang, Dali L., and Wei, Houkai, “Rural Enterprise Development and Regional Policy in China.”Asian Persp.20, 1:71–94, Spring-Summer 1996.

Zheng, Jinghai, Liu, Xiaoxuan, and Bigsten, Arne, “Ownership Structure and Determinants ofTechnical Efficiency: An Application of Data Envelopment Analysis to Chinese Enterprises(1986–1990).”J. Comp. Econom.26, 3:465–484, Sept. 1998.

CHINA’S RURAL INDUSTRIAL GROWTH 497