China's Economic Growth 1978-2025

27
China’s Economic Growth 1978–2025: What We Know Today About China’s Economic Growth Tomorrow CARSTEN A. HOLZ * Princeton University, NJ, USA Hong Kong University of Science and Technology, Kowloon, Hong Kong Summary. This paper examines China’s future growth prospects and the potential drivers of fu- ture growth using two approaches. It first asks in how far China’s recent economic development matches standard growth patterns identified in development economics and trade theory. Second, GDP is decomposed into income components, which in turn are explained, for the reform period, by the quantity and quality of labor; future GDP can then be re-composed from future labor data available today. Overall, China’s economic growth is likely to continue at current rates through 2015 before it gradually slows. Such a growth has numerous implications. Ó 2008 Elsevier Ltd. All rights reserved. Key words — economic growth, growth accounting, growth forecasts, development theories, edu- cation, China 1. INTRODUCTION The rapid economic growth of China since the beginning of economic reforms in 1978 has captured the imagination of Western com- mentators and researchers. The responses range from outright pessimism about China’s future to fear of a strong China. Brown (1995) won- dered who will feed China. Chang (2001) an- nounced the coming collapse of China. Henderson (1999) saw China as on the ‘‘brink,’’ while Terrill (2003) wrote of the ‘‘illusory nat- ure of the market in most of the Chinese econ- omy’’ and that ‘‘a crash looms because the Leninist core of the regime is unchanged from Mao’s construction of it in Yan’an six decades ago’’ (pp. 329, 313). At the other end of the spectrum are those who envisage a strong China. Murray (1998) de- scribed China as the next superpower. A number of authors view an all-powerful China as a threat (Gertz, 2000, or Timperlake et al., 2002). 1 News items imploring the ‘‘devastation Chinese com- petitors are inflicting on US industries, from kitchenware and car tires to electronic circuit boards’’ and the ‘‘futility of trying to match the China price’’ have become common fare. 2 What is undeniable is China’s rapid economic growth over the past 27 years, from the begin- ning of economic reforms in 1978–2005, of, measured in gross domestic product (GDP), on average 9.6% per year. In economic size, Chi- na in 2005 was surpassed only by the United States, Japan, and Germany. 3 Its share in glo- bal growth 1995–2002 has been estimated at 25%, compared to 20% for the United States. 4 While China’s future economic growth has received much attention in the popular litera- ture, research literature is scarce and more lim- ited in scope. Douglas (2002) documented the transformation in four industries in China and predicted the imminent emergence of world-class Chinese brands. Zeng and William- son (2003) examined the international growth prospects of Chinese firms (positive). Carter and Rozelle (2001) asked if China will become * The work described in this paper was supported by a Grant from the Research Grants Council of Hong Kong (Project HKUST6423/05H). I gratefully acknowledge feedback received from discussants and participants at a dozen seminars and conference presentations in 2003– 07. Final revision accepted: September 24, 2007. World Development Vol. 36, No. 10, pp. 1665–1691, 2008 Ó 2008 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter doi:10.1016/j.worlddev.2007.09.013 www.elsevier.com/locate/worlddev 1665

Transcript of China's Economic Growth 1978-2025

Page 1: China's Economic Growth 1978-2025

World Development Vol. 36, No. 10, pp. 1665–1691, 2008� 2008 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matter

doi:10.1016/j.worlddev.2007.09.013www.elsevier.com/locate/worlddev

China’s Economic Growth 1978–2025: What

We Know Today About China’s Economic

Growth Tomorrow

CARSTEN A. HOLZ *

Princeton University, NJ, USAHong Kong University of Science and Technology, Kowloon, Hong Kong

Summary. — This paper examines China’s future growth prospects and the potential drivers of fu-ture growth using two approaches. It first asks in how far China’s recent economic developmentmatches standard growth patterns identified in development economics and trade theory. Second,GDP is decomposed into income components, which in turn are explained, for the reform period,by the quantity and quality of labor; future GDP can then be re-composed from future labor dataavailable today. Overall, China’s economic growth is likely to continue at current rates through2015 before it gradually slows. Such a growth has numerous implications.� 2008 Elsevier Ltd. All rights reserved.

Key words — economic growth, growth accounting, growth forecasts, development theories, edu-cation, China

* The work described in this paper was supported by a

Grant from the Research Grants Council of Hong Kong

(Project HKUST6423/05H). I gratefully acknowledge

feedback received from discussants and participants at a

dozen seminars and conference presentations in 2003–

07. Final revision accepted: September 24, 2007.

1. INTRODUCTION

The rapid economic growth of China sincethe beginning of economic reforms in 1978has captured the imagination of Western com-mentators and researchers. The responses rangefrom outright pessimism about China’s futureto fear of a strong China. Brown (1995) won-dered who will feed China. Chang (2001) an-nounced the coming collapse of China.Henderson (1999) saw China as on the ‘‘brink,’’while Terrill (2003) wrote of the ‘‘illusory nat-ure of the market in most of the Chinese econ-omy’’ and that ‘‘a crash looms because theLeninist core of the regime is unchanged fromMao’s construction of it in Yan’an six decadesago’’ (pp. 329, 313).

At the other end of the spectrum are those whoenvisage a strong China. Murray (1998) de-scribed China as the next superpower. A numberof authors view an all-powerful China as a threat(Gertz, 2000, or Timperlake et al., 2002). 1 Newsitems imploring the ‘‘devastation Chinese com-petitors are inflicting on US industries, fromkitchenware and car tires to electronic circuitboards’’ and the ‘‘futility of trying to match theChina price’’ have become common fare. 2

166

What is undeniable is China’s rapid economicgrowth over the past 27 years, from the begin-ning of economic reforms in 1978–2005, of,measured in gross domestic product (GDP),on average 9.6% per year. In economic size, Chi-na in 2005 was surpassed only by the UnitedStates, Japan, and Germany. 3 Its share in glo-bal growth 1995–2002 has been estimated at25%, compared to 20% for the United States. 4

While China’s future economic growth hasreceived much attention in the popular litera-ture, research literature is scarce and more lim-ited in scope. Douglas (2002) documented thetransformation in four industries in Chinaand predicted the imminent emergence ofworld-class Chinese brands. Zeng and William-son (2003) examined the international growthprospects of Chinese firms (positive). Carterand Rozelle (2001) asked if China will become

5

Page 2: China's Economic Growth 1978-2025

1666 WORLD DEVELOPMENT

a major force in world food markets (yes, ifstructural transformation occurs).

A number of studies try to explain China’spast economic growth. Much of the transitionliterature focuses on institutional changes to ex-plain past reform successes—usually, if onlyimplicitly, measured as economic growth (e.g.,Qian, 2000, 2003, or Woo, 1994, 1999). Linet al. (2003) explained China’s reform successwith the adoption of a ‘‘comparative-advantagefollowing strategy,’’ allowing full play to Chi-na’s plentiful endowment with labor. Othersanalyze China’s past economic growth withinan aggregate production function framework(Wang & Yao, 2003; Wu, 2004; Young,2003). Chow and Li (2002) obtained GDP val-ues through 2010 by extending their 1952–98estimation using variations of past factor inputgrowth rates; Chow (2002) further included anestimate of the year 2020 GDP. A 1997 WorldBank policy analysis of current economic issuesis supplemented by an estimate of China’s GDPin 2020 based on a closed-economy model. Asystematic study of future economic growth inChina that makes full use of hard facts aboutthe future that are available today, however,is still lacking. 5

Answers to the question of what we know to-day about the economic size of China 10 or 20years from now are relevant not only for thepopular discourse on the new superpower andfor China threat theorists and military strate-gists, but also for economists. For example,conventional economic wisdom holds that freetrade benefits, by the laws of comparativeadvantage, all countries in the long run. Thus,a growing China, and growing trade betweenChina and the United States, benefits the Uni-ted States. But as Samuelson (2004) within theframework of standard trade theory showed,a productivity gain in one country in the pro-duction of the good in which the other countrypreviously had a comparative advantage may,in fact, cause a permanent loss in per capita realincome in the other country (though not rela-tive to autarky); references to China, as thecountry with a productivity gain, abound.

This paper examines China’s future economicgrowth and some potential drivers of thisgrowth. The first step is a linear extrapolationof the stable, past growth trend into the future.A number of scenarios are played through in acomparison with the United States. But extrap-olation is not a particularly convincing researchtool. Why should past trends continue into thefuture?

One approach is to examine how China fareswith respect to standard growth patterns identi-fied by development economics and trade the-ory. These are structural changes, catchingup, and factor price equalization. China’s pasteconomic growth fits well with all three. Fur-thermore, China’s reform period growth, with-in these three growth patterns, matches that ofJapan, Korea, and Taiwan at an earlier stage oftheir development. Obviously, these four coun-tries differ, as do the domestic and internationalcircumstances under which they experienced aparticular stage of development. But the differ-ences need not be systematic with respect to theimpact on the dependent variable economicgrowth. In as far as these growth patterns havealready played out for Japan, Korea, and Tai-wan in the past, they bode well for China’s fu-ture economic growth.

A second approach is to decompose GDPinto its income components and to explainthese, for the reform period, with the quantityand quality of labor. We know with near-cer-tainty the size of China’s working age popula-tion through 2015, and can project with highreliability for several years after. In as far asgains in high school and university enrollmentare unlikely to be reversed, a bottom line sce-nario for the quality of China’s future laborerscan be developed. Future GDP growth canthen be re-composed, using the established rela-tionships, from data on the future quantity andquality of labor available today. One result ofthe decomposition of GDP growth is the poten-tial importance of future labor quality in driv-ing continued economic growth in China. Aseparate section directly compares educationlevels and education trends in China and theUnited States.

Projections cannot predict the future withcertainty. Potential complications abound,from political risks to non-performing loansin China’s banking system, loss-making state-owned enterprises, pilfered pension funds, localgovernment budget deficits, rural–urbaninequality, and environmental degradation.This paper works within these limitations. Itdoes not attempt to provide an exhaustive listof complications, examine each of them, quan-tify their potential impact, and to attribute alikelihood to their occurrence. It predicts Chi-na’s future economic growth under theassumption that current and future problemscontinue to be resolved as they arise, and thatevents of catastrophic dimensions do notoccur. 6

Page 3: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1667

2. EXTRAPOLATION OF PAST GROWTHINTO THE FUTURE

China’s year 2005 nominal GDP of 18.308trillion RMB at the official average annual ex-change rate of 8.1917 RMB per USD translatesinto USD 2.235 trillion, compared to the year2005 US GDP of 12.456 trillion. 7 The USeconomy in 2005, thus, was almost six timeslarger than the Chinese economy. However,such a comparison neglects to take into accountthe differing price levels in the two countries.The Penn World Table 6.2 (PWT 6.2), whichreports comparable GDP data in ‘‘internationaldollars,’’ assumes that the general price level inthe United States in 2004, the most recent yearcovered, was 3.86 times higher than the Chineseprice level. Applying the 2004 price adjustmentfactor to the 2005 GDP data, the US economy,today, in purchasing power terms, is only 43%larger than China’s economy.

Table 1. First year in which Ch

Price adjustment? Aggre

Exchange rate adjustment? GD

1. US price level = 3.86 * PRC price level 202. +3% annual RMB appreciation 203. US price level = 3* PRC price level 204. +3% annual RMB appreciation 205. No price adjustment 206. +3% annual RMB appreciation 20

Base year for the projections is 2005. Annual growth rates urates of China and the United States in 1978–2005. The precent year for which the PWT 6.2 provides data (The value3.859263).‘‘China coast’’ refers to the coastal provinces of China (in coincludes Beijing, Tianjin, Hebei, Liaoning, Shanghai, JiGuangxi (the small province Hainan was omitted due to a laprovince). ‘‘China 5 prov.’’ refers to the five fastest-growinShandong, and Guangdong). The share of the coast in Chin66.2%; the corresponding shares of the five provinces are 2Nominal and real GDP values of 1978 and 2005 are thoseand by the Bureau of Economic Analysis in the United Saggregate GDP by the mid-year population. United StatesChinese provincial-level population data of 1978 are basedwith mid-1978 values obtained by extrapolation using the i2000.The average annual real growth rate of economy-wide GDPit is 2.9687%. The average annual growth rate of real GDP pis 1.8821%. The average annual population growth rate inThe average exchange rate in 2005 is 8.1917 yuan RMB/USources: PWT 6.2. Census 2000 Major Figures, pp. 16, 31. SBureau of Economic Analysis, at <https//www.bea.gov/beaGDP, XLS;’’ accessed on 30 Nov. 06). US population data apopclockest.txt> and for 2000–05 from <http://www.censusNovember 30, 2006).

In what year will China’s GDP exceed that ofthe United States, assuming that the past realgrowth rates continue into the future? The an-swer, reported in Table 1, is obtained by apply-ing the average annual real growth rate of theperiod 1978–2005 of China and the UnitedStates to their respective year 2005 GDP values,and then noting the year in which China’s GDPvalue exceeds that of the United States 8 (theinitial year, 1978, reflects the beginning of Chi-na’s economic reforms). In the first row shownin Table 1, it is assumed that the PWT 6.2 priceadjustment factor of 3.86 (in 2004) continues toapply into the indefinite future. In this scenario,China’s GDP begins to exceed that of the Uni-ted States, in purchasing power terms, in 2011.If one further incorporates a 3% annual appre-ciation of the RMB, reflecting the most recentexchange rate developments, the year whenChina’s GDP exceeds that of the United States,in purchasing power terms, moves further to

ina’s GDP exceeds US GDP

gate Per capita GDP

P National China coast China 5 prov.

11 2035 2024 202109 2026 2019 201715 2040 2027 202312 2029 2021 201933 2057 2041 203624 2041 2031 2028

sed in the projections are the average annual real growthrice adjustment factor of 3.86 is that of 2004, the mosts of 2000–04 are: 4.217306, 4.223668, 4.264024, 4.186857,

mparison to the national average in the United States); itangsu, Zhejiang, Fujian, Shandong, Guangdong, andck of data for 1978 when Hainan was part of Guangdongg coastal provinces in China (Jiangsu, Zhejiang, Fujian,a’s population in mid-2005 is 42.5%, and in annual GDP5.9% and 43.3%.published by the National Bureau of Statistics in Chinatates. Per capita GDP values are obtained by dividingpopulation data for the year 2005 is the official estimate.on the July 1, 1982 and the November 1, 2000 censuses,mplicit average annual population growth rate of 1982–

in China in 1978–2005 is 9.6358%; in the United States,er capita in China is 8.3840%, and in the United States it

China is 1.1549% and in the United States it is 1.0665%.SD.tatistical Yearbook 1991, p. 79; 2006, pp. 57, 60, 99, 734./dn/home/gdp.htm> (choose ‘‘Current-dollar and ‘‘real’’re from <http://www.census.gov/popest/archives/1990s/.gov/popest/states/NST-ann-est.html> (both accessed on

Page 4: China's Economic Growth 1978-2025

1668 WORLD DEVELOPMENT

the present, to 2009 (second row shown inTable 1).

The price adjustment factor in the case ofChina carries ‘‘substantial uncertainty’’ (Hes-ton, 2001, p. 1). The multiplicative factor of3.86 could be considerably off in either direc-tion. 9 A price adjustment factor of three wouldimply a cross-over year of 2015 or, with RMBappreciation, of 2012 (third and fourth rowsshown in Table 1). At the extreme, making noprice adjustments whatsoever, China’s GDPequals that of the United States 25 years inthe future, in 2033, or, with RMB appreciation,in 2024 (fifth and sixth rows shown in Table 1).

In terms of per capita GDP, the year whenChina surpasses the United States is 20–50years in the future (see additional columnsshown in Table 1). But what is far in the futurefor the average Chinese person moves closer tothe present for the richer areas of China. One totwo decades from now, coastal China (as de-fined in the notes to Table 1), accounting for42% of China’s population in 2005 and exceed-ing the US population by 87%, may well be asrich, per capita, as the average US citizen.Focusing only on the five fastest-growing prov-inces in coastal China, with a joint populationin 2000 that is 11% larger than the US popula-tion, moves the cross-over year a few years fur-ther to the present. In other words, by around2020, a share of China’s population that ex-ceeds the size of the US population could enjoythe same average standard of living as the Uni-ted States; in addition, China has a poorer hin-terland with roughly three times the populationin the richer areas. 10

In these calculations, all variables are heldconstant except those explicitly allowed to vary.In particular, in the extrapolation of aggregateGDP, the use of an average annual past realgrowth rate reflects the product of the realGDP growth rate per capita and the populationgrowth rate. In the short run, such as one dec-ade, changes in the population growth rate arelikely to be minor, but this may not hold in thelong run. Similarly, the exchange rate was heldconstant in three scenarios, as was the priceadjustment factor throughout (at the three dif-ferent levels). 11 The further into the future, theless valid are the assumptions. A more sophisti-cated approach would, furthermore, focus onthe employed instead of the population. Butnone of these adjustments is likely to have a sig-nificant impact in the short run, and the extrap-olations at the aggregate level suggest thatmuch will change in the short run. 12

Extrapolations do not constitute a reliable re-search tool. The following two sections employtheories and tools of economics to examineChina’s future growth and its potential deter-minants.

3. DEVELOPMENT THEORIES ANDASIAN PRECEDENTS

Explanations of China’s economic growthtend to focus on economic transition. The ‘‘suc-cess’’ of the reform process is explained by tran-sition facts and strategies, where success isusually taken to imply economic growth (or arise in living standards). For example, Woo(1994) listed as crucial the creation of non-statefirms in every sector of the economy, a highsavings rate, good initial conditions (such as asmall extent of central planning, unemploymentin the countryside that could be taken up bytownship and village enterprises, or a limitedsocial security net), historical conditions, andthe Chinese Diaspora. Qian (2000, 2003) as-cribed much of China’s reform success to theunorthodox economic policy measures adoptedby China’s leadership; the key to China’s eco-nomic growth was the unleashing of incentivesand competition while making reform interest-compatible for those in power.

Yet these explanations of past economic re-form successes (and thus economic growth) lacka counterfactual. Wing-Thye Woo offers cross-country comparisons, but the number of explan-atory variables appears larger than the numberof comparison countries. Qian Yingyi makesan in part historical argument, but offers noexplicit time series evaluation that examines thestatus before and after a particular reform mea-sure was implemented. These explanations, thus,are not as strong as one might wish them to be. 13

If one takes the view that economic transition‘‘in total’’ has caused China’s past economicgrowth, then one could argue that the key ele-ments of transition have been in place sincethe early 1990s (price and domestic trade liber-alization, the entry of private enterprises) andthat therefore the gains from transition have al-ready been exhausted. Consequently, economicgrowth should since have slowed, which it didnot, or it could slow any time soon. Alterna-tively, one could argue that past transition mea-sures impact on economic growth over anextended period of time, or that transition isas yet incomplete, with further measures togo, in which cases the gains continue.

Page 5: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1669

Growth patterns identified by economic theo-ries of development and trade perhaps offer fir-mer ground. Economic transition could then beviewed as the removal of constraints that pre-vented well-known development patterns fromunfolding. Furthermore, if China’s reform per-iod growth patterns match those exhibited byother countries at stages in their economicdevelopment similar to China’s development le-vel in the reform period, then China’s futuregrowth patterns may also match those of theother economies later. The argument has twofoundations: one is a focus on standard growthpatterns, the other is a cross-country compari-son with countries with which a comparison islikely to be meaningful.

The growth patterns are structural change,catching up, and factor price equalization. 14

These three patterns are not independent ofeach other, nor do they hold irrespective ofthe larger economic environment; they do notalways come with perfectly clear-cut causality(though one causality, perhaps the most plausi-ble one, is used to present each pattern). Thepatterns have the advantage that they do notrely on individual transition measures and thatthey have been identified by development eco-nomics and trade theory as of relevance. In asfar as China is at the very early stages of theseaccepted development patterns, with good rea-sons for these patterns to continue to unfold,future economic growth is likely.

The comparison countries are Japan, Korea,and Taiwan. These three countries are relevantfor China only if they do not differ systemati-cally with respect to the relationship betweenkey variables. While these four countries differ,as do the domestic and international circum-stances under which they experienced a partic-ular stage of development, that does notnecessarily invalidate the assumption of a simi-lar relationship between key variables.

The choice of countries to compare China to isa subjective choice, motivated only by the desireto make comparisons across a relatively homo-geneous group of countries. China may sharesome economic growth patterns with Japan,Korea, and Taiwan due to cultural similarities,geographic location, similar economic develop-ment strategies, or, in the case of Japan, rela-tively large size of the domestic economy.Limiting the analysis to Japan, Korea, and Tai-wan allows the careful compilation of the neces-sary data from each individual country’sstatistical office, with a very few missing datapoints obtained from the World Bank Develop-

ment Indicators database, the InternationalFinancial Statistics (IFS), and the PWT. 15 Anattempt was made to cover the years 1950–2005 but data are often available only since the1960s or 1970s. The earlier in the developmentprocess of Japan, Korea, and Taiwan, the closerto the case of China the initial conditions mayhave been. Data on China are for the years ofthe reform period (the years since 1978).

Throughout, the variable to be explained isreal GDP growth per laborer, because GDP isproduced only by the active working popula-tion, namely laborers. 16

(a) Structural change

As labor shifts from low-productivity agricul-ture to higher-productivity industry and ser-vices, economy-wide real GDP per laborer,that is, real (partial) labor productivity, in-creases, if only because those laborers who haveshifted sectors now produce a multiple of theirformer output value. Figure 1 shows just howbig, and increasing, the labor productivity dif-ferences are between the three economic sectorsin China. Consequently, one would expect tosee relatively high aggregate (economy-wide)labor productivity growth in those years whena relatively large number of laborers shift outof agriculture.

Figure 2 shows that this is indeed the case.Each data point reflects a particular countryin a particular year. In the years with a highabsolute reduction in the share of laborers inagriculture, the growth rate of (real, likewisebelow) labor productivity is high. 17 This rela-tionship holds equally for all four countriesand is remarkably similar for China, Taiwan,and Korea. The pattern of decline over timein the share of laborers in agriculture is alsosimilar across all countries, with Japan the firstto have embarked on this process, followed byTaiwan, Korea, and then China (Figure 3). 18

At China’s 1978–2005 rate of decline in theshare of laborers in agriculture from just above70% to 45%, with an annual reduction in theshare of laborers in agriculture by, on average,an absolute value of 0.01 every year (thus, e.g.,from 0.7 to 0.69 in one year), China has an-other 35 years to go before its agricultural laborshare reaches the 10% level at which the agri-cultural labor shares of Japan, Korea, and Tai-wan stabilize. But this implies that China facesanother 35 years of structural change as asource of economic growth, a long time spanin the context of the extrapolations.

Page 6: China's Economic Growth 1978-2025

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

1978 1981 1984 1987 1990 1993 1996 1999 2002 2005Val

ue-a

dded

per

labo

rer

(yua

n R

MB

, cur

rent

pric

es)

Primary

Secondary

Tertiary

Figure 1. Sectoral labor productivity in China. Employment data are mid-year values (as average of end-year values).

All employment data (sectoral, total) during 1989–1990 experience a statistical break; end-year 1990 total employment

exceeds that of 1989 by 17.03%. Official (unadjusted) employment data are used. The secondary sector comprises

industry and construction.

-4-3-2-10123456789

1011121314

-0.035 -0.030 -0.025 -0.020 -0.015 -0.010 -0.005 0.000 0.005 0.010

Annual absolute change inshare of agriculture in employment

Gro

wth

rat

e of

rea

l GD

P p

er la

bore

r (in

%)

Japan

Korea

Taiwan

China

China

Japan

Korea

Taiwan

Figure 2. Structural change and labor productivity growth. Japan 1956–2005, Korea 1971–2005, Taiwan 1967–2005,

and China 1979–2005. (Pre-1991 employment data for China are adjusted as explained in appendix.) Two data points

are omitted in order to keep the chart compact: Korea 1972 (0.0224, 0.9448) and 1998 (0.0125, �4.6146).

1670 WORLD DEVELOPMENT

The pattern of structural change is likely tocontinue to unfold in China. Rural areas stillsuffer from underemployment in agriculture;the number of agricultural laborers per landarea is more than one hundred times higher

than in the U.S. School leavers in rural areas al-most invariably do not enter agriculture, andthis is unlikely to change. The development ofrural industries is officially encouraged. Urbanincome, including for unskilled labor, is still

Page 7: China's Economic Growth 1978-2025

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005

Shar

e of

agr

icul

ture

in e

mpl

oym

ent

Japan Korea

Taiwan China

Figure 3. Share of agricultural laborers in economy-wide employment. Employment data are average annual values (in

the case of China, mid-year values). Values for Japan only comprise agriculture (presumably farming and animal

husbandry) and forestry; they exclude fishery due to the lack of time series data.

CHINA’S ECONOMIC GROWTH 1978–2025 1671

significantly higher than the income in manyrural areas, creating incentives for rural–urbanmigration. The constraints that limited rural–urban labor flows in the pre-reform period havebeen gradually relaxed and official policy is torelax them further. The development of ruralland markets is on the political agenda, and,once completed, will likely remove anotherimpediment to rural–urban migration. 19

(b) Catching up

A latecomer to economic development hasthe advantage that production techniques andtechnologies that have already been inventedcan be copied. Taking the United States asthe leader in research and development, andproxying the level of technological developmentby labor productivity, the distance between aparticular country’s labor productivity (inUSD) and US labor productivity serves as ameasure of the potential scope for catchingup. One would expect to see relatively high realGDP growth per laborer in a country whosedistance to the leading country (the UnitedStates) is relatively high, that is, when the po-tential for catching up is large.

Figure 4 confirms the hypothesis. Japanexperienced high growth rates when its laborproductivity was 10–50% of US labor produc-tivity. As the gap closed, growth rates in Japanfell. 20 Korea and Taiwan exhibit constant highgrowth rates of labor productivity at an aver-

age 4% to 5% throughout all years. These twocountries’ level of economic development hasso far remained in the range of 5% to 50% ofthe US level, a range in which Japan also exhib-ited high and constant labor productivitygrowth rates. Korea and Taiwan may yet haveto experience the slow-down that comes whenthe potential gains from catching up are ex-hausted. China has remained within such a nar-row range of (low) development, with laborproductivity at only 1.2 to 3.4% of the US level(using the official exchange rate to translateChinese yuan RMB values into USD values),that a meaningful pattern is not yet discernible(Figure 5). At China’s highest level of catchingup so far, the growth rate of labor productivitywas a relatively high 9%.

Independent of the choice of data, China is ata very low development level compared to theUnited States. It appears to be at a stage of eco-nomic development where other Asian coun-tries started out more than 30 years ago. 21

Even if China were to move along the patternonly as far as Korea and Taiwan have, theextrapolations presented earlier still apply.

The pattern of catching up is likely to con-tinue to unfold in China as China’s integrationinto the world economy continues. Foreignfirms bring their knowledge of production pro-cesses if not their innovative capacity to China.Chinese firms buy foreign firms abroad or setup factories abroad, gaining more immediateaccess to foreign knowledge. At the same time,

Page 8: China's Economic Growth 1978-2025

123456789

1011121314

0.012 0.014 0.016 0.018 0.020 0.022 0.024 0.026 0.028 0.030 0.032 0.034

GDP per laborer (in USD) per U.S.GDP per laborer (in USD)

Gro

wth

rate

of r

eal G

DP

per l

abor

er (i

n %

)

Figure 5. Catching up and labor productivity growth in China, 1979–2005.

-3-2-10123456789

1011121314

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

GDP per laborer (in USD) per U.S.GDP per laborer (in USD)

Gro

wth

rat

e of

rea

l GD

P p

er la

bore

r (in

%)

Japan

Korea

Taiwan

China

Japan

Taiwan

Korea

China

Figure 4. Catching up and labor productivity growth. Japan 1960–2005, Korea 1971–2005, Taiwan 1967–2005, and

China 1979–2005.

1672 WORLD DEVELOPMENT

foreign knowledge becomes more accessiblewithin China due to the emergence of a Chinesegeneration that is both significantly better edu-cated than previous generations and increas-ingly fluent in English. A growing contingentof Chinese has studied abroad. 22

(c) Factor price equalization

The factor price equalization theorem (orHeckscher-Ohlin-Samuelson theorem) statesthat factor prices, such as skill-specific wages,should equalize between two countries aslong as a range of assumptions are met, 23

that is, the skill-specific wage rate of onecountry divided by that of the other countryshould equal unity. The slightly less restric-tive, relative version of the factor priceequalization theorem focuses on the relativeprices of factors of production. Thus, forexample, a country with labor that is cheaprelative to capital should see demand for itslabor rise. As underemployed laborersbecome fully employed (or labor shifts outof low-productivity agriculture), labor pro-ductivity rises.

In the absence of a reliable price of capital,the price of investment goods is used as a

Page 9: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1673

proxy. The price of investment goods of anyparticular country, relative to the UnitedStates, is available in the PWT.

Figure 6 reveals a clear downward trend forJapan. Relative factor price equalization, at10–40% of the US level, appears to have littleeffect on Korea and Taiwan, which experiencecontinuously stable labor productivity growth,with perhaps a slight decline in the most recentyears when approaching 50% of the US level.The observations for China are within such anarrow range, at only 2–7% of the correspond-ing US ratio, that a clear pattern does not yetemerge. (If the China data are presented in achart of their own, omitted here, that chartlooks similar to Figure 5.) China’s level of rela-tive factor price equalization is well below theearly levels of the other three Asian countries.If China were again at the beginning of along-run trajectory, then the patterns of theother three Asian countries suggest that Chi-na’s potential for economic growth from rela-tively low labor costs will continue to exist foranother 30 years. 24

China’s relative wages are currently still solow that they are likely to remain internation-ally attractive for many years to come. Accessto world trade translates the low relative wagesinto investment in labor-intensive industries inChina and thus into industrialization and ulti-mately labor productivity growth. As long asChina participates in world trade, there appears

-5-4-3-2-10123456789

1011121314

0.0 0.1 0.2 0.3

Relative wage rates, divided b

Gro

wth

rate

of r

eal G

DP

per l

abor

er (i

n %

)

Taiwan

Korea

China

Figure 6. Relative factor price equalization and labor product

1967–2004, and China 1979–2003. For China, 2004 inc

remuneration, needed to derive wages) are not available. The

2004 as the final year for which data are available. Wages

no reason why the gradual process of factorprice equalization should not continue to un-fold in China.

This section applied three explanations ofeconomic growth from development economicsand trade theory to the case of China. China’spast economic growth matches these standarddevelopment patterns, as does economicgrowth in Japan, Korea, and Taiwan. In asfar as China—in comparison to Japan, Korea,and Taiwan—is located at the early stages ofeach pattern, there remains much scope for fu-ture gains in labor productivity and therebygrowth.

These broad development patterns are obvi-ously not etched in stone and ignore numerousother factors that may impact on labor produc-tivity growth. One example is short-run devia-tions from the trend (as exhibited in thefigures) due to such factors as business cyclesor discrete decisions on the systemic featuresof the economy. Similarly, what was the casefor Japan, or Korea, or Taiwan may not holdfor China, or may not occur at the same stageof development in China as it has in the othercountries. One example would be labor mobil-ity (but the restrictions on rural–urban labormobility in China are falling), or weaknessesin the legal system that may limit adoption ofnew technologies. Nevertheless, as the figuresattest, these broad patterns tend to play outin the long run. 25

0.4 0.5 0.6 0.7 0.8

y relative investment prices (PWT)

JapanKoreaTaiwanChina

Japan

ivity growth. Japan 1960–2004, Korea 1971–2004, Taiwan

ome approach GDP data (with the component labor

relative investment prices (only) are from the PWT, with

and investment prices are relative to the United States.

Page 10: China's Economic Growth 1978-2025

1674 WORLD DEVELOPMENT

4. GROWTH ACCOUNTING 26

Growth accounting decomposes GDP growthinto growth of variables other than GDP, thatis, growth of GDP is ‘‘explained’’ by theweighted growth rates of other variables. 27 IfGDP growth and the growth of the variablesinto which it is decomposed exhibited a stablerelationship in the past, and if this stable rela-tionship is likely to continue into the future,then information about the future values ofthe variables into which GDP has been decom-posed allows derivation of future GDP growth.

One way to decompose GDP is to assume theexistence of an aggregate production functionand to explain GDP growth by the growth offactor inputs and of a residual, with the weightsof factor input growth given by their estimatedoutput elasticity. Estimation of an economy-wide production function for China in 1978–2002 yields insignificant coefficients, that is, inthe typical Cobb-Douglas production function,the output elasticities are not constant overtime. Perhaps for good reason, authors fromWang and Yao (2003) to Young (2003) didnot actually estimate a production function,but assumed weights. They assumed constantweights and set them approximately equal tothe factor shares; this implies the assumptionsof constant output elasticities (despite the evi-dence to the contrary) and perfect competition(unlikely to be met in a planned economy). 28

A different growth decomposition that is freeof assumptions follows from the definition ofGDP on the income side:

GDP� labor remuneration + depreciation +net taxes on production + operating surplus.

The data on these income components areavailable for China, for all years of the reformperiod, at the provincial but not at the nationallevel. Summing across provinces, and applyingthe sum provincial shares of the different incomecomponents to national GDP yield an approxi-mation of national income components. 29

Net taxes on production reflect taxes on pro-duction, such as the value added tax, less pro-duction subsidies, such as subsidies for policylosses, price subsidies for the grain system,and tax refunds for exporting enterprises. Thelargest part of this income component is likelyto depend on value added. Net taxes on pro-duction are in the following taken to be a func-tion of value added (GDP).

Depreciation depends on the value of the notyet fully depreciated fixed assets. It is in the fol-lowing taken to be a function of capital.

The operating surplus approximately reflectseconomy-wide business profit. 30 It is the returnon equity. Consolidating balance sheets acrossthe economy and eliminating the financial sec-tor so that deposits at banks combined withbank loans to enterprises translate into house-hold and firm claims on production units, oper-ating surplus constitutes the return on fixedassets, intangible assets, and inventories. Sinceinventories are not productive, operating sur-plus is largely a function of fixed and intangibleassets. With data on land values not available(and presumably even today much smaller thanthe value of fixed assets), only fixed assets re-main as feasible argument.

The national income identity, that is, the def-inition of GDP in the income approach, canthen be rewritten as

P tY t � wtLt þ dtKt þ stP tY t þ stKt; or ð1Þ

Y st � Y tð1� stÞ �

wt

P tLt þ ðdt þ stÞ

Kt

P t; ð2Þ

where P denotes the price level, Y the realGDP, w the wage rate, L the quantity of labor,K the nominal fixed asset stock (‘‘capital’’), anddt, st, and st the depreciation rate, the net taxrate on production, and the surplus rate. Tak-ing derivatives with respect to time, and thendividing by GDP less net taxes on production,while expanding right-hand side terms, yields

dY st

dt

Y st

�dY tdt ð1� stÞY tð1� stÞ

�Y t

dstdt

Y tð1� stÞ

�d

wtPt

dtwtP t

Lt

wtP t

Y st

þ wt

P t

Lt

Y st

dLtdt

Ltþ

ddtdt

dt

Kt

P t

dt

Y st

þdstdt

st

� Kt

P t

st

Y st

þ ðdt þ stÞd

KtPt

dtKtP t

KtP t

Y st

: ð3Þ

Using hats to denote growth rates and abbrevi-ating the shares of labor, depreciation, andoperating surplus in ‘‘GDP less net taxes onproduction’’ by as

t ; dst , and es

t lead to the incomegrowth identity

bY st � bY t �

dstdt

ð1� stÞ

� ast

bwt

P t

� �þ as

tbLt þ ds

t dt

þ est st þ ðds

t þ est Þ

bK t

P t

!: ð4Þ

Page 11: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1675

Like the behavioral growth accounting equa-tion derived from an assumed aggregate pro-duction function, the income growth identityrepresents one form of GDP decomposition,only now in the form of a definitional identity.Compared to the traditional growth accountingequation, the income version newly introducesthe growth rates of the real wage rate, of thedepreciation rate, of the surplus rate (or ‘‘rentalrate of capital’’), and of the net tax rate on pro-duction; in exchange, there is no room for aresidual. 31

In the period 1978–2002, nominal GDPdeflated by the implicit deflator as firstpublished grew by 827%, and nominal GDPless net taxes on production, deflated by the im-plicit deflator as first published, grew by814%. 32 The difference of only thirteen per-centage points implies that the changes in thenet rate of taxes on production are negligiblysmall. The income decomposition, using theaverage income shares of 1978–2002, that is,weighting factor growth with the mean of thefirst and last period’s income share values, isas follows:

bY st � as

t

bwt

P t

� �þ as

tbLt þ ds

t dt

þ est st þ ðds

t þ est Þ

bK t

P t

!814% ffi 0:5981 � 495%þ 0:5981 � 60%

þ 0:1370 � 16%þ 0:2649 � ð�50%Þþ 0:4019 � 1185%

ffi 296%þ 36%þ 2%–13%

þ 476% ¼ 797%

In the period 1978–2002, growth in constant-price capital accounted for approximately60% of real GDP growth (476/797) and growthin the real wage for 37%; growth in the quantityof labor contributed just below 5%, with theremainder due to very small positive and nega-tive contributions of changes in the deprecia-tion rate and in the surplus rate.

The income growth identity only holds withperfect accuracy in the instantaneous case whengrowth rates are infinitesimally small and thefactor shares at a given point of time thereforeare exactly applicable. Measuring growth ratesover 24 years introduces a discrepancy of 814%versus 797% due to the need to apply averagefactor shares. The discrepancy is small, thatis, using mid-period weights has little impacton the outcome.

The share of labor remuneration in nominalGDP less net taxes on production was ratherconstant in the period 1978–2002, with a meanof 0.5981 and a standard deviation of 0.0110(Figure 7); consequently, the weight of real cap-ital—unity less the labor share, or the sum ofthe shares of depreciation and operating sur-plus—was also near-constant.

The shares of depreciation and of the operat-ing surplus varied over time, but they appearindividually only in the products with thegrowth rate of the depreciation rate and ofthe surplus rate, and these products are rela-tively small. The product of the depreciationshare and depreciation rate growth in the in-come decomposition of GDP contributed only0.25% of GDP growth in 1978–2002 (2% di-vided by 797%), and the product of surplusshare and surplus rate growth only (negative)1.63%. These two terms, thus, can be ignoredwithout any significant impact on the explana-tion of GDP.

The growth decomposition of 1978–2002 sug-gests that when forecasting future GDPgrowth, (i) the growth of ‘‘GDP less net taxeson production’’ is a good proxy for real GDPgrowth and (ii) future GDP growth can be re-composed solely from the growth rates of thefuture real wage, labor, and real capital, that isbY t ffi bY s

t

ffi ast

bwt

P t

� �þ bLt

� �þ ðds

t þ est Þ

bK t

P t

!

¼ ast

bwt

P t

� �þ bLt

� �þ ð1� as

t ÞbK t

P t

!: ð5Þ

The future growth rates of the quantity of la-bor are readily available if one is willing tomake a few, plausible assumptions. Those aged15 in 2015 (the minimum working age) were al-ready born at the time of the year 2000 popula-tion census, that is, forecasts through 2015 arehighly reliable, and in as far as trends in repro-duction rates have been stable in recent years,forecasts through later years are likely to onlygradually lose reliability. 33

What remains is to obtain future growthrates of the real wage and of real capital. Ide-ally, these growth rates are related as much aspossible to variables whose future values are al-ready known today—the quantity of labor, thequality of labor, and age measures. When theunit of analysis is the individual, these are nat-ural explanatory variables for the real wage. 34

Page 12: China's Economic Growth 1978-2025

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Labor remuneration

Depreciation

Operating surplus

Figure 7. Income factor shares in ‘‘GDP less net taxes on production,’’ 1978–2003. The values of all income components

are by necessity derived from provincial-level data published in GDP 1952–95, GDP 1996–2002, and for the years since

the mid-1990s in the corresponding year’s Statistical Yearbook. 2004 values are not available. For each year, the share of

each income component in ‘‘GDP less net taxes on production’’ is obtained as the sum across provinces of the values of a

particular income component divided by the sum across provinces of ‘‘GDP less net taxes on production.’’

1676 WORLD DEVELOPMENT

Since they are the only hard facts that we knowabout the future, they are also used to explainreal capital. This is plausible at least for thequality of capital, because more sophisticatedcapital is likely to require a higher level of laborquality. To some extent, it may even be plausi-ble for the quantity of capital, if agriculture re-quires little education and little capital, andthose who move into other sectors with a high-er capital-labor ratio are the most educated.

Examining three dozen potentially wage-re-lated variables in the period 1978–2002, fromaverage years of schooling and share of labor-ers with a particular level of schooling (pri-mary, lower middle, upper middle, college,university, post-graduate) to age characteristicsand potentially relevant non-labor variables,such as the share of foreign direct investmentin total domestic investment, most of these ser-ies are unit root processes.

The solution then is to find a long-run rela-tionship using the time series data of the period1978–2002. Table 2 reports two long-run rela-tionships embedded in vector error correction(VEC) models, one with variables in levels,the other with variables in logarithms. Thesame procedure was used for real capital andthose results are also reported in Table 2. 35

The first significant long-run relationship(cointegrating equation) reported in the tableis between the real wage, the share of laborerswith upper middle school education, and the

share of laborers with college level educationor above. For simplicity of exposition, therelationship is expressed with the wage asleft-hand side variable. The coefficient signsare plausible and the coefficients are highlysignificant. Forecast error variance decomposi-tion shows that much of the movement in thewage series is due to its own shocks, but overa 10-year forecast horizon, 22% of the forecasterror variance of the wage series is due toshocks to the share of laborers with collegelevel education or above. 36 The other threecointegration equations reported in the tablealso have plausible coefficient signs and thecoefficients are highly significant. 37 The twocases with variables in natural logarithmscome with strong Granger causality from theeducation variables to the wage and capitalseries, and over a 10-year forecast horizonthe forecast error variance of capital is mostlyaccounted for by shocks to one of the educa-tion variables. 38

The first four data columns of Table 3 reportthe average annual growth rates (over five-yearperiods) that are obtained by fitting the fourcointegration equations of the previous tableto the forecast labor quantity and quality val-ues through 2025. When the cointegrationequations are estimated in logarithms, theresulting growth rates tend to be higher, espe-cially far into the future. The forecast quantityof labor series (fifth data column) has a falling

Page 13: China's Economic Growth 1978-2025

Table 2. Cointegration equations, 1978–2002/03

‘‘Dependent’’ variable All variables

in levels

All variables in

natural logarithms

A. Wage B. Capital A. Wage B. Capital

Regression number I II III IV

Constant 4387.96 4144.38 6.2426 7.6163

C. Average years of schooling

minus 5

0.7171 0.2829(12.84) (3.18)

D. Average years of schooling

minus 5, squared

3126.44(23.20)

E. Share of laborers with primary

school education

�1.3593(�5.02)

F. Share of laborers with upper

middle school education

�39066.18(�4.07)

G. Share of laborers with college

level education or above

152080.0(18.80)

H. Number of laborers �13973.42

(�2.55)

I. Trend 238.63 0.0946

(2.39) (15.23)

Lag length of the VEC 2 2 1 2

Trace, maximum eigenvalue test (%) 1, 5 1, 1 1, 1 5, 5

Normality (10% level) OK OK C 1%, E 10% OK

Granger causality ‘‘Y: X (Z%)’’ None H:D(10) A: E(0.1) B:I(1)

C: E(5)

Variance decomposition period 3 A: 97/1/2 B: 87/11/2 A: 82/3/15 B: 87/13

F: 34/66/0 D: 19/77/5 C: 0/94/6 C: 8/92

G: 1/78/21 H: 83/1/17 E: 0/61/38

Variancedecomposition period 10 A: 66/12/22 B: 29/71

F: 16/81/3

G:19/60/21

B: 25/70/5 A: 65/4/31C: 6/94D: 23/72/5

H:33/45/22

C: 11/89/10

E:13/60/27

Number of observations 23 22 24 22

Wage per laborer: labor compensation in the national income accounts divided by the number of laborers including military

personnel (in yuan per laborer); deflated by the GDP deflator (in 2000 prices).

Capital: measure of capital’s contribution to production, in billion yuan, deflated by the GDP deflator (in 2000 prices).

Number of laborers is in billion.

Values in parentheses underneath coefficient estimates are asymptotic t-values (with standard errors corrected for degrees of free-

dom). All coefficients are significant at the 1% level, except the ones of H and I in the second data column (significant at the 5% level).

Trend: the cointegrating vector always includes a constant (as does the VAR); depending on the results of the trace test and the

maximum eigenvalue test, a trend is included or not included.

The logic behind the use of the variable ‘‘average years of schooling minus five’’ (squared in one case) is that a primary school

education is standard, and wage differentiation then depends only on the number of years of schooling in excess of primary school

education. While primary school lasts for six years, the 1978 average years of schooling is just below 6 years; in order to have a

positive variable throughout (necessary to take logarithms), 5 years is used as a benchmark.

Trace, maximum eigenvalue test: significance level of the existence of one eigenvector.

Normality in VEC equations: Jarque-Bera values are used (based on Cholesky (Lutkepohl) orthogonalization). When normality is

rejected for one of the VEC equations at the 10% level, the significance level is stated together with a letter which identified the

variable that, in first differences, is the dependent variable of the particular equation.

Granger causality ‘‘Y:X (Z%)’’ means that ‘‘Y is Granger-caused by X (at the 0.1%, 1%, 5%, or 10% significance level).’’

Variance decomposition: a Cholesky ordering of variables as listed above in the particular table column was used, and the forecast

error variance decomposed annually over periods of 3 and 10 years. Values are percentages, ordered as the variables appear in the

column. For example, if the three variables involved are A, C, and D, and the variance concerned is that of A, then A/C/D % of the

variance of A is explained by variation in A, C, and D. Percentages may not add up to 100 due to rounding.

CHINA’S ECONOMIC GROWTH 1978–2025 1677

Page 14: China's Economic Growth 1978-2025

Table 3. Growth forecasts (average annual growth rates in %)

Individual variables GDP

Wage Capital Wage Capital Labor Eqn. (5) Labor shareassumed fixed

(I) (II) (III) (IV) (V)

I, II, V III, IV, V I, V III, V

2000–05 4.15 6.98 5.18 10.91 0.83 5.80 8.09 5.01 6.052005–10 8.18 6.37 6.29 10.85 0.71 7.84 8.83 8.95 7.052010–15 7.76 5.63 7.81 10.75 0.20 7.06 9.42 7.98 8.032015–20 6.36 5.40 11.79 10.71 �0.80 5.50 10.85 5.51 10.902020–25 4.63 4.14 14.22 10.54 �0.62 4.06 12.02 3.98 13.512000–25 6.20 5.70 9.01 10.75 0.06 6.05 9.83 6.27 9.07

Cumulative

2000–25 350.04 299.69 763.49 1184.75 1.58 333.77 942.78 357.15 777.14

The numbers I–IV refer to the previous table. All calculations are done annually before averaging to five-year or 25-year values. The antilogarithm was applied to the annual forecast values of the wage rate (III) and capital (IV) beforecalculating growth rates.GDP calculated via the definition of the labor share, with an assumed fixed labor share in GDP including net taxes

on production (of 0.5138, year 2000), uses one of the two wage series plus the labor series to calculate annual GDPand then the five-year average GDP growth rate. (Alternatively, one could simply add up the wage and labor growthrates as provided in the table.) GDP calculated using Eqn. (5), in addition, uses one of the two capital series.When calculating the year 2001 GDP growth rate using Eqn. (5), the actual year 2000 labor share in GDP less net

taxes on production (0.5985) is used to weight the 2001 (over 2000) growth rates of the wage, of labor, and of capitalto obtain 2001 GDP. The 2001 labor share is then calculated from the 2001 labor remuneration and GDP values, andused in the calculation of the 2002 GDP growth rate, etc. Using variables I+II, the minimum labor share in theperiod 2000–25 is 0.5741, the maximum labor share is 0.6336; using variables III+IV, the two values are 0.4615 and0.5985.For comparison, the average annual GDP growth rate in 1978–2003 was 9.37% (using the official real GDP growth

rates) or 9.72% using the latest revised nominal GDP data combined with the first published, implicit GDP deflator.

1678 WORLD DEVELOPMENT

positive growth rate that eventually (by 2014)turns negative.

Future GDP growth can be obtained by sum-ming the weighted growth rates of wages, cap-ital, and labor (sixth and seventh datacolumns). In each year, the previous-year valuesof the tax-adjusted labor share, and of oneminus this share, are used as weights; that is,weights are allowed to change in accordancewith the growth forecast.

In 2000–05, depending on which cointegra-tion equation is used (with variables in levelsor in logarithms), the resulting average annualGDP growth rate is 5.80% or 8.09% (sixthand seventh data columns shown in the table).With actual real growth in this period ofapproximately 9.5%, the second series, withthe cointegration based on variables in loga-rithms, seems more appropriate. In the periods2005–10 and 2010–25, the two series yieldrather similar estimates, in the range of 7.06–9.42%. At this level of economic growth, theextrapolations of Table 1 suggest that Chinawill surpass the US economy in size, in pur-chasing power terms, by around 2010.

Further into the future, the predictions di-verge significantly, with one GDP growth seriesfalling off, and the other rising. Due to the re-cent explosion in education, the out-of-sampleforecasts use education values that are far fromsample average values. The comparison of thetwo GDP growth series shows well the degreeof uncertainty involved in predicting furtherinto the future than 2015.

An alternative approach to predicting GDPgrowth is to assume a constant labor share.This appears plausible given the worldwideexperience that labor shares tend to be constantover time. 39 In China, the share of labor inGDP (now not less net taxes on production)has been near-constant throughout the reformperiod, with a mean of 0.5210 and a standarddeviation of 0.0111 in 1978–2002, that is, thecoefficient of variation is only 2%. Given thedefinition of the labor share as labor remuner-ation divided by GDP, that is, as

at �wtP t

Lt

Y t; ð6Þ

Page 15: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1679

assuming the labor share at to be constantyields the growth rate of real GDP asbY t �

wt

P tþ bLt: ð7Þ

That is, under the assumption of a constant la-bor share, real GDP growth, by definition, isthe sum of real wage growth and labor growth.Neither capital series nor weights are needed.For the period 1978–2002, the relationship be-tween the real wage and the quantity and qual-ity of labor was identified in Table 2 and can beapplied to the data on the future quantity andquality of labor constructed through 2025. Ta-ble 3 in the last two columns reports the result-ing real GDP growth rates. They carry no newor different insights but confirm the findingsfollowing the more elaborate decompositionof GDP growth by income categories.

The above growth forecasts rely solely onfacts about the future quantity and quality oflaborers in China, facts that are near-certain to-day, and some assumptions. The assumptionsare that the past relationship between wageand labor variables (and, in the case of the in-come growth identity, also between capitaland labor variables) continues to hold in the fu-ture, that the annual change in the net tax rateon production continues to be relatively small,and that, in the case of the income growth iden-tity, the products of the depreciation share anddepreciation growth rate and of the surplusshare and surplus growth rate continue to berelatively small.

The projections, by necessity, reflect onlylong-run trends. The model does not allow forbusiness cycles or any other sources of volatil-ity. Koren and Tenreyro (2007) in a cross-coun-try study found that as countries develop, theirproduction structure moves from volatile toless volatile sectors and the volatility of coun-try-specific macroeconomic shocks falls, bothof which would add credibility to the long-runtrends obtained here for China.

The long-run trends in the income growthidentity are driven by the number of laborersand human capital formation. Other variables,from average age to non-demographic vari-ables, do not exhibit a long-run stable relation-ship with the real wage (or capital) in 1978–2002. With little change in China’s labor forceover the next 20 years, the quality of labor be-comes of key interest. The next section exam-ines in more detail the most recentdevelopments in human capital formation inChina.

5. HUMAN CAPITAL

Change in human capital formation has beenrapid in reform period China. A comparisonwith the United States puts China’s human cap-ital measures into perspective, while the devel-opment of education in China over timereveals the scale of changes underway.

Table 4 compares the educational level of theChinese versus US population as reported inthe population censuses of 1990 and 2000 inboth countries. 40 As a percentage of the totalpopulation, a far higher proportion of the USpopulation has achieved a secondary or tertiarylevel of education. For example, in 1990, 75.2%of the US population had completed highschool (or above), but only 9.52% of the Chi-nese population had; 20.3% of the US popula-tion had completed a bachelor’s degree (orabove), but only 0.56% of the Chinese popula-tion had. But because the US population in1990 was only 28% of the size of the Chinesepopulation (in 2000, 24%), the differenceshrinks by a factor of approximately four oncetotal population numbers are considered.

During 1990–2000, China narrowed the rela-tive gap. While in 1990 the number of US citi-zens with high school education was 1.65 timesthe number in China, by 2000 this ratio hadfallen to 1.17 (last column shown in Table 4).For the BA degree, the ratio fell from 7.59 to4.33. At the level above the bachelor’s degree,the ratio in 2000 was still 21.42 in favor ofthe United States (with no such figure availablefor 1990).

The population total of China masks drasticdifferences in education levels of different agecohorts. Figure 8 shows that in 2000 more thanhalf the population aged 65 or above had a le-vel of education below primary school, while ofthe age cohort 20–29 years, only 2.06% had; forsome form of secondary education, the two per-centages are 10.55% (age 65 or above) versus69.44% (aged cohort 20–29), and for tertiary le-vel education 1.50% versus 7.53%. The age co-horts in between reveal a smooth transitionfrom the less educated older generation to theever better educated younger age cohorts. 41

The Chinese age cohort of 20–29 years in2000, at the primary and secondary school lev-els, comes close to the United States populationaverage, but in tertiary level education still fallsfar short.

The picture changes drastically if one consid-ers current education data. In terms of current-year new enrollment at the BA level (including

Page 16: China's Economic Growth 1978-2025

Table 4. Educational level (or higher) of population aged 25 or older (Census data)

This degree or above in % Absolute number(mio.)

Ratio UnitedStates/China

United States China United States China

2000

High school 80.4 16.5 146.498 125.137 1.17Some college 51.8 94.386College-level associate degree 4.3 32.457 2.91Bachelor’s degree 24.4 1.4 44.460 10.258 4.33Advanced degree 8.9 0.1 16.217 0.757 21.42Reference: total population aged 25 or higher 182.212 760.480 0.24

1990

High school 75.2 9.52 119.469 72.421 1.65Some college 45.2 71.809College-level associate degree 1.51 11.517 6.24Bachelor’s degree 20.3 0.56 32.250 4.249 7.59Advanced degree 7.2 n.a. 11.439 n.a. n.a.Reference: total population aged 25 or higher 158.868 571.589 0.28

For the classification, see appendix. Advanced degree holders in the United States refer to the three degrees aboveB.A., in China to Master’s or Doctorate degree holders. No data on advanced degree holders in China for 1990 areavailable. If Bachelor’s degree and above in the case of the United States were compared to College-level associatedegree and above in China, the ratio of the United States to China in 1990 is 2.80 (32.250/11.517), and in 2000 it is1.37 (44.460/32.457).United States data report degree obtained, while Chinese data include those who have already obtained the

particular degree as well as those who currently study for the particular degree. Since the US age classification islimited to those aged 25 and older (and the Chinese classification has been reduced here to the US classification), thenumber of Chinese who are currently studying toward a particular degree should be small, except possibly in the caseof advanced degrees.Sources: United States: Educational Attainment 2000. China: Census 2000, pp. 593–602; Census 1990, Vol. 1, p. 380,Vol. 2, pp. 112f. and 132f.

20-2930-39

40-4950-60

60-6465+

No school/ lit. class

Primary school

Secondary school

Tertiary level

0

20

40

60

80

%

Figure 8. Share of educational level in individual age groups, China, 2000 (in %). ‘‘No school / lit. class’’ refers to no

schooling or only basic literacy class (below primary school level). Source: Census 2000 (nationality), pp. 125–127.

1680 WORLD DEVELOPMENT

Page 17: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1681

college-level associate degrees) China has nowcaught up with the United States (Figure 9).In the reform period, new enrollment in Chinafirst languished, but began to expand graduallyin the early 1980s and then rapidly after 1998.In 1998, new enrollment was at approximatelytwice its 1978 level. In the course of the follow-ing seven years, during 1998–2005, new enroll-ment quintupled. Since 2001, new enrollmentin Chinese BA programs at (only) regular insti-tutions of higher education exceeds the numberof freshmen in the United States. In 2005, Chi-na’s new enrollment figure was twice that of the

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

1978 1981 1984 1987 1990 1

mill

ion

China: BA+ Associate degre

U.S.: Freshmen (2- and 4-ye

China: MA+ PhD --- right sca

Figure 9. Tertiary level new enroll

‘‘BA’’ in the case of China refers to (current-year) new enro

includes BA and college-level associate degree programs of

enrollment in college-level associate degree programs of ‘‘non-

the United States, only data on ‘‘first-time freshmen’’ are avai

year programs, presumably associate and BA programs, and

graduation figures in US MA/PhD programs suggest new enro

year (assuming no drop-outs). In China, in 2005, 54.35% of

was in regular BA programs (benke), while the rest was in co

46.86% of new enrollment in regular institutions of higher educ

2004, the latest year for which the data are available, 62.10%

in 4-year (versus 2-year) programs, and 56.54% of total first

in 4-year pr

Data for the United States for the years through 1995 cover ‘

granting institutions’’ (which include some additional institu

education institutions that did not award associate or higher

data are available for 1996 and 1997: data in the latter classifi

earlier class

Sources: China: Statistical Yearbook 1990, pp. 708f.; 2003,

800f.United States: <http://nces.ed.gov/programs/digest/d05/

15 20

United States, and in 2006 it was scheduled torise by another 5%. 42

Data on the joint category of MA and PhDnew enrollment in China (also Figure 9) followthe same trend as new enrollment for the BA/college-level associate degree. Lacking data onnew enrollment at the graduate level in the Uni-ted States, and therefore assuming that the1998 US MA and PhD graduation number of476,174 and the 2005 number of 609,200 are agood indicator of current new enrollment, USenrollment numbers at the graduate level wereat least seven times the Chinese figure of

993 1996 1999 2002 20050.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.55

million

e

ar programs)

le

ment, China, and United States.

llment in regular institutions of higher education, that is,

regular institutions of higher education; no data on new

regular’’ institutions of higher education are available. For

lable; the coverage is unclear, but seems to be 2-year and 4-

possibly including professional degree programs. Stable

llment in these programs of at least 400,000 to 500,000 per

total enrollment in regular institutions of higher education

llege-level associate degree programs (zhuanke); but only

ation was in regular BA programs. In the United States, in

of total fall enrollment in degree-granting institutions was

-time freshmen enrollment in undergraduate programs was

ograms.

‘institutions of higher education,’’ and since 1996 ‘‘degree-

tions, primarily 2-year colleges, and exclude a few higher

degrees); for first-time freshmen enrollment, both sets of

cation are 3–4 percentage points larger than the data in the

ification.

pp. 720f.; 2004, pp. 780–2; 2005, pp. 692f.; 2006, pp. 798,

tables/dt05_179.asp>, and . . ./dt195.asp, accessed on Nov.

06.

Page 18: China's Economic Growth 1978-2025

1682 WORLD DEVELOPMENT

72,508 in 1998, but perhaps only one-thirdabove the Chinese figure of 364,831 in 2005.

Graduation numbers mimic enrollment num-bers with a time lag (Figure 10). The absolutenumber of undergraduate level graduates inChina began to exceed that of the United Statesin 2004, and by 2005 it was 50% higher. 43

Based on the enrollment numbers of 2005(and the significant drop-out rates in the UnitedStates), by 2008/09 the number of undergradu-ate level graduates in China should be 2.5 timesthe US number. At the MA and PhD level(with only joint data available for China), theUnited States in 1998 had a 10-fold lead inabsolute graduation numbers (Figure 10). 44

But by 2005, the gap narrowed to a threefoldlead, and by 2008, given the 2005 enrollmentnumbers, may shrink to a less than 30–40%lead. 45

The explosion in education is only happeningsince the late 1990s. Much of the upgrading inthe education level of Chinese laborers may

0.0

0.4

0.8

1.2

1.6

2.0

2.4

2.8

3.2

1978 1981 1984 1987 1990

mill

ion

China: BA + Associate de

U.S.: BA + Associate & Pr

China: MA + PhD

U.S.: MA + PhD

Figure 10. Tertiary level gradua

For the coverage of ‘‘BA’’ in China, see the previous figure.

degrees, BA degrees, and first professional degrees. If first pro

the United States ‘‘BA’’ category would have been (projected

would have been larger by 13.95%. US data for 2005 are offi

through 1995 cover ‘‘institutions of higher education,’’ and sin

Figure

In China, in 2005, 47.78% of the ‘‘BA + Assoc.’’ graduates

degree. In the United States, in 2005, BA degrees are projec

668,00

Sources: China: Statistical Yearbook 1990, pp. 708f.; 2003,

801.United States: <http://nces.ed.gov/programs/digest/

not be felt until some years later, when freshgraduates have gained some job experience,and not all the additional graduates may beas qualified as the average graduate of earliergenerations.

Head count data do not necessarily translateinto comparable accumulation of human capi-tal because they reveal nothing about the qual-ity of education. 46 One may suspect thateducation in China contains much ‘‘rote learn-ing’’ or is dominated by a ‘‘cult of authority,’’but how this translates into GDP growth atChina’s current stage of economic developmentis unclear. Education in China through the sec-ondary school level may not be all that inferiorto the United States. Even at the tertiary level,a provincial railway ministry college in Chinacould match a US community college. At theMA and PhD level, the conclusion is likely tobe less ambiguous; but at that level Chinese stu-dents probably make good use of the US insti-tutions.

1993 1996 1999 2002 2005

gree

ofessional degree

tes, China and United States.

‘‘BA’’ in the case of the United States refers to associate

fessional degrees had been included with MA/PhD, in 2005

to be) smaller by 3.92%, and the US MA/PhD category

cial projections. Data for the United States for the years

ce 1996 ‘‘degree-granting institutions.’’ (Also see notes to

9.)

received a BA degree, the others a college-level associate

ted to outnumber associate degrees 2:1 (1,416,000 versus

0).

pp. 720f.; 2004, pp. 780-2; 2005, p. 693.; 2006, pp. 798,

d05/tables/dt05_246.asp>, accessed on Nov. 15 2006.

Page 19: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1683

6. CONCLUSIONS AND IMPLICATIONS

Extrapolation of China’s reform period eco-nomic growth into the future suggests that thesize of China’s economy will exceed that ofthe United States, in purchasing power parityterms, by about 2010. Per capita, the point oftime when China catches up with the UnitedStates is much further into the future, 20 to50 years from now, although the coastal areas,especially in the past the fastest growing fiveprovinces, with a population exceeding that ofthe United States, may catch up in as little asone to two decades.

China’s economic development in the reformperiod fits well with the broad developmentpatterns of structural change, catching up,and factor price equalization, not least in com-parison with other Asian countries earlier intheir economic development. Following allthree patterns, China faces another 30 yearsof continued growth. Re-composing China’seconomic growth from growth in income com-ponents indicates economic growth during2005–15 in the range of 7–9%, high enoughfor China to catch up with the United States,in purchasing power terms, within about fiveyears.

In the growth decomposition, human capitalhad significant predictive power. Commonsense similarly suggests that if talent is ran-domly distributed among the world populationand if China’s education system is able to iden-tify the brightest students, then China has a lar-ger pool of talent to draw from than any othercountry in the world—for every ‘‘genius’’ in theUnited States, China potentially has four. 47 Ifinnovation depends on the agglomeration oftalent (geographically, nationally, culturally,or linguistically), then China is in an excellentposition to innovate and grow. The explosionin tertiary level education in China may yetsupply the productivity gains that Paul Samuel-son identifies as the potential causes of reduc-tions in the United States per capita realincome. 48

China’s talent may initially be focused on afew projects or industries, such as the state-sup-ported space program. In industry, it is onlywhen profits can no longer be made by usingexisting technologies or upgrading to knowntechnologies that pressure for indigenous inno-vation sets in. Because not all industries sectorsdevelop at the same pace, China is unlikely totechnologically advance on all fronts simulta-neously.

Demographics also matters in terms of mar-ket size—and 80% of China’s population livesin the densely populated Eastern part of thecountry. Size of the domestic market should al-low unprecedented variety and economies ofscale. Economies of scale lead to low cost. Alarge number of suppliers suggest a high degreeof competition and pressure on prices; eventu-ally, decreasing profit margins create incentivesto innovate in order to either lower costs fur-ther or be able to offer a more differentiatedproduct with higher profit margins. The pricewars of the late 1990s are evidence of the degreeof price pressures Chinese markets are capableof. 49

One immediate implication of China’s eco-nomic growth on people’s livelihood is thatone-fifth of the world population will con-tinue to experience significant improvementsin their living standard. A share of China’spopulation that exceeds the size of the USpopulation will enjoy living standards closeto the level of developed countries in the nearfuture. In the Chinese hinterland, the rise outof poverty continues. Consumption wavessimilar to those previously experienced inthe West are unfolding in China, albeit withsome variations, such as the early introduc-tion of the cell phone. In production, withfirst labor shortages in Guangdong provincein 2005, wages of unskilled labor have begunto rise, and labor standards are increasinglybeing enforced. From the creation of capitalmarkets to the establishment of a country-wide social security system, what has takenthe West centuries has been condensed intodecades in China.

Internationally, the growth of China impliesthat the center of world economic activity shiftsto an Asia that besides China also includes In-dia, South East Asia, Japan, and Korea.Shocks to the United States and Europeaneconomies, or stagnation of these economies,will have less and less impact on Asia that con-tinues to integrate. In 2005, China’s ratio of‘‘exports of goods and services’’ to GDP was37%, compared to 10% for the United States. 50

Of China’s exports of goods in 2005, 21% wentto the United States and 22% went to Europe,but 48% went to Asia; 67% of China’s importscame from Asia. 51

The West plays an important entrepreneurialrole in China. In 2004, the most recent year forwhich the data are available, 19% of grossoutput value of industrial enterprises with an-nual sales revenue in excess of 5 m yuan

Page 20: China's Economic Growth 1978-2025

1684 WORLD DEVELOPMENT

RMB (USD 0.6 m) was produced by foreign-funded enterprises, a figure which excludes anadditional 11% in enterprises funded by HongKong, Macao, and Taiwanese entrepreneurs, 52

that is, non-domestic entities control one-thirdof Chinese industry. To the extent that Westernpension funds invest in these firms, often multi-national corporations, Western citizens directlybenefit from China’s continued economicgrowth.

The foreign-funded industrial enterprisesalone, with their approximately one-fifth shareof industrial output in China, accounted for58% of China’s exports in 2005 (and 59% ofChina’s imports). 53 This suggests that China’sexports to the West largely reflect the West‘‘importing’’ from ‘‘itself,’’ where the West’snationals, or firms incorporated in the West,use Chinese labor in their productionprocesses.

China’s economic growth affects relativeprices around the world, thereby promotingan increasing division of labor and specializa-tion worldwide. This implies structuralchanges in countries around the world. Therestructuring extends beyond the traditionalrealms of manufacturing, to services; for exam-ple, the newly rich in China’s coastal provincesare potential tourists for the West, while back-room banking services may gradually shiftfrom the West to China. China’s currentstrength in unskilled labor and mass produc-tion implies a wage constraint for unskilledlaborers employed in Western industries pro-ducing tradables.

China’s economic growth is in some waysa rerun of the industrial revolution in theUK two hundred years earlier, but at amuch swifter pace. It is accompanied bysimultaneous changes in the patterns of pro-duction, ownership, and sources of income ineconomies around the world. The concept ofa national economy, and, following fromthat, of a new economic ‘‘superpower,’’ ap-pears far-fetched; China is integrated intothe world economy to an extent that is sim-ilar to that of a European country withinEurope. What emerges is one world economyorganized along dimensions of comparativeadvantage, where the construct of a ‘‘nation’’is, in the economic realm, reduced to issuesof enterprise registration, regulation, and tax-ation.

One may wonder if China will not, oneday, be overwhelmed by obstacles to further

economic growth, ranging from bad loans inthe banking system to an under-funded pen-sion insurance scheme, increasing incomeinequality, the lack of a rural health care sys-tem, bankrupt local governments, and envi-ronmental pollution. In 1998, Lardy stressedthe large economic problems and the unprec-edented potential for social unrest due to theever more indebted state-owned enterprises,the extent of nonperforming loans, and a de-cline in government revenue. By 2005, thestate-owned enterprises have largely been re-formed, the problem of nonperforming loanshas been addressed, and government revenueis rising. There is a rationale for the Chineseleadership to not try to shoot for optimaltransition in every respect (not feasible, andtoo costly), but to let problems emerge andin part resolve themselves, and to step in onlywhen the need for action becomes urgent.Where Western observers tend to predict Chi-na’s doom, China’s leadership muddlesthrough by trial and error. The same patternappears to be repeated yet again when itcomes to inequality, social security, or envi-ronmental pollution.

Income inequality has now reached the top ofthe agenda of policymakers and first signs haveemerged that some aspects of it may be aboutto go into reverse. 54 Reform of the urbanhealth care and pension system is underway.The systematic construction of a rural socialsecurity system that includes health care andpension has begun.

The current five-year plan (2006–10) listsobligatory targets to reduce the use of water(per unit of value added by 30%) and the vol-ume of effluents (overall reduction by 10%,with specific targets to be set by the StateEnvironmental Protection Administration).The Five-Year Plan also stipulates an obliga-tory reduction of energy use per unit ofGDP by 20%. 55 Environmental protection,energy reduction, and land conservation tar-gets have been written into cadres’ evaluationsystem as obligatory. With little success inenvironmental preservation and energy conser-vation in the first year of the current Five-Year Plan (2006), the year 2007 has startedwith a plethora of detailed plans, regulations,and exhortations that include targets handeddown to individual enterprises. There can beno doubt that drastic action is being taken.How successful that action will be remains tobe seen.

Page 21: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1685

NOTES

1. Broomfield (2003) provided an overview over the‘‘China threat’’ literature. Lubman (2004) documentedthe negative views of China in the US congressionaldebates.

2. The quotes are from ‘‘The China Price’’ in Business

Week, 6 December 2004 (issue 3911), pp. 102ff.

3. Statistical Yearbook (2006, pp. 734, 1025).

4. See The Economist, November 15, 2003, p. 67,quoting the Bank Credit Analyst, a Canadian researchfirm.

5. Perkins and Rawski (2006) perhaps came closest.They rely on the growth accounting equation derivedfrom the Cobb-Douglas production function, usingincome shares as weights. Their labor calculations aresimilar to those made here; their future capital series areassumed series.

6. This paper is also not concerned about how mean-ingful the concept of GDP data is for measuring ‘‘well-being’’ or ‘‘standard of living.’’ One may furtherquestion if value added created in China contributes asmuch to well-being as does the value added created inthe United States, or vice versa. Since the use of GDP asa measure of economic size, or, in per capita terms, as aproxy for standard of living, dominates in the literatureas well as in popular discussions, this paper focuses onGDP.

7. See Statistical Yearbook (2006, pp. 57, 734); <http://www.bea.gov/bea/newsrel/gdpnewsrelease.htm> (ac-cessed November 30, 2006).

8. The GDP data are those published by NationalBureau of Statistics in China (for China) and thosepublished by the Bureau of Economic Analysis in theUnited States (for the United States). The PWT alsoinclude GDP values; in the case of China these incor-porate manipulations of the official values in accordancewith adjustments to official values suggested by Madd-ison. Holz (2006a) argued that Maddison’s adjustmentsare not justified (and questions Angus Maddison’s(2006) reply in Holz (2006c)).

9. The PWT 6.1 for the year 2000 (the most recent yearcovered in the PWT 6.1) report a factor of 4.32; in thePWT 6.2, for 2000, it is 4.22. This suggests a rather smallmargin of error, except that the PWT 6.2 values seem toincorporate very little new information.

10. One could allow the future real growth rate to differfrom the past one. For example, assuming a 7% realgrowth rate for China (instead of the 1978–05 average of9.64%) and a 3% real growth rate for the United States(instead of 2.97%), China’s GDP exceeds that of theUnited States by 2015; with a 3% RMB appreciation thischanges to 2011. Per capita, reducing the growth rate bythree percentage points results in the years 2059 coun-trywide, 2033 for the coastal provinces, and 2027 for thefive fastest-growing provinces (or, with RMB apprecia-tion, 2035, 2023, and 2020). In contrast, increasingChina’s aggregate real growth rate to 12% (or increasingthe per capita growth rate by two percentage points)yields 2010 (2029, 2020, 2018) or, with RMB appreci-ation, 2009 (2023, 2017, 2015). The impact of suchchanges to the growth rate is small in the short, butsubstantial in the long run.

11. One particular alternative would be the projectionsbased on nominal values combined with assumptionsabout the price adjustment factor and/or the exchangerate.

12. An appendix with further details on extrapolationsis available, as all appendices are, at <http://iho-me.ust.hk/~socholz/Growth/Appendices-Holz-China-Growth-2025–3July07.pdf>.

13. In a different approach, Woo (1999) examined theresults of sector-level reforms in China, thus breakingdown the case of China into several sectoral observa-tions (which, however, are not independent). Rawski(1999) traced the interaction between different reforms;in total, these are taken to add up to the reform success.The World Bank (1997, p. 19) listed a variety of‘‘strengths and advantages’’ considered to promoteeconomic growth, and a variety of ‘‘risks and chal-lenges’’ considered to hamper economic growth.

14. The first two are standard fare in developmenteconomics; see, for example, Cypher and Dietz (1997,pp. 27ff, 267, 403). The third is a basic theorem in tradetheory as a corollary of the Heckscher-Ohlin theorem;see, for example, Salvatore (1998), chapter 5.

15. The individual data sources, with explanations, aredocumented in an appendix, which also offers somefurther thoughts on the inclusion of other countries (seenote 12 for the web location of the appendix).

16. Total and sectoral employment data for Chinaexperience a statistical break during 1989–90; total

Page 22: China's Economic Growth 1978-2025

1686 WORLD DEVELOPMENT

employment, following the 1990 population census,increases by 17.03%. In the following, total employmentdata for the years prior to 1990 are adjusted; for details,see Holz (2006d), for a summary explanation see anappendix. Sectoral employment data and the share ofemployment in the primary sector are the officialvalues.

17. Figure 2 suggests that the typical annual rate ofstructural change contributes approximately three tofour percentage points to China’s annual growth rate.

18. While doubts are occasionally raised in the litera-ture about the accuracy of the Chinese data on thenumber of laborers in agriculture, suggesting that theofficial data constitute an over-count, the 1996 agricul-tural census (with a distinction between pure farmersand part-time farmers) as well as the 1990 and 2000population censuses (with unambiguous definitions thatshould lead to a proper counting of migrant laborers)implies that the official primary sector labor data, usedhere, are, if anything, an underestimate.

19. Among the three patterns, the direction of causalityis perhaps most ambiguous in the case of ‘‘structuralchange.’’ One could probably also argue for laborproductivity growth to cause structural change; a morespecific argument would be for labor productivitygrowth in non-agriculture that is different from laborproductivity growth in agriculture to cause structuralchange. On the other hand, if much of rural–urbanmigration happens in response to a reduction inconstraints on such migration, then the past andongoing relaxation of these constraints would explainmigration and, that, in turn average economy-wide laborproductivity growth (also see Holz, 2006b). No matterwhich direction the causality runs (probably in both), onboth accounts there is pressure for the pattern tocontinue to unfold.

20. A second-degree polynomial is imposed on thetrend lines shown in Figure 4 and in the followingfigures. In contrast, a linear trend was used in Figure 2because one extreme outlier in the cases of Taiwan andJapan would have resulted in (misleadingly) sharpinverse-U shaped curves for these two countries. Thesensitivity of second-degree polynomials to some typesof outliers is also seen in the upturn of the Japanesetrend shown in Figure 4.

21. If the scope for catching up were measured by thelabor productivity differences in the industrial sectoronly (rather than economy-wide), China’s highest levelof catching up would probably be slightly higher, butnot yet in the range where the corresponding growthrates of real GDP per laborer fall off significantly.

22. As Abramovitz (1986) elaborated, catching updepends, more broadly, on the ‘‘social capability’’ tocatch up and on the factors controlling the rate ofrealization of the potential for catching up (facilities forthe diffusion of knowledge, conditions facilitating orhindering structural change, and macroeconomic andmonetary conditions). In all these respects, China’scapabilities have been (and are still) improving.

23. These assumptions are two countries; two com-modities (goods); two factors of production; same,constant returns to scale technology in use in bothcountries; across both countries one commodity is laborintensive, the other capital intensive; incomplete special-ization in production in both countries; tastes are equalin both countries; perfect competition in both goods andfactor markets in both countries; perfect labor mobilitywithin each country but no international factor mobility;no transportation costs, tariffs, or other obstructions tothe free flow of international trade; all resources are fullyemployed in both countries; international trade betweenthe two countries is balanced (Salvatore, 1998, p. 110).

24. A chart with annual wages relative to the UnitedStates, based on the absolute factor price equalizationtheorem, is very similar to Figure 4 (with GDP perlaborer as a measure of catching up) and omitted; this isdue to the fact that labor shares in GDP tend to beconstant over time and similar across countries. (Wagesare obtained by dividing the labor remuneration com-ponent of GDP by the number of laborers.)

25. All three patterns also well predict Japan’s relativestagnation since the 1990s.

26. While all other sections of this paper use the mostupdated data available, including the revised 1978/93–2004 production approach GDP values following the2004 economic census, this section uses the pre-eco-nomic census national accounts data. This is becauserevised income approach data, which are at the core ofthis section, are not available. An appendix explains theissue of data availability in more detail and explores thepotential implications of using pre-economic census data(none, or only very minor ones). Two separate appen-dices provide all data used in this section.

27. Barro and Sala-i-Martin (1995, p. 352) stressed that‘‘growth accounting does not . . . constitute a theory ofeconomic growth because it does not attempt to explainhow the changes in inputs and the improvements in totalfactor productivity relate to elements—such as aspectsof preferences, technology, and government policies—that can reasonably be viewed as fundamentals.’’ Even ifsuch relationships could be specified, the next questionthat arises is what explains preferences, technology, and

Page 23: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1687

government policies? Trying to trace economic growthback to original causes presents a problem of infiniteregress. The decision on when to stop, that is, whichlevel to view as ultimate ‘‘explanation,’’ appears arbi-trary.

28. For the results of production function estimationswith the data used here, see appendix.

29. Provincial gross value added in the incomeapproach in most years, for most provinces, exactlyequals provincial gross value added in the productionapproach, China’s official approach to calculating GDP.But the sum of provincial gross value added only closely,not perfectly matches China’s national GDP value; theNational Bureau of Statistics does not solely rely onprovincial data when deriving the national GDP value.In addition, a very few provinces do not report incomecomponents for some of the early reform years. As longas the discrepancy between sum provincial GDP andnational GDP does not reflect a particular bias in thesize of one or more of the different income components(and there is no reason why it should), the procedureused here yields accurate national income componentvalues. The provincial data are from GDP 1952–95 forthe years 1978–95, and from GDP 1996–2002 for theyears 1996–2002; since the mid-1990s, each issue of theannual Statistical Yearbook also reports the data of theprevious year.

30. Business profit equals profit net of investmentreturns and non-business revenues. For details on thelink between the operating surplus and profit (orbusiness profit), see Holz (2003, p. 88f). The definitionof operating surplus in China follows the standardSystem of National Accounts promulgated by theUnited Nations. In the case of individual-owned enter-prises with seven or fewer employees (getihu) and ofagricultural households, labor remuneration incorpo-rates the operating surplus.

31. What is called ‘‘technological progress’’ (the resid-ual) in the traditional growth accounting equation isnow measured as (and is identical to the) weighted rateof change in the real wage, in the depreciation rate, andin the surplus rate, plus a particular form of change inthe net tax rate on production. Harberger (1998, p. 1)postulated a growth equation that attributes ‘‘to eachfactor a marginal product measured by its economicreward’’ in the form of �pDy ¼ �wDLþ ð�qþ dÞDK þ R,with the variables denoting initial general price level,change in output (GDP), initial real wage, change inlabor input, initial real rate of return to capital, rate ofreal depreciation of capital, change in capital stock, andthe residual interpreted as ‘‘real cost reduction.’’ Com-pared to the income growth identity (and incorporating

the price level in the coefficients of capital rather than incapital), the real cost reduction, once divided by output,is identical to the weighted growth rates of real wages,the real depreciation rate, the real rate of return, and aparticular form of change in the rate of net taxes onproduction.

32. Since the right-hand side variables are deflatedusing the implicit GDP deflator as first published, thereported real growth rates of GDP are obtained bydeflating nominal GDP by the same deflators. NominalGDP values for all years are taken from a recent source(see table of data in the appendix) and may includerevisions of earlier years’ nominal GDP; in contrast,official real GDP growth rates of earlier years aretypically not revised. Official real GDP grew by 759%.

33. Annual data on laborers by age and education forthe period 1978–2003 are obtained/calculated from the1982, 1990, and 2000 population censuses together withthe 1987 and 1995 1% population sample surveys.Projections for the years 2000–25 make use of the year2000 census data and past patterns and trends. Anappendix on labor data provides general explanations onthe derivation of the labor data; all calculations andestimations are explained in detail in Holz (2005).

34. Young (2003), for example, regressed individuals’wages on gender, age (for which he has a total of 11categories), and education (total of 4 categories).

35. Details on the estimation of the VEC models arereported in an appendix.

36. Similarly, the largest share of the forecast errorvariance of the share of laborers with upper middleschool education is due to its own shocks, but 60% of theforecast error variance of the share of laborers withcollege level education or above is explained by variationin the share of laborers with upper middle schooleducation. The forecast error variance decompositiondepends, in this particular instance, much on theCholesky ordering of the variables. Reversing the order,63% of the variance of the wage is explained by variationin the share of laborers with upper middle schooleducation.

37. Non-education variables are typically not cointe-grated with the wage rate. Age, or age squared, does nothave a unit root. The difference in the choice of variablesfor the cointegration equations in levels versus inlogarithms is necessitated by the fact that some variableshave unit roots when the variable is measured in levels,but not when it is measured in logarithms, and vice

versa. For example, in levels, ‘‘average years of schoolingminus five’’ is not I(1) but I(2), while the share of

Page 24: China's Economic Growth 1978-2025

1688 WORLD DEVELOPMENT

laborers with college level education or above inlogarithms is I(2). Average years of schooling is I(2)independent of if the variable is in levels or inlogarithms.

38. In the case of the first cointegration equation, withthe wage in levels, a different cointegration equation thatinvolves the share of laborers with primary schooleducation rather than the share of laborers with uppermiddle school education exhibits strong Granger cau-sality from both education variables to the wage series.In a visual comparison of the actual wage series andthese two fitted series, the fitted series reported in thetable with the share of laborers with upper middle schooleducation better matched the actual series.

39. This is a standard assumption in productionfunction based growth accounting, where, furthermore,the constant labor share, and one minus the labor share,is assumed to equal output elasticities.

40. The United States and Chinese education systemsare not perfectly comparable. An appendix establishesthe equivalence of different degrees in the two countriesand provides additional details.

41. The educational level of the age cohorts in the late1940s may be exaggerated in that some upper middleschool graduates in China during the Cultural Revolu-tion may have enjoyed no more than a lower middleschool education under the label of an upper middleschool education.

42. The 5% increase is reported in a news item in China

Infobank (<http://www.chinainfobank.com>) of May26, 2006, titled ‘‘2006 nian zhongguo gaoxiao dezhaosheng guimo zai qunian 504 wan ren de jichushangzai zengjia 5%’’ (The 2006 tertiary level new enrollmentfigure is 5% higher than last year’s basis of 5.04 m).

43. This assumes a narrow definition of undergraduatelevel graduates in China and a broad definition for thecase of the United States (See notes to Figure 10).

44. If the US graduates with a first professional degreeare included with the US figure of MA and PhDrecipients, the comparison with China is slightly furtherin favor of the United States. But with Chinese medicaldoctors and lawyers primarily holding a BA degree (inthe medical doctor’s case a 5-year BA program), it seemsinappropriate to include graduates of medical and lawschools in the US number but not in the Chinesenumber.

45. These data ignore Chinese students studyingabroad because no specific degree and program comple-

tion data are available for these. Assuming Chinesestudents abroad were all studying in MA and PhDprograms, their number in 2002 was equivalent toapproximately half of the year 2005 domestic MA andPhD graduates; the 2005 number of returning students,after completing studies, was one-fifth the 2005 numberof domestic MA and PhD graduates. (For the datasources, see those listed in Figure 9). At the high schoollevel, the number of Chinese graduates has well exceededtheir US counterpart in absolute numbers throughoutthe reform period. In 2005, China had 2.69 times morehigh school graduates than the United States. Foradditional data, see appendix.

46. The head count comparison is furthermore incom-plete because it excludes education in China outside theregular institutions of higher education, while the USdata make no distinction between ‘‘official’’ and ‘‘unof-ficial’’ education. For example, adult education outsidethe regular university system in China is not included inthe tertiary level education data presented. In 2001,316,367 persons graduated with a degree from a tertiarylevel adult education institution. This number comparesto 3,068,000 graduates from regular institutions ofhigher education (at undergraduate level) and corre-sponds to just below 1& of the population age 25–40 onNovember 1, 2000 (Census 2000, Vol. 1, pp. 215–7;Education Yearbook 2002, p. 107).

47. There is little doubt that the Chinese educationsystem is indeed capable of identifying the brightestchildren, with lower middle school to university admis-sion now routinely based on entrance examinations.Heckman (2005) expressed concern about financialhurdles for children from low-income families.

48. Why has China not ‘‘overtaken’’ the United Statesa long time ago? The immediate answer is China’s choiceof economic system prior to 1978. In terms of education,with uniform wages and access to higher-level educationseverely limited, neither were incentives for humancapital accumulation in place nor the opportunitiesavailable. For example, tertiary level education almostceased during the Cultural Revolution. In 1971, a since1949 low of 600 students are reported to have graduatedfrom a regular institution of higher education, downfrom 186,000 in 1965; only by 1977 did the figure againreach the 1965 level, with 194,000 graduates. (StatisticalYearbook 1990 p. 709.) This means that among theolder generations, only the generation born in or beforethe 1940s had some access to tertiary level education.

49. Holz (2003), Chapter 8, provided evidence of theprice wars of the late 1990s among state-owned enter-prises.

Page 25: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1689

50. See Statistical Yearbook 2006, pp. 57, 76, 734. Theshare in France was 26%, in Germany 40%, in the UK26%, and in the relatively small South Korea 43%. Datafor all countries except China are from the IFS. Thevalue of China’s exports of goods and services is notreported in the IFS, only the net export value. The dataon net exports and GDP for China in the IFS arevirtually identical to those reported in China’s balanceof payments and national income accounts in theStatistical Yearbook series; the balance of paymentsdata on goods, in turn, are near identical to the import-export data in the trade section of the StatisticalYearbook.

51. For the Data, see Statistical Yearbook 2006, pp.740–3.

52. Statistical Yearbook 2006, p. 505. These data arefrom the economic census 2004. Value added data for2004 are not available. The data for 2005 do not comewith an ownership breakdown; in 2005, industrialenterprises with annual sales revenue in excess of 5 myuan RMB accounted for 93.86% of total industrial

value added (pp. 57, 511). Based on pre-economic census(unrevised) values, foreign-funded enterprises plus HongKong/Macao/Taiwan enterprises together in 2005 ac-counted for 28.57% of the value added produced inindustrial enterprises with annual sales revenue in excessof 5 m yuan RMB (Statistical Abstract 2006, p. 133).

53. Statistical Yearbook 2006, pp. 734, 751. Foreign-funded plus Hong Kong/Macao/Taiwan enterprisesaccounted for 70.84% of total exports by industrialenterprises with annual sales revenue in excess of 5 myuan RMB in 2005 (Statistical Abstract 2006, p. 133).

54. See, for example, Riskin (2007) on recent decreasesin rural inequality and in urban inequality.

55. Work on a green GDP has begun. Qi (2004)presented an overview of progress toward the calcula-tion of green GDP for China. No economy-wide figureappears as yet available, but work by the NationalBureau of Statistics and China’s National Environmen-tal Protection Office has begun to put a framework forgreen GDP into place within 3–6 years.

REFERENCES

Abramovitz, M. (1986). Catching Up, Forging Ahead,and Falling Behind. Journal of Economic History,46(2), 385–406.

Barro, R. J., & Sala-i-Martin, X. (1995). Economicgrowth. New York, NY: McGraw-Hill.

Broomfield, E. V. (2003). Perceptions of Danger: TheChina Threat Theory. Journal of ContemporaryChina, 12(35), 265–284.

Brown, L. R. (1995). Who will feed China? Wake-up callfor a small planet. New York: W.W. Norton & Co..

Carter, C. A., & Rozelle, S. (2001). Will China Become aMajor Force in World Food Markets?. Review ofAgricultural Economics, 23(2), 319–331.

Census 1990 (1993). Zhongguo 1990 nian quanguo renkoupucha ziliao (National population census material1990). Compiled by the State Council populationcensus office and the National Bureau of Statisticspopulation division. Beijing: Zhongguo tongji chu-banshe.

Census 2000 (2002). Zhongguo 2000 nian quanguo renkoupucha ziliao (National population census material2000). Compiled by the State Council populationcensus office and the National Bureau of Statisticspopulation, society, and science division. Beijing:Zhongguo tongji chubanshe.

Census 2000 Major Figures (2001) 2000 nian di wu ciquanguo renkou pucha zhuyao shuju (Major figuresfrom the 2000 population census of China). Beijing:Zhongguo tongji chubanshe.

Census 2000 (nationality) (2003). 2000 nian renkou puchazhongguo minzu renkou ziliao (National populationcensus material 2000 by nationality). Issued by theNational Bureau of Statistics population, society,and science division. Beijing: Zhongguo tongji chu-banshe.

Chang, G. G. (2001). The coming collapse of China. NewYork: Random House.

Chow, G. C. (2002). China’s economic transformation.Malden, MA: Blackwell Publishing.

Chow, G., & Li, K.-W. (2002). China’s EconomicGrowth 1952–2010. Economic Development and Cul-tural Change, 51(1), 247–256.

Cypher, J. M., & Dietz, J. L. (1997). The process ofeconomic development. London: Routledge.

Douglas, A. (2002). Meet the New China. InternationalJournal of Business, 7(3), 35–49.

Education Yearbook (Annual). Zhongguo jiaoyu nianj-ian (China education yearbook). Beijing: Zhongguojiaoyu chubanshe.

Educational Attainment 2000 (2004). At <http://www.census.gov/prod/2003pubs/c2kbr-24.pdf> Ac-cessed 1.01.04.

GDP 1952–95 (1997). Zhongguo guonei shengchan zon-gzhi hesuan lishi ziliao 1952–1995 (Historical data onChina’s gross domestic product 1952–1995). Dalian:Dongbei caijing daxue chubanshe.

GDP 1996–2002 (2004). Zhongguo guonei shengchanzongzhi hesuan lishi ziliao 1996–2002 (Historical data

Page 26: China's Economic Growth 1978-2025

1690 WORLD DEVELOPMENT

on China’s gross domestic product 1996–2002). Bei-jing: Zhongguo tongji chubanshe.

Gertz, B. (2000). The China threat how the people’srepublic targets America. Washington D.C: RegneryPublishing.

Harberger, A. C. (1998). A Vision of the GrowthProcess. American Economic Review, 88(1), 1–32.

Heckman, J. J. (2005). China’s Human Capital Invest-ment. China Economic Review, 16(1), 50–70.

Henderson, C. (1999). China on the brink: The Mythsand realities of the World’s largest market. NewYork: McGraw-Hill.

Heston, A. (2001). Treatment of China in PWT 6. Draft,December 2001. <http://pwt.econ.upenn.edu/Docu-mentation/China.PDF> Accessed 16.12.03.

Holz, C. A. (2003). China’s industrial state-owned enter-prises: Between profitability and bankruptcy. Singa-pore: World Scientific Publishing Co.

Holz, C. A. (2005). The quantity and quality of labor inChina 1978–2000–2025. Mimeo, Hong Kong Uni-versity of Science and Technology. <http://iho-me.ust.hk/~socholz>.

Holz, C. A. (2006a). China’s Reform Period EconomicGrowth: How Reliable Are Angus Maddison’sEstimates? Review of Income and Wealth, 52(1),85–119.

Holz, C. A. (2006b). Why China’s Rise Is Sustainable.Far Eastern Economic Review, 169(3), 41–46.

Holz, C. A. (2006c). China’s Reform Period Eco-nomic Growth: How Reliable Are AngusMaddison’s Estimates? Response to Angus Maddi-son’s Reply. Review of Income and Wealth, 52(3),471–475.

Holz, C. A. (2006d). Measuring Chinese productivitygrowth, 1952–2005. Mimeo, Hong Kong Universityof Science and Technology. <http://ihome.ust.hk/~socholz>.

IFS (Annual). International financial statistics yearbook.Washington, DC: International Monetary Fund.

Koren, M., & Tenreyro, S. (2007). Volatility andDevelopment. The Quarterly Journal of Economics,122(1), 243–287.

Lardy, N. R. (1998). China’s unfinished economic revo-lution. Washington D.C.: The Brookings Institution.

Lin, Justin Yifu, Cai, Fang, & Li, Zhou (2003). TheChina miracle: Development strategy and economicreform (revised edition). Hong Kong: The ChineseUniversity Press.

Lubman, Stanley (2004). The dragon as demon: Imagesof China on capitol hill. Journal of ContemporaryChina, 13(40), 541–565.

Maddison, A. (2006). Do Official Statistics ExaggerateChina’s GDP Growth? A Reply to Carsten Holz.Review of Income and Wealth, 52(1), 121–126.

Murray, G. (1998). China: The next superpower. Dilem-mas in change and continuity. Richmond, Surrey:China Library (Curzon Press).

Perkins, D. H., & Rawski, T. G. (2006). ForecastingChina’s Economic Growth over the Next TwoDecades. Draft, 29, 2006, June.

PWT6.1. Heston, A., Summers, R., & Aten, B. (2002).Penn World Table Version 6.1, Center for Interna-tional Comparisons at the University of Pennsylva-nia (CICUP). <http://pwt.econ.upenn.edu/>.

PWT6.2 Heston, A., Summers, R., & Aten, B. (2006),Penn World Table Version 6.2, Center for Interna-tional Comparisons of Production, Income andPrices at the University of Pennsylvania. <http://pwt.econ.upenn.edu/>.

Qi Yuanjun (2004). Guoneiwai luse GDP yanjiu de zongtijinzhan (Overall progress in research on green GDP inChina and abroad). Jingji yanjiu cankao (88), 25–9,34.

Qian, Yingyi (2000). The Process of China’s MarketTransition (1978–1998): The Evolutionary, Histori-cal, and Comparative Perspectives. Journal of Insti-tutional and Theoretical Economics, 156(1), 151–171.

Qian, Yingyi (2003). How Reform Worked in China. InD. Rodrik (Ed.). Search of prosperity: Analyticnarratives on economic growth (pp. 297–333). Prince-ton, NJ: Princeton University Press.

Rawski, T. G. (1999). Reforming China’s Economy:What Have We Learned? The China Journal, (41),139–156.

Riskin, C. (2007). Has China Reached the Top of theKuznets Curve?. In V. Shue, & C. Wong (Eds.),Paying for progress in China (pp. 29–45). MiltonPark, UK: Routledge.

Salvatore, D. (1998). International economics. UpperSaddle River, NJ: Prentice Hall.

Samuelson, P. (2004). Where Ricardo and Mill Rebutand Confirm Arguments of Mainstream EconomistsSupporting Globalization. Journal of Economic Per-spectives, 18(3), 135–146.

Statistical Abstract (Annual). Zhongguo tongji zhaiyao(China Statistical Abstract). Beijing: Zhongguo ton-gji chubanshe.

Statistical Yearbook (Annual). Zhongguo tongji nianjian(Statistical Yearbook of China). Beijing: Zhongguotongji chubanshe.

Terrill, R. (2003). The new Chinese empire: And what itmeans for the United States. Basic Books.

Timperlake, E., & Triplett, W. C. Jr., (2002). Red dragonrising: Communist China’s military threat to America.Washington, D.C.: Regnery Publishing.

Wang, Yan, & Yao, Yudong (2003). Sources of China’sEconomic Growth 1952–1999: Incorporating HumanCapital Accumulation. China Economic Review,14(1), 32–52.

World Bank Development Indicators. Cross-countrydatabase. <http://www.worldbank.org/data>.

World Bank (1997). China 2020. Washington, DC:World Bank.

Woo, W. T. (1994). The Art of Reforming CentrallyPlanned Economies: Comparing China, Poland, andRussia. Journal of Comparative Economics, 18(3),276–308.

Woo, W. T. (1999). The Real Reasons for China’sGrowth. The China Journal, (41), 115–137.

Wu, Yanrui (2004). China’s economic growth: A miraclewith Chinese characteristics. London and New York:RoutledgeCurzon.

Page 27: China's Economic Growth 1978-2025

CHINA’S ECONOMIC GROWTH 1978–2025 1691

Young, A. (2003). Gold into base metals: Productivitygrowth in the People’s Republic of China during thereform period. Journal of Political Economy, 111(6),1220–1261.

Zeng, M., & Williamson, P. J. (2003). The HiddenDragons. Harvard Business Review, 81(10), 92–99.

Available online at www.sciencedirect.com