1、ASIAN DEVELOPMENT BANKASIAN DEVELOPMENT BANK6 ADB Avenue,Mandaluyong City1550 Metro Manila,Philippineswww.adb.orgThe Long-Term Growth Prospects of the Peoples Republic of ChinaEconomic growth in the Peoples Republic of China has moderated over the past decade,even before the coronavirus disease(COVI

2、D-19)pandemic.A critical question is to which level of economic growth the country will return to after COVID-19 and,more broadly,how its long-term growth prospects are,given a rapidly aging society,the ongoing high dependence of growth on investment,and a changed international environment.This pape

3、r estimates the countrys potential gross domestic product growth up to 2040 based on a Cobb-Douglas production function and provides policy recommendations on structural reforms to boost potential growth.About the Asian Development BankADB is committed to achieving a prosperous,inclusive,resilient,a

4、nd sustainable Asia and the Pacific,while sustaining its efforts to eradicate extreme poverty.Established in 1966,it is owned by 68 members 49 from the region.Its main instruments for helping its developing member countries are policy dialogue,loans,equity investments,guarantees,grants,and technical

5、 assistance.ADB EAST ASIA WORKING PAPER SERIESNO.54December 2022THE LONG-TERM GROWTH PROSPECTS OF THE PEOPLES REPUBLIC OF CHINADominik Peschel and Wenyu LiuASIAN DEVELOPMENT BANKADB East Asia Working Paper SeriesThe Long-Term Growth Prospects of the Peoples Republic of China Dominik Peschel and Weny

6、u LiuNo.54|December 2022Dominik Peschel is head of the Economics Unit at the Asian Development Banks Resident Mission in the Peoples Republic of China.Wenyu Liu is a consultant at the Asian Development Banks Resident Mission in the Peoples Republic of China.Creative Commons Attribution 3.0 IGO licen

7、se(CC BY 3.0 IGO)2022 Asian Development Bank6 ADB Avenue,Mandaluyong City,1550 Metro Manila,PhilippinesTel+63 2 8632 4444;Fax+63 2 8636 2444www.adb.orgSome rights reserved.Published in 2022.Publication Stock No.WPS220567-3DOI:http:/dx.doi.org/10.22617/WPS220567-3The views expressed in this publicati

8、on are those of the authors and do not necessarily reflect the views and policies ofthe Asian Development Bank(ADB)or its Board of Governors or the governments they represent.ADB does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequenc

9、e of their use.The mention of specific companies or products of manufacturers does not imply that they are endorsed or recommended by ADB in preference to others of a similar nature that are not mentioned.By making any designation of or reference to a particular territory or geographic area,or by us

10、ing the term“country”inthis document,ADB does not intend to make any judgments as to the legal or other status of any territory or area.This work is available under the Creative Commons Attribution 3.0 IGO license(CC BY 3.0 IGO)https:/creativecommons.org/licenses/by/3.0/igo/.By using the content of

11、this publication,you agree to be bound bytheterms of this license.For attribution,translations,adaptations,and permissions,please read the provisions andterms of use at https:/www.adb.org/terms-use#openaccess.This CC license does not apply to non-ADB copyright materials in this publication.If the ma

12、terial is attributed toanother source,please contact the copyright owner or publisher of that source for permission to reproduce it.ADB cannot be held liable for any claims that arise as a result of your use of the material.Please contact pubsmarketingadb.org if you have questions or comments with r

13、espect to content,or if you wish toobtain copyright permission for your intended use that does not fall within these terms,or for permission to use theADB logo.Corrigenda to ADB publications may be found at http:/www.adb.org/publications/corrigenda.Notes:In this report,“$”refers to United States dol

14、lars,and“CNY”refers to Chinese yuan.ADB recognizes“China”as the Peoples Republic of China.The ADB East Asia Working Paper Series is a forum for stimulating discussion and eliciting feedback on ongoing and recently completed research and policy studies undertaken by the East Asia Department of the As

15、ian Development Bank(ADB)staff,consultants,or resource persons.The series deals with key economic and development problems,as well as conceptual,analytical,or methodological issues relating to project/program economic analysis,and statistical data and measurement.The series aims to enhance the knowl

16、edge on Asias development and policy challenges;strengthen analytical rigor and quality of ADBs country partnership strategies,and its subregional and country operations;and improve the quality and availability of statistical data and development indicators for monitoring development effectiveness.T

17、he ADB East Asia Working Paper Series is a quick-disseminating,informal publication whose titles could subsequently be revised for publication as articles in professional journals or chapters in books.The series is maintained by the East Asia Department.CONTENTSTABLES,FIGURES,AND BOXES ivACKNOWLEDGM

18、ENTS viCURRENCY EQUIVALENTS viiABBREVIATIONS viiEXECUTIVE SUMMARY viiiI.Introduction 1II.Diverging Views on the PRCs Long-Term Growth 3A.Conditional Convergence 3B.Approaches to Long-Term Economic Growth Forecasts 4III.Model Specification and Results 7A.Functional Form 7B.Labor 8C.Human Capital 9D.C

19、apital 10E.Total Factor Productivity 13F.Results 18G.Potential Impact of Recent Developments 19IV.Reforms to Boost Long-Term Growth 23A.Reform Areas 23B.Labor 23C.Human Capital 24D.Capital 26E.Total Factor Productivity 36V.Outlook on Policy Challenges and Reforms 39Appendixes 41Appendix 1:Barros Est

20、imate of Convergence 41Appendix 2:Human Capital Index 43Appendix 3:Comparison of Results 45Appendix 4:Estimating Capital Stock Growth 46Appendix 5:Improving the Quality of Education 47Appendix 6:Summary of Policy Recommendations 48References 50TABLES,FIGURES,AND BOXESTables1 Selected Long-Term Growt

21、h Forecasts for the PRC 42 Comparison of Selected Long-Term Growth Estimates for the PRC 19A1 Barros Explanation of the PRCs Per Capita Growth Pattern 42A3.1 Comparison of Potential GDP Growth Estimates with Bailliu et al.(2016)45A3.2 Contributions to Potential GDP Growth,20202040 45A6 Summary of Po

22、licy Recommendations 48Figures1 Real GDP Growth,20002021 12 Real Growth in Income and Consumption,20002021 23 Growth in Investment by Category,20042021 24 Japans and the Republic of Koreas Catch-Up Process to the United States,1960s2019 35 PRCs Per Capita GDP in Comparison 46 Long-Term Growth Foreca

23、sts,20132038 67 Input Factors of the Cobb-Douglas Function 78 Working-Age Population,20002050 89 Impact of Fertility Assumptions on Working-Age Population Projections up to 2050 810 Tertiary School Enrollment,20102040 1011 Human Capital Index,20002040 1012 Capital Stock Estimates with Different Defl

24、ators,20002019 1013 Estimated Capital Stock with and without Housing,20002016 1014 Gross Fixed Capital Formation,19602021 1215 Capital Formation and Labor Supply,19602040 1216 Real Growth in FDI Stock,20002040 1417 Real Import Growth,20012040 1418 R&D Expenditure,20002040 1519 Employment by Sector,2

25、0002040 1520 Share of Agriculture in Total Employment 1621 Total Factor Productivity Growth,20012019 1722 Decomposition of TFP Growth,20012040 17The Long-Term Growth Prospects of the Peoples Republic of Chinav23 Decomposition of Contributions to Potential GDP Growth,20012040 1824 Birth Rate,19602021

26、 2325 Fertility Rate for Selected Countries,19602020 2326 Comparison of Average Years of Schooling,19502015 2527 Comparison of Education Attainment,2015 2528 Years of Schooling,19852019 2629 Urban and Rural School Completion Rates,2005,2015,and 2019 2630 Incremental Capital Output Ratio,20002019 273

27、1 Fixed Asset Investment by Ownership Type,20042017 2732 Employment by Category of Employer,19802019 2833 SOE Assets,and Return on Assets by Sector,20032020 2834 Selected Financing Sources,20022021 3135 Shadow Bank Financing,20022021 3136 Loans Outstanding by Category,20092021 3237 Maturity of Finan

28、cing,20132020 3238 Market Share by Type of Bank,20032021 3239 Key Interest Rates,20082022 3240 Household Loans by Type,20072022 3341 Corporate Debt,20072021 3342 Trade Openness,19802021 3643 Processing Trade,19932021 3644 FDI Net Inflow into the PRC,20002020 3745 FDI Restrictiveness Index,2003,2006,

29、2010,2015,and 2020 37B1 Debt Structure,20072022 21B2 Credit Gap and Change in Debt,20082021 21B3 Public Debt and Gross Regional Product Per Capita by Province,2021 22B4 Industrial State Holding Enterprises 30B5 Sector Share in GDP,19802021 35B6 Household Income and Consumption,and Investment,1980202

30、1 35A1 Real Per Capita GDP Growth in the PRC:Barros Estimate of Convergence,19602020 41Boxes1 GDP Growth,Investment,and Debt 212 Industrial State Holding Enterprises 303 Structural Change 35ACKNOWLEDGMENTSInsightful comments were provided by Yolanda Fernandez Lommen,advisor,and Peter Rosenkranz,fina

31、ncial sector specialist,East Asia Regional Department,Asian Development Bank(ADB);and Marcel Schroder,economist,Economic Research and Regional Cooperation Department,ADB.Valuable research assistance was provided by Wen Qi,associate economics officer,Peoples Republic of China Resident Mission,ADB.(As

32、 of 1 December 2022)Currency unit CNY CNY1.00=$0.1410$1.00=CNY7.0924ABBREVIATIONSCURRENCY EQUIVALENTSADB Asian Development BankBIS Bank for International Settlements COVID-19 coronavirus diseaseFAI fixed asset investmentFDI foreign direct investmentGDP gross domestic productGFCF gross fixed capital

33、formationIMF International Monetary FundICOR incremental capital output ratioIP intellectual propertyNBS National Bureau of StatisticsOECD Organisation for Economic Co-operation and DevelopmentPRC Peoples Republic of ChinaPPP purchasing power parityR&D research and developmentSOE state-owned enterpr

34、iseSME small and medium-sized enterpriseTFP total factor productivityUN United NationsUS United States EXECUTIVE SUMMARYEconomic growth in the Peoples Republic of China(PRC)has moderated over the past decade,even before the coronavirus disease(COVID-19)pandemic.A critical question is to which level

35、of gross domestic product(GDP)growth the country will return to after COVID-19 and how its long-term growth prospects are,given its rapidly aging society,a continued dependence of growth on investment,and a changed international environment.This paper estimates the countrys potential GDP growth up t

36、o 2040 based on a Cobb-Douglas production function,estimating all input factors separately with newest available data.Many assumptions in the past,e.g.,on the demographic development,have often been too optimistic,resulting in inflated growth estimates.According to our estimate,potential GDP growth

37、averages 5.3%in 20202025,declining gradually to 2.0%in 20362040.Compared to three earlier papers that also use the production function approach(each using different specifications)Bailliu et al.(2016),Roberts and Russell(2019),and Higgins(2020)our estimate of future potential growth is lower than Ba

38、illiu et al.(2016)and closer to the other two.Our estimates of the different input factors allow us to analyze the drivers of the growth moderation and to use these insights for policy recommendations.Capital and total factor productivity(TFP)are identified as the major contributors to growth in the

39、 future.Meanwhile,a shrinking working-age population will increasingly weigh on growth,while the contribution of human capital to growth is comparatively small.To increase potential growth,this paper suggests reforms in four areaslabor,human capital,capital,and TFP:Labor.Given rapid demographic agin

40、g,mitigating the impact of the declining labor force is key.One measure is to raise the retirement age,in particular the actual average retirement age that is in the mid-50s.Other measures to cope with an aging working force include improving health care,raising female workforce participation rates,

41、and increasing labor mobility.Human capital.The PRC has room to catch up with advanced economies in both years of and quality of education,especially in rural areas.Furthermore,education can be improved by expanding early child development,enhancing the quality of tertiary education,strengthening te

42、chnical and vocational education and training,and increasing on-the-job training.Capital.Changes to the allocation of capital and credit are considered to be one of the biggest factors to raise potential growth.State-owned enterprise(SOE)reforms are needed.This requires clarifying SOEs scope and fun

43、ction,leveling the playing field for the private sector,separating social functions from SOEs,and improving SOE management.Moreover,credit allocation needs to be shifted in favor of the private sector,in particular micro and small businesses.To this end,the implicit guarantees of SOEs need to be rem

44、oved,while banks must strengthen their credit risk assessment capabilities and increase their operational efficiency.The Long-Term Growth Prospects of the Peoples Republic of ChinaixTotal factor productivity.First,to facilitate the sectoral labor shift,residence permit restrictions should be loosene

45、d to facilitate labor migration to urban areas and the access to basic social services should be made easier for labor migrants.Second,to increase the countrys attractiveness for foreign direct investment(FDI)inflows,the negative list for FDI should be shortened and joint venture requirements reduce

46、d.Third,advancing trade and investment agreements could help increase openness and facilitate domestic reforms.Finally,to improve the countrys research and development(R&D)effectiveness,the prevailing misuse of R&D funds must be addressed,the share of basic research increased,the quality of patents

47、issued raised,and intellectual property(IP)rights protection strengthened.I.INTRODUCTIONThe long-term economic growth prospects of the PRC matter,not only for the country but also for the world.While some worry about the PRC“growing old before becoming rich,”others see the PRC emerging as the worlds

48、 biggest economy.1 A third group highlights adverse effects on global growth in case the PRCs economic growth slows(IMF 2016;Ahmed et al.2019).However,there is no consensus on the PRCs long-term growth prospects.2 This paper discusses the assumptions and findings of different approaches to long-term

49、 growth forecasts for the PRC,estimates the countrys potential growth until 2040,and provides policy recommendations to raise it.The PRC has had a long stretch of high growth from the end of the 1970s until 2010,which is exceptional in international perspective.3 At the same time,domestic debt has i

50、ncreased sharply over the past decade,and economic growth has been on a moderating trend,even before the COVID-19 pandemic hit the economy.4 The big question is which growth rate the PRC can sustain over the next 2 decades and how the decline in potential growth evolves.Answers to these questions wo

51、uld be helpful for economic policy making and social planning.Figure 1:Real GDP Growth,2000202416GDPGDP per capita%change year on year20002000420052006200720082009200001920202021GDP=growth domestic product.Note:In recent years,the population siz

52、e virtually stagnated so that GDP growth and per capita GDP growth aligned.Source:National Bureau of Statistics(NBS)via CEIC Data Company(accessed 10 June 2022).Economic growth moderated in the past decade,well before the COVID-19 pandemic in 2020(Figure 1).After years of two-digit growth rates,GDP

53、growth slowed from 9.0%in 2011 to 6.0%in 2019 in line with potential growth moderating(as our estimates in part III show).On the demand side,growth in both household income and consumption has softened in the decade before COVID-19,while investment growth has been on a declining trend(Figures 2 and

54、3).1 See for instance Economist(2019)as an example for the discussion on aging,and Zhu and Orlik(2022)on the PRCs changes to emerge as the worlds biggest economy.2 For economists,usually not the absolute size of an economy matters,but GDP per capita.Thus,the important question is not if the PRC take

55、s over as the worlds biggest economy in the coming decades,but how fast GDP per capitaa measure for the stage of development of an economy when adjusted for purchasing power parity(PPP)evolves.Dollar,Huang,and Yao(2020)forecast the PRCs per capita GDP(PPP adjusted)relative to the US to rise from abo

56、ut one-quarter in 2018 to around two-thirds in 2049.3 See for instance Summers and Pritchett(2014).Johnson and Papageorgiou(2020)provide a table of the fastest-growing economies per decade.The PRC comes first for the 1980s and 1990s,and third for the 2000s(following two commodity exporters).4 In thi

57、s paper,projections start in 2020 to avoid distortions from the COVID-19 pandemic.2ADB East Asia Working Paper Series No.54Figure 2:Real Growth in Income and Consumption,20002021-4-2024681012142000%change year on year200042005200620072008200920000192020202

58、1Income per capitaConsumption per capitaGFCFFAIManufacturingInfrastructureReal estate010%change year on year200420052006200720082009200001920202021Sources:NBS via CEIC Data Company(accessed 27 May 2022);and ADB calculations.Figure 3:Growth in Investmen

59、t by Category,20042021FAI=fixed asset investment,GFCF=gross fixed capital formation.Note:Growth rates in GFCF and FAI differ,among others,due to different scope of data.A breakdown by sector is only available for FAI.Sources:NBS via CEIC Data Company(accessed 26 May 2022);and ADB calculations.Unders

60、tanding the countrys future potential growth is critical for designing economic policies.Potential growth is a theoretical construct,an estimate of the output that the economy would have produced if labor and capital had been employed at their maximum sustainable rates,i.e.,rates that are consistent

61、 with steady growth and stable inflation(Powell,Sheiner,and Wessel 2021).5 Part II will discuss different approaches to estimating long-term growth.The production function approach,the method we chose for our model in part III,first estimates input factorslabor,human capital,capital stock,and total

62、factor productivity(TFP)and then combines them into one production function.While this approach yields a better understanding of the drivers of potential growth,it requires estimating the trends in many variables,which can be a challenge(Arsov and Watson 2019).6 Lardy(2019)points out the challenges

63、as follows:(i)it is extremely data intensive and requires innumerable judgement and assumptions;(ii)it requires reasonable accurate historical data(on the capital stock,labor force,and human capital);and(iii)projections require many judgements,with capital stock growth and improvements in human capi

64、tal being difficult to estimate,and future TFP growth being an even greater challenge.Our estimate of potential growth indicates that,GDP growth will decline notably within this decade without major reforms,which are needed in particular to the countrys capital and credit allocation,as are measures

65、to address the shrinking labor force.Furthermore,TFP growth will be much lower in future,calling for improvements in R&D efficiency.The remainder of this paper is organized as follows:Part II discusses the diverging views on the PRCs long-term economic growth potential,including different approaches

66、 to forecasting long-term growth.Part III specifies a production function and estimates potential growth for 20202040.Part IV develops policy recommendations based on the findings in the previous part.Part V concludes with an outlook on policy challenges and reforms to boost potential growth.5 For t

67、he PRC,such approach faces limitations given the drivers of inflation in the PRCa detailed discussion can be found in ADB(2021a)and the fact that labor market statistics have shortcomings(ADB 2020a).6 For different ways to estimate potential growth,see Guisinger,Owyang,and Shell(2018)for the US.II.D

68、IVERGING VIEWS ON THE PRCS LONG-TERM GROWTHA.Conditional ConvergenceThe neoclassical growth model predicts poor countries to grow faster than rich ones,leading to a convergence toward rich countries levels of per capita GDP over time.In modern forms of the model,the convergence arises due to diminis

69、hing returns to investments in physical and human capital,and poor countries ability to imitate advanced countries technological innovations.However,this convergence holds empirically only conditionally,i.e.,poor countries tend to grow rapidly only if they first create reasonably well-functioning le

70、gal,political,and market institutions(Barro 2017).7 Barro(2016)predicts convergence to prevail in the long run and a growth rate of 3%4%over the next 23 decades.For his model of the PRCs past growth rates and his explanation of the countrys historical growth pattern,see Appendix 1.In East Asia,Japan

71、 and the Republic of Korea initially caught up rapidly to the US,before the process slowed after their growth peaked in 1969 and 1995,respectively(Figure 4).Figure 4:Japans and the Republic of Koreas Catch-Up Process to the United States,1960s2019Japans Per Capita GDP as Share of the United States,1

72、9602019Republic of Koreas Per Capita GDP as Share of the United States,95060708090246840506070GDP per capita as share of the US(%)GDP per capita as share of the US(%)GDP per capita(%change year on year)GDP per capita(%change year on year)G

73、DP=gross domestic product,US=United States.Note:Growth in per capita GDP(at 2017$,purchasing power parity adjusted)is a 10-year moving average.Red dots indicate the rising growth rates and blue dots declining ones.Sources:Jiang and Yi(2015);Penn World Table 10.0;and ADB calculations.Thanks to faster

74、 GDP growth,the PRC caught up and narrowed the distance in terms of GDP per capita(adjusted for purchasing power parity)to Japan and the Republic of Korea over time,especially in the past 2 decades,as well as to the US(Figure 5).7 For a recent discussion,see Johnson and Papageorgiou(2020),and specif

75、ically on the PRC,Barro(2016).4ADB East Asia Working Paper Series No.54B.Approaches to Long-Term Economic Growth ForecastsThere is a range of opinions on the PRCs long-term growth.Table 1 summarizes a selection of forecasts from different years,using different approaches and time frames.Table 1:Sele

76、cted Long-Term Growth Forecasts for the PRCStudyApproachYearsResultSummers and Pritchett(2014)Presents basic evidence on regression to the mean in country growth rates.Then uses regressions that predict countries growth rates based on their past growth20132033Using regressions analysis,the authors p

77、redict GDP growth per capita to be 3.9%on average over a 20-year horizon.For a 10-year horizon(different regression),they predict 5.0%in 20132023 and 3.3%in 20232033.Barro(2016)Conducts conditional convergence and cross-country growth regressions to determine economic growth20152035GDP growth per ca

78、pita is estimated at 3%4%over the next 2 to 3 decadesBailliu et al.(2016)Uses a Cobb-Douglas production function to forecast the PRCs potential growth until 203020152030GDP growth is forecast to decline gradually over 15 years to around 5%Zhu et al.(2019)Predicts potential growth by examining past s

79、ectoral transitions 20182030Potential growth in the baseline scenario will slow,but remain robust at around 4%by 2030Roberts and Rus-sell(2019)Extrapolates growth trends in input factors(labor,capital,and TFP)20202030The PRCs GDP growth is estimated to around 4%5%in 2020-2030Higgins(2020)Growth acco

80、unting of contributions from labor force growth,capital accumulation,and TFP20182038In the“pretty good”scenario,the PRCs real per capita income growth averages 3.8%in 20182028,slowing to 2.1%in 20282038Figure 5:PRCs Per Capita GDP in ComparisonCompared to Japan and Republic of Korea,19802021As Share

81、 of US Per Capita GDP,19622019GDP per capita as share of the US(%)GDP per capita(%change year on year)01020304050PRCJapanRepublic of Korea2017$PPP(000)05209622010152025PRC=Peoples Republic of China,PPP=purchasing power parity.Estimates start after 20

82、20 for the PRC and the Republic of Korea,and 2015 for Japan.Source:IMF,World Economic Outlook Database 2022.GDP=gross domestic product,US=United States.Notes:Growth in per capita GDP(at 2017$PPP)is a 10-year moving average.Red dots indicate rising growth rates and blue dots declining ones.GDP growth

83、 rates in Penn World Table differ from official ones.Sources:Jiang and Yi(2015);Penn World Table 10.0;and ADB calculations.The Long-Term Growth Prospects of the Peoples Republic of China5StudyApproachYearsResultSasaki et al.(2021)Follows approach in Zhu,Zhang,and Peng(2019),but adopts different indu

84、stry-level convergence rates rather than a fixed rate20202035The size of the PRCs economy can potentially double by 2035(implying an average growth rate of about 4.8%until 2035)as long as the country follows the catch-up process achieved by other East Asian economies.GDP=gross domestic product,PRC=P

85、eoples Republic of China,TFP=total factor productivity.Note:Only studies that are further discussed in this paper.Source:Authors compilation.There are three main approaches in the literature cited in Table 1:8(i)Conditional convergence:Poor countriesonce they have created reasonably well-functioning

86、 legal,political,and market institutionstend to grow faster than rich ones,leading to a convergence toward rich countries levels of per capita GDP over time(Barro 2017).For instance,Barro(2016)uses a cross-country data set on convergence,taking into account several factors,to forecast the PRCs growt

87、h rate,and concludes 3%4%growth in the next 2 to 3 decades.Summers and Pritchett(2014)forecast 3.9%growth over the next 2 decades.At the same time,the convergence approach can only provide a rough estimate of potential growth(Maliszewski and Zhang 2015).Maliszewski and Zhang(2015)further claim that

88、the convergence path of the PRC might be different due to the size of the economy.Zhu,Zhang,and Peng(2019)argue that it is unlikely that the PRC will follow the East Asian economies growth paths due to the changing global environment and the size of its economy.However,it seems unlikely that in the

89、long run a country can sustain an economic growth rate exceeding all other countries experiences and keep constantly growing at a high rate(Summer and Pritchett 2014;Barro 2016).Hence,the convergence approach should be able to provide at least a rough idea of how the PRCs long-run growth will evolve

90、.Also,other East Asian economies growth path might be the best comparator given their geographical proximity and their similar development pattern via industrialization and export-led growth in earlier stages.In this context,Higgins(2020)points out to the similarities between the PRC and Japan and t

91、he Asian Tigers with respect to the development of the share in global manufacturing exports as well as similarities with respect to their high investment share in GDP.9(ii)Sector-based approach:Another approach,using sectoral catch-up,deploys a similar idea as convergence:dividing the economy into

92、different sectors,productivity of those that are further from the international frontier(i.e.,frequently those in advanced economies)will gradually catch up.Zhu,Zhang,and Peng(2019)highlight that sectoral transition is a key driver of the PRCs potential growth.Sasaki et al.(2021)follow this approach

93、,but use differentiated industry-level convergence rates.Both papers reach similar conclusions,though the estimate by Sasaki et al.(2021)comes in slightly higher.As catch-up processes and their speed depend on many variables,projections based on these processes tend to be highly uncertain(Higgins 20

94、20).Also,as seen above,results are sensitive to the selection of the convergence rates in different sectors.8 Another approach to estimate potential output uses statistical techniques to decompose output into trend and cyclical components,with the trend component assumed to be potential output(Arsov

95、 and Watson 2019)see World Bank(2019 and 2021)applying this approach to the PRCs case.Growth accounting using a Cobb-Douglas function is the approach mostly used to estimate potential growth for advanced economies(Arsov and Watson 2019).9 We will discuss this in the part III when looking at capital

96、formation and the expansion path of the capital stock.6ADB East Asia Working Paper Series No.54(iii)Extended Solow model10:Bailliu et al.(2016),Roberts and Russell(2019),and Higgins(2020)adopt a Cobb-Douglas production function with labor(quality and quantity),capital,and TFP as input factors to for

97、ecast long-term growth.The challenge of this approach is that it needs several assumptions on the trend of the input variables.Here,the three papers take different routes:Roberts and Russell(2019)extrapolate trends based on the past 10 years,Higgins looks at comparator groups and develops different

98、scenarios,and Bailliu et al.(2016)estimate the trend of each input factor separately.In Figure 6,we cluster the seven selected studies according to the approach used.Figure 6:Long-Term Growth Forecasts,2056782002820332038Summers and Pritchett(2014)Sasaki et al.(2021)Bailliu et

99、al.(2016)Roberts and Russell(2019)Barro(2016)Zhu et al.(2019)Higgins(2020)PercentNote:Convergence approach in orange,sector-based approach in blue,and production function approach in green.Barro(2016)provides a range of 3%4%for the entire period(split in two periods in the chart for better compariso

100、n).Sources:Authors compilation from Summers and Pritchett(2014);Barro(2016);Bailliu et al.(2016);Zhu,Zhang,and Peng(2019);Roberts and Russell(2019);Higgins(2020);Sasaki et al.(2021).The results using the convergence approach(Summers and Pritchett 2014;Barro 2016)vary only slightly for the future and

101、 are lower than the forecasts from the sector-based approach(Zhu,Zhang,and Peng 2019;Sasaki et al.2021)that are within a range of 4%5%.However,estimates based on a production function differ more widely,given different assumptions and data availability at different points in time,all of which can in

102、fluence the estimates of input factors.The Cobb-Douglas production function warrants further research given that the estimates of the studies cited above differ most widely and given recent data updates,such as the 2020 census results,updated capital stock estimates(IMF 2021a),and a new data set on

103、educational attainment(Barro and Lee 2021).Given the moderation in actual GDP growth in recent years,even before the COVID-19 shock on the economy,it seems warranted to re-estimating potential growth using these data updates(and others).Using the approach by Bailliu et al.(2016),we modify it as need

104、ed to integrate data updates and newer findings and developments.10 See Mankiw,Romer and Weil(1992)for a version with human capital.III.MODEL SPECIFICATION AND RESULTSA.Functional FormWe adopt a Cobb-Douglas production function to estimate GDP growth:11 (1),where Y is GDP,A is productivity,K is(phys

105、ical)capital stock,L is labor,h is human capital,is the share of capital income in GDP,and(1-)is the share of labor income in GDP.By log-linearization,we get (2),where TFP is total factor productivity.Then,taking the first difference of equation(2),we obtain growth rates:(3).Equation(3)derives the G

106、DP growth rate.To abstract from the business cycle,all factor inputs are assessed at their trend level:(4),where X*denotes the trend level of variable X.Figure 7 illustrates the input factors of GDP growth in the model.Figure 7:Input Factors of the Cobb-Douglas FunctionGDP Growth Working-Age Populat

107、ion Human Capital Index Capital StockSectoral Labor Shift FDI Stock Imports R&D ExpenditureLaborHuman CapitalCapitalTotal Factor ProductivityInput FactorEstimateFDI=foreign direct investment,GDP=gross domestic product,R&D=research and development.Source:Authors.11 See for instance Bailliu et al.(201

108、6).8ADB East Asia Working Paper Series No.54In the following,we discuss the input factors and their estimates and forecasts.Challenges relate to estimating the capital stock and(future)TFP.A possible approach to estimate TFP,but only looking backward,is to calculate the residual,12 i.e.,the differen

109、ce between actual economic growth and the part of growth explained by capital,labor,and human capital input.Conversely,we need to use a combination of variables that can be forecast to obtain an estimate of future TFP growth.13 B.Labor As for the growth of working-age population,we assume in line wi

110、th Bailliu et al.(2016)thatat full employment and an unchanged labor participation ratethe trend labor growth is measured by the growth rate of the working-age population,as shown in the equation:(5).We consider the working-age population as the group of people between the ages of 15 and 59.While th

111、ere are plans in the 14th Five-Year Plan to gradually raise the legal retirement age,the actual retirement age in the PRC is far below 60(Economist 2021)so that in our model an extension of work life beyond 60 does not seem warranted.The PRCs working-age population started to fall in 2012.From its p

112、eak in 2011 until 2021,the working-age population declined by 42 million.In 2021,it accounted for 64%of total population,down from an average of 69%in 20002010.The ratio is projected to fall to 61%in 2030 and further to 52%in 2050(Figure 8).Figure 8:Working-Age Population,20002050 200020052010201520

113、202025f2030f2035f2040f2045f2050fAge 0 to 14Age 15 to 59Age 60 and aboveMillionMillion650750850950700800900Actual working-age populationLow fertility variantMedium fertility variantHigh fertility variant0300600900520025f2030f2035f2040f2045f2050f2000Note:No data available between

114、 2000 and 2005.Until 2021 actual figures from NBS,from 2022 based on forecast from United Nations(UN)(2019),low fertility variant(using linear extrapolation).Sources:NBS via CEIC Data Company(accessed 2 December 2022);UN(2019),World Population Prospects;and ADB calculations.Figure 9:Impact of Fertil

115、ity Assumptions on Working-Age Population Projections up to 2050Note:Actual figures in blue from NBS until 2020,forecast from UN(2019),with variants overlapping until 2035.NBS figures take into account the 2020 Census(with its downward revisions).Sources:NBS via CEIC Data Company(accessed 2 December

116、 2022);and UN(2019),World Population Prospects.12 So-called Solow residual,named after the economist Robert Solow.13 Here,we need to briefly address concerns about GDP growth data itself.The latest World Penn Tables data set shows lower GDP growth(especially in recent years)than official statistics,

117、which we use.Chen et al.(2019)find that GDP growth rates were overstated for 20082016.On the other hand,Clark,Pinkovskiy,and Sala-i-Martin(2020)find for up to 2015 that official GDP growth figures may be understated.The Long-Term Growth Prospects of the Peoples Republic of China9Projections from UN(

118、2019)show that the PRCs working-age population will decline in all(i.e.,low,medium,and high)fertility scenarios(Figure 9).In the low-fertility scenario,the working-age population will continue to decline fast after 2035,while in the other two scenarios the decline decelerates.According to data from

119、NBS,the PRCs birth ratethe number of live births per 1,000 of population per yearfell to a historical low of 7.5 in 2021,down from 14.6 in 2012,which was the highest rate reported within the past 2 decades.This sharp decline happened despite attempts to increase it by adopting the Two-Child Policy i

120、n 2015 and even a Three-Child Policy in 2021.Thus,we use the low fertility variant from the UN for our estimates.C.Human CapitalThe qualitative component of human capital is determined by years of schooling and returns to education.For educational attainment,we use the Barro-Lee data set(2021)ending

121、 in 2015.14 We construct five age groups:1524,2534,3544,4554,and 5459.Based on the PRCs education system,we assume 6,12,and 16 years of education for the completion of primary,secondary,and tertiary education.In line with Bailliu et al.(2016),we assume those who have attended but not completed prima

122、ry,secondary and tertiary education received 3,9,and 14 years of education,respectively(and the assumed return for no education is zero).As for returns to education,we assume different annual returns to education for primary,secondary and tertiary education as follows:7.69%,8.92%,and 13.38%(Bailliu

123、et al.2016),which averages 10%.15 While the returns-to-education approach provides a detailed picture,several steps are needed to obtain the human capital index and its forecast(Appendix 2).16 One critical assumption is higher education attainment rates and their development.Bailliu et al.(2016)fore

124、cast secondary and tertiary school attainment on the relationship between school enrolment rates and per capita income of high-income countries.However,the PRCs tertiary school enrolment rate has grown rapidly from 24.2%in 2010 to 58.4%in 2020(Figure 10).High-income countries achieved a tertiary sch

125、ool enrolment rate of about 58%in 2001,with GDP per capita closed to$40,000(in 2017$,PPP-adjusted),while the PRCs purchasing-power-adjusted GDP per capita was only$16,000 in 2020when the PRC already had reached 58%.We account for that in our estimate(see Appendix 2 for details).The human capital ind

126、ex is shown in Figure 11.The trend of the index is rising from 2000 to 2040.According to our estimate,a 11%average return to educationinstead of the assumed 10%returnwould increase the human capital index by 0.27 on average.14 In the data set,data are available every 5 years(i.e.,for 2000,2005,2010,

127、and 2015)for five age groups:1524,2534,3544,4554,and 5464.Gaps for years are filled by linear extrapolation.Furthermore,we approximate values for the age group 5459 as Barro-Lee(2021)extended the age group to 64.15 Ma and Iwasaki(2021)conducted a meta-analysis of 213 studies on returns to education,

128、confirming that a majority of previous studies reported about 10%(while some more recent estimates vary greatly).16 Roberts and Russell(2019)estimate trend growth of labor quality by adjusting for average years of schooling and use that estimate as the combined contribution from labor and human capi

129、tal.Higgins(2020)includes average hours worked and total employment as labor inputs besides human capital.10ADB East Asia Working Paper Series No.54Figure 10:Tertiary School Enrollment,2030405060708020025203020352040PRCHigh-income countriesUpper middle-income countries2.02.12.2

130、2.32.42.52.62.72.82.93.02000200520025203020352040Human Capital IndexPRC=Peoples Republic of ChinaSources:World Bank,World Development Indicators Database 2022;and authors estimates.Figure 11:Human Capital Index,20002040Source:Authors estimates.D.Capital The capital stock and its growth ra

131、te need to be estimated for the production function.The so-called perpetual inventory method is frequently used to determine the capital stock each year:capital stock at a starting point plus investment(each year)minus depreciation of the capital stock.Estimating the PRCs capital stock is challengin

132、g for the following reasons:first,the capital stock must be estimated for the starting date(no related statistics exist);second,estimates of the depreciation rate differ widelyHerd(2020)provides an overview.Finally,fixed asset investment(FAI)data frequently differs from that on gross fixed capital f

133、ormation,which can be only partly explained by their different methodology(Brooks and Barnett 2006).We use the capital stock estimate by the International Monetary Fund(IMF),which covers the period 19602019.Since these estimates are in current CNY,we have to address price effects.To do so,we deflate

134、 the capital stock growth with the GDP deflator(Figure 12),which results in an estimate close to that in Bailliu et al.(2016)for 20002014.Using the FAI index as deflator would yield a higher estimatewe further discuss deflator choices and cross-checks in Appendix 4.Figure 12:Capital Stock Estimates

135、with Different Deflators,200020200250Using GDP deflatorUsing FAI indexTrillion 2000CNY200020022004200620082000022004200620082002330500Share of housing in total(RHS)Including housingExcluding housingTrillion 2010CNYPercentFAI=fixed ass

136、et investment,GDP=gross domestic product.Sources:IMF Investment and Capital Stock Dataset(2021);NBS via CEIC Data Company(accessed 26 May 2022);and authors calculations.Figure 13:Estimated Capital Stock with and without Housing,20002016Note:In constant 2010 prices;different from Figure 12.Sources:He

137、rd(2020)via World Banks Development Data Hub(https:/datacatalog.worldbank.org/);and authors calculations.The Long-Term Growth Prospects of the Peoples Republic of China11Another potential challenge comes into bearing when separating the capital stock into housing and nonhousing.The argument behind t

138、his is that housing is not productive to produce goods and services,and hence should be excluded from the capital stock(Bailliu et al.2016).For that reason,Higgins(2020)uses a subset from Penn World Table that estimates the flow of services from the capital stock,but concedes that the results would

139、be basically the same using the capital stock series.We use the growth rate of the entire capital stock(including housing)for 20012019.Figure 13 shows Herds(2020)estimates of capital stock with and without housing as well as the share of housing in the capital stock,which varies over time.In Appendi

140、x 4,we provide a discussion on that issue.17 Furthermore,Box 1 will discuss overinvestment.As mentioned,the future growth rate of the capital stock is key for our estimate and this rate is expected to decline.In 20012019,the GDP-deflated capital stock grew annually by 10.4%on average.In the future,t

141、he growth rate of the capital stock should decline as the capital stock ages and depreciation increases,while investment as a share in GDP is forecast to fall(Figure 14).In order to estimate the future growth rate of the capital stock,we apply two methods.First,we look at the historical investment p

142、attern of Japan and the Republic of Korea(Figures 14 and 15)and estimate the PRCs share of investment in GDP based on the development pattern of these two countries.Second,we forecast the PRCs future capital-labor ratio in line with the previous development of the comparators and compare the results

143、 with the first method as a cross-check.Share of investment in GDP.Though the share of gross fixed capital formation(GFCF)in GDP has been higher in the PRC than in the Republic of Korea and Japan,it started to decline after 2010,broadly following the experience of the two comparators(Figure 14).18 T

144、wenty years after their respective peak in Japan and the Republic of Korea,the share of GFCF in GDP was about 30%on average and only slightly lower28%on average30 years after.According to our projection for the PRC,the share of GFCF in GDP will reach 32%by 2030(i.e.,20 years after the peak)and 29%by

145、 2040(or 30 years after),taking into account that it peaked at a higher rate in the PRC(Figure 14).19 This implies an annual growth rate of the capital stock of 4.2%on average in 20202040.17 The cross-checks in Appendix 4 show that the deflator choice matters for the capital stock growth rate,while

146、the exclusion of housing hardly affects the estimate of the growth rate.18 Dollar and Wei(2007)argue that the PRCs capital allocation has been inefficient and could be improvedwe will discuss both issues in part IV.19 Backing out the(growth in)capital stock from a projection of the capital-labor rat

147、io,Bailliu et al.(2016)forecast a decline in the investment rate to around 25%of GDP by 2030.According to our estimate,the decline until 2030 will be less pronounced,as GFCF still stood at 42%of GDP in 2020 and 2021.12ADB East Asia Working Paper Series No.54Figure 14:Gross Fixed Capital Formation,19

148、6020254045t-50t-45t-40t-35t-30t-25t-20t-15t-10t-5Peakt+5t+10t+15t+20t+25t+30t+35t+40t+45t+50PercentPRCJapanRepublic of KoreaGDP=gross domestic product,PRC=Peoples Republic of China.Note:Gross fixed capital formation as a share in GDP peaked in Japan in 1973,in the Republic of K

149、orea in 1991,and in the PRC in 2010.Data for the PRC starts in 1962(to exclude distortions from the Great Leap Forward).Sources:NBS and OECD via CEIC Data Company(accessed 31 August 2022);and authors calculations.Figure 15:Capital Formation and Labor Supply,20304050607019601970 1980199020

150、002010 2020 20302040PercentGross domestic savings(%of GDP)Gross fixed capital formation(%of GDP)Working-age population(%of total population)00701960 1970 1980 1990 2000 2010 2020 2030 2040PercentJapanPRC00701960 1970 1980 1990 2000 2010 2020 2030 2040PercentRepublic of KoreaGDP

151、=gross domestic product,PRC=Peoples Republic of China.Note:Working-age population refers to age 1559 for all three countries.Gross domestic savings for Japan is only available from 1970 onward.Sources:NBS,OECD and World Bank via CEIC Data Company(accessed 31 August 2022);UN(2019),World Population Pr

152、ospects(low fertility variant);and authors calculations.In the two comparator countries,the decline in GFCF as share in GDP happened quickly.Such a drop in GFCF(as share in GDP)has not happened in the PRC.However,such adjustment might happen in the PRC in the future as accumulated debt has risen sha

153、rply after 2008,forcing adjustment to the investment-led growth model of the 2010s(Box 1 at the end of this part).The Long-Term Growth Prospects of the Peoples Republic of China13Capital-labor ratio.Another way to estimate the future growth rate of the capital stock goes via the capital-labor ratio.

154、The PRCs current capital-labor ratio is close to that of Japan and higher than that of the Republic of Korea at the comparable GDP per capita level.In order to forecast the future capital stock,we use Japans and the Republic of Koreas expansion paths of their capital-labor ratios at similar GDP per

155、capita levels,but assume slightly higher ratios for the PRC given the historical precedent.This yields an annual capital stock growth of 4.2%on average in 20202040.Result.The forecast of capital stock derived from the share of GFCF in GDP approach and from the capital-labor ratio both yield virtuall

156、y the same average growth rate of the capital stock for 20202040,only the dynamics over time differ slightly.The contributions to GDP growth are initially higher with the first method(GFCF)which we usebut they also decline faster.E.Total Factor Productivity To explain the past trend of the contribut

157、ion of total factor productivity(TFP)to growth and to forecast TFP growth over the next 2 decades,we follow Bailliu et al.(2016)and decompose TFP improvement into four factors:(i)foreign direct investment(FDI),(ii)import growth,(iii)research and development(R&D)expenditure,and(iv)sectoral labor shif

158、t.Foreign direct investmentWe look at FDI as a driver of TFP growth.Lee,Lee,and Kim(2011)found a positive correlation between growth in FDI stock and TFP of 0.0218,i.e.,a 10%increase in the FDI stock improves TFP by about 0.2%.Following Bailliu et al.(2016),we deflate the nominal FDI stock,denominat

159、ed in US dollar,by the consumer price index(CPI)for the US to obtain an estimate of real FDI stock growth.We assume the same coefficient from Lee,Lee,and Kim(2011)for the forecast and assume that growth in the PRCs FDI stock will slow gradually to the simple average of the current real growth rates

160、in Japans and the Republic of Koreas FDI stock,which averaged close to 5%in 20152020.We assume the PRCs real growth in FDI stock will slow from 8.4%in 20152020 to 5.0%by 2030 and then marginally ease further to 4.0%in 2040(Figure 16).Import growthWe use the import index from the CPE Netherlands Bure

161、au for Economic Policy Analysis to get the growth rate of imports in real terms.In 20012019,annual real import growth averaged 3.8%(Figure 17).20 It is forecast to decline to 2.5%by 2030 and further to 2.0%by 2040.We further assume,in line with Lee,Lee,and Kim(2011),that a 1%increase in real import

162、growth raises TFP by 0.068 percentage points.21 20 The index is based on 2010 prices.Using the GDP deflator to calculate real import growth in 20032013,Bailliu et al.(2016)forecast a slightly higher contribution of real import growth to future TFP growth than we do.21 We use the regression result fo

163、r the biggest sample,i.e.,all countries covered in Lee,Lee,and Kim(2011).14ADB East Asia Working Paper Series No.54Figure 16:Real Growth in FDI Stock,20002040051015202530FDI stock%change year on year2000200520025203020352040FDI=foreign direct investment.Note:In 2000$.Forecast starts in 20

164、21.Possible distortions from COVID-19 for 2020 do not affect the assumed trend.Sources:United Nations Conference on Trade and Development via CEIC Data Company(accessed 26 May 2022);and authors estimates.Figure 17:Real Import Growth,200202530FDI stock%change year on year200020052010201520

165、202025203020352040200002820340%change year on year-15-10-5051015202530ImportsAverage 20012019Note:Index is based 2010=100.Forecast starts in 2020 to avoid distortions from COVID-19,which would be sizable.Sources:CPE Netherlands Bureau for Economic Policy A

166、nalysis,World Trade Monitor;and authors estimates.Research and developmentR&D is the major contributor to TFP improvement.In the PRC,annual R&D expenditure increased rapidly from less than 1%of GDP in the early 2000s to 2.2%of GDP in 2019.We assume annual R&D to reach 2.5%of GDP in 2030,and,by 2040,

167、the current 2.6%-of-GDP level of high-income countries(Figure 18).We assume that a 1%increase in R&D spending raises TFP by 0.4 percentage points for 20002019 and 0.23 points in 20202040.The elasticity chosen has a substantial effect on the TFPs contribution to growth because R&D is its biggest comp

168、onent.However,estimates of the elasticity of R&D expenditure on GDP growth vary widely.22 For 20002019,we use a coefficient of 0.4,which lies in between Griffith et al.(2004)and other estimates that are much lower.23 For 20202040,we assume a coefficient of 0.23,whichexcept for one outlieris the high

169、est elasticity reported in the compilation by McMorrow and Roeger(2009).24 As access to technology,industry-specific knowledge,and some markets has become more restrictive in recent years,the return to R&D should decline given complementarities of inputs for technological 22 Griffith et al.(2004)fou

170、nd a coefficient 0.669 on lagged level of R&D intensity,which Bailliu et al.(2016)uses.Haider,Kunst,and Wirl(2020)refute the finding in Griffith et al.(2004)that the elasticity is higher for countries that are further away from the international frontier.Also,the mentioned elasticity of nearly 0.67

171、seems to be very high and pertains only to the manufacturing sector(McMorrow and Roeger 2009;CBO 2005).Other estimates are much lower,frequently roughly ranging 0.10.2 for studies using economy-wide data(McMorrow and Roeger 2009).23 The choice of elasticity does not affect past TFP growth,which is e

172、stimated as a smoothed residual.Using a lower coefficient would only increase the unexplained part of the TFP estimate for 20012019.24 CBO(2005)states that the evidence is mixed on the question of whether the return to R&D has changed over time.The Long-Term Growth Prospects of the Peoples Republic

173、of China15products.25 In addition,top-down research decisions impose costs on research quality(Acemoglu,Yang,and Zhou 2022).Figure 18:R&D Expenditure,20002040 0.00.51.01.52.02.53.02000 2005 2010 2015 2020 2025 2030 2035 2040PRCHigh-income countries%of GDP 00702000 2005 2010 2015 2020 2025

174、 2030 2035 2040AgricultureIndustryServices%of total employmentGDP=gross domestic product,PRC=Peoples Republic of China,R&D=research and development.Sources:National Bureau of Statistics via CEIC Data Company;World Bank,World Development Indicators(both accessed 28 May 2022);and authors estimates.Fig

175、ure 19:Employment by Sector,20002040 Sources:Ministry of Human Resources and Social Security via CEIC Data Company(accessed 28 May 2022;and authors estimates.Sectoral labor shiftThe share of employment in primary industry has been falling from 50.0%in 2000 to 22.9%in 2021.Meanwhile,the share of empl

176、oyment in tertiary industry rose from 27.5%to 48.0%during the same period(Figure 19).We estimate the historical contribution of sectoral labor shifts by conducting a counterfactual scenario in two steps following Bailliu et al.(2016).First,we derive the productivity for primary,secondary,and tertiar

177、y industries from GDP and employment of each sector.Second,we conduct a counterfactual scenario by keeping the employment shares for all three sectors unchanged at 2000 levels while applying the productivity derived in the first step.We then calculate GDP growth in the actual and the counterfactual

178、scenario.The difference(actual growth minus counterfactual)is the estimated contribution of sectoral labor shift to TFP growth.To estimate the future contribution of the sectoral labor shift to growth,we need to forecast the development of each sectors employment share.To do so,we draw on the experi

179、ence of selected countries,including Japan,the Republic of Korea,and Brazil to forecast the share of agriculture in employment(Figure 20).26 With rising GDP per capita,the agricultural share of employment tends to fall.Data from selected countries suggest that when GDP per capita reaches$30,000(in 2

180、017$PPP),the agricultural share of employment is around 7%.However,as the PRC still has a larger share of workforce in the primary sector than these countries,the decline in the share of employment in agricultural will likely be more gradually,falling to 15%in 2030 and further to 11%in 2040 accordin

181、g to our estimates.25 PRC companies prefer to focus on developing new technologies and products with clear commercial application(Zheng,Zhuang,and Wang 2020).26 Brazil is chosen to compare adjustment in the primary sector at a per capita GDP in the range of$10,000$15,000(PPP).16ADB East Asia Working

182、 Paper Series No.54Further,we assumeunlike Balliu et al.(2016)but in line with newer research by Rozelle et al.(2020)that the employment share of the secondary sector,having peaked in 2012,will continue to decline at the average rate in 20122021.Finally,the share of tertiary employment,calculated as

183、 the residual,is expected to rise from 48%in 2021 to 57%in 2030 and 63%by 2040.Figure 20:Share of Agriculture in Total Employment 00010,00020,00030,00040,00050,000PRCRepublic of KoreaBrazilJapanPercentGDP per capita,2017$PPP GDP=gross domestic product,PPP=purchasing power parity,PRC=Peopl

184、es Republic of China.Note:Blue dashed line shows the forecast for the PRC up to 2040.Sources:World Bank,World Development Indicators;and authors estimates.To estimate the future contribution of the sectoral labor shift to growth,we create an alternative scenario,under which the productivity of the t

185、hree sectors is frozen at the 2019 levelthe year 2019 is chosen to avoid distortions by the COVID-19 shock on the economywhile the employment shares of each sector change.We assume that the total employment will decline at the same rate as the working-age population.We then calculate a baseline scen

186、ario of the level of GDP,where both the sectoral employment shares and the sectoral productivity are fixed at 2019 levels(while the decline in total working-age population is taken into account).Finally,we get the contribution of the sectoral labor shift to TFP growth as the difference in GDP growth

187、 between the alternative and the baseline scenario.For 20012019,we obtain TFP growth as the residual from equation(3)above.From the actual GDP growth,we deduct the contribution of capital,labor,and human capital to growth,assuming a capital and labor share of output of 0.5(Bailliu et al.2016;ILO 201

188、9).27 Then,we apply an Hodrick-Prescott(HP)filter to the residual and get the trend TFP(Figure 21).28 As illustrated in Figure 21,TFP growth slowed in 20012019.Based on our estimates,the slowdown stems mainly from contribution of the sectoral labor shift fading away.This is consistent with findings

189、by 27 Lardy(2019)discusses the labor share based on NBS data.The range quoted was 54%in 2002,falling to 47%in 2011.The share recovered to 52%by 2015.However,wage growth in the second half of the 2010s was lower than that of nominal GDP(Peschel 2021).The assumed 0.5 takes this into account.28 Bailliu

190、 et al.(2016)show similar results for(the residual measuring)TFP growth up to 2015 when their data end.Using data from World Penn Table(that assumes lower GDP growth rates for the PRC),IMF(2022a)finds a similar pattern of change in TFP growthi.e.,a notable decline in the 2020sbut at lower rates.Furt

191、hermore,estimates by Brandt et al.(2022)confirm that,comparing the 2000s to the 2010s,annual TFP growth per worker has declined by 2 percentage points on average.The Long-Term Growth Prospects of the Peoples Republic of China17Rozelle et al.(2020)showing that employment in the secondary sectorhaving

192、 a higher productivity than the service sectordeclined after 2012,while low-skill services absorbed part of the laid-off industrial workers.While we assume the sectoral labor shift to continue,it should slow down after 2030 when the employment share of agriculture declines at a slower rate,resulting

193、 in a downward shift in TFP growth(chart 22).Figure 21:Total Factor Productivity Growth,20012019Trend TFP(HP-filtered)TFP(Solow residual)Average 2000sAverage 2010sSectoral labor shiftFDIImportsR&DTrend TFP-10123456Percent200002820340012345678Percent2001200

194、3200520072009200172019HP=Hodrick-Prescott,TFP=total factor productivity.Source:Authors calculations.Figure 22:Decomposition of TFP Growth,20012040FDI=foreign direct investment,R&D=research and development,TFP=total factor productivity.Note:Forecast starts in 2020.Source:Authors estimates.

195、TFP growth averaged 3.2%in 20012019,with the sum of the four components adding up to 2.7%on average(Figure 22).The sectoral labor shift contributed on average 1.6 percentage points and R&D 0.7 points.Real imports and FDI together only contributed half a percentage point,with imports slightly more th

196、an FDI(0.3 versus 0.2 points),while half of a percentage point remains unexplained on average.29 For 20202040,potential annual TFP growth is estimated to average 1.3%,mainly driven by R&D(0.6%),followed by the sectoral labor shift,real imports,and FDI.The contribution of R&D to TFP stems from steady

197、 growth in R&D expenditure,resulting in R&D accounting on average for nearly half of TFP growth in 20202040,with a higher share in the 2030s when the contribution from the sectoral labor shift wanes.30 Our TFP forecast is lower than that of Bailliu et al.(2016),mainly due to different assumptions ad

198、opted with respect to the elasticity of R&D on TFP growth and regarding the sectoral labor shift.Our forecast is roughly in line with that by Roberts and Russell(2019)extrapolating long-term trends.Higgins(2020)projects in the“pretty good”scenario TFP growth to ease from 1.4%in 20182028 to 1.0%in 20

199、282038our estimates are only slightly higher for comparable periods.29 The unexplained part mostly disappears when looking from 2003 onwardFigure 22 shows a big gap between sum of the components and the overall for 20012002.30 CBO(2005)notes that research by Griliches(1991)argued that R&D spending,a

200、ssuming high social returns from R&D,could account for almost three-quarters of TFP growth during the postwar period.18ADB East Asia Working Paper Series No.54F.Results We use equation(4)to model GDP growth in 20012019 and then forecast potential growth for 20202040:31 (4),where is the model-fitted

201、GDP growth,and and(1-)are capital and labor share of GDP,with assumed to be 0.5;is the trend growth in capital stock;is the trend labor growth(i.e.,the growth rate of the working-age population);and is the trend human capital growth,which is derived from the growth rate of human capital index,whilei

202、s the trend TFP growth derived from an HP-filtered TFP growth estimate.32 The decomposition of contributions to GDP growth for 20012019 is illustrated in Figure 23.As in other studies(Bailliu et al.2016,Roberts and Russell 2019,and Higgins 2020),capital and TFP are key drivers of growth in 20012019.

203、The contribution of human capital was relatively small.For the 2010s,a declining population aged 1559(Figure 8)ceases the demographic dividend of the 2000s(and earlier).The adverse effects of a declining labor-age population will show more strongly after 2025.Figure 23:Decomposition of Contributions

204、 to Potential GDP Growth,20CapitalLaborPotential GDP growthTFPHuman capitalActual GDP growthPercent20000252026203020340GDP=gross domestic product,TFP=total factor productivity.Note:Forecast starts in 2020.The future contribution of capital

205、 to growth is derived from the projected share of gross fixed capital formation in GDP.Source:Authors estimates.Actual GDP growth averaged 9.0%in 20012019,the same average as potential growth according to our model.For the period 20062010 our estimate of potential growthwhich abstracts from business

206、 cyclesis below actual GDP growth.This is consistent with actual growth exceeding its medium-term potential before the global financial crisis(World Bank 2019;Lardy 2019).31 Due to the significant impact of the COVID-19 on the economy in 20202022,we use actual data until 2019 to avoid distortions fr

207、om the pandemic.32 In growth accounting,the contribution of TFP growth to GDP growth is calculated as a residual i.e.,the difference between actual GDP growth and the part of GDP growth explained by changes in capital and labor(including human capital).However,when estimating future potential GDP gr

208、owth,TFP growth cannot be calculated as a residual.Hence,we estimate the past TFP trend growth by smoothing the residual by an HP filter and find a set of proxy variables that match the TFP trend growth.We can then forecast TFP growth based on the projections of these variables.The Long-Term Growth

209、Prospects of the Peoples Republic of China19As for the forecast,the main findings are as follows:(i)Potential GDP growth is estimated at 5.3%on average in 20202025 and then to decline gradually to 2.0%on average in 20362040.(ii)Consistent with results from Bailliu et al.(2016),Roberts and Russell(20

210、19),and Higgins(2020),capital and TFP are major contributors to potential growth in the long run.The contribution of capital is forecast to gradually decline from 3.4 percentage points on average in 20202025 to 1.2 percentage points on average in 20362040.(iii)The contribution of TFP to growth is es

211、timated to be smaller in the next 2 decades because of a much smaller contribution from the sectoral labor shift and lower returns to R&D.Despite that,R&D expenditure will be the main driver of TFP growth in 20202040.33 (iv)A shrinking working-age population will increasingly weigh on growth,while h

212、uman capital will contribute on average about 0.4 percentage points to growth in 20202040.In a nutshell,the decline in potential growth stems from a lower contribution of capital to growth,a shrinking working-age population dragging on growth,and lower TFP growth.Table 2 shows that our estimate come

213、s in lower than Bailliu et al.(2016),see Appendix 3 for details.For the period up to 2030,our estimate is broadly in line with Roberts and Russell(2019),while being higher than Higgins(2020).At the same time,our forecast for 20312040 averages 2.3%,which is only slightly higher than that of Higgins(2

214、020)for 20292038.Table 2:Comparison of Selected Long-Term Growth Estimates for the PRCStudyPeriodGDP growthBailliu et al.(2016)202120255.9%202620304.9%Roberts and Russell(2019)202020304%5%Higgins(2020)(pretty good scenario)201820283.8%202920382.1%Authors forecast202020255.3%202620303.5%203120352.7%2

215、03620402.0%GDP=gross domestic product,PRC=Peoples Republic of China.Note:Only studies that apply a similar methodology.Sources:Bailliu et al.(2016);Roberts and Russell(2019);Higgins(2020);and authors estimates.G.Potential Impact of Recent DevelopmentsDespite the declining trend in potential growth,a

216、ctual GDP growth will almost certainly be below the forecast potential for the period 20202025.COVID-19 has weighed on economic growth in recent years and will likely continue to do so in 2023(ADB 2022a).An ailing property market has also dragged down growth,especially in 2022.In addition,the trade

217、conflict and technological competition with the US is another point adversely impacting GDP growth.These three factorsand how they are taken into accountcould also influence the potential growth estimate.33 As mentioned,there is a high uncertainty about the elasticity of R&D spending on TFP growth.2

218、0ADB East Asia Working Paper Series No.54COVID-19.The economic effects of COVID-19 are assumed to be temporary.We assume that the countrys zero-COVID policy will be loosened at one point in time,and thus expect the adverse effects of COVID-19-related restrictions be confined to the 20202025 period o

219、f our estimate.However,there are possible channels for COVID-19 to affect long-term growth.First,there could be scarring effects from human capital loss via the impact of COVID-19-related lockdowns on schooling.Second,the pandemic could delay investment given increased uncertainty.Finally,COVID-19 m

220、ight have adversely impacted labor force participation rates,possibly even in the longer run.The adverse effects of COVID-19 on schooling have been less pronounced in the PRC than in other countries in the region(ADB 2022b).34 Also,in the case of the PRC,it seems unlikely that COVID-19 much delayed

221、investment decisions as FDI recovered quickly from an initial shock in 2020 and domestic manufacturing investment,driven by an export boom,grew solidly in the past couple of years.The housing market slump is not primarily caused by COVID-19 but by borrowing curbs on developers(Box 1).As for labor ma

222、rket participation rates,the effects should be limited as state-owned enterprises tend not to shed staff during economic downturns,while labor migrantsthough hit hard by COVID-19-related restrictions will likely remain in the labor market,waiting for economic conditions to brighten.However,COVID-19

223、adversely affected household consumption,which in turn hit services and,together with the export boom and strong growth in industry in recent years,partly reversed the economys transition to services.35 Property market.The recent property market downturn continues to weigh on economic growth.Rogoff

224、and Yang(2022)estimate that the direct value added of the real estate sector(domestic content only)accounted for 11.8%of the economy in 2021,but when taking its connections to other sectors into account,the figure more than doubles to 25.4%(including imported content accounting for about 3 percentag

225、e points).Thus,property market investment affects economic growth beyond construction.The current housing market downturn in the PRC mainly stems from government deleveraging efforts by imposing borrowing curbs on developers.In our model,the decline in potential growth in the latter half of the 2020

226、s is mostly driven by a lower contribution from capital to GDP growth.While the future workforce can be determined with a high degree of certainty(absent pension reforms),the contribution of capital to growth could fall faster in case the deleveraging the property market leads to a faster-than-expec

227、ted decline in overall investment(as a share in GDP).For a discussion on the role of investment and debt,see Box 1.Technological competition and potential decoupling.Our assumptions take into account that barriers to acquire technology have increased and that there will likely be less knowledge exch

228、ange between the PRC and the rest of the world,lowering returns to R&D expenditure(and TFP growth)in the future.While the discount applied is already substantial,a technological decoupling could have adverse effects on potential growth stronger than assumed.36 Comparison.Our model predicts potential

229、 growth of 5.3%on average in 20202025(from 6.8%in 20162019),declining to 3.5%in 20262030.37 On average,our estimate of potential growth is 4.4%for 20202030.This is broadly in line with recent estimates that could take COVID-19 and a more restricted access to technology into account.Wang and Zhang(20

230、22)see potential GDP growth at 4.0%4.5%in this decade.Zhu,Ng,and Ge(2021)estimate potential growth at 5.3%in 20212025 and 4.4%in 20262030(with a higher contribution of TFP than our estimate).34 In the model,returns to education are assumed to be fix over time(but differ by level of schooling);adjust

231、ing them over time(possibly accounting for effects from COVID-19)is beyond the scope of this paper.At the same time,the contribution of human capital to growth is small so that the results of the model would not change substantially.35 See discussion in Box 3 later in part IV.36 Technological decoup

232、lingi.e.,the undoing of cross-border trade in high-tech goods and servicescould have a substantial adverse impact on the PRCs GDP growth depending on the coalitions of countries(Cerdeiro et al.2021).37 In our model,we assume a gradual decline in the share of GFCF in GDP by 2030.As we know(and take i

233、nto account)the actual GFCF in 20202021(at 42%of GDP),the effects of a falling GFCF share will show more strongly after 2025,explaining the drop in potential growth in 20262030.The Long-Term Growth Prospects of the Peoples Republic of China21Box 1:GDP Growth,Investment,and DebtAccording to our estim

234、ate,potential growth averages 4.4%in 20202030 and 2.3%in 20302040.An important question is if potential growth can be achieved.In theory,actual growth should be able to reach potential growth.While not being an input factor in the estimate of potential growth,debtespecially at an excessive levelcan

235、potentially hurt growth(Reinhart and Rogoff 2010;Cecchetti,Mohanty,and Zampolli 2011).a In the Peoples Republic of China(PRC),debt more than doubled from 139%of gross domestic product(GDP)at the end of 2008 to 287%by the end of 2021(Figure B1).It seems unlikely that the PRC can continue to accumulat

236、e debt at the speed witnessed in 20092019(without COVID-19),when debt rose by 124 percentage points of GDP,or on average over 11 points per year.Extrapolating this trend would result in a debt-to-GDP ratio of about 500%by 2040.Even Japan,the highest-indebted country covered by the Bank for Internati

237、onal Settlements(BIS)statistics,only had debt equivalent to 420%of GDP by the end of 2021.Figure B1:Debt Structure,20072022Figure B2:Credit Gap and Change in Debt,200820200250300Nonfinancial corporationsGeneral governmentHouseholds%of GDP-15-10-5051015202530-15-10-5050082009201

238、020000920000022Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q1Q0202021Credit-to-GDP gapDebt(RHS)Percentage point%change year on yearGDP=gross domestic product.Source:Bank for International Settlements.https:/www.bis.org/statistic

239、s/totcredit/totcredit.xlsx(accessed 2 December 2022).GDP=gross domestic product.Note:Credit-to-GDP gap is the difference between the credit-to-GDP ratio(for the private nonfinancial sector)and its long-run trend;debt of private nonfinancial sector.Source:Bank for International Settlements(accessed 6

240、 October 2022).Economic growth in the 2010s could have been driven by unsustainable overinvestment.The BIS provides estimates of the credit gapi.e.,the difference between the ratio of total credit relative to GDP and its long-run trendfor several countries,including the PRC(Figure B2).In our context

241、,credit gap analysis can be useful as overinvestment must be financed.A positive credit gap can be a sign of overinvestment.The credit gap shot up after 2008 and came down rapidly in 20172018.So,we might have seen about a decade of overinvestment.In 2018 infrastructure investment slowed sharply(Figu

242、re 3 in the main text)and new asset management rulesaiming to curb shadow bank financingwere announced(Economist 2018),which could help explain the quick narrowing of the gap.As for debt,deleveraging efforts targeting state-owned enterprise debt showed some effects in the second half of the 2010s be

243、fore the COVID-19 shock on the economy drove up debt again(Figures B1 and B2).Continued on next page22ADB East Asia Working Paper Series No.54For several reasons it seems unlikely that the PRC can continue their investment pattern for another 2 decades.First,housing demand is about to peak in 2022(Y

244、ao 2022).In this respect,one can interpret the government curbs on developers borrowing as an attempt to prevent them from overbuilding as well as to rein in growth generated by inefficient investment.b Second,the countrys physical infrastructure(airports,high-speed railway,highways,etc.)is already

245、well developed.Continuing to invest in it at high rates will not only yield declining returns to investment in terms of GDP growth,but would also overstretch local governments capacity for additional debt.As local governments receive a substantial share of their revenue from land sales,the current p

246、roperty market downturn puts pressure on local government finances.In addition,many infrastructure projects with high returns have been undertaken already so that remaining projects are less attractive.Finally,many provincesespecially poorer oneshave already accumulated a sizable stock of public deb

247、t,which makes it more difficult for them to shoulder additional debt(Figure B3).cFigure B3:Public Debt and Gross Regional Product Per Capita by Province,2021Debt-to-GRP ratio(%)GRP per capita(CNY 000)AHBJCQGSFJGZGDGXHEHLHAHNIMHNHBJSJXJLLNQHNXSNSDSHSXSCTJXZXJYNZJ0070809020406080

248、0180200AH=Anhui,BJ=Beijing,CQ=Chongqing,FJ=Fujian,GD=Guangdong,GRP=gross regional product,GS=Gansu,GX=Guangxi Zhuang Autonomous Region,GZ=Guizhou,HA=Henan,HB=Hubei,HE=Hebei,HI=Hainan,HL=Heilongjiang,HN=Hunan,IM=Inner Mongolia Autonomous Region,JL=Jilin,JS=Jiangsu,JX=Jiangxi,LN=Liaoning,NX=Ningxia Hu

249、i Autonomous Region,QH=Qinghai,SC=Sichuan,SD=Shandong,SH=Shanghai,SN=Shaanxi,SX=Shanxi,TJ=Tianjin,XJ=Xinjiang Uygur Autonomous Region,XZ=Tibet Autonomous Region,YN=Yunnan,ZJ=Zhejiang.GRP=Gross Regional Product.Note:Official public debt(general and special local government debt outstanding)only.Sourc

250、es:Ministry of Finance and National Bureau of Statistics via CEIC Data Company(accessed 10 October 2022);and ADB calculations.In sum,the growth model of the 2010s cannot be extended for another 2 decades.Reforms are needed,including improving investment efficiency and reducing the capital intensity

251、of growth.Notes:a See Pettis(2022)on ways how debt can hurt GDP growth in the case of the PRC.b Pettis(2022)sees this as an example of agents in an economic sector adjusting behavior to avoid potential losses.Before the government clampdown there was the assumption prevailing that the debt of develo

252、pers would be resolved through implicit government guarantees;now property developers,homebuyers,suppliers,contractors,banks,local governments,etc.,change behavior to protect themselves from defaults(such as withholding payments),which would lead to losses on their side.These behavioral changes caus

253、e economic growth to slow.c Specifically,more local governments are approaching the 10%threshold.In 2016,the PRC created rules requiring fiscal consolidation in case local government interest payments breach thresholds:(i)interest payments on general government bonds exceed 10%of finance spending in

254、 the general public budget,or(ii)interest payments on(local government)special bonds exceed 10%of spending in the government fund budget(He 2022;State Council 2016).Box 1 continuedIV.REFORMS TO BOOST LONG-TERM GROWTHA.Reform AreasThis part outlines possible reforms to enhance the countrys long-term

255、growth potential.The advantage of using a Cobb-Douglas production function is that it allows for a breakdown of the drivers of economic growth.Our model provide a baseline for long-term growthabsent major reforms and shocks38which can help guide policy making.Potential growth,though hard to change,c

256、an be increased,in particular by structural reforms.Mirroring the models input factors,we focus on reforms pertaining to labor,human capital,capital stock,and TFP.B.LaborThe model assumes that the work force grows at the same rate as the working-age population.If the retirement age is raised,the wor

257、king-age population would increase as more people were captured by the working-age bracket.In practice,such reforms are more complex because currently the retirement age for men is 60 years,and for women in blue-collar and white-collar jobs 50 and 55 years,respectively,while the average retirement a

258、ge is only in the mid-50s(Economist 2021).Another issue is demographics.Given the current low birth rates,an increase in fertility could slow the speed of decline of the workforce in the future,but not reverse trend in the foreseeable future.As we only look at until 2040not a long horizon in terms o

259、f demographicsand birth rates are known up to 2021,a gradual increase in the birth rate will hardly affect our projections.Also,the introduction of the universal Two-Child Policy in 2015 had shown no lasting effect(Figure 24),and the experience from other countries shows that the fertility rate rema

260、ins low once it has declined(Figure 25).Figure 24:Birth Rate,0045Births in the PRCPer 1,000 population050200520960505201234567PRCUSOECDRepublic of KoreaJapanBirths per womanPRC=Peoples Repub

261、lic of China.Note:Live human births per 1,000 population.Source:NBS via CEIC Data Company(accessed 2 December 2022).Figure 25:Fertility Rate for Selected Countries,19602020OECD=Organisation for Economic Co-operation and Development,PRC=Peoples Republic of China,US=United States.Source:World Bank,Wor

262、ld Development Indicators(accessed 11 July 2022).38 In the baseline model,we assume no major structural reforms and do not take into account business cycles.24ADB East Asia Working Paper Series No.54Against this background,the government should take measures that could mitigate the impact of demogra

263、phic aging on the labor force:(i)Extend work life.While the legal retirement age should be increased,it is important that people actually work longer.Reforms must start with people actually continuing to work beyond their mid-50sotherwise an extension of the statutory retirement age by some years co

264、mes to nothing.39 (ii)Improve health conditions of the workforce.Improved health care and promoting healthy aging is needed to cope with an aging workforce.While the focus should be on a safe and secure working environment,occupational health services,and occupational disease prevention,efforts are

265、needed to scale up screening,early diagnosis and controlling chronic conditions,as well as detecting cognitive impairment and depression at the primary care level(Bai and Lei 2020).(iii)Raise female workforce participation.40 Possible measures include strengthening equal employment opportunities,inc

266、reasing maternity leave,and improving support for childcare,elderly care,and single mothers(ADB 2021b).(iv)Increase labor mobility.This includes further relaxing the household registration system(hukou),improving access to social services for labor migrants(including the access to education for thei

267、r children),and ensuring the portability of social security benefits so that people do not have substantial disadvantages when moving.41(v)Increase family support.A massive shift of policy priorities and resources must take place to support families in order to raise the birth rate.C.Human CapitalRe

268、forms focusing on increasing human capital can be rolled out faster than those aiming at the labor force.Despite significant improvements over time,the PRCs educational attainment lags behind those in advanced economies,suggesting room for catching up.While the forecast takes into account an increas

269、e in the years of educationwhich serve as a proxy for the increase in human capitalquality of education is also important.An increase in human capital based on betternot necessarily moreeducation would contribute to higher potential growth.42 As mentioned,increasing average years of schooling is imp

270、ortant for raising human capital.The PRC saw significant improvement in education attainment levels over the past decades:the average years of schooling has increased substantially from 5.7 years in 1980 to 8.7 years in 2015(Figure 26).However,there is still a gap to advanced economies like Japan,th

271、e Republic of Korea,and the US.As a result,in 2015 only 37.4%of the PRCs labor force had completed upper secondary education,compared with 65.5%across G20 countries and 100%in Japan(Figure 27).Similarly,only 18.2%of the PRCs labor force 39 Such reforms tend to raise the statutory retirement age only

272、 by some years and are usually implemented gradually.40 According to household survey data by the Southwestern University of Finance and Economics,the labor force participation rate of women aged 1664 declined from 61.6%in 2011 to 58.2%in 2017,while that of men showed a much smaller decline from 76.

273、6%to 76.0%during the same period(Dong and Joffre 2019).41 Immigration of skilled people into areas with labor shortage would mitigate such shortages.Though unpopular,extending working hours would also mitigate the effect of a shrinking workforce.We do not further elaborate on these options as immigr

274、ation into the PRC hasleaving other challenges asidehigh language barriers.As for extending working hours,diminished productivity in some areas might stem from insufficient work or inadequate qualification of employees,which cannot be addressed by simply extending work hours.42 Quality of education

275、is hard to measure.Though it is not directly an input factor in the modelwhile years of schooling areimproving education increases potential growth because it raises work force skills and thereby productivity.The Long-Term Growth Prospects of the Peoples Republic of China25had completed tertiary edu

276、cation in 2015,which is lower than across G20 countries at 29.8%and much lower than in Japan,the Republic of Korea,and the US.However,tertiary education attainment has been rising in the PRC,reflected in a growing number of new college graduates.A record number of 10.76 million was expected by the g

277、overnment in 2022(Cheng 2022).Figure 26:Comparison of Average Years of Schooling,68101214Average years of schooling050520102015PRCG20OECDGermanyRepublicof KoreaUSJapan18.228.629.836.546.447.751.400Labor force with completed tertiary e

278、ducationPercentPRCG20OECDGermanyRepublicof KoreaUSJapan37.465.579.386.5 87.690.60100PercentLabor force with completed upper secondary educationPRCRepublic of KoreaG20JapanUSGermanyPRC=Peoples Republic of China,US=United States Note:For age group 1564.Source:Barro and Lee(2021).Figure 27:C

279、omparison of Education Attainment,2015OECD=Organisation for Economic Co-operation and Development,PRC=Peoples Republic of China,US=United States.Source:Li et al.(2017)There is a disparity in years of schooling between rural and urban areas in the PRC.The average years of schooling for the urban work

280、ing-age population was 11.4 years,versus 9.1 years in rural areas in 2019(Li 2021).43 The urbanrural gap in years of schooling has been narrowing since 1985,but has widened again after 2015(Figure 28).Also,there is a gap in the working-age populations senior secondary and tertiary school completion

281、rates(Figure 29).43 The China Human Capital Report defines working-age population in the PRC as ages 1654 for female and 1659 for male.26ADB East Asia Working Paper Series No.54Figure 28:Years of Schooling,68101214Urbanrural gapTotalUrbanRuralUrbanRuralAverage years of schooling36.148.354

282、.613.523.830.49.420.321.61.15.35.6002005200152019At least senior secondaryTertiaryPercent1985 1990 1995 2000 2005 2010 2015Note:Working-age population defined as 1654 for female and 1659 for male.Source:Li(2021).Figure 29:Urban and Rural School Completion Rates,2005,2015,and 20

283、19Note:Working-age population defined as 1654 for female and 1659 for male.Source:Li(2021).Against this background,the following policy recommendations are made:(i)Increase years of schooling.Greater efforts are needed to increase years of schooling,especially in rural areas,to bolster higher educat

284、ional attainment.Experiences from high-income countries show that those without post-secondary qualifications are most likely to face job loss and encounter long-term unemployment(ADB 2020b).(ii)Improve rural education and support children of migrant worker.More funding from the central government s

285、hould be allocated to rural education and children of migrant workers should get more support from the government to get easier access to education in cities(Li et al.2017).(iii)Improve the quality of education and qualification of the workforce.This includes expanding early child development,improv

286、ing the quality of tertiary education,strengthening technical and vocational education and training(TVET),and increasing on-the-job and lifelong training opportunities(see Appendix 5 for a discussion of these recommendations).D.CapitalOverviewChanges to the allocation of capital and credit might be

287、one of the biggest factors raising potential growth in the PRC,but they are challenging to implement because they require adjustments to economic structures as well as policy changes.The banking system is predominately state-owned,with large commercial banks providing about half of all loans.Bank lo

288、ans,in turn,account for the vast majority of company financing.Moreover,the state-owned banking sector dominates credit allocation and prefers to lend to state-owned enterprises(SOEs)as these frequently benefit from government support.44 44 Credit flows to SOEs were 83%of total credit to nonfinancia

289、l enterprises in 2016 latest available data(Lardy 2019).The Long-Term Growth Prospects of the Peoples Republic of China27In the following,we look at capital allocation and its efficiency as well as at credit allocation.Box 3 at the end of this section will cover structural change.Returns to investme

290、ntThe impact of investment on growth has declined in the PRC after the global financial crisis.The incremental capital output ratio(ICOR),which measures how much capital is needed to generate one extra unit of output,has increased.In 20002009,the ratio averaged 3.7,followed by 5.8 on average in 2010

291、2019(Figure 30).45 In 2019,it exceeded 7.46 Thus,more capital is needed to generate growth.An obvious question is why returns to capital have decreased and the ICOR has increased.Marginal productivity of capital generally declines with a higher capital stock per worker.Real return on capital has tre

292、nded down in the PRC the last quarter century(IMF 2022a).The decline,in particular over the last decade,could be driven by the composition of investment47 and/or the ownership of investment.The state holding share in total investment picked up after 2015,while that of privately held companies slight

293、ly eased,raising concerns over allocation efficiency(Figure 31).Figure 30:Incremental Capital Output Ratio,2000201902468ICOR007020042000200042005200620072008200920000007200820092000162017State holdingPrivate hold

294、ingPercentICOR=incremental capital output ratio.Sources:NBS via CEIC Data Company(accessed 20 September 2022);and ADB calculations.Figure 31:Fixed Asset Investment by Ownership Type,20042017Note:Shares do not add up to 100%as there are other categories in the statistic.Sources:NBS via CEIC Data Comp

295、any(accessed 26 September 2022);and ADB calculations.45 Calculated as investment ratio(GFCF as%of GDP)divided by GDP growth(Kwan 2004).46 As GDP growth slowed sharply in 2020,while investment was the main contributor to growth as consumption took a hit from COVID-19,the ratio spiked that year,but re

296、versed in 2021(not shown in chart).47 The available statistics on investment are sketchy.Estimates from Herd(2020)until 2016 show that the share of capital formation in the business sector declined over time and the shares of infrastructure and housing increased,while the capital-output ratio in the

297、 infrastructure sector increased the most,followed by the housing and the business sector,which explains the rise in the ICOR(Brandt et al.2020).28ADB East Asia Working Paper Series No.54State-owned enterprises and supply-side structural reforms The share of employees working at SOEs has declined ov

298、er time(Figure 32).At the same time,SOE assets(as%of GDP)are several times higher in the PRC than in other countries(IMF 2021b)and have increased after the global financial crisis(Figure 33).However,information is only readily available on industrial state holding enterprises,i.e.,enterprises where

299、the state has a dominant equity share or otherwise actual control.48 These tend to be less efficient than private enterprises and about one in four of them is loss-making(Box 2).Figure 32:Employment by Category of Employer,4060802019S

300、OEsCollectivePrivateOtherPercent0500300020042005200620072008200920000192020AgricultureConstructionSOE assets(RHS)IndustryServicesPercent%of GDPSOE=state-owned enterprise.Sources:Ministry of Human Resources and Social Security via CEIC Data Compa

301、ny(accessed 16 September 2022);and ADB calculations.Figure 33:SOE Assets,and Return on Assets by Sector,20032020GDP=gross domestic product,SOE=state-owned enterprise.Note:Nonfinancial SOEs only.No data on return on assets of industrial SOEs in 20172018.Sources:Ministry of Finance and NBS via CEIC Da

302、ta Company and WIND(accessed 2 December2022);and ADB calculations.The return on assets(ROA)of SOEs in the service sector has first risen before the global financial crisis andafter a rise followed by a long-stretched declinestood in 2020 again approximately at its 2003 value(Figure 33).The ROA of in

303、dustrial SOEs has declined over time until the mid-2010s,and then improved by nearly 1 percentage point from 2.0%in 2015 to 2.9%in 20182019,reflecting supply-side structural reforms in 20162018.49 Supply-side structural reforms focused on the steel,coal and electricity sectors,and zombie enterprises

304、 therein.In 2016,the National Development and Reform Commission released detailed 5-year plans for the steel and coal industries,aiming at consolidating the sector.Other industries targeted for capacity reduction included cement,glass,and agriculture(Boulter 2018).At the same time,SOEs in the servic

305、e sector have gained less attention and have not seen their ROA rise(Figure 33).Positive effects of structural reforms included closing some zombie firms and highly environment-polluting,energy-inefficient companies.It should also be noted that the government invested in industrial upgrading so that

306、 supply-side reforms were not limited to closing firms.The reforms also contributed,as part of a 48 SOEs are enterprises whose entire assets are state-owned and are registered as an unincorporated economic organization.49 Kroeber(2020,chapter 14)notes that supply-side structural reforms started to c

307、ollide with economic growth targets in 2018 and were shelved.Boulter(2018)notes that the profit margin in the steel sector improved in 20162018.The Long-Term Growth Prospects of the Peoples Republic of China29broader deleverage campaign,to a temporary stabilization in corporate debt(Boulter 2018 and

308、 Figure 41).The overall positive experience of supply-side structural reforms suggests exploring possibilities to extend them to other industries.There remain several SOE-related issues that need to be addressed.Data available on industrial state holding enterprises paints the following picture(Box

309、2):(i)The share of state holding enterprises in industry declined rapidly in the early 2000s,while in the 2010s this development decelerated.(ii)The share of loss-making industrial state holding enterprises declined in the first decade of the 2000s,but this trend stopped in the 2010s and the share h

310、as hovered around at one in four.Annual losses by industrial state holding enterprises averaged 0.5%of GDP in the 2010s.(iii)Industrial state holding enterprises have,on average,lower returns on assets and on equity than their private peers.(iv)Industrial state holding enterprises used to have a hig

311、her liabilities-to-asset ratio as well as a higher liabilities-to-equity ratio than their private peers.However,industrial state holding enterprises have reduced their liabilities-to-asset ratio in recent years,while that of private industrial companies rose sharply in 2018.There are reform needs wi

312、th respect to SOEs.Four areas are frequently highlighted:50(i)Clarify SOEs scope and function.While the function and scope of SOEs needs clarifications,in particular the areas in which the government wants to retain SOEs should be small and well-specified.Competition should be strengthened by making

313、 market entry easier for private companies in the other areas.This could increase pressure on inefficient SOEs to increase their efficiency.Monopolies should be kept to a minimum and the use of administrative monopolies(i.e.,exclusive rights granted by regulations)reduced.(ii)Level the playing field

314、.There should be a level playing field for the private sector to compete with SOEs,in particular in terms of access to land and credit.This includes removing implicit guarantees for SOEs and their preferred access to bank credit.(iii)Separate social functions from SOEs.In the PRC,many SOEs also perf

315、orm social functions,such as stabilizing employment and providing social services that are in other countries frequently fall into the category of basic public service.Social functions should be separated from SOEs to increase their competitiveness and better realize economies of scale and scope.(iv

316、)Improve SOE management.The management and supervision of SOEs must be strengthened to raise their efficiency.Elements such as aligning incentive mechanisms and gradually adopting market-based salaries are mentioned in the 14th Five-Year Plan.50 See for instance more comprehensive discussions in Wor

317、ld Bank and DRC(2013 and 2019);IMF(2019,2021b,and 2021c);ADB(2021d);and OECD(2022a).30ADB East Asia Working Paper Series No.54Box 2:Industrial State Holding Enterprises Figure B4:Industrial State Holding EnterprisesThe share of state holding enterprises in industrial assets fell sharply in the 2000s

318、.0.00.20.40.60.81.01.205540Share of loss-making state holding enterprises Losses by state holding enterprises(RHS)Percent%of GDP303540455055606570Share of state holding enterprisesPercent2000200220042006200820000020022004200620082001620182020Since the glob

319、al financial crisis,the share of loss-making state holding enterprises hovers around one-quarter.Private industrial enterprises have higher returns on assets and on equity.200020022004200620082001020304050PercentState holding enterprisesPrivate enterprises00200220042

320、00620082001620182020ROA state holding enterprisesROA private enterprisesROE state holding enterprisesROE private enterprisesPercentState holding enterprises pay a higher share of their pre-tax profits on interest payments.The liabilities-to-asset ratio of central SOEs has been higher than

321、 those of industrial state holding enterprises after 2008.20002002200420062008200000150160170PercentState holding enterprisesPrivate enterprises20002002200420062008200055606570PercentState holding enterprisesPrivate enterprisesCentral SOEsLiabil

322、ities-to-equity ratios of state-holding enterprises and private enterprises have converged in recent years.GDP=gross domestic product,ROA=return on asset,ROE=return on equity,SOE=state-owned enterprise.Note:The data is not controlled for industry.IMF(2022b)controls capital productivity in manufactur

323、ing for industry composition of SOEs,with similar results to the ones in the chart above on return on assets.Central SOEs under the State Council include non-industrial SOEs.Sources:NBS and Ministry of Finance via CEIC Data Company and WIND(accessed 16 September 2022);and ADB calculations.The Long-T

324、erm Growth Prospects of the Peoples Republic of China31Credit allocationWhile credit allocation can theoretically be analyzed from various angles,there is only limited credit-related information available for the PRC.Furthermore,many relevant statistics end in 2019(by the latest).Thus,in the followi

325、ng we analyze broad trends in credit,using available information on business loans,the banking sector structure,and the evolution of key interest rates and debt.From the statistics on total social financinga broad credit aggregate including bank loans,shadow bank financing,equity financing,and gover

326、nment and corporate bonds(and some smaller other items)some broad trends relevant to company financing can be inferred.Figure 34:Selected Financing Sources,200220204200520062007200820092000000220032004200520062007200820092001320142015

327、20020406080100LoansShadow bank financingCorporate bondsEquityPercent-750510152025Share in credit outstandingShadow bank financing outstanding(RHS)%increase year on yearNote:Loans by financial institutions.The sum of the four categories is normalized to 100%.Shadow ba

328、nk financing comprises entrust loans,trust loans,and banks acceptance bills.Sources:Peoples Bank of China via CEIC Data Company(accessed 20 September 2022);and ADB calculations.Figure 35:Shadow Bank Financing,20022021Note:Credit defined as the sum of loans by financial institutions and shadow bank f

329、inancing outstanding.Definition of shadow bank financing as in Figure 34.Sources:Peoples Bank of China via CEIC(accessed 20 September 2022);and ADB calculations.The share of equity financing has stayed very low,while bond financing could gain share over time(Figure 34).Secondly,shadow bank financing

330、judging from the regulated part covered by the statisticshas been reined in since the mid-2010s(Figure 35).The share of shadow bank financing in overall creditdefined here as sum of bank loans plus regulated shadow bank financinghas fallen to below 9%in 2021,down from about 20%in 2013 and 2014.Busin

331、ess loans.In 2019,outstanding business loans equaled to 56.4%of bank loans(or 87.3%of GDP).Among business loans outstanding,the top three categories were manufacturing(9.6%);transport,storage,and postal services(9.1%);and leasing and commercial service(8.4%).Three other categories had a share more t

332、han 5%:real estate(6.4%),water conservancy,environment and utility management(6.2%),and wholesale and retail trade(6.1%).These top six categories accounted for nearly half(45.8%)of business credit outstanding.Total loans have risen faster than business loans up to 2019,the latest available year(Figu

333、re 36).Small and medium-sized enterprise(SME)loans grew as percentage of GDP but,after 2012,expanded only broadly in line with business credit so that its share in business loans stagnated and even declined as a share in total 32ADB East Asia Working Paper Series No.54loans.In 2017,there was a changed in statistical categories and SME loans were discontinued.Loans to micro and small business,a new