1、ASIAN DEVELOPMENT BANKASIAN DEVELOPMENT BANK6 ADB Avenue,Mandaluyong City1550 Metro Manila,Philippineswww.adb.orgADB ECONOMICSWORKING PAPER SERIESNO.715February 2024Factors Affecting Micro,Small,and Medium-Sized Enterprise Development in Developing AsiaFindings from a Probabilistic Principal Compone

2、nt Analysis To identify factors affecting micro,small,and medium-sized enterprise(MSME)development,this paper proposes a probabilistic principal component analysis method that works despite current data limitations.The estimation results suggest that sound MSME credit markets,diversified financing o

3、ptions,support for new businesses and job creation,and active MSME participation in global marketplaces play a critical role in ensuring a smooth business recovery from various crises and shocks affecting developing Asia and the Pacific.About the Asian Development BankADB is committed to achieving a

4、 prosperous,inclusive,resilient,and 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 investment

5、s,guarantees,grants,and technical assistance.FACTORS AFFECTING MICRO,SMALL,AND MEDIUM-SIZED ENTERPRISE DEVELOPMENT IN DEVELOPING ASIAFINDINGS FROM A PROBABILISTIC PRINCIPAL COMPONENT ANALYSISShigehiro Shinozaki,Daisuke Miyakawa,and Romeo ArahanASIAN DEVELOPMENT BANKThe ADB Economics Working Paper Se

6、ries presents research in progress to elicit comments and encourage debate on development issues in Asia and the Pacific.The views expressed are those of the authors and do not necessarily reflect the views and policies of ADB or its Board of Governors or the governments they represent.ADB Economics

7、 Working Paper SeriesShigehiro Shinozaki,Daisuke Miyakawa,and Romeo ArahanNo.715|February 2024Shigehiro Shinozaki(sshinozakiadb.org)is a senior economist and Romeo Arahan(rarahan.consultantadb.org)is a consultant at the Economic Research and Development Impact Department.Daisuke Miyakawa (damiyakwas

8、eda.jp)is a professor at Waseda University,and chief economist of UTokyo Economic Consulting Inc.(UTEcon).Factors Affecting Micro,Small,and Medium-Sized Enterprise Development in Developing Asia:Findings from a Probabilistic Principal Component AnalysisCreative Commons Attribution 3.0 IGO license(CC

9、 BY 3.0 IGO)2024 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 2024.ISSN 2313-6537(print),2313-6545(electronic)Publication Stock No.WPS240032-2DOI:http:/dx.doi.org/10.22617/WPS240032-

10、2The views expressed in this publication 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 accept

11、s no responsibility for any consequence 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

12、 territory or geographic area,or by using the term“country”inthis publication,ADB does not intend to make any judgments as to the legal or other status of any territory or area.This publication is available under the Creative Commons Attribution 3.0 IGO license(CC BY 3.0 IGO)https:/creativecommons.o

13、rg/licenses/by/3.0/igo/.By using the content of 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 c

14、opyright materials in this publication.If the material 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 pubsmarketing

15、adb.org if you have questions or comments with respect 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.Not

16、e:In this publication,“$”refers to United States dollars.ABSTRACT Limited data on micro,small,and medium-sized enterprises(MSMEs)make it difficult for governments to design appropriate MSME policies in Asia and the Pacific.To identify factors affecting MSME development and promote evidence-based pol

17、icymaking,we propose a probabilistic principal component analysis method that works despite current data limitations.The study uses time-series MSME data collected from 25 developing member countries of the Asian Development Bank(ADB)through the Asia Small and Medium-Sized Enterprise Monitor project

18、.The estimation results suggest that sound MSME credit markets,diversified financing options,support for new businesses and job creation,and active MSME participation in global marketplaces play a critical role in ensuring a smooth business recovery from various crises and shocks affecting developin

19、g Asia and the Pacific.Keywords:SME development,access to finance,financial inclusion,SME policy,probabilistic principal component analysis,Southeast Asia,South Asia,Central and West Asia,the Pacific JEL codes:D22,G20,L20,L501.Introduction Developing Asian economies continue to recover from the coro

20、navirus disease(COVID-19)pandemic that began in March 2020,although economic growth differs by country.Continuous global economic uncertainty,however,has amplified downside risksincluding high inflation,currency depreciation,and global supply chain disruptions accelerated by regional political turbu

21、lence.In Southeast Asia,a recovery in tourism partly contributed to the regions 5.6%growth in 2022;but it is forecast to drop to 4.6%in 2023 given continued weak exports.In South Asia,economic and political crises in Pakistan and Sri Lanka pushed the regions growth down from 6.7%in 2022 to 5.4%in 20

22、23.In Central and West Asia,the ongoing impact from the Russian invasion of Ukraine helped lower the regions growth from 5.1%in 2022 to a forecast 4.6%in 2023.In the Pacific,a strong post-pandemic tourism rebound energized the regions sharp economic recovery to 6.1%growth in 2022;but it is forecast

23、to slow to 3.5%in 2023 partly due to labor shortages accelerated by emigration from small island countries to Australia and New Zealand(ADB 2023a).Micro,small,and medium-sized enterprises(MSMEs)help drive growth across developing Asia and the Pacific,given their large share of business enterprises,j

24、ob creation,and economic output.Given their impact,governments in the region have taken several policy measures to promote MSME development.They commonly promote entrepreneurial development(especially for youth and women),use of technology that encourages business innovation,expanded market access b

25、y internationalizing MSMEs,human capital and skills development,and better access to finance.But constraints on MSME development remain in most countries.These include a lack of an entrepreneurial culture,high dependence on cash transactions that stymie innovation,a large percentage of unregistered

26、or informal businesses,limited exports or participation in global markets,skilled labor shortages,and structural problems limiting access to formal financial services for working and growth capital.This raises the question how governments can enhance policies and their implementation to promote MSME

27、 development toward more inclusive,resilient growth.Better understanding the MSME business environment and structural problems that inhibit growth is critical before designing a feasible policy framework on MSME assistance.However,the lack of data on MSMEs makes this extremely difficult.To help gove

28、rnments promote evidence-based MSME policymaking,the Asian Development Bank(ADB)has,since 2020,provided benchmark indicators on MSME development and access to finance through its annual Asia Small and Medium-Sized Enterprise Monitor(ASM).As of November 2023,the ASM covers MSMEs in 25 ADB developing

29、members in Southeast Asia,South Asia,Central and West Asia,and the Pacific.Insufficient data,however,remains a major problem.A solid quantitative evaluation on MSME development using sufficient,accurate,and comparable data remains a challenge both nationally and regionally.Incomplete data on MSMEs l

30、ed global institutionssuch as the Organisation for Economic Co-operation and Development(OECD),the Economic Research Institute for ASEAN and East Asia(ERIA),and the International Trade Centre(ITC)to propose a qualitative approach using assessment matrices for performance ratings or median comparison

31、s based on available data to evaluate MSME development and competitiveness,both nationally and regionally.The ASM project has also explored a new way to quantitatively identify factors affecting MSME development through its ASM database.In 2021,it developed a new trial that deals with MSME data limi

32、tationsa variant of a standard principal component analysis(PCA)that supplements some missing MSME dataa probabilistic PCA(ADB 2022).The pilot test covered 15 countries 2 from Southeast Asia and South Asia along with a firm-level data analysis for Viet Nam.While this contributed to the new MSME deve

33、lopment index,insufficient data limited the proposed models ability to estimate more accurately the factors that represent MSME activities.More test-runs for additional countries are needed to produce a reliable index conducive to evidence-based policy design on MSMEs in the region.In 2023,we succes

34、sfully compiled time-series MSME data covering 25 countries.With this new dataset,this study re-estimates factors that explain the MSME development path by region and country and rethinks how to develop a quantitative approach to better assess MSME development.Section 2 summarizes the MSME landscape

35、 in developing Asia,extracted from ADB(2023b).Section 3 reviews global MSME data initiatives in Asia and the Pacific.Section 4 explains the methodology and dataset used for analysis.Section 5 discusses the estimation results in four groups(i)all countries,(ii)Southeast Asia,(iii)South Asia,and(iv)Ce

36、ntral and West Asia.This is followed by associated policy implications in Section 6 and conclusions in Section 7.2.MSME Landscape in Developing Asia MSMEs dominate the private sector in Asia and the Pacific.According to ADB(2023b),based on available data in participating countries through 2022,MSMEs

37、 in Asia and the Pacific accounted for an average 96.6%of all enterprises,55.8%of the total workforce,and 28%of a countrys economic output(Table 1).Data collected depend on the national MSME definition of each country.Most MSMEs serve small domestic markets,with many engaged in distributive trade an

38、d informal business.Cash dominates their business model and there is little incentive to grow furthercategorized as“stability-oriented”firms.With a large base of informal businesses,the official MSME contribution to a countrys economic output is likely well below its actual impact.Nonetheless,“growt

39、h-oriented”and innovative firms that want to expand into global markets have gradually increased across the region,although they remain a small fraction of MSMEs.Based on available data through 2022,MSME exports accounted for an average 26.3%of total export value.And MSME export growth is slowing,ma

40、inly due to the weak export environment globally.Low business diversification limits a countrys growth potential,suggesting the need for creating more innovative and globalized small firms,startups,and an entrepreneurial base,both nationally and regionally.Limited access to finance remains a chronic

41、 barrier to MSME growth.The MSME credit market remains small in Asia and the Pacific.ADB(2023b)reported that bank loans to MSMEs averaged 10.6%of a countrys gross domestic product(GDP)and 22%of total bank lending.The pandemic response boosted commercial bank lending to MSMEs,provided government emer

42、gency financial assistance or strengthened new lending to MSMEs through subsidized loan programs,refinancing facilities,and special credit guarantees.Despite this,MSME nonperforming loans remained high,averaging 7.2%of total MSME bank loans in the region.The lack of alternative financing options bey

43、ond traditional bank credit limits innovation and business opportunities for viable MSMEs,startups,and entrepreneurs.3 Table 1:MSMEs in Developing Asia and the Pacific(percentage share)All Countries Southeast Asia South Asia Central and West Asia MSME development Number of MSMEs to total enterprises

44、 96.6%98.0%99.6%99.2%MSME employees to total employees 55.8%66.4%76.6%51.9%MSME contribution to economic output 28.0%41.2%17.7%41.5%MSME exports to total export value 26.3%13.3%37.4%28.3%Access to finance(bank credit)MSME loans to national GDP 10.6%13.3%5.2%11.1%MSME loans to total bank loans 22.0%1

45、2.3%12.5%33.1%MSME NPLs to total MSME loans 7.2%5.3%12.1%4.3%GDP=gross domestic product,MSME=micro,small,and medium-sized enterprise,NPL=nonperforming loan.Notes:Reporting countries only.Data based on latest available data until 2022.Data for all countries cover 25 countries:10 from Southeast Asia;5

46、 from South Asia;7 from Central and West Asia;and 3 from the Pacific.Source:Asia SME Monitor 2023 database.3.Global MSME Data Initiatives Several global initiatives are developing indices to measure specific aspects of MSMEssuch as access to markets,infrastructure,finance,skills development,use of t

47、echnology and innovation,business operations and administration,competitiveness,and policy and regulatory frameworks(Table 2).Multilateral organizations such as the OECD,ERIA,ITC,and World Bank Group have been using various analytical approaches to overcome the lack of sufficient MSME data.The OECD

48、produces two related reports on SME development:(i)the SME and Entrepreneurship Outlook and(ii)Financing SMEs and Entrepreneurships(OECD Scoreboard).Launched in 2002,the Entrepreneurship Outlook reviews 6 dimensions with 29 subdimensions using cross sectional data for median comparison.Dimensions in

49、clude(i)institutional and regulatory frameworks,(ii)market conditions,(iii)infrastructure,(iv)access to finance,(v)access to skills,and(vi)access to innovation assets(OECD 2023).The subdimensions include(i)regulations,courts and laws,land and housing,public governance,competition,and taxation;(ii)do

50、mestic markets,global markets,public procurement,and trade and investment;(iii)logistics,energy,research and development(R&D)and innovation,the internet and information and communications technology(ICT);(iv)self-funding,debt,the financial system,and alternative instruments;(v)adult literacy,the lab

51、or market,entrepreneurial culture,training,and education;and(vi)technology,R&D,organization and processes,marketing,and data.It covers OECD members,including,from Asia,Australia,Japan,New Zealand,and the Republic of Korea.The OECD Scoreboard,launched in 2012,is an annual report focusing on trends in

52、 SME financing and policies for 48 countries.In 2022,it included 11 countries from AsiaAustralia,Georgia,Indonesia,Japan,Kazakhstan,Malaysia,New Zealand,the Peoples Republic of China(PRC),the Republic of Korea,Thailand,and Trkiye.It reviews 5 financial dimensions with 25 subdimensions(indicators):(i

53、)allocation and structure of bank credit to SMEs;(ii)extent of public 4 support for SME finance;(iii)credit costs and conditions;(iv)nonbank sources of finance;and(v)financial health(OECD 2022).The OECD constructs the indicators mainly using supply-side data from standardized forms filled in by bank

54、s,other financial institutions,statistics offices,and government agencies.The core indicators include total lending(overall and SMEs),new lending(overall and SMEs),short-versus long-term SME loans,direct government SME loans,government loan guarantees,interest rates(overall and SMEs),collateral(SMEs

55、),and bankruptcies(SMEs),among others.The OECD and ERIA produced an ASEAN SME Policy Index in 2014 and 2018 outlining the policy landscape for SME development.It evaluates the scope and intensity of SME development policies through 8 dimensions with 25 subdimensions:(i)productivity,technology,and in

56、novation;(ii)environmental policies targeting SMEs;(iii)access to finance;(iv)access to markets and internationalization;(v)institutional framework;(vi)legislation,regulation,and taxes;(vii)entrepreneurial education and skills;and(viii)social enterprises and inclusive entrepreneurship(OECD and ERIA

57、2018).These are measured in three stages:(i)planning and design;(ii)implementation;and(iii)monitoring and evaluation.Participating governments share their SME data and assess SME policies.They also conduct surveys of key stakeholders and private sector representatives to help supply missing informat

58、ion needed for qualitative analysis.For each subdimension,respondents score the strengths and weaknesses of current SME policies on a scale from 1 to 6,with higher scores indicating a better level of policy development and implementation.The ITCs SME Competitiveness Outlook annually reviews SME deve

59、lopment and financing conditions in 85 countries including several from Asia(ITC 2022).It aims to facilitate implementation of United Nations Sustainable Development Goals 8 and 9.The report produces an SME Competitiveness Index based on 3 dimensions with 39 subdimensions:(i)firm capabilities(SMEs a

60、bility to manage resources under its control);(ii)business ecosystem(resources and competencies needed to enhance a firms competitiveness);and(iii)national environment(government functionality and policy implementation).Each dimension is measured on three abilities:(i)capacity to compete(enterprise

61、efficiency);(ii)capacity to connect(information and knowledge gathering/exploitation);and(iii)capacity to change(human and financial capital investments).The index assesses the competitive strengths and weaknesses by firm size on a 0100 scale,analyzing time-series data obtained from secondary data i

62、ncluding(i)the World Banks Enterprise Surveys,Ease of Doing Business Index,and Logistics Performance Index;(ii)the International Monetary Funds(IMF)World Economic Outlook;and(iii)the ITCs Market Access Map.Firm size classifications use the definition from World Bank Enterprise Surveys.1 Strengths an

63、d weaknesses are measured based on a reference level of per capita GDP.As mentioned,the World Bank Group regularly releases three related reports:(i)the International Finance Corporation(IFC)MSME Finance Gap Report;(ii)the Enterprise Surveys;and(iii)the Doing Business report.The latest IFC report wa

64、s released in 2017(with updates as needed),covering 128 countries including 29 ADB developing members(IFC 2017).Data cover general indicators such as MSME landscape,bank lending,and nonbank finance data.It estimates the potential demand for financing in emerging economies compared with current suppl

65、y,and calculates the“finance gap.”The report is considered a benchmark of MSME financing needs in 10 advanced economies.MSME categories include industry(manufacturing,services,or retail),1 The Enterprise Surveys define small firms as having 5-19 employees,medium firms 20-99,and large firms 100+emplo

66、yees.https:/www.enterprisesurveys.org/en/methodology.5 size,and age.The World Bank Enterprise Surveys furnish the data necessary in conjunction with benchmarking for estimating the potential demand for MSME finance.The World Banks Doing Business report offers thematic firm-level data for more than 1

67、30 countries covering five dimensions:(i)starting a business;(ii)hiring and firing workers;(iii)enforcing a contract;(iv)getting credit;and(v)closing a business.The report was discontinued in 2021 to be replaced by the Business Enabling Environment(BEE)Project(as of December 2022).BEE will focus on

68、similar indicators,adding an indicator on market competition and removing the section on protecting minority investors.BEE will add analyses on digital adoption,environmental sustainability,and gender equality.Although not focusing on MSMEs,other global initiatives on developing relevant indices inc

69、lude(i)the Global Competitiveness Index from the World Economic Forum(WEF);(ii)the Global Innovation Index from the World Intellectual Property Organization(WIPO);and(iii)the Global Entrepreneurship Index from the Global Entrepreneurship and Development Institute(GEDI).The Global Competitiveness Ind

70、ex,started in 2005,covers 141 economies including several in Asia(WEF 2019).It analyses four dimensions of competitiveness,including the enabling environment,markets,human capital,and innovation ecosystem.It uses aggregate data sourced from international organizations such as the World Bank and resu

71、lts of the WEF Executive Opinion Survey conducted for business executives.The Global Innovation Index was launched in 2007 as a measuring tool for innovation in a society by using cross-sectional data for median comparisons(WIPO 2022).It is subdivided into two subindicesinnovation inputs and innovat

72、ion outputs.The Innovation inputs sub-index offers a snapshot of societys enabling environment for innovation and innovative activities.Five areas are monitored:(i)institutions;(ii)human capital and research;(iii)infrastructure;(iv)market sophistication;and(v)business sophistication.The innovation o

73、utputs sub-index measures the results of innovative activities by evaluating(vi)knowledge and technology outputs and(vii)creative outputs.The average scores from both indices comprise the overall score.The 2022 edition covered 132 countries including several in Asia.The Global Entrepreneurship Index

74、 measures a countrys entrepreneurial ecosystem,also using cross-sectional data for median comparisons(GEDI 2019).The ecosystem is the prevailing environment an entrepreneur faces.It examines entrepreneurships in terms of attitudes,abilities,and aspirations,all predicated on the societys entrepreneur

75、ial frameworkwhich includes market structure,infrastructure,the R&D system,financial sector,corporate sector,government,and education system.Fourteen areas are measured.A composite score is produced,and then compared nationally and regionally.The 2019 report covered 137 countries including several i

76、n Asia.Another trial for examining factors affecting MSME development The global MSME data initiatives reviewed above mainly use qualitative scoring methods based on national surveys with descriptive analyses and/or use median comparisons from secondary data.The limited availability of MSME data mak

77、es it difficult for direct comparisons across countries.Thus,they have tried to describe MSME conditions and identify constraints on MSME development primarily using scoring methods based on evaluation matrices that supporting governments use to design MSME policies.This study has the same purpose,b

78、ut apart from these indices,applies a more quantitative approach by using panel data obtained under the Asia SME Monitor project.6 Table 2:Summary of Global MSME Data Initiatives in Asia and the Pacific ADB=Asian Development Bank;ASEAN=Association of Southeast Asian Nations;ERIA=Economic Research In

79、stitute for ASEAN and East Asia;GEDI=Global Entrepreneurship and Development Institute;ITC=International Trade Centre;Lao PDR=Lao Peoples Democratic Republic;MSME=micro,small,and medium-sized enterprise;OECD=Organisation for Economic Co-operation and Development;PRC=Peoples Republic of China;WEF=Wor

80、ld Economic Forum;WIPO=World Intellectual Property Organization.Source:Authors.Updates as of 30 October 2023.ItemAsia Small and Medium-Sized Enterprise Monitor(ASM)SME and Entrepreneurship OutlookFinancing SMEs and Entrepreneurs(OECD Scoreboard)ASEAN SME Policy IndexSME Competitiveness OutlookGlobal

81、 Competitiveness IndexGlobal Innovation IndexGlobal Entrepreneurship IndexLead organizationADBOECDOECDOECD and ERIAITCWEFWIPOGEDIYear launched20202002201220142015200520072006Latest edition20232023202220182022201920222019Dimension365834731MSME developmentInstitutional and regulatory frameworkAllocati

82、on and structure of bank credit to SMEsProductivity,technology,and innovationFirm capabilitiesEnabling environmentInstitutionsEntrepreneurial attitudes2Access to financeMarket conditionsExtent of public support for SME financeEnvironmental policies targeting SMEsBusiness ecosystemMarketsHuman capita

83、l and researchEntrepreneurial abilities3Policies and regulationsInfrastructureCredit costs and conditionsAccess to financeNational environmentHuman capitalInfrastructureEntrepreneurial aspirations4Access to financeNonbank sources of financeAccess to market and internationalizationInnovation ecosyste

84、mMarket sophistication5Access to skillsFinancial healthInstitutional frameworkBusiness sophistication6Access to innovation assetsLegislation,regulation,and taxKnowledge and technology ouputs7Entrepreneurial education and skillsCreative outputs8Social enterprises and inclusive entrepreneurshipSub-dim

85、ension1529252539128114DataCross-sectional and time seriesCross-sectional and time seriesCross-sectional and time seriesCross-sectional(intent to create a time series)Cross-sectional and time seriesCross-sectional and time seriesCross-sectionalCross-sectionalMethodologyQuantitative and qualitative na

86、tional surveysMedian comparisonDescriptive national surveysQualitative national surveysMedian comparisonQuantitative and qualitative national surveysMedian comparisonMedian comparisonParticipating economies25 developing economies:(i)Southeast Asia(10);(ii)South Asia(5);(iii)Central and West Asia(7);

87、and(iv)Pacific(3).OECD member economies(including Australia,Japan,New Zealand,and the Republic of Korea).48 economies(including Australia,Georgia,Indonesia,Japan,Kazakhstan,Malaysia,New Zealand,the PRC,the Republic of Korea,Thailand,and Trkiye).10 ASEAN member states85 economies(including Armenia,Az

88、erbaijan,Bangladesh,Bhutan,Cambodia,Georgia,Indonesia,Kazakhstan,the Kyrgyz Republic,the Lao PDR,Mongolia,Myanmar,Nepal,Pakistan,the Philippines,Tajikistan,Timor-Leste,Trkiye,and Viet Nam).141 economies(including Armenia;Australia;Azerbaijan;Bangladesh;Brunei Darussalam;Cambodia;the PRC;Georgia;Hong

89、 Kong,China;Indonesia;India;Japan;Kazakhstan;the Republic of Korea;the Kyrgyz Republic;the Lao PDR;Malaysia;Mongolia;Nepal;Pakistan;the Philippines;Singapore;Sri Lanka;Tajikistan;Thailand;Trkiye;and Viet Nam).132 economies(including Armenia;Australia;Azerbaijan;Bangladesh;Brunei Darussalam;Cambodia;

90、the PRC;Georgia;Hong Kong,China;Indonesia;India;Japan;Kazakhstan;the Republic of Korea;the Kyrgyz Republic;the Lao PDR;Malaysia;Mongolia;Nepal;Pakistan;the Philippines;Singapore;Sri Lanka;Tajikistan;Thailand;Trkiye;Uzbekistan;and Viet Nam).137 economies(including Armenia;Australia;Azerbaijan;Banglad

91、esh;Brunei Darussalam;Cambodia;the PRC;Georgia;Hong Kong,China;Indonesia;India;Japan;Kazakhstan;the Republic of Korea;the Kyrgyz Republic;the Lao PDR;Malaysia;Myanmar;Pakistan;the Philippines;Singapore;Tajikistan;Thailand;Trkiye;and Viet Nam).RemarksCountry and regional reviewsSME performance and th

92、e degree of entrepreneurshipFinance and entrepreneurship scoreboardPolicy landscape relates to SME development and policy implementationSME competitiveness at the macro levelCompetitiveness and economic productivityInnovation trends and analysis Entrepreneurship ecosystem7 4.Methodology and Dataset

93、Given MSME data limitations,a model was developed using a probabilistic principal component analysis(P-PCA),combined with estimating the expectation-maximization(EM)algorithm to compensate for missing data.The exercise used the ADB Asia SME Monitor 2023 database.2 The following sections explain the

94、structure of the data used and the model specifications.4.1.Data Structure The Asia SME Monitor database stores various MSME-related aggregate variables,covering 25 countries as of November 202310 in Southeast Asia(Brunei Darussalam,Cambodia,Indonesia,the Lao Peoples Democratic Republic Lao PDR,Mala

95、ysia,Myanmar,the Philippines,Singapore,Thailand,and Viet Nam);5 in South Asia(Bangladesh,India,Nepal,Pakistan,and Sri Lanka);7 in Central and West Asia(Armenia,Azerbaijan,Georgia,Kazakhstan,the Kyrgyz Republic,Tajikistan,and Uzbekistan),and 3 in the Pacific(Fiji,Papua New Guinea,and Samoa).3 The dat

96、abase covers three dimensions:(i)the MSME landscape14 data categories including the number of MSMEs,those employed by MSMEs,contribution to economic output(whether in GDP or gross value added),and MSME export/import values;(ii)MSME access to bank credit8 data categories including MSME bank loans out

97、standing,nonperforming MSME loans,and guaranteed loans;and(iii)MSME access to nonbank and market-based finance8 data categories including nonbank finance institution(NBFI)finance(including microfinance institutions,finance companies,credit cooperatives,and pawnshops),nonperforming financing,and the

98、market capitalization MSMEs can tap.All local currency data were converted into US dollars,referring to end-of-year currency rates from the IMF International Financial Statistics for designated years.The data covers 20072022.Data with sufficient observations are used as independent variables to esti

99、mate the latent variable“MSME development”under the P-PCA model.4 There are two groups of variables incorporated into the model:nonfinance and finance data.For nonfinance data,the variable“number of MSMEs”is the number of enterprises meeting the MSME criteria for each country and year.It indicates t

100、he net data provided by national statistics agencies and does not show details of a firms“scrap-and-build”conditions,but an increased number roughly suggests newly created businesses.The variable“number of employees”measures the number of workers employed by the MSMEs in each country and year.“MSME

101、output”measures the sum of value-added produced by MSMEs in each country and year.The variables“MSME exports”and“MSME imports”show the value of products exported and imported by MSMEs in each country and year.Finance-related variables measure the state of corporate financing in each country and year

102、.The variables“MSME loans outstanding”and“nonperforming MSME loans”correspond to the outstanding amounts of bank loans to MSMEs and the amount of nonperforming loans.Given that data on MSME bank loans are unavailable for some countries,“bank loans outstanding”and“nonperforming bank loans”are include

103、d in the datasets,as these include MSME borrowers.The variables“NBFI loans outstanding”and“nonperforming NBFI loans”also refer to loans from nonbank finance institutions available for MSMEs.The variable“market capitalization”is the 2 ADB Asia SME Monitor 2023 database.https:/data.adb.org/dataset/202

104、3-asia-small-and-medium-sized-enterprise-monitor.3 Myanmar was excluded after 2020 for the data update.Effective 1 February 2021,ADB placed a temporary hold on sovereign project disbursements and new contracts in Myanmar.4 In the case of an extremely small number of observations,P-PCA estimates disp

105、lay“errors.”8 market value of listed companies on dedicated MSME market boards or equity markets that MSMEs can tap in each country and year.For countries without dedicated MSME markets or where MSME market data are unavailable,main market data are used,given that equity financing is an important ex

106、ternal financing source for MSMEs.Appendix 1 summarizes each variable for each country.The number of available variables varies by country.Although the range of available data has largely improved compared with the 2021 pilot test,even country-level aggregate variables are not commonly available for

107、 all countries.In addition,the number of observations for each variable varies for each country,creating missing values in certain variables(Appendix 2).These facts support the use of the P-PCA method for this study.4.2.Regression Models Probabilistic principal component analysis Missing data is alw

108、ays an issue for data analysis.The P-PCA,while a derivation of principal component analysis,has been used to solve problems and issues relating to missing data across different sectors in social science and engineering by analyzing,predicting,or detecting variables of interest.Several studies have f

109、ocused on the effectiveness of P-PCA in imputing for missing data by running experiments and comparing results by using other methods of imputing missing data.For instance,Hegde et al.(2019)conducted an experiment where some data points were deliberately omitted.P-PCA was used to estimate whether th

110、e predicted values would be closest to the original data and then compared the results using multiple imputation using chained equations(MICE).The experiment showed that P-PCA was the better statistical tool for imputing missing completely at random(MCAR)data than MICE.Another experiment run by Jene

111、lius and Koutsopoulos(2017)tried to predict taxi travel times in urban networks using P-PCA and k-nearest neighbors.It revealed that the results of P-PCA provided more accurate predictions.P-PCA was also used to input missing data for image analysis and reconstruction.Yu et al.(2010)utilized P-PCA t

112、o impute missing data that would help reconstruct 3D images.Employing an algorithm using P-PCA and expectation maximization proved an effective way to reconstruct 3D images.Cao,Liu,and Yang(2008)used P-PCA to help detect small infrared targets by helping map the input vector from the image onto a su

113、bspace.It better predicted the possibility of the input being a target.Other experiments utilized P-PCA to detect and filter out abnormal data and outliers.This was pivotal in addressing data issues,such as identifiability issues and removal of bias in the analysis.Qu et al.(2009)used robust PCA to

114、filter out abnormal traffic flow data and compared it to other methods such as the nearest/mean historical imputation method and local interpolation/regression method.Compared with other methods,it showed P-PCA reduced the root-mean-square imputation error by at least 25%.Chen,Martin,and Montague(20

115、09)successfully used P-PCA as a tool to detect outliers and were able to run contribution analysis after yielding the missing data.Their research showed that P-PCA was critical in identifying the source of the outliers,thereby improving their analysis.Xiang,Zhong,and Gao(2015)used P-PCA to better de

116、tect rolling element bearing faults and then conducted spectral kurtosis to determine the optimal center frequency and bandwidth.Ma et al.(2021)used P-PCA to create a base model to detect anomalies and identify structural damage in buildings.The experiment proved that P-PCA is effective in recoverin

117、g missing data to conduct the analysis.9 Advances in data science and analysis require using machine learning and neural networks to support sophisticated processes.This allows P-PCA to be used in conjunction with more advanced tools to achieve research goals.Dixit,Bhagat,and Dangi(2022)were success

118、ful in integrating P-PCA in detecting fake news.P-PCA was utilized to improve the filtering process of news after initial manual filtering.By automating the process,the authors were able to detect and classify fake news using long short-term memory networks.Overall,P-PCA has been an effective tool i

119、n imputing missing data that allows for predictive modelling,anomaly and outlier detection,and in improving data analysis across different sectors in social sciences and engineering.It has worked better than other missing data estimation techniques.Combined with other sophisticated data analysis met

120、hodologies,P-PCA helps increase accuracy and reduces errors.Concept of P-PCA This section provides a detailed explanation of the P-PCA model developed for this study,referring to Tipping and Bishop(1999),Bishop(2006),and Hastie et al.(2009).5 The P-PCA is a variant of a standard PCA that allows for

121、some missing data,assuming that every observed data?correspond to a latent variable?and are generated by the following linear model:where matrix W?relates the latent variable to the observed data,(?)is the mean of this model,and is the noise.The distribution of is the dimensional standard Gaussian(0

122、,),while comes from the Gaussian(0,?).When there is observed data?,the latent variable and a noise corresponding to?are written as?and?,respectively.For simplicity,=(?,?,?)?and =(?,?,?)?,and each?is regarded as identically and independently sampled from model(1).Thus,model(1)assumes the observed dat

123、a is realized by the low dimensional(?)latent variable.The goal is to find optimal parameters(,?)to maximize the posterior likelihood.Before applying the analysis,is regularized into mean 0 and variance 1 for each column.Under these premises,the observed variable?follows its marginal distribution?(,

124、?+?)(independent and identically).Thus,the following log likelihood function is generated,where is regularized as the zero-mean in the following transformation:where the independent terms for the maximum likelihood estimation are omitted.According to Tipping and Bishop(1999),the optimal parameters t

125、hat attain the maximal of can be explicitly written.Here,the eigenvalue decomposition of the covariance matrix of is used.5 The methodology is explained in ADB(2022,pp.47).As the same analytical estimation process was followed using expanded datasets through 2022,the same explanation applies here.10

126、 Let(?,1),(?,?)be sets of eigenvector and eigenvalue of?sorted in order of increasing eigenvalues.With =(?,?)and =(?,?),these notations bring the following solution for the maximum likelihood estimation:Using expectationmaximization algorithm Apart from the explicit solution to equation(3),there are

127、 several useful iterative algorithms to solve optimization problems.The gradient descent method is probably the most popular algorithm for optimization.The expectation-maximization(EM)algorithm introduced here assures that does not decrease in each step.In the following exposition,we denote a set of

128、 parameters(,?)as,and subscript the parameters of the k-th iteration as?(=1,2,).However,a more generalized setting is considered where the random variables x and z follow a joint distribution(,|),but only can be observed.We draw data?identically and independently from(,|).To apply the maximum likeli

129、hood estimation to derive?,the objective function can be written as follows:The EM algorithm aims to maximize the first term so that the following equation holds:The second term is no less than zero and attains its minimum at =?.Thus,the EM algorithm assures that(?)(?)holds for all =1,2,as follows:1

130、1 The second equality holds as(?,?)(=1,)is independent.In summary,the EM algorithm alternately repeats two steps:an expectation step for log(,|)with regard to(|,?)and to calculate the k-th target function?(),and a maximization step to maximize it.Although there is no guarantee of obtaining a global

131、optimal solution,convergence of its likelihood is guaranteed by its derivation.This method is particularly useful for estimating parameters in latent variable models where the optimization of simultaneous distributions is difficult.Applying the EM algorithm to the probabilistic PCA model,the simulta

132、neous distributions can be written as follows:Then,the conditional distribution(|?,?)of with kth parameters is given by the following:This object means that the mean and the covariance of under(|?,?)can be written respectively as follows,with?+?denoted as?:By extracting terms which relate to from?,w

133、e get the kth target function:12 Finally,?is calculated by differentiating the target function by(,?)and finding a unique stationary point:under the condition?=0.When the set of data is missing some values,each?is decomposed into the following two terms for easier explanation:The two variables?and?c

134、orrespond to the observed coordinates and missing coordinates,respectively.When?consists of observed coordinates and v missing coordinates,(?)?is the th observed coordinates of?,and(?)?is the th missing coordinates of?.Therefore,?and?are -dimensional and -dimensional,while?and?are defined as follows

135、:Then,the simultaneous distribution of,under fixed?and?can be written as follows:Also,there are conditional distributions about and under the observed:13 We define and calculate and with the mean and variance of and under the fixed.Using these distribution functions,we can derive the EM algorithm fo

136、r data with missing values by regarding both and as latent variables.5.Estimation Results This section presents the estimation results for(i)all 25 countries,(ii)Southeast Asia,(iii)South Asia;and(iv)Central and West Asia.A regional estimation for the Pacific was not conducted as there were only 3 c

137、ountries included in the model.A robustness test was done by applying another variant of P-PCA to aggregate data of the 25 countries(Appendix 4).5.1.All Countries The P-PCA was applied to country-level panel data of 25 countries to see the time-series dynamics of MSME development in Asia and the Pac

138、ific.Three factors were obtainedprincipal component(PC)1 to PC3(Figure 1,Table 3).PC1 makes the largest contribution to the variation in country-level panel data(59%),followed by PC2(15%)and PC3(7%),explaining 80%in total(Table 4).Factor loadings are sorted in descending order(Table A3.1).A darker r

139、ed color indicates a positive impact to the trend in the principal component,while a darker blue means a negative impact.Each factor is orthogonal to each other and related to each variable with specific factor loadings.Key factors that form each PC can thus be extracted.PC1-PC3 formed three differe

140、nt trend curves on MSME development(Figure 1).PC1 traces a low line until the middle of the sample period,then rises from 2015 until it slows after 2020.It suggests that MSMEs felt the effects of the aftermath of the 20082009 global financial crisis(GFC)until 2014,when recovery accelerated until dev

141、elopment slowed after the COVID-19 pandemic spread in 2020.Overall,it indicates“a slow recovery against the shocks.”PC2 remains low until 2011,moves up until the 2015 peak,then declines afterward.It suggests that MSMEs made relatively rapid recovery from the GFC,then decelerated development around t

142、he latter part of the 20142016 Russian Financial Crisis(RFC),and shifted to the negative after 2019,accelerated by the 20202021 COVID-19 pandemic.It indicates“relatively fast recovery against the shocks but sensitive to the shocks.”The PC3 curve is more complicated,rising soon after the GFC with its

143、 first peak in 2011,bottoming out in 2016(RFC),and rising again afterward.It suggests“a quick recovery against the shocks but very sensitive to the shocks.”In PC1(slow recovery),the main factors slowing MSME development were nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,Pakistan,Kazakhstan,B

144、runei Darussalam,Malaysia,and Viet Nam).Negative factor loadings also indicated“MSME loans(e.g.,Tajikistan,Kazakhstan,Papua New Guinea,and Georgia),which means that increased MSME loans in some countries lowered the level of MSME development under PC1.This suggests that the delivery of low quality M

145、SME loans with increased nonperforming loans in the countries mentioned likely contributed to MSMEs slow GFC recovery until 2014.In contrast,the main factors that boosted MSME development were(i)bank loans(e.g.,the Philippines,Fiji,the Kyrgyz Republic,and India),(ii)number of MSMEs(e.g.,Indonesia,Ne

146、pal,Georgia,Viet Nam,and the Kyrgyz Republic),and(iii)MSME output(e.g.,Indonesia and Pakistan).This suggests that steady delivery of bank loans to businesses likely catalyzed the increase in number of MSMEs(new small business creation)and output,bringing MSMEs back to their growth path after 2015.In

147、 PC2(relatively fast recovery),key drivers that lowered MSME development were also nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,Georgia,Fiji,Bangladesh,14 Uzbekistan,the Philippines,Thailand,Papua New Guinea,the Kyrgyz Republic,Viet Nam,Cambodia,and Sri Lanka).On the other hand,factors that

148、 accelerated MSME development were(i)number of MSME employees(e.g.,India,Uzbekistan,Viet Nam,Georgia,Tajikistan,Malaysia,Indonesia,and the Philippines),(ii)MSME output(e.g.,Tajikistan,Uzbekistan,and Georgia),(iii)MSME loans(e.g.,the Lao PDR,the Philippines,Thailand,and Malaysia),and(iv)equity market

149、 capitalization(e.g.,the Lao PDR,Sri Lanka,and Pakistan).PC2 suggests that nonperforming MSME loans likely slowed MSME development after the GFC and the COVID-19 pandemic,while improved delivery of MSME loans and the recovery of market-based finance resulted in a better environment for new MSME jobs

150、 and enhanced output.This likely supported the relatively fast recovery and growth of MSME businesses after the GFC.In PC3(quick but sensitive recovery),nonperforming loans by banks and for MSMEs(e.g.,the Lao PDR,Indonesia,and Tajikistan)were again the main drivers slowing MSME development.Factors t

151、hat raised the MSME development level were(i)bank loans(e.g.,Malaysia,Thailand,Armenia,and India),(ii)market capitalization(e.g.,Indonesia,Papua New Guinea,and Bangladesh),(iii)MSME output(e.g.,Malaysia,Kazakhstan,Azerbaijan,Georgia,and the Kyrgyz Republic),and(iv)MSME exports and/or imports(e.g.,In

152、donesia and the Kyrgyz Republic).Although PC3 showed a small contribution to explaining MSME development,it suggests that expanded bank lending and capital markets likely helped the rapid recovery of MSME exports and output and quickly accelerated MSME development.But due to weak financial markets a

153、nd international trade for MSMEs,it remained highly sensitive to shocks like the GFC,RFC,and COVID-19 pandemic.Overall,the estimation results for all countries show that sound MSME credit markets,diversified financing options(market-based finance),support for new business development and job creatio

154、n,along with active MSME participation in global markets play a critical role for the smooth recovery from crises and shocks in developing Asia and the Pacific.Sound,resilient finance sector development is indispensable for sustainable MSME growth nationally.Figure 1:Time Series Plots of Estimated P

155、rincipal ComponentsAll Countries PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.-0.6-0.4-0.20.00.20.40.62007200820092010201120122013201420152016201720182019202020212022PC1PC2PC315 Table 3:Time Series Plots of Estimated Principal ComponentsAll Countries PC=princi

156、pal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.Table 4:Contribution of Each Estimated Principal ComponentAll Countries PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.5.2.Southeast Asia For Southeast Asia,three PC factors(PC1-PC3)were

157、 also obtained(Figure 2,Table 5).PC1 makes the largest contribution to the variation in country data(59%),followed by PC2(16%)and PC3(7%),explaining 82%in total(Table 6).Factor loadings are sorted in descending order(Table A3.2).PC1-PC3 in Southeast Asia followed similar trend curves as in“all count

158、ries”but with somewhat more complicated shapes(Figure 2).PC1 remains low until 2014,then rises to a 2017 peak,before decelerating growth with a drop in 2022.It indicates a slow recovery path from the GFC.PC2 also remains low until 2011,moves up until the peak in 2014,and then declines afterwardreach

159、ing its peak a year earlier than in“all countries.”It indicates a fast recovery from the GFC but sensitive to shocks such as the RFC and the pandemic.PC3 generated a complicated shape with frequent ups and downs during 20072022.It bottoms out in 2011,moves up to a peak in 2014,then falls until 2018

160、before rising again to a peak in 2021.It indicates that MSME development is very sensitive to shocks such as the GFC,RFC,and the pandemic,while also quickly recovering.In PC1(slow recovery),the negative curve until 2014 is explained by negative factor loadings denoted by nonperforming loans by banks

161、 and for MSMEs(e.g.,Brunei Darussalam,Malaysia,Viet Nam,and the Philippines).In contrast,the positive curve after 2015 is explained by positive factor loadings denoted by(i)bank loans and NBFI loans(e.g.,the Philippines,Viet Nam,the Lao PDR,Thailand,and Singapore),(ii)number of MSMEs(e.g.,Indonesia

162、and Viet Nam),and(iii)MSME output(e.g.,Indonesia,Brunei Darussalam,and Thailand).PC1 suggests that increased nonperforming loans likely contributed to their slow recovery from the GFC.But improved lending by banks and NBFIs likely facilitated new small business creation and a rebound in output,allow

163、ing a return to development growth after 2015.In PC2(fast recovery),the negative curve until 2011 and after 2019 is also explained by negative factor loadings denoted by nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,Thailand,Viet Nam,Myanmar,Cambodia,and the Philippines).The positive curve p

164、eaking in 2014 is explained by positive factor loadings denoted by(i)MSME loans(e.g.,the Lao PDR,the Philippines,Thailand,and Malaysia),(ii)market capitalization(e.g.,the Lao PDR,Singapore,and Thailand),and(iii)number of MSME employees(e.g.,Viet Nam,Malaysia,the Philippines,and Indonesia).PC2 sugges

165、ts that the high level of nonperforming loans in the finance sector likely Year2007200820092010201120122013201420152016201720182019202020212022PC1-0.17-0.21-0.25-0.27-0.32-0.30-0.25-0.170.030.190.260.300.330.310.270.19PC2-0.23-0.25-0.19-0.17-0.050.130.300.420.440.330.260.09-0.05-0.17-0.25-0.26PC3-0.

166、42-0.34-0.310.280.370.350.200.03-0.14-0.22-0.070.070.150.060.210.29ItemPC1PC2PC3Contribution ratio0.590.150.07Cumulative contribution rate0.590.740.8016 impeded MSME development around the GFC and the pandemic,while expanded MSME lending,the capital market recovery(including dedicated MSME equity ma

167、rkets such as Catalist in Singapore and mai in Thailand),along with more MSME jobs likely helped the relatively fast MSME development post GFC.In PC3(sensitive recovery),the negative curve around two points in 2011 and 2018 is explained by negative factor loadings denoted by(i)nonperforming loans by

168、 banks,NBFIs,and for MSMEs(e.g.,the Lao PDR,Singapore,Brunei Darussalam,Thailand,and Viet Nam)and(ii)MSME loans(e.g.,Thailand,the Philippines,and the Lao PDR).The positive curve around two points in 2014 and 2021 is explained by positive factor loadings denoted by(i)loans by banks,NBFIs,and MSME len

169、ding(e.g.,Singapore,Brunei Darussalam,and Cambodia)and(ii)market capitalization(e.g.,Malaysia ACE and LEAP markets and the Philippines SME Board).PC3 generated a different curve than for“all countries”,more pronounced in the effect of access to finance.It suggests that low quality MSME loans with in

170、creased nonperforming loans in countries such as Thailand and the Lao PDR likely kept MSME development suppressed in the region(especially in 20162018 amid the global economic slowdown).But diversified financing options from bank credit along with nonbank and market-based finance likely helped MSMEs

171、 recover from the shocks smoothly,while volatile financial markets held back resilience.Figure 2:Time Series Plots of Estimated Principal ComponentsSoutheast Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.Table 5:Time Series Plots of Estimated Principal Com

172、ponentsSoutheast Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.-0.6-0.4-0.20.00.20.40.62007200820092010201120122013201420152016201720182019202020212022PC1PC2PC3Year2007200820092010201120122013201420152016201720182019202020212022PC1-0.28-0.28-0.25-0.27-0.32

173、-0.29-0.15-0.070.070.260.310.270.300.270.270.17PC2-0.21-0.23-0.26-0.12-0.010.230.420.460.390.190.160.02-0.12-0.16-0.24-0.28PC30.140.040.00-0.26-0.29-0.200.160.300.28-0.24-0.32-0.41-0.200.230.310.2617 Table 6:Contribution of Each Estimated Principal ComponentSoutheast Asia PC=principal component.Sour

174、ce:Calculated based on ADB Asia SME Monitor 2023 database.5.3.South Asia In South Asia,three factors were obtained from the P-PCA,but PC1 and PC2 trends were swapped(Figure 3,Table 7).PC1 makes the largest contribution to the variation in country data(59%),followed by PC2(22%)and PC3(8%),explaining

175、89%in total(Table 8).Factor loadings are sorted in descending order(Table A3.3).PC1-PC3 in South Asia generated different trend curves from“all countries”estimates(Figure 3).PC1 remains low until 2010,moves up until its 2014 peak,and then declines afterward.It indicates a fast recovery from the GFC(

176、a year earlier than the trend in“all countries”)but was sensitive to the global economic slowdown and the pandemic.PC2 remains low until 2015,then rises through 2019,before declining until 2021.It indicates a slow recovery from the GFC and South Asias economic slowdown.The PC3 trend was somewhat rev

177、ersed from the trend in“all countries.”It bottoms out in 2013,moves up to a peak in 2017,then drops afterward bottoming out in 2022.It indicates MSME development was sensitive to the regions economic and political instability(e.g.,India before the current administration started in 2014,and economic

178、and political crises in Pakistan and Sri Lanka from around 2019)as well as the shock from the pandemic.In PC1(fast recovery),the negative curve until 2010 and after 2019 is explained by negative factor loadings denoted by nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,India,Bangladesh,Pakista

179、n,and Sri Lanka).MSME loans in India and Pakistan were also identified as negative factors,suggesting that the delivery of low quality MSME loans with high levels of nonperforming loans in these countries likely impeded MSME development in the region.The positive curve peaking in 2014 is explained b

180、y positive factor loadings denoted by(i)NBFI loans and market capitalization(e.g.,Sri Lanka,Pakistan,and Bangladesh)and(ii)number of MSME employees(e.g.,India and Nepal).A recovery in nonbank and market-based finance,along with increased MSME jobs,likely supported a smooth shift back to growth.But t

181、he MSME funding environment was likely sensitive to economic and political environment changes,especially after 2019.In PC2(slow recovery),the negative curve until 2015 is explained by negative factor loadings denoted by MSME loans and nonperforming loans by banks,NBFIs,and for MSMEs in Pakistan and

182、 India,suggesting that increased MSME loans accompanying rising nonperforming loans in these countries likely made MSMEs recover slowly from the GFC and the regions stagnant economic growth.The positive curve after 2016 is explained by positive factor loadings denoted by(i)loans by banks and for MSM

183、Es(e.g.,Bangladesh,Pakistan,Sri Lanka,and India),(ii)number of MSMEs(e.g.,Nepal),and(iii)MSME output(e.g.,Bangladesh and Pakistan).After 2016,improved bank lending and MSME loans along with an environment conducive to new small businesses and better productivity likely boosted MSME development.In PC

184、3(sensitive recovery),the negative,downward trend during 20102013 and after 2019(economic crises in Pakistan and Sri Lanka)is explained by negative factor loadings denoted by ItemPC1PC2PC3Contribution ratio0.590.160.07Cumulative contribution rate0.590.750.8218 nonperforming loans by banks and NBFIs(

185、e.g.,Sri Lanka,Pakistan,and Bangladesh).The positive curve before the GFC and during 20162018(linked to the new administration in India)is explained by positive factor loadings denoted by MSME loans and NBFI loans(e.g.,Pakistan,Bangladesh,and India).PC3 was more affected by access to finance for MSM

186、E development.It suggests that high levels of nonperforming loans by banks and NBFIs likely led to a slowdown in MSME development.But once the MSME credit market and the nonbank finance industry expanded,MSME development quickly turned positive,although its growth pattern was likely highly sensitive

187、 to shocks,such as regional economic crises,political conditions,and the pandemic.Figure 3:Time Series Plots of Estimated Principal ComponentsSouth Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.Table 7:Time Series Plots of Estimated Principal ComponentsSou

188、th Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.Table 8:Contribution of Each Estimated Principal ComponentSouth Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.-0.6-0.4-0.20.00.20.40.620072008200920102011201220132

189、01420152016201720182019202020212022PC1PC2PC3Year2007200820092010201120122013201420152016201720182019202020212022PC1-0.22-0.34-0.26-0.090.020.200.320.440.410.380.260.08-0.004-0.10-0.14-0.08PC2-0.08-0.15-0.30-0.36-0.35-0.33-0.23-0.13-0.0060.080.220.270.360.270.240.27PC30.440.260.08-0.07-0.14-0.20-0.22

190、-0.003-0.0010.250.370.28-0.03-0.33-0.32-0.35ItemPC1PC2PC3Contribution ratio0.590.220.08Cumulative contribution rate0.590.800.8919 5.4.Central and West Asia As in other regions,three factors were obtained in Central and West Asia from the P-PCA(Figure 4,Table 9).PC1 makes the largest contribution to

191、the variation in country data(63%),followed by PC2(15%)and PC3(8%),explaining 85%in total(Table 10).Factor loadings are sorted in descending order(Table A3.4).PC1-PC3 show similar trends on MSME development as those in“all countries”(Figure 4).PC1 remains low until 2014,rising in 20152018 before slo

192、wing afterwards,indicating a slow recovery from the GFC.PC2 remains low until 2010,rises to a 2015 peak,and then declines with a negative curve after 2019,indicating a relatively rapid recovery from the GFC but sensitive to shocks like the RFC and the pandemic.PC3 bottoms out in 2009(GFC),peaks in 2

193、011,then drops until 2016(RFC).It rises afterward with erratic movement during the pandemic,suggesting a quick but very sensitive recovery from shocks.In PC1(slow recovery),the negative curve until 2014 is explained by negative factor loadings denoted by(i)nonperforming bank loans(e.g.,Kazakhstan,Az

194、erbaijan,and Uzbekistan)and(ii)MSME loans(e.g.,Kazakhstan,Tajikistan,and Georgia).The positive curve after 2015 is explained by positive factor loadings denoted by(i)bank loans and those for MSMEs(e.g.,the Kyrgyz Republic,Armenia,and Georgia),(ii)number of MSMEs(e.g.,Georgia,the Kyrgyz Republic,and

195、Kazakhstan),and(iii)number of MSME employees(e.g.,Azerbaijan,Kazakhstan,and Georgia).PC1 suggests that increased MSME lending with high levels of nonperforming loans in countries such as Kazakhstan likely slowed the recovery from the GFC.But improved bank lending likely supported creating new MSMEs

196、and jobs in countries such as Georgia,helping them shift to growth.In PC2(fast recovery),the negative curve until 2010 and after 2019 is explained by negative factor loadings denoted by nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,Uzbekistan,Georgia,and the Kyrgyz Republic).The negative cur

197、ve largely reflected the trends in Uzbekistan.The positive curve during 20112018peaking in 2015is explained by positive factor loadings denoted by(i)number of MSME employees(e.g.,Uzbekistan,Georgia,and Tajikistan),(ii)MSME output(e.g.,Tajikistan,Uzbekistan,and Georgia),and(iii)MSME exports and/or im

198、ports(e.g.,Uzbekistan and the Kyrgyz Republic).PC2 suggests that a high level of nonperforming loans likely impeded MSME development.Increased job creation,higher output,and internationalization of MSMEs likely helped drive MSME development.In PC3(sensitive recovery),the downward curve around the GF

199、C and RFC is also explained by negative factor loadings denoted by nonperforming loans by banks,NBFIs,and for MSMEs(e.g.,Tajikistan,Uzbekistan,and Armenia).The negative curve in the PC3 largely reflected the trends in Tajikistan.The positive curve during 20102014 and after 2017 is explained by posit

200、ive factor loadings denoted by(i)MSME exports and/or imports(e.g.,the Kyrgyz Republic and Georgia),(ii)MSME output(e.g.,Kazakhstan,Azerbaijan,Georgia,and the Kyrgyz Republic),and(iii)NBFI loans(the Kyrgyz Republic).PC3 suggests that MSMEs felt the hard impacts from financial crises(GFC and RFC)with

201、poor access to quality bank credit and NBFI loans,more pronounced in Tajikistan.However,higher MSME foreign trade,output,and improved access to NBFI loans likely encouraged MSME development,yet it remained very sensitive to shocks like the GFC,RFC,and the pandemic.20 Figure 4:Time Series Plots of Es

202、timated Principal Components Central and West Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.Table 9:Time Series Plots of Estimated Principal Components Central and West Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 databa

203、se.Table 10:Contribution of Each Estimated Principal Component Central and West Asia PC=principal component.Source:Calculated based on ADB Asia SME Monitor 2023 database.-0.6-0.4-0.20.00.20.40.62007200820092010201120122013201420152016201720182019202020212022PC1PC2PC3Year20072008200920102011201220132

204、01420152016201720182019202020212022PC1-0.22-0.23-0.24-0.25-0.29-0.28-0.27-0.190.050.150.230.340.320.300.270.23PC2-0.27-0.28-0.19-0.170.030.100.230.360.410.340.300.12-0.06-0.19-0.28-0.29PC3-0.33-0.25-0.350.240.400.350.250.04-0.30-0.320.030.150.180.030.210.07ItemPC1PC2PC3Contribution ratio0.630.150.08

205、Cumulative contribution rate0.630.770.8521 6.Policy Implications The P-PCA estimates found three patterns of MSME development:(i)slow growth,(ii)fast growth,and(iii)quick growth but sensitive to shocks.The key factors affecting MSME development vary by the response to shocks and by region.Based on a

206、ll estimation results,the slow MSME development pattern(22%63%explained,Figure 5A)found the key growth drivers to be(i)a sound market for bank credit,(ii)an environment favoring new business creation(number of MSMEs),and(iii)enhanced productivity(MSME output).However,these were unable to counter the

207、 major factors impeding MSME developmentinsufficient delivery or low quality of MSME lending.The result was slow MSME growth and development.Southeast Asia,South Asia,and Central and West Asia followed a similar structure,although the strongest growth drivers differed slightly by region.To avoid thi

208、s slow growth scenario,policy support should focus more on developing sound MSME credit markets(e.g.,through credit enhancement schemes such as credit guarantees and secured lending),increasing alternative financing options for MSMEs that go beyond traditional bank credit(i.e.,nonbank and market-bas

209、ed finance),and promoting MSME internationalization.In the fast MSME development pattern(15%59%explained,Figure 5B),the estimations for all countries and regions found the key growth drivers to be(i)job creation(MSME employees),(ii)enhanced productivity(MSME output),(iii)more and better quality MSME

210、 loans,and(iv)alternative financing options(market-based finance).These addressed negative factors such as low quality MSME loans.The more positive factors led to a relatively fast recovery from the GFC that supported MSME development.The availability of diverse financing options such as nonbank fin

211、ance and capital markets likely helped.However,as a solid base of growth-oriented MSMEs does not exist nationally,they are likely sensitive to new shocks.Greater policy support in promoting MSME participation in foreign markets(exports or participation in global value chains)would boost development.

212、This fast MSME growth pattern features strengthened capital markets as one of the growth drivers.In Southeast Asia,for example,there are several dedicated MSME equity marketsThailands mai(market for alternative investment,since 1998);the Philippine SME Board(2001);Singapores Catalist(2007);Malaysias

213、 ACE(Access,Certainty,Efficiency)(2009)and LEAP(Leading Entrepreneur Accelerator Platform)(2017);and Indonesias Acceleration Board(2019).These markets provide growth capital to qualified MSMEs in those countries.In the estimations for all countries and regions showing MSME development sensitive to s

214、hocks(7%8%explained,Figure 5C),key growth drivers were(i)a generally sound bank credit market,(ii)capital market development,(iii)enhanced productivity(MSME output),and(iv)internationalization of MSMEs(exports/imports).The growth and development drivers largely unaddressed included alternative finan

215、ce and MSME internationalization.The results showed a quick recovery from shocks,but one highly sensitive to new shocksgiven its frequently changing trend curve.The negative factors estimated included a lack ofor low qualityMSME credit market.It suggests that the financial system remains weak and fr

216、agile against shocks in developing Asia and the Pacific.To sum up,the balanced development of sound MSME credit markets and expanding alternative financing options(nonbank finance and capital markets)would be critical to support sustainable MSME growth in developing Asia and the Pacific.These should

217、 be promoted and supported further within national financial inclusion strategies.Policy support for new business development(startups and entrepreneurship development),quality job creation(skilled labor training),and increasing MSME participation in global markets(promoting MSME exports/imports or

218、22 participation in global value chains)also plays a critical role.Together,they will help facilitate a smooth recovery of MSME growth and development from crises and shocks.And they will help create a solid base of growth oriented MSMEs nationally,contributing to resilient,inclusive growth across t

219、he region.Figure 5A:Slow MSME Development Path Source:Authors.Figure 5B:Fast MSME Development Path Source:Authors.MSME/Bank/NBFI NPLs MSME loans MSME output Number of MSMEs Bank loans Bank/MSME NPLs MSME output Number of MSMEs Bank/NBFI loans All Countries Southeast Asia MSME/Bank/NBFI NPLs MSME loa

220、ns MSME output Number of MSMEs Bank/MSME loans South Asia Bank NPLs MSME loans MSME employees Number of MSMEs Bank/MSME loans Central and West Asia+(positive factors)MSME/Bank/NBFI NPLs MSME loans MSME output MSME employees Bank/MSME/NBFI NPLs MSME employees Capital markets MSME loans All Countries

221、Southeast Asia NBFI/MSME/Bank NPLs MSME employees NBFI loans&capital markets South Asia NBFI/MSME/Bank NPLs MSME exports/imports MSME output MSME employees Central and West Asia Capital markets MSME loans-(negative factors)+(positive factors)-(negative factors)23 Figure 5C:Sensitive MSME Development

222、 Path Source:Authors.7.Conclusion In a new trial for estimating factors affecting MSME development,we applied a variant of the PCA methodor P-PCAby using time series MSME data from the ADB Asia SME Monitor 2023 database.The P-PCA estimated three factors that commonly explain various variables for MS

223、ME activities.Through these,we obtained three types of MSME development in developing Asia and the Pacificslow growth,fast growth,and growth sensitive to shocks.They explain the different growth drivers and factors impeding MSME development,and the mismatches that may disrupt sustainable MSME growth

224、.The study found P-PCA useful where there are missing MSME data.However,continuous test-runs for various countries and strengthening datasets(especially business performance data)are crucial to improve the P-PCA model.Given that data limitation remains a critical concern,surveys should be designed t

225、o help fill missing information on MSMEs.As a next step,the use of granular firm-level data for the P-PCA should be examined to identify demand-side factors for business development.Once the validity of the P-PCA model is confirmedas an analytical tool on MSME developmentit could be used to forecast

226、 MSME development and help governments design more feasible,proactive,and evidence-based MSME policies.This is our future target.Bank/MSME NPLs MSME output Capital markets Bank loans Bank/MSME NPLs MSME loans Capital markets MSME/Bank/NBFI loans All Countries Southeast Asia Bank/NBFI NPLs NBFI/MSME

227、loans South Asia Bank/MSME/NBFI NPLs NBFI loans MSME output MSME exports/imports Central and West Asia MSME exports/imports+(positive factors)-(negative factors)24 Appendix 1:MSME-Related Variables,20072022 Table A1.1:Southeast Asia Lao PDR=Lao Peoples Democratic Republic;MSME=micro,small,and medium

228、-sized enterprise;NBFI=nonbank finance institution;obs=observations;S.D.=standard deviation.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariableUnitObsMeanMedianS.D.MinMaxMSME output$million103,292 3,383 565 2,506 3,972 Bank loans outst

229、anding$million134,210 4,328 289 3,727 4,718 Nonperforming bank loans$million13228 218 54 161 341 NBFI loans outstanding$million131,241 1,176 137 1,097 1,533 Nonperforming NBFI loans$million1313 13 4 6 19 Bank loans outstanding$million1614,853 10,427 13,572 1,584 45,157 Nonperforming bank loans$milli

230、on16358 220 349 55 1,396 NBFI loans outstanding$million163,417 2,449 3,265 154 9,890 Nonperforming NBFI loans$million1658 18 78 0 254 Market capitalization$million10324 252 203 117 687 Number of MSMEs1059,001,245 58,579,247 4,418,366 52,764,750 65,465,497 Number of employees10110,698,728 113,486,346

231、 9,308,079 96,193,623 123,229,386 MSME output$million10482,190 449,210 110,994 330,258 689,214 MSME exports$million1018,456 19,186 3,615 13,481 24,400 Bank loans outstanding$million12335,731 337,748 58,820 242,622 408,338 Nonperforming bank loans$million128,353 8,862 2,581 4,594 12,143 MSME loans ou

232、tstanding$million1267,251 66,671 12,997 49,949 85,742 Nonperforming MSME loans$million122,510 2,593 551 1,601 3,277 Market capitalization$million16403,428 423,522 138,691 98,310 603,848 Bank loans outstanding$million165,021 5,455 3,044 406 8,436 Nonperforming bank loans$million9209 219 49 125 266 MS

233、ME loans outstanding$million81,565 1,683 340 932 1,922 NBFI loans outstanding$million13276 179 264 6 686 Market capitalization$million121,046 1,093 309 571 1,479 Number of MSMEs91,074,391 1,151,339 187,220 638,790 1,226,494 Number of employees157,222,467 7,106,000 829,309 6,133,000 9,069,000 MSME ou

234、tput$million15107,410 110,035 18,369 77,910 135,312 MSME exports$million1238,279 41,065 5,143 29,377 44,020 Bank loans outstanding$million16352,497 367,883 86,771 194,634 458,655 Nonperforming bank loans$million167,573 7,044 1,906 5,403 12,561 MSME loans outstanding$million1662,278 65,548 14,382 38,

235、666 82,303 Nonperforming MSME loans$million162,665 2,513 444 1,976 3,650 NBFI loans outstanding$million16392 350 153 180 749 Market capitalization$million143,812 2,848 2,351 1,548 8,875 Number of MSMEs1347,962 40,396 13,741 38,590 75,116 Bank loans outstanding$million1162,118 11,845 101,770 3,368 31

236、4,562 NBFI loans outstanding$million8326 161 395 58 1,228 Nonperforming NBFI loans$million82 1 4 0 10 Number of MSMEs16911,709 929,002 106,438 758,436 1,105,143 Number of employees164,662,845 4,885,508 836,204 3,355,742 5,714,262 Bank loans outstanding$million15132,784 123,401 53,721 61,158 205,881

237、Nonperforming bank loans$million153,972 3,022 2,207 2,415 8,912 MSME loans outstanding$million159,019 9,325 1,776 5,563 11,605 Nonperforming MSME loans$million13656 574 179 521 1,037 NBFI loans outstanding$million135,704 6,582 5,291 135 12,998 Market capitalization$million16322 221 349 7 1,216 Numbe

238、r of MSMEs9266,356 262,800 20,475 242,900 298,500 Number of employees92,467,778 2,470,000 62,805 2,360,000 2,590,000 MSME output$million9156,700 152,593 26,133 129,286 211,215 Bank loans outstanding$million13847,714 870,379 160,722 496,942 1,025,553 Nonperforming bank loans$million1412,890 11,698 7,

239、741 -26,357 MSME loans outstanding$million1360,861 56,205 18,347 35,987 101,108 Nonperforming MSME loans$million131,788 1,607 1,259 441 3,617 NBFI loans outstanding$million139,374 9,398 1,393 6,247 11,237 Market capitalization$million156,537 6,890 1,992 2,474 9,567 Number of MSMEs162,898,833 2,904,6

240、37 226,194 2,366,227 3,187,378 Number of employees1511,569,702 11,747,093 1,367,048 8,900,567 13,950,241 MSME output$million16141,978 131,488 29,202 103,128 196,381 MSME exports$million1638,068 30,908 14,333 22,258 67,433 MSME imports$million1643,567 37,855 16,399 29,799 84,502 Bank loans outstandin

241、g$million16517,849 507,787 165,642 237,067 758,489 Nonperforming bank loans$million1411,802 11,026 3,227 8,097 17,706 MSME loans outstanding$million16148,950 155,560 42,169 77,586 218,289 Nonperforming MSME loans$million145,851 5,629 1,539 4,079 8,222 Market capitalization$million166,986 7,279 4,830

242、 635 15,490 Number of MSMEs15422,093 393,915 179,963 143,622 699,859 Number of employees154,968,626 5,321,882 944,386 2,834,950 6,205,320 Bank loans outstanding$million9256,219 248,450 84,589 148,401 383,940 Nonperforming bank loans$million85,872 5,959 233 5,424 6,084 NBFI loans outstanding$million1

243、12,763 2,827 1,037 1,252 4,212 Nonperforming NBFI loans$million106 0 12 0 31 Market capitalization$million1419,776 8,239 21,478 237 61,286 Brunei DarussalamCambodiaIndonesiaLao PDRMalaysiaMyanmarPhilippinesSingaporeThailandViet Nam25 Table A1.2:South Asia MSME=micro,small,and medium-sized enterprise

244、;NBFI=nonbank finance institution;obs=observations;S.D.=standard deviation.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariableUnitObsMeanMedianS.D.MinMaxMSME output$million93,942 3,927 795 2,826 5,230 Bank loans outstanding$million 139

245、2,427 88,779 38,578 43,935 151,724 Nonperforming bank loans$million 1311,163 7,891 11,347 2,767 46,951 MSME loans outstanding$million 1319,485 21,024 6,576 9,478 28,066 NBFI loans outstanding$million 116,508 7,115 1,594 3,364 7,912 Nonperforming NBFI loans$million 11729 547 519 174 1,699 Market capi

246、talization$million 1636,419 37,965 13,967 6,939 59,940 Number of MSMEs1145,758,425 44,764,000 7,799,256 36,176,000 63,387,673 Number of employees1199,962,545 101,169,000 12,467,165 80,523,000 117,132,000 MSME output$million 12524,532 525,836 57,853 445,618 615,535 MSME exports$million8141,178 137,97

247、5 11,433 127,986 159,189 Bank loans outstanding$million 161,051,381 1,026,220 337,711 540,122 1,540,214 Nonperforming bank loans$million 1668,102 56,813 46,769 12,722 132,150 MSME loans outstanding$million8225,942 235,876 28,126 182,149 255,880 Nonperforming MSME loans$million821,087 21,955 2,797 17

248、,433 24,512 NBFI loans outstanding$million8258,910 291,440 107,343 38,953 382,720 Nonperforming NBFI loans$million815,079 15,447 8,708 2,241 27,983 Market capitalization$million 102,403 2,621 1,785 52 6,402 Number of MSMEs10248,979 235,538 75,819 156,343 390,493 Number of employees122,195,870 2,149,

249、225 315,984 1,761,863 2,808,052 Bank loans outstanding$million 1615,690 12,011 10,606 -35,691 Market capitalization$million 1512,118 10,607 8,135 3,783 33,042 MSME output$million 163,783 4,045 1,497 1,524 5,883 Bank loans outstanding$million 1649,848 48,494 8,039 40,363 62,093 Nonperforming bank loa

250、ns$million 164,929 4,908 919 2,964 6,364 MSME loans outstanding$million 163,566 3,176 1,076 2,387 6,725 Nonperforming MSME loans$million 16721 686 230 349 1,133 NBFI loans outstanding$million 16577 560 158 367 820 Nonperforming NBFI loans$million 1626 26 9 14 41 Market capitalization$million 1652,23

251、8 50,396 19,593 23,498 91,866 Bank loans outstanding$million 1630,910 30,471 12,670 13,739 52,285 Nonperforming bank loans$million 161,435 1,077 761 714 3,476 NBFI loans outstanding$million 164,490 4,562 2,119 977 7,253 Nonperforming NBFI loans$million 14410 345 240 53 867 Market capitalization$mill

252、ion 1616,447 17,687 5,968 4,320 27,386 BangladeshIndiaNepalPakistanSri Lanka26 Table A1.3:Central and West Asia MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution;obs=observations;S.D.=standard deviation.Source:Authors calculation based on ADB Asia Small and Medium-Sized E

253、nterprise Monitor 2023 database.CountryVariableUnitObsMeanMedianS.D.MinMaxNumber of MSMEs1581,523 76,589 21,862 59,267 132,923 Number of employees14306,576 269,773 99,645 168,185 471,394 MSME output$million153,707 3,562 571 2,833 4,851 Bank loans outstanding$million164,723 4,280 2,415 1,344 9,916 No

254、nperforming bank loans$million16153 158 104 15 360 NBFI loans outstanding$million16389 368 191 126 725 Market capitalization$million16240 195 141 105 679 Number of MSMEs15212,011 196,972 61,548 135,353 355,906 Number of employees15198,780 115,035 111,155 90,182 357,840 MSME output$million153,229 2,3

255、95 2,358 531 8,209 Bank loans outstanding$million1611,105 9,937 4,532 5,389 22,790 Nonperforming bank loans$million16697 777 322 119 1,245 NBFI loans outstanding$million11312 225 184 148 687 Number of MSMEs1581,549 70,491 39,754 32,961 139,275 Number of employees15381,627 405,637 93,643 226,064 491,

256、668 MSME output$million153,417 3,744 1,232 1,400 4,814 MSME exports$million81,814 1,774 467 1,311 2,703 MSME imports$million84,930 4,821 1,001 3,801 7,011 Bank loans outstanding$million167,545 6,754 4,118 2,830 16,575 Nonperforming bank loans$million16523 509 184 76 977 MSME loans outstanding$millio

257、n82,724 2,310 1,671 1,508 6,735 Nonperforming MSME loans$million13127 82 83 54 297 NBFI loans outstanding$million16400 427 269 35 1,061 Market capitalization$million11861 764 243 553 1,283 Number of MSMEs161,051,333 1,016,599 338,534 643,376 1,818,764 Number of employees162,924,749 2,988,877 581,227

258、 2,121,198 4,109,741 MSME output$million1641,463 38,440 18,161 11,393 81,284 Bank loans outstanding$million1650,063 50,388 13,103 34,075 73,293 Nonperforming bank loans$million1510 4 9 2 27 MSME loans outstanding$million168,911 9,025 2,236 5,512 13,003 NBFI loans outstanding$million161,690 326 2,086

259、 121 5,825 Number of MSMEs1513,720 13,505 2,171 9,852 17,050 Number of employees1591,773 89,400 5,928 86,000 107,800 MSME output$million152,692 2,695 641 1,702 3,811 MSME exports$million15585 578 123 386 773 MSME imports$million152,645 2,837 634 1,139 3,398 Bank loans outstanding$million161,307 1,28

260、8 632 560 2,381 Nonperforming bank loans$million1574 65 50 18 205 MSME loans outstanding$million15982 1,000 452 456 1,648 NBFI loans outstanding$million16248 236 76 139 381 Market capitalization$million16277 174 228 72 941 Number of MSMEs1534,382 31,144 10,025 24,459 55,750 Number of employees14232,

261、316 231,273 38,669 187,000 302,000 MSME output$million144,096 4,498 1,044 2,242 5,336 Bank loans outstanding$million15990 914 249 701 1,463 Nonperforming bank loans$million12244 237 152 52 557 MSME loans outstanding$million11182 144 74 101 329 Nonperforming MSME loans$million1137 33 24 6 86 NBFI loa

262、ns outstanding$million15288 219 164 99 647 Nonperforming NBFI loans$million159 7 7 1 27 Number of MSMEs16247,815 201,559 120,778 129,211 523,556 Number of employees169,518,725 9,908,250 898,698 7,743,100 10,541,500 MSME output$million1630,280 30,455 11,562 8,974 48,168 MSME exports$million162,925 3,

263、120 1,026 1,334 4,715 MSME imports$million167,187 5,793 4,003 2,155 14,972 Bank loans outstanding$million819,523 18,176 6,338 13,276 30,116 Nonperforming bank loans$million8435 268 476 121 1,566 NBFI loans outstanding$million860 46 34 24 123 Nonperforming NBFI loans$million86 3 6 1 18 Market capital

264、ization$million133,735 2,897 1,831 1,825 8,408 KazakhstanKyrgyz RepublicTajikistanUzbekistanGeorgiaArmeniaAzerbaijan27 Table A1.4:Pacific MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution;obs=observations;S.D.=standard deviation.Source:Authors calculation based on ADB Asi

265、a Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariableUnitObsMeanMedianS.D.MinMaxBank loans outstanding$million162,571 2,641 783 1,566 3,568 Nonperforming bank loans$million16109 99 69 28 282 MSME loans outstanding$million16216 190 138 35 415 Market capitalization$million16854 545

266、 537 405 1,921 Bank loans outstanding$million144,535 5,420 2,170 494 6,273 Nonperforming bank loans$million14166 160 123 0 365 MSME loans outstanding$million81,195 1,034 410 928 2,170 NBFI loans outstanding$million11509 511 46 430 586 Nonperforming NBFI loans$million1156 51 18 23 83 Market capitaliz

267、ation$million725,573 23,541 12,499 10,981 41,408 Number of MSMEs114,472 4,614 585 3,277 5,218 Bank loans outstanding$million16374 362 71 246 465 Nonperforming bank loans$million1619 18 3 15 25 MSME loans outstanding$million11128 123 42 63 200 SamoaFijiPapua New Guinea28 Appendix 2:Missing Data,20072

268、022 Table A2.1:Southeast Asia Lao PDR=Lao Peoples Democratic Republic;MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution;n/a.,=not available.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariable2007 2008 2009 2

269、010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022MSME outputn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstandingn/a.,n/a.,n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,Nonperforming NBFI loansn/a.,n/a.,n/a.,Bank loans outstanding Nonperforming bank lo

270、ans NBFI loans outstanding Nonperforming NBFI loans Market capitalizationn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of employeesn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,MSME outputn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,MSME exportsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstandi

271、ngn/a.,n/a.,n/a.,n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.,n/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,n/a.,Market capitalization Bank loans outstanding Nonperforming bank loansn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.,n/a.,n/

272、a.,n/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,Market capitalizationn/a.,n/a.,n/a.,n/a.,Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of employees n/a.,MSME output n/a.,MSME exportsn/a.,n/a.,n/a.,n/a.,Bank loans outstanding Nonperforming bank loans MSME loans outstanding Nonperfor

273、ming MSME loans NBFI loans outstanding Market capitalizationn/a.,n/a.,Number of MSMEs n/a.,n/a.,n/a.,Bank loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEs Number of

274、 employees Bank loans outstandingn/a.,Nonperforming bank loansn/a.,MSME loans outstandingn/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,Market capitalization Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of employeesn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,MSME

275、 outputn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstandingn/a.,n/a.,n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,Market capitalizationn/a.,Number of MSMEs Number of employees n/a.,MS

276、ME output MSME exports MSME imports Bank loans outstanding Nonperforming bank loansn/a.,n/a.,MSME loans outstanding Nonperforming MSME loansn/a.,n/a.,Market capitalization Number of MSMEs n/a.,Number of employees n/a.,Bank loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming bank loansn

277、/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Market capitalizationn/a.,n/a.,MyanmarBrunei DarussalamCambodiaIndonesiaLao PDRMalaysiaPhilippinesSingaporeThailandViet Nam29 Table A2.2:South Asia MSME=micro,

278、small,and medium-sized enterprise;NBFI=nonbank finance institution;n/a.,=not available.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariable2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022MSME outputn/a.,n/

279、a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstandingn/a.,n/a.,n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,Market capitalization Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,

280、Number of employeesn/a.,n/a.,n/a.,n/a.,n/a.,MSME outputn/a.,n/a.,n/a.,n/a.,MSME exportsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstanding Nonperforming bank loansMSME loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,

281、NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Market capitalizationn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of employeesn/a.,n/a.,n/a.,n/a.,Bank loans outstandingn/a.,Market capi

282、talizationn/a.,MSME output Bank loans outstanding Nonperforming bank loans MSME loans outstanding Nonperforming MSME loans NBFI loans outstanding Nonperforming NBFI loans Market capitalization Bank loans outstanding Nonperforming bank loans NBFI loans outstanding Nonperforming NBFI loans n/a.,n/a.,M

283、arket capitalization BangladeshIndiaNepalPakistanSri Lanka30 Table A2.3:Central and West Asia MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution;n/a.,=not available.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.Country

284、Variable2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022Number of MSMEs n/a.,Number of employees n/a.,n/a.,MSME output n/a.,Bank loans outstanding Nonperforming bank loans NBFI loans outstanding Market capitalization Number of MSMEs n/a.,Number of employees n/a.,MSME o

285、utput n/a.,Bank loans outstanding Nonperforming bank loans NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEs n/a.,Number of employees n/a.,MSME output n/a.,MSME exportsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,MSME importsn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstanding Nonp

286、erforming bank loans MSME loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,NBFI loans outstanding Market capitalizationn/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEs Number of employees MSME output Bank loans outstanding Nonperforming bank loansn/a.,MSME loan

287、s outstanding NBFI loans outstanding Number of MSMEs n/a.,Number of employees n/a.,MSME output n/a.,MSME exports n/a.,MSME imports n/a.,Bank loans outstanding Nonperforming bank loansn/a.,MSME loans outstanding n/a.,NBFI loans outstanding Market capitalization Number of MSMEsn/a.,Number of employees

288、 n/a.,n/a.,MSME outputn/a.,n/a.,Bank loans outstanding n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming MSME loansn/a.,n/a.,n/a.,n/a.,n/a.,NBFI loans outstanding n/a.,Nonperforming NBFI loans n/a.,Number of MSMEs Number of employees MSME o

289、utput MSME exports MSME imports Bank loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming bank loansn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Market capitali

290、zationn/a.,n/a.,n/a.,GeorgiaArmeniaAzerbaijanKazakhstanKyrgyz RepublicTajikistanUzbekistan31 Table A2.4:Pacific MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution;n/a.,=not available.Source:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 202

291、3 database.CountryVariable2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022Bank loans outstanding Nonperforming bank loans MSME loans outstanding Market capitalization Bank loans outstandingn/a.,n/a.,Nonperforming bank loansn/a.,n/a.,MSME loans outstandingn/a.,n/a.,n/a.

292、,n/a.,n/a.,n/a.,n/a.,n/a.,NBFI loans outstandingn/a.,n/a.,n/a.,n/a.,n/a.,Nonperforming NBFI loans n/a.,n/a.,n/a.,n/a.,n/a.,Market capitalizationn/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,n/a.,Number of MSMEsn/a.,n/a.,n/a.,n/a.,n/a.,Bank loans outstanding Nonperforming bank loans MSME loans outstandingn

293、/a.,n/a.,n/a.,n/a.,n/a.,SamoaFijiPapua New Guinea32 Appendix 3:Factor Loadings Based on Probabilistic Principal Component Analysis Table A3.1:All Countries Continued on next page CountryVariablePC1CountryVariablePC2CountryVariablePC3PhilippinesBank loans outstanding0.98TajikistanMSME output0.88Malay

294、siaBank loans outstanding0.76IndonesiaNumber of MSMEs0.97IndiaNumber of employees0.87IndonesiaMarket capitalization0.72NepalNumber of MSMEs0.96UzbekistanNumber of employees0.86MalaysiaMSME output0.68IndonesiaMSME output0.92AzerbaijanNonperforming bank loans0.85Papua New Guinea Market capitalization0

295、.66FijiBank loans outstanding0.92Viet NamNumber of employees0.82KazakhstanMSME output0.63PhilippinesNBFI loans outstanding0.92Lao PDRMarket capitalization0.80ThailandBank loans outstanding0.57Kyrgyz RepublicBank loans outstanding0.91Lao PDRMSME loans outstanding0.76BangladeshMarket capitalization0.5

296、7GeorgiaNumber of MSMEs0.88PhilippinesMSME loans outstanding0.76AzerbaijanMSME output0.53Kyrgyz RepublicMSME loans outstanding0.87Sri LankaNBFI loans outstanding0.76IndonesiaMSME exports0.50CambodiaNBFI loans outstanding0.87ThailandMSME loans outstanding0.75Kyrgyz RepublicNBFI loans outstanding0.50V

297、iet NamNumber of MSMEs0.86PakistanMarket capitalization0.73ArmeniaBank loans outstanding0.49IndiaBank loans outstanding0.84GeorgiaNumber of employees0.72GeorgiaMSME output0.48Viet NamNBFI loans outstanding0.84TajikistanNumber of employees0.70PhilippinesNumber of MSMEs0.47Kyrgyz RepublicNumber of MSM

298、Es0.82UzbekistanMSME output0.70Kyrgyz RepublicMSME output0.44IndiaNonperforming bank loans0.82TajikistanNonperforming NBFI loans 0.70IndiaBank loans outstanding0.44NepalNumber of employees0.79GeorgiaMSME output0.68ThailandNumber of employees0.39AzerbaijanNumber of employees0.79Papua New Guinea Bank

299、loans outstanding0.67Kyrgyz RepublicMSME imports0.39Lao PDRBank loans outstanding0.79MalaysiaMSME loans outstanding0.67ArmeniaNBFI loans outstanding0.39PakistanMSME output0.78SingaporeMarket capitalization0.67MyanmarNonperforming NBFI loans 0.39NepalBank loans outstanding0.78TajikistanNonperforming

300、bank loans0.65Viet NamNonperforming bank loans0.38Viet NamBank loans outstanding0.77MalaysiaNumber of employees0.64SamoaBank loans outstanding0.37PakistanBank loans outstanding0.77IndonesiaNumber of employees0.60Sri LankaMarket capitalization0.35Sri LankaBank loans outstanding0.77TajikistanNonperfor

301、ming MSME loans0.59MyanmarNBFI loans outstanding0.34GeorgiaBank loans outstanding0.77ThailandMarket capitalization0.57PhilippinesNumber of employees0.33FijiMSME loans outstanding0.77PhilippinesNumber of employees0.56NepalBank loans outstanding0.33KazakhstanNumber of employees0.76Lao PDRBank loans ou

302、tstanding0.56UzbekistanNumber of MSMEs0.32BangladeshMSME output0.74UzbekistanMSME exports0.54Kyrgyz RepublicNonperforming bank loans0.32Sri LankaNonperforming NBFI loans 0.74Sri LankaMarket capitalization0.54Viet NamNumber of MSMEs0.31MyanmarNumber of MSMEs0.72Brunei DarussalamNonperforming NBFI loa

303、ns 0.52GeorgiaBank loans outstanding0.31ThailandBank loans outstanding0.72Brunei DarussalamNBFI loans outstanding0.48MalaysiaMSME loans outstanding0.31ArmeniaBank loans outstanding0.72AzerbaijanBank loans outstanding0.45PakistanNonperforming bank loans0.30Papua New Guinea Nonperforming bank loans0.7

304、1PakistanMSME output0.44NepalNumber of employees0.29BangladeshMSME loans outstanding0.71IndiaNumber of MSMEs0.43GeorgiaMSME imports0.29KazakhstanNumber of MSMEs0.69PakistanNonperforming bank loans0.43Papua New Guinea NBFI loans outstanding0.29SamoaBank loans outstanding0.69TajikistanBank loans outst

305、anding0.41ThailandMSME loans outstanding0.28CambodiaBank loans outstanding0.68Kyrgyz RepublicMSME imports0.39Viet NamNumber of employees0.27ArmeniaNBFI loans outstanding0.68BangladeshMarket capitalization0.38UzbekistanMSME imports0.27MyanmarNonperforming NBFI loans 0.68MalaysiaBank loans outstanding

306、0.37Sri LankaNonperforming bank loans0.27Viet NamMarket capitalization0.68Sri LankaBank loans outstanding0.35ThailandMSME imports0.25BangladeshBank loans outstanding0.65GeorgiaNBFI loans outstanding0.35Kyrgyz RepublicMarket capitalization0.25UzbekistanMSME imports0.64ArmeniaNonperforming bank loans0

307、.34IndonesiaMSME output0.24MyanmarNBFI loans outstanding0.64AzerbaijanNBFI loans outstanding0.34IndiaMSME output0.24KazakhstanNBFI loans outstanding0.64SamoaBank loans outstanding0.32GeorgiaMSME exports0.24ArmeniaNonperforming bank loans0.63MalaysiaMSME output0.31UzbekistanMSME exports0.24IndiaNumbe

308、r of MSMEs0.62FijiBank loans outstanding0.31PhilippinesMSME loans outstanding0.24ThailandMSME output0.62FijiMSME loans outstanding0.31Viet NamNonperforming NBFI loans 0.24SingaporeNonperforming MSME loans0.62PhilippinesNumber of MSMEs0.30MalaysiaMSME exports0.23FijiMarket capitalization0.62SamoaNonp

309、erforming bank loans0.30Kyrgyz RepublicMSME exports0.23Lao PDRNBFI loans outstanding0.61NepalNumber of employees0.29GeorgiaNBFI loans outstanding0.23SingaporeBank loans outstanding0.60IndiaNonperforming bank loans0.28KazakhstanNonperforming bank loans0.22Brunei DarussalamMSME output0.57TajikistanMSM

310、E loans outstanding0.27CambodiaBank loans outstanding0.20GeorgiaNumber of employees0.55MalaysiaMSME exports0.26AzerbaijanNonperforming bank loans0.19IndonesiaBank loans outstanding0.55Kyrgyz RepublicMSME loans outstanding0.24IndiaMSME loans outstanding0.19ThailandNonperforming MSME loans0.55Pakistan

311、NBFI loans outstanding0.24GeorgiaMSME loans outstanding0.19PakistanNBFI loans outstanding0.55IndonesiaMarket capitalization0.24NepalNumber of MSMEs0.18BangladeshNBFI loans outstanding0.54ThailandBank loans outstanding0.22GeorgiaNumber of employees0.18Kyrgyz RepublicMSME output0.51Viet NamNBFI loans

312、outstanding0.22IndiaNumber of MSMEs0.17PhilippinesNumber of employees0.49GeorgiaMarket capitalization0.22ThailandMSME output0.17IndonesiaNonperforming bank loans0.49IndonesiaNumber of MSMEs0.21UzbekistanMSME output0.17ThailandNumber of MSMEs0.48PhilippinesMarket capitalization0.21UzbekistanNonperfor

313、ming bank loans0.16Papua New Guinea Bank loans outstanding0.48GeorgiaNumber of MSMEs0.20TajikistanMSME output0.15Sri LankaNBFI loans outstanding0.48Viet NamNumber of MSMEs0.18CambodiaNonperforming bank loans0.15NepalMarket capitalization0.47SingaporeNumber of employees0.18SingaporeMarket capitalizat

314、ion0.15IndonesiaNumber of employees0.46ThailandNumber of employees0.18IndiaMSME exports0.14ThailandNonperforming bank loans0.46KazakhstanNonperforming bank loans0.17AzerbaijanBank loans outstanding0.14MalaysiaMSME loans outstanding0.46PakistanBank loans outstanding0.16Kyrgyz RepublicNumber of MSMEs0

315、.13IndonesiaNonperforming MSME loans0.45KazakhstanNumber of employees0.15FijiNonperforming bank loans0.13IndonesiaMSME exports0.44SingaporeBank loans outstanding0.15Papua New Guinea MSME loans outstanding0.11ThailandNumber of employees0.43Kyrgyz RepublicMSME output0.12PakistanMSME output0.10Indonesi

316、aMSME loans outstanding0.43AzerbaijanNumber of employees0.11ThailandMSME exports0.10UzbekistanNumber of MSMEs0.42IndiaBank loans outstanding0.11MyanmarBank loans outstanding0.10SamoaNumber of MSMEs0.42PakistanNonperforming MSME loans0.11CambodiaNBFI loans outstanding0.10AzerbaijanMSME output0.42Kyrg

317、yz RepublicNBFI loans outstanding0.10Kyrgyz RepublicBank loans outstanding0.10UzbekistanNumber of employees0.40KazakhstanNumber of MSMEs0.08SingaporeMSME output0.10PhilippinesMSME loans outstanding0.40SingaporeNBFI loans outstanding0.08UzbekistanNumber of employees0.09CambodiaNonperforming NBFI loan

318、s 0.40Kyrgyz RepublicBank loans outstanding0.05PhilippinesNonperforming bank loans0.09SamoaMSME loans outstanding0.39NepalNumber of MSMEs0.04Brunei DarussalamNonperforming bank loans0.09CambodiaMarket capitalization0.39KazakhstanBank loans outstanding0.03Kyrgyz RepublicMSME loans outstanding0.08Mala

319、ysiaMSME output0.38BangladeshNonperforming bank loans0.03KazakhstanNumber of MSMEs0.08IndiaMSME exports0.38SingaporeNonperforming bank loans0.02Sri LankaBank loans outstanding0.08MalaysiaBank loans outstanding0.37Brunei DarussalamMSME output0.02GeorgiaNonperforming bank loans0.08CambodiaNonperformin

320、g bank loans0.36IndonesiaMSME output0.02Brunei DarussalamBank loans outstanding0.08SingaporeNBFI loans outstanding0.36ArmeniaNBFI loans outstanding0.00PakistanNonperforming NBFI loans 0.08GeorgiaMSME output0.35SamoaNumber of MSMEs-0.01KazakhstanNumber of employees0.08SingaporeMarket capitalization0.

321、35TajikistanNumber of MSMEs-0.03AzerbaijanNBFI loans outstanding0.07ArmeniaMarket capitalization0.35Kyrgyz RepublicNumber of MSMEs-0.03CambodiaNonperforming NBFI loans 0.07IndiaNumber of employees0.34KazakhstanMSME output-0.04GeorgiaMarket capitalization0.07IndonesiaMarket capitalization0.34Brunei D

322、arussalamNonperforming bank loans-0.05TajikistanMSME loans outstanding0.07PhilippinesNumber of MSMEs0.33ArmeniaNumber of employees-0.07GeorgiaNumber of MSMEs0.07AzerbaijanNumber of MSMEs0.32IndiaNBFI loans outstanding-0.08PhilippinesBank loans outstanding0.07MalaysiaNumber of MSMEs0.31PhilippinesBan

323、k loans outstanding-0.09PakistanNonperforming MSME loans0.06ThailandMSME loans outstanding0.31ArmeniaBank loans outstanding-0.12KazakhstanNBFI loans outstanding0.0633 continued Lao PDR=Lao Peoples Democratic Republic;MSME=micro,small,and medium-sized enterprise;NBFI=nonbank finance institution.Sourc

324、e:Authors calculation based on ADB Asia Small and Medium-Sized Enterprise Monitor 2023 database.CountryVariablePC1CountryVariablePC2CountryVariablePC3GeorgiaNBFI loans outstanding0.31Kyrgyz RepublicMSME exports-0.13MalaysiaNumber of employees0.05IndiaMarket capitalization0.31MalaysiaNBFI loans outst

325、anding-0.13Sri LankaNonperforming NBFI loans 0.04Kyrgyz RepublicMarket capitalization0.31TajikistanNBFI loans outstanding-0.13KazakhstanBank loans outstanding0.04ThailandMarket capitalization0.29PhilippinesNBFI loans outstanding-0.14MalaysiaMarket capitalization0.03BangladeshMarket capitalization0.2

326、7GeorgiaNonperforming bank loans-0.15FijiMSME loans outstanding0.03Kyrgyz RepublicNonperforming bank loans0.26NepalMarket capitalization-0.18UzbekistanNBFI loans outstanding0.03Lao PDRNonperforming bank loans0.23Papua New Guinea Nonperforming bank loans-0.19MalaysiaNonperforming MSME loans0.03Viet N

327、amNonperforming NBFI loans 0.23Brunei DarussalamBank loans outstanding-0.20FijiMarket capitalization0.02TajikistanMSME output0.21MyanmarNumber of MSMEs-0.21ThailandNumber of MSMEs0.01MalaysiaMarket capitalization0.20ThailandMSME imports-0.21Lao PDRBank loans outstanding0.01KazakhstanMSME output0.19B

328、angladeshMSME loans outstanding-0.22UzbekistanBank loans outstanding0.01UzbekistanMSME exports0.16SingaporeNonperforming MSME loans-0.22ArmeniaMSME output0.00SingaporeNonperforming bank loans0.16Lao PDRNonperforming bank loans-0.23IndonesiaNumber of MSMEs-0.01PhilippinesNonperforming bank loans0.15V

329、iet NamNonperforming bank loans-0.24MyanmarNumber of MSMEs-0.01BangladeshNonperforming bank loans0.15KazakhstanMSME loans outstanding-0.25SingaporeNumber of MSMEs-0.02PhilippinesMarket capitalization0.14MalaysiaNumber of MSMEs-0.26Sri LankaNBFI loans outstanding-0.02Papua New Guinea Nonperforming NB

330、FI loans 0.13PakistanMSME loans outstanding-0.26Papua New Guinea Bank loans outstanding-0.03Sri LankaNonperforming bank loans0.11UzbekistanMSME imports-0.27Brunei DarussalamNBFI loans outstanding-0.03Viet NamNumber of employees0.09GeorgiaBank loans outstanding-0.27UzbekistanMarket capitalization-0.0

331、3Kyrgyz RepublicNumber of employees0.08ThailandNumber of MSMEs-0.28SingaporeNumber of employees-0.03SingaporeMSME loans outstanding0.08ThailandMSME output-0.28ArmeniaMarket capitalization-0.03GeorgiaNonperforming MSME loans0.07IndonesiaNonperforming MSME loans-0.30IndiaNumber of employees-0.04Tajiki

332、stanNonperforming NBFI loans 0.05Sri LankaNonperforming NBFI loans-0.31Viet NamMarket capitalization-0.04GeorgiaNonperforming bank loans0.05ThailandMSME exports-0.32SingaporeNonperforming bank loans-0.04IndiaMSME output0.03ArmeniaMarket capitalization-0.32PhilippinesNonperforming MSME loans-0.05Taji

333、kistanNonperforming bank loans-0.04Papua New Guinea MSME loans outstanding-0.35Papua New Guinea Nonperforming bank loans-0.05BangladeshNonperforming NBFI loans-0.06ArmeniaNumber of MSMEs-0.36Brunei DarussalamMSME output-0.06TajikistanNumber of employees-0.09ArmeniaMSME output-0.36ThailandMarket capitalization-0.07PakistanMarket capitalization-0.09CambodiaNBFI loans outstanding-0.36BangladeshNonper