SME policies as a barrier to growth of SMEsDaisuke Shimizu, Hirofumi Uchida, Iichiro Uesugi, Wako Watanabe, Peng Xu, Noriyuki Yanagawa, and Makoto Yano for many valuable comments.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
/Published online: 12 November 2018
(2020) 54:1067–1106Small Bus Econhttps://doi.org/10.1007/s11187-018-0119-0
Abstract We investigate whether firms have incen-tives to retain their status as small and medium enter-prises (SMEs) to benefit from various SME policies.To examine this issue, we use the exogenous changesin the SME Basic Act in Japan. The SME BasicAct describes aims of SME policies in Japan, whichinvolve enhancing the growth of SMEs using policy.The Act also contains the requirements and defini-tions of SMEs that are the target of the policies. Onlyfirms that satisfy the definitions of SMEs under theSME Basic Act can participate in the SME policy pro-
This study was conducted as part of the Project “Study onCorporate Finance and Firm Dynamics” at the ResearchInstitute of Economy, Trade, and Industry (RIETI). Thestudy uses microdata from the Surveys for the FinancialStatements Statistics of Corporations by Industry HoujinKigyou Toukei Chosa conducted by the Ministry of Finance.This study was supported by a Grant-in-Aid for ScientificResearch (C) #16K03753 from the Japan Society for thePromotion of Science. The author would also like to thankFumio Akiyoshi, Hikaru Fukanuma, Yasuo Goto, KaoruHosono, Daisuke Miyakawa, Masayuki Morikawa, Yoshi-aki Ogura, Hiroshi Ohashi, Arito Ono, Kuniyoshi Saito,Daisuke Shimizu, Hirofumi Uchida, Iichiro Uesugi, WakoWatanabe, Peng Xu, Noriyuki Yanagawa, and Makoto Yanofor many valuable comments. The seminar participants atRIETI also provided useful comments. All remaining errorsare mine.
D. Tsuruta (�)College of Economics, Nihon University, 1-3-2 Misaki-cho,Chiyoda-ku, Tokyo 102-8360, Japane-mail: [email protected]
grams in Japan. The requirements regarding capitalstock in the SME Basic Act were changed in 1999.By focusing on the change in the requirements forcapital stock as an exogenous event, we show thatfirms are less likely to increase their capital stockso that they can continue to satisfy the requirementsthat retain their status as SMEs. Furthermore, firmsthat increase their capital stock increase their size. Asfirms have a disincentive to increase capital stock sothat they can keep their SME status, this indicates thatthe SME requirements are significant constraints onfirm growth. Many studies show that individual SMEpolicies (such as R&D subsidies and public creditguarantees) stimulate firm growth and activities thatare consistent with the aim of these SME policies.Although the purpose of the SME Basic Act is theenhancement of the growth of SMEs, we show thatthe requirements under this law impede the growth ofSMEs.
Keywords SME policy · Firm growth · Equitycapital · Small businesses
JEL Classification L53 · L25 · G32 · L26
1 Introduction
We investigate whether policies for small and mediumenterprises (SMEs) impede incentives to graduatefrom SME status. Economic theories justify the use
of SME policies when market failure occurs (Storey1994). Berger and Udell (1998) argue that the infor-mation gap between lenders and borrowers is severe,which means that credit rationing is serious forSMEs.1 To enhance credit supply for SMEs, Mankiw(1986) argues that the government can establish pub-lic lending and credit guarantee schemes, which canimprove social welfare.2 In addition, Lenihan (2011)argues that dynamic externalities (e.g., knowledgespillovers and network externalities) should be pro-moted by SME policies because of systemic failures.If the spillover of knowledge and innovation fromR&D investment is significant,3 the benefits frominnovation spread to other firms that do not pay thecost of investment. Arrow (1962) argues that the socialbenefits of innovation are larger than the private ben-efits for firms, which leads to underinvestment inR&D. If the positive externalities relating to spilloversfrom innovation are significant, policy intervention forunderinvestment is justified.
However, as Storey (1994) argues, the aim of actualSME policies is ambiguous from an economic view-point. Many SME policies exist even if they are notjustified by market failures, and an excessive numberof SME policies have been adopted in many countries.One of the reasons for the excessive use of SME poli-cies is that many government departments feel thatSMEs should be supported by economic policies andregard SMEs as “their responsibility” (Storey 2008).As shown in Table 1, which provides a list of SMEpolicies compiled by the Ministry of Economy, Tradeand Industry (METI) in Japan, the menu of SME poli-cies is large. In addition to these policies, the Ministryof Finance reduces the corporate tax rate for firmsthat satisfy SME requirements. Often, the governmentimplements SME policies even when market failureis not serious. For example, Table 1 shows that theprogram “Trade practices and public procurement”increases the opportunity for SMEs to win contractsin government offices. Because SMEs cannot alwayssupply a higher quality of goods or services comparedwith large firms, SME policy should not increase their
1Some previous studies (e.g., De Meza 2002; Dawson et al.2014) argue that under asymmetric information, firms bor-row more funds compared with the situation of symmetricinformation.2In Japan, the social cost of credit guarantees is also significant,as argued by Saito and Tsuruta (2018).3Acs and Szerb (2007), for instance, argue that this is the case.
opportunities to win contracts because this reducesmarket efficiency. In addition, the corporate tax rateis reduced for all SMEs, which is another policy thatis not justified by market failure. Although a publiccredit guarantee is justified by severe market failureas Mankiw (1986) argues, the amount of credit guar-antee provided can often be excessive. OECD (2016)points out that “SMEs receive substantial governmentsupport, particularly through a large credit guaranteesystem, which supports about 40% of Japanese SMEs”(p. 16).4
In sum, SMEs receive substantial government sup-port through a wide variety of SME policies and, asOECD (2016) argues, these policies can reduce incen-tives for SME growth.5 Firms cannot benefit from thehuge range of SME policies listed in Table 1 if theyoutgrow their SME status and become large firms.6
In addition, if a firm graduates from an SME to alarge firm, it cannot access the public credit guaran-tee. During periods of financial crisis (in particular,the global financial crisis in 2008), additional publiccredit guarantee programs commenced, which enabledSMEs to access guaranteed loans more easily, therebyenhancing their liquidity. Furthermore, large firms donot enjoy a reduced corporate tax rate; therefore, theymust pay a higher rate if they graduate from SME sta-tus. Because they wish to retain their access to thevarious SME policies available, we predict that firmsdo not have the incentive to graduate from SME status,which impedes firm growth.
Many studies empirically investigate the effects ofseveral individual SME policies. A number of stud-ies (e.g., Kang and Heshmati 2008; Oh et al. 2009;Craig et al. 2007; Uesugi et al. 2007) show empiricallythat the public credit guarantee program has posi-tive effects on employment, sales, and local growth,and that it reduces the default and bankruptcy rates
4According to OECD (2013), the volume of credit guaranteesas a percentage of GDP is 7.3% in Japan, the highest among thelisted countries.5OECD (2016) notes that “small companies in Japan tend toremain small, in part because high public support discouragessmall firms from growing because they would lose the benefitsassociated with SME status” (p. 11).6For example, the policies result in SMEs winning contractseven if their production capacity and efficiency are not as highas those of larger firms, so the policies enhance the profitabilityof SMEs. In other words, if firms grow from SMEs into largefirms, they cannot participate in these policy programs and thisreduces their opportunities to win contracts.
1068
SME policies as a barrier to growth of SMEs
Table 1 List of major SME policies in Japan
Management Support
Start-ups and ventures Assists those planning to start a business or venture owners trying toimprove their operations in financing and obtaining relevant information.
Business innovation Assists SMEs undergoing business innovation in financing, handling taxes,and cultivating markets.
New collaboration Supports collaboration between SMEs to enter new areas of business byproviding subsidies, advice, and financing assistance.
Business revitalization Supports SMEs in their efforts to revitalize their business through the SMERevitalization Support Committee.
Employment and human resources Supports SMEs with human resources development and the resolutionof business challenges by implementing the Small and Medium-sizedEnterprise Consultants system, offering training, and dispatching experts.
Globalization Provides information and advice to help SMEs to move production overseasor find markets abroad.
Trade practices and public procurement Promotes fair subcontracting practices and the development of small andmedium-sized subcontractors and thereby increases the opportunity forSMEs to win contracts.
Business stability Assists SMEs in maintaining stable operations by supporting them dur-ing bankruptcy, new pandemic influenza, and earthquakes and other naturaldisasters, as well as by assisting them to develop business continuity plans.
Mutual aid system Helps small companies to prepare for business closure and retirement, andSMEs to prepare for the bankruptcy of their major customers.
Small businesses Provides managerial and financial support to small businesses with 20 orfewer employees (five or fewer for those in the commerce or service sector).
Small and medium manufacturers Supports R&D and human resources development at SMEs with key man-ufacturing technologies. Selects “300 of Japan’s Exciting Monozukuri(Manufacturing) SMEs.”
Technological innovation, IT,and energy efficiency
Assists SMEs committed to technological development, IT utilization, andhigher energy efficiency by providing subsidies, financial assistance, andrelevant information.
Intellectual property Supports SMEs with intellectual property strategies by implementing mea-sures to protect intellectual property and measures to combat damagecaused by counterfeiting.
SME Assistance Centers Dispatches experts to assist SMEs in addressing difficult or specializedbusiness challenges (e.g., launch of new operations or business succession)and otherwise helps SMEs directly or via support institutions.
Financial Support
Safety-net guarantee program Supports SMEs whose business stability is threatened by external factors(e.g., a major customer’s restricted operations or application for rehabilita-tion procedures, the impact of a disaster, or the failure of the main bank) bymaking additional credit guarantees available.
Safety-net loans Makes loans to SMEs temporarily facing cash flow problems owing toa radical change in the business environment, the bankruptcy of a majorcustomer, or the streamlining of the main bank.
Fiscal Support
Taxation Gives information and advice on various tax measures to support SMEs.
Accounting Gives information and advice on “SME accounting,” which helps SMEsto enhance their capability to analyze management, ensure financing, andincrease order intake.
Companies Act Gives information and advice on the new Companies Act, which addition-ally includes systems that bring significant benefits to SMEs, such as theaccounting adviser system.
1069
D. Tsuruta
Table 1 (continued)
Business succession Gives information and advice on measures to support SMEs’ smoothbusiness succession.
Commerce and Regional Support
Revitalization of commerce Supports efforts to improve the attractiveness of small and medium mer-chants, shopping districts, and city centers.
Regional industries Invigorates regional industries, such as locally based industries and tradi-tional handicraft industries, by providing subsidies and low-interest loans.
Collaboration between agriculture,commerce, and industry
Comprehensively assists business activities conducted by organic partner-ships between SMEs and those engaged in agriculture/forestry/fisheriesthrough the effective use of their business resources.
“Meet and Experience RegionalAttractiveness” campaign
Actively supports and increases the publicity for attractive regional prod-ucts.
Source: Website of the Small and Medium Enterprise Agency (http://www.chusho.meti.go.jp/sme english/outline/04/01.html)
Note: This table shows the list of SME policies implemented by the Ministry of Economy, Trade and Industry (METI) in Japan
of SMEs. Focusing on the R&D policy, many stud-ies (e.g., Cantner and Kosters 2012; Czarnitzki andDelanote 2015; Cowling 2016; Lokshin and Mohnen2012; Koga 2005) investigate the effects of R&Dtax credits and subsidies on R&D investment andinnovation of SMEs, which are common policies foraddressing the underinvestment problem. Some stud-ies (e.g., Foreman-Peck 2013; Cin et al. 2017) focuson the policy effects of R&D tax credits and subsi-dies on firm growth or productivity of SMEs. Otherstudies (e.g., Bernini and Pellegrini 2011; Cerqua andPellegrini 2014; Ipinnaiye et al. 2017) argue that pub-lic subsidies (not limited to R&D subsidies) enhancethe growth of SMEs or regional firms. Some papers
investigate the effects of location-based tax incen-tives on the growth of regional firms. Location-basedtaxes are used to stimulate the growth of regionalfirms in many countries, which aims at the regionaldevelopment of less-developed areas through policyintervention by local governments. Hanson and Rohlin(2011) and Chaurey (2017) show empirically that firmgrowth can be enhanced in the target areas, which isthe aim of the policy.
Although a number of reports focus on the effectsof individual SME policies, few studies investigatewhether the substantial government support for SMEsprovides an incentive to retain SME status, which hasa negative effect on firm growth. In this paper, we
Table 2 Definition of SMEs under the Corporation Tax Act and the SME Basic Act in Japan
Panel A: Definition under the Corporation Tax Act
Capital
Industry Stock Employees
All 100 million or less No requirement
Panel B: Definition under the SME Basic Act
Before 1999 After 2000
Capital Capital
Industry Stock Employees Stock Employees
Definition 1 Manufacturing, etc. 100 million or less 300 or fewer 300 million or less 300 or fewer
Definition 2 Wholesale 30 million or less 100 or fewer 100 million or less 100 or fewer
Definition 3 Retail 10 million or less 50 or fewer 50 million or less 50 or fewer
Definition 4 Service 10 million or less 50 or fewer 50 million or less 100 or fewer
Note: This table shows the definitions of SMEs under the Corporation Tax Act (in Panel A) and the SME Basic Act (in Panel B).“Manufacturing, etc.” includes all industries except the wholesale, retail, and service industries
investigate whether the various SME policy programsin Japan impede the growth of firms from SMEs tolarge firms. To examine this issue, we employ twostrategies. First, we focus on the definitions of SMEsunder the Corporation Tax Act, which defines SMEsas firms with capital stock of 100 million yen or less.We investigate whether SMEs that are close to this capfor capital stock (i.e., 100 million yen) are less likelythan other firms to increase their capital stock so thatthey can remain an SME and retain their access to theSME policy programs.
Second, for our analysis, we use the changes inthe Small and Medium-sized Enterprise Basic Act(the SME Basic Act) in Japan that came into effectin December 1999. The SME Basic Act is the lawthat establishes the basic principles and aims of SMEpolicies in Japan, but does not describe the con-tent of individual SME policies. This law describesthe aim of SME policies as the development of theJapanese economy through the achievement of sig-nificant growth of SMEs using SME policies. Thelaw also describes the definition and requirements ofthe SMEs that are the targets of the SME policies inJapan. In Japan, only firms that satisfy the definition ofSMEs under the SME Basic Act can participate in theSME policy programs listed in Table 1. Before 1999,SMEs were defined as firms with 100 or fewer regularemployees or with 100 million yen of capital stock orless (except for firms in the wholesale, retail, and ser-vice industries). Following revision of the SME BasicAct, the requirement for capital stock was changed to300 million yen or less, so that firms could increasecapital stock but still satisfy the requirements of theSME Basic Act. As Table 2 shows, the definitionsof SMEs differ for firms in the wholesale, retail, andservice industries.
By focusing on the change of the SME Basic Actas an exogenous event, we can test whether firms haveincentives to retain their SME status even if they cangraduate to large firm status. Firms that did not grad-uate from SME status when they could have are likelyto have increased their capital stock after the changein the SME Basic Act, which relaxed the capital stockrequirement. Furthermore, by focusing on the differ-ence in the requirements for SMEs between industries,we can test the hypothesis using the difference-in-differences approach. As Table 2 shows, the capi-tal stock requirement changed from 100 million yenor less to 300 million yen or less for a firm in a
manufacturing industry with the change in the SMEBasic Act. However, in the wholesale, retail, andservice industries, this change in the capital stockrequirement was not adopted either before or after thechange in the SME Basic Act. Therefore, we can usethe subsample of firms in the wholesale, retail, andservice industries as a control group and those in theother industries as a treatment group.
Similarly, by focusing on the changes in the cap-ital stock requirements for the wholesale industry(changed from 30 million yen or less to 100 mil-lion yen or less) and those for the retail and serviceindustries (changed from 10 million yen or less to50 million yen or less), we can use firms in thoseindustries as a treatment group. As these changes inthe capital stock requirement were not adopted in theother industries, we can use a subsample of firms inother industries as a control group. These heteroge-neous changes in the definition of SMEs are rarelyadopted in other countries, so the Japanese case is anideal natural experiment to test the hypothesis.
The main findings of this paper are as follows. First,firms with capital stock of 100 million yen or less (i.e.,SMEs according to the definitions in the CorporationTax Act and the SME Basic Act before it was alteredin 1999) are less likely to increase capital stock com-pared with firms with capital stock of over 100 millionyen. This implies that SMEs have a disincentive toincrease their capital stock because they benefit fromkeeping their SME status.
Second, before the change in the definition ofSMEs under the SME Basic Act in 1999, a firmincreased its capital stock less if the firm’s capitalstock was close to the capital stock requirement underthe original SME Basic Act. After the change in thedefinition, which involved an increase in the capi-tal stock limit, firms that satisfied the previous SMErequirement then increased their capital stock. Thiseffect is larger if a firm’s capital stock is close tothe capital stock requirement under the original SMEBasic Act. These effects are robust because they aresupported if we estimate them using a different treat-ment group. Furthermore, the distributions of capitalstock after the policy change are right-shifted com-pared with those without the policy change. Theseimpacts are large for firms with capital stock close to100 million yen that are larger SMEs.
Third, firms that increased their capital stock alsoincreased their firm size (in terms of asset growth).
1071
D. Tsuruta
As firms have a disincentive to increase their cap-ital stock so that they can keep their SME status,this indicates that the SME requirements are sig-nificant constraints on firm growth. The mitigationof the requirements of SMEs enhanced asset growthfor manufacturing industries by 0.149% on average,while the actual average percentage asset growth was1.932% from 2000 to 2007. Additionally, we showthat firms decreased their debt by increasing theirequity. This implies that firms were able to adjust toan optimal capital structure after the relaxation of theircapital stock requirements.
This study differs from those that focus on severalSME policies. Previous work shows that each indi-vidual policy enhanced the growth of SMEs, which isconsistent with the aim of those policies. The SMEpolicy that we study in Japan aims to promote thedevelopment of small businesses.7 Although the aimof the policy is the growth of SMEs, we show that theSME policy impeded firm growth of small businessesthrough decreases in equity capital.
Our study is related to work that uses the calibra-tion of a theoretical model to argue that policies thatdepend on firm size cause distortions of firm size. Forexample, Garicano et al. (2016) and Gourio and Roys(2014) focus on the many labor laws in France thatare binding for firms with 50 employees or more andestimate the welfare costs of these regulations. Usingthe Lucas model, Guner et al. (2006) and Guner et al.(2008) show that size-dependent laws, such as Japan’sLarge Scale Retail Location Law, distort firm-sizedistributions. Garcıa-Santana and Pijoan-Mas (2014)focus on the Small Scale Reservation Laws in Indiathat reserve several products for production by small-scale industries. They also use the Lucas model toshow that this policy decreases average output perworker by 2% in the economy. Hosono et al. (2017)investigate whether the distribution of firm size isdistorted using the SME Basic Act in Japan.
Similarly to these studies, we find that size-dependent policies impede firm growth by small busi-nesses. However, this paper differs from the exist-ing literature in two ways. First, whereas previouswork employs simulations from theoretical models,we employ a difference-in-differences approach usingthe change in the SME Basic Act as an exogenous
7See http://www.chusho.meti.go.jp/soshiki/ninmu.html regard-ing the aim of the Small and Medium Enterprise Agency inJapan.
event and utilize not a macroeconomic model but aneconometric model, using firm-level data of smallbusinesses. Second, we focus not only on firm growth,but also on the financial activities of small businesses.8
The remainder of the paper is organized as follows.Section 2 provides the definitions of SMEs underthe Corporation Tax Act and the SME Basic Act.Section 3 describes the data set. Section 4 introducesthe empirical strategy and hypotheses for the relation-ships between SME policies, capital stock, and firmgrowth. Section 5 provides the estimation results forthe hypotheses. Section 6 concludes the paper.
2 Definitions of SMEs in Japan
2.1 Corporation Tax Act
There are several definitions of SMEs in Japan, withthe major definitions being those of the SME BasicAct and the Corporation Tax Act. Under the Cor-poration Tax Act, SMEs are defined as firms with100 million yen of capital stock or less (Panel A ofTable 2). The corporate tax rate is reduced for firmsthat satisfy the definition of an SME under the Corpo-ration Tax Act. For example, the corporate tax rate forSMEs is 22%, which is applied to incomes under eightmillion yen. The corporate tax rate for large firmsbetween 1999 and 2012 was 30%.9 Therefore, if firms
8Hosono et al. (2017) also investigate the effects of changingthe SME Basic Act of Japan using firm-level data. However,the data used in Hosono et al. (2017) include firms with 30 mil-lion yen or more of capital stock, and therefore do not containthe data of smaller-sized SMEs. Our data include various-sizedSMEs, so we can investigate the effects on micro firms thatneed the support of SME policies. According to EconomicCensus for Business Frame in 2014 by the Statistics Bureau ofthe Ministry of Internal Affairs and Communications (MIC),the ratio of the number of firms with 30 million yen or less ofcapital stock to all firms is 86.7% (see the website of the MIC(http://www.stat.go.jp/data/e-census/2014/pdf/kaku gaiyo.pdf[last date accessed: September 2018]) for more detail), so wecan investigate the effects of the SME Basic Act more accu-rately. However, the data used in Hosono et al. (2017) are paneldata, not pooled cross-section data, and can therefore investi-gate the effects on the postperformance of SMEs over severalyears.9See the website of the Ministry of Finance in Japan (https://www.mof.go.jp/tax policy/summary/corporation/082.pdf (in Japanese,last date accessed: September 2018)) regarding the corporate taxrate trends.
satisfy the definition of an SME under the Corpora-tion Tax Act, they can pay a low corporate tax rate andincrease their cash flow.
2.2 SME Basic Act
The SME Basic Act was implemented in 1963. Theaim of this law is “to promote in a comprehensivemanner measures for small and medium enterprisesby establishing the basic principles, basic policies andother basic matters relating to measures for SMEs andclarifying the responsibilities, etc. of the State and oflocal public entities, so as to contribute to the sounddevelopment of the national economy and improve-ment in the quality of life of the people.” To realize thisaim, this law establishes “the basic principles, basicpolicies and other basic matters relating to measuresfor SMEs”.10
This law does not describe the contents of indi-vidual SME policies. Instead, it describes the basicprinciples and purpose of SME policies in Japan,which relate to the development and growth of SMEs.Article 3, Basic Principles notes that the “developmentof SMEs must be encouraged by promoting busi-ness innovation and start-ups among them, strengthen-ing their business fundamentals, and smoothing theiradaptation to changes in social or economic conditionsso as to foster the autonomous efforts of indepen-dent SMEs.”11 In sum, this law aims to develop theJapanese economy through the encouragement of thegrowth of SMEs using various SME policies.
The key part of this law is the definition andrequirement of SMEs that are used in the SME poli-cies of the Small and Medium Enterprise Agency.This law determines the target firms of SME policiesby defining the requirements of firms to be classi-fied as “SMEs”. This requirement is adopted in almostall policies implemented by the Small and MediumEnterprise Agency, so firms that satisfy the require-ment of the SME Basic Act can utilize individual SMEpolicies.
10See Article 1 on the website of the Small and MediumEnterprise Agency (http://www.chusho.meti.go.jp/sme english/outline/08/01 01.html) (last date accessed: September 2018) .11See Article 3 on the website of the Small and MediumEnterprise Agency (URL is in footnote 10).
As shown in Panel B of Table 2, the definition ofSMEs in the SME Basic Act is complex. The defini-tions of SMEs before December 1999 are as follows.i) SMEs under the SME Basic Act are defined as firmswith 100 million yen of capital stock or less and/or 300or fewer regular employees. ii) SMEs in the whole-sale industry are defined as firms with 30 million yenof capital stock or less and/or 100 or fewer regularemployees. iii) SMEs in the retail industry and theservice industry are defined as firms with 10 millionyen of capital stock or less and/or 50 or fewer regularemployees.
In December 1999, the SME Basic Act was revisedand the capital stock requirement was relaxed. Therevised requirement for SME status after December1999 is as follows. i) SMEs under the SME Basic Actare defined as firms with 300 million yen of capitalstock or less and/or 300 or fewer regular employ-ees. ii) SMEs in the wholesale industry are definedas firms with 100 million yen of capital stock or lessand/or 100 or fewer regular employees. iii) SMEs inthe retail industry are defined as firms with 50 millionyen of capital stock or less and/or 50 or fewer regu-lar employees. iv) SMEs in the service industry aredefined as firms with 100 million yen of capital stockor less and/or 100 or fewer regular employees.
According to Nakata (2013), the main reasons forrelaxing the capital stock requirements are as follows.First, the original requirement for the capital stockwas established in 1963, which is not consistent withJapan as a developed country. Second, personal guar-antees by business owners were required to borrowfunds from banks for over half of the firms with a cap-ital stock of 300 million yen or less. In contrast, iffirms’ capital stock was over 300 million yen, personalguarantees by business owners were rarely required.Third, the number of public firms with 300 millionyen of capital stock or less was small, while the num-ber of firms with over 300 million yen was large. Thisimplies that credit constraints were more severe forfirms with 300 million yen of capital stock or less, sothe capital stock requirement was relaxed.
3 Data
We use annual firm-level data from the Surveys forthe Financial Statements Statistics of Corporationsby Industry (hereafter FSSC; Houjin Kigyou Toukei
Chosa in Japanese) conducted by the Ministry ofFinance. According to the website of the Ministry ofFinance,12 the FSSC are “one part of the fundamen-tal statistical surveys under the Statistics Act and havebeen conducted as sampling surveys so as to ascertainthe current status of business activities of commer-cial corporations in Japan.” The target firms of theFSSC are all commercial corporations in Japan. Allfirms with capital stock of 600 million yen or moreare included. Those with capital stock of between 100million and 600 million yen are randomly selectedwith equal probability. Those with less than 100 mil-lion yen of capital are randomly sampled every fiscalyear. Therefore, of the firms with less than 100 mil-lion yen in capital, a different sample of target firms isselected each fiscal year. The response rates for eachfiscal year are around 80%. The questionnaire formsare available at the Ministry of Finance (MOF) web-site.13 A list of all variables is available from thiswebsite. 14
Data on firms’ balance sheets are available at thebeginning and end of each fiscal year. The data atthe end of fiscal year t are set equal to the data atthe beginning of fiscal year t + 1. We use observa-tions from FY1991 to FY2007. To exclude large firms,the sample is limited to firms with 500 million yenor less of capital stock. We choose the sample periodFY1991–FY2007 to exclude the effects of the bubbleeconomy before 1990 and the global financial crisisafter 2008. The number of full firm-year observationsis 306,353 during the period FY1991–FY2007.15
12For details of the survey see: https://www.mof.go.jp/english/pri/reference/ssc/index.htm (last date accessed: September2018).13https://www.mof.go.jp/pri/reference/ssc/outline.htm#questionnaire (in Japanese, last date accessed: September2018).14The items in the annual survey are similar to those in thequarterly survey. All quarterly items in English are avail-able at the following website: https://www.mof.go.jp/english/pri/reference/ssc/historical.htm (last date accessed: September2018).15The database in this paper is pooled-cross section data, so theentry–exit data are unavailable. The observations are randomlyselected in every year, so firms that are likely to exit are includedin our sample. This implies that sample selection bias is not sosevere. However, we cannot control remaining sample selectionbias using econometric models, which is a shortcoming of ourdatabase.
4 Empirical strategy
4.1 Effects of the cap on capital stock
4.1.1 Hypothesis
As argued by Baumol (1962), firms determine thegrowth rate that maximizes the net revenue and costof growth. Without a cost of growth (i.e., the admin-istrative cost of growth), firms have an incentive togrow infinitely. However, the cost of growth is sig-nificant, so an optimal growth rate exists. In general,public support lowers the cost of growth for SMEs, soSME policy could stimulate firm growth and increasethe number of firms transitioning from a SME to alarge firm. If they do not satisfy the requirements ofSME status, the cost of growth increases discontinu-ously because they cannot receive public support andreduced corporate tax.16 Therefore, the growth ratesof firms are lower after transition from SME status.
Firms cannot receive public support, however, ifthey grow and transition from SME status. In this case,the cost of firm growth, which includes the opportu-nity cost of receiving SME policies, is higher for firmsnear to the cap of the requirements of SMEs. Aftertransition from SME status, the cost of firm growthdoes not include the opportunity cost, so the optimalgrowth rate increases.
In sum, when the positive effects of public supportof firm growth or the benefit of being a large firmare large, the growth rate of SMEs is not lower, eventhough firms are close to the capital limit for beingclassified as an SME. In this case, the SME policyis not a significant barrier to firm growth. However,if the opportunity cost of receiving SME policies islarge, the growth rate of those SMEs is lower. In thiscase, the SME policy is a significant barrier to firmgrowth. We need to test which cases are supportedusing empirical analysis.
As described in the previous section, the cap oncapital stock in the definition of SMEs under the Cor-poration Tax Act is 100 million yen. If firms have anincentive to retain their SME status and observe theSME requirements to save corporate tax, they will notincrease their capital stock over 100 million yen. We
16This is an upward shift in the cost of growth in Figure 1 ofBaumol (1962) when public support is terminated or tax ratesrise.
Capital Stock 20M equals one if a firm’s capital stock is >10–20 million yen 26,499 0.087
Capital Stock 30M equals one if a firm’s capital stock is >20–30 million yen 18,839 0.062
Capital Stock 40M equals one if a firm’s capital stock is >30–40 million yen 8,334 0.027
Capital Stock 50M equals one if a firm’s capital stock is >40–50 million yen 17,455 0.057
Capital Stock 60M equals one if a firm’s capital stock is >50–60 million yen 4,680 0.015
Capital Stock 70M equals one if a firm’s capital stock is >60–70 million yen 2,954 0.010
Capital Stock 80M equals one if a firm’s capital stock is >70–80 million yen 4,839 0.016
Capital Stock 90M equals one if a firm’s capital stock is >80–90 million yen 3,783 0.012
Capital Stock 100M equals one if a firm’s capital stock is >90–100 million yen 16,422 0.054
Capital Stock 110M equals one if a firm’s capital stock is >100–110 million yen 1,857 0.006
Capital Stock 120M equals one if a firm’s capital stock is >110–120 million yen 2,719 0.009
Capital Stock 130M equals one if a firm’s capital stock is >120–130 million yen 1,540 0.005
Capital Stock 140M equals one if a firm’s capital stock is >130–140 million yen 1,447 0.005
Capital Stock 150M equals one if a firm’s capital stock is >140–150 million yen 4,101 0.013
Capital Stock 200M equals one if a firm’s capital stock is >150–200 million yen 14,155 0.047
Capital Stock 300M equals one if a firm’s capital stock is >200–300 million yen 22,918 0.075
Capital Stock 400M equals one if a firm’s capital stock is >300–400 million yen 18,626 0.061
Capital Stock 500M equals one if a firm’s capital stock is >400–500 million yen 26,540 0.087
Note: This table shows the definitions of Capital Stock Dummies used in Table 5
predict that firms with a capital stock close to 100 mil-lion yen are less likely to increase their capital stock ifthe requirements are a significant constraint. In addi-tion, under the SME Basic Act, the caps on capitalstock in the definitions of SMEs are 10, 30, 100, or300 million yen, depending on the industry and theyear. We predict that firms with a capital stock closeto these caps are less likely to increase their capitalstock. However, as we argued, if SME policies reducethe cost of firm growth significantly and the oppor-tunity cost of SME policy is negligible, the estimatedeffects of these caps are insignificant.
4.1.2 Equation
To test our hypothesis, we estimate the followingequation:
�Capital Stocki,t =∑
αj
1Capital Stock Dummyj
i,t−1
+ Xi,t α2 + ζt + ηi + θi,t (1)
where θi,t is the error term of firm i in fiscal year t, ηi
is industry fixed effects of 45 industries, and ζi is year
fixed effects from FY1991 to FY2007. We use twodefinitions of �Capital Stocki,t . One is a dummyvariable that has a value of one if capital stock at theend of fiscal year t is larger than at the beginning offiscal year t (additional capital stock dummy, shownas “Dummy” in tables). The other is the differencein capital stock at the end of fiscal year t comparedwith that at the beginning of fiscal year t (amountof �capital stock, shown as “Amount” in tables). Xincludes leverage at the beginning of fiscal year t, tan-gible fixed assets at the beginning of fiscal year t, andoperating incomes in fiscal year t.
We drop observations where there is a decrease inthe capital stock because there is only a very smallnumber of such observations. The percentage of obser-vations where there is a decrease in capital stock is0.74%, while that with an increase in capital stockis 4.67% and that with unchanged capital stock is94.59%. Furthermore, the distribution of �CapitalStock is complicated if we include the observationswith a decreasing capital stock. This is because mostfirms did not change their capital stock. Therefore, tofocus on the firms that increased their capital stock,
1075
D. Tsuruta
we do not use observations for which the capital stockdecreased. By dropping the observations where thecapital stock decreased, we can apply a tobit modelto estimate the policy effects of the cap on the capitalstock.
We focus on the effects of 18 types of Capital
Stock Dummy at the end of fiscal year t-1. The defi-nitions of each dummy variable are shown in Table 3.If firms have an incentive to satisfy the SME require-ments under the Corporation Tax Act and the SMEBasic Act, firms with capital stock close to 100 mil-lion yen are less likely to increase their capital stock.Therefore, in this case, the coefficient of capital stock100M is negative. In addition, compared with theeffects of the capital stock 110M dummy and capitalstock dummies for similar amounts, the magnitudesof the negative effects should be larger. Similarly,the caps of 30 and 300 million yen under the SMEBasic Act have significant effects on additional capitalstock, and are both negative. On the other hand, theseeffects are not significant if the constraint caused bythe requirement is not severe.
According to Ou and Haynes (2006), funds foracquiring additional equity capital in SMEs are deter-mined by firm age, size, sales growth, financial con-dition, internal financial sources (such as owner loansor personal and business credit cards), and loanswith traditional or nontraditional institutions. We useleverage, cash holdings, tangible fixed assets, operat-ing incomes, total factor productivity, and year andindustry dummies as control variables. We do notemploy firm size because this is highly correlatedwith Capital Stock Dummy. Leverage is a proxy forfinancial condition and loans with traditional or non-traditional institutions. Highly leveraged firms haveeasier access to loans from traditional or nontradi-tional institutions. On the other hand, very highlyleveraged firms are financially distressed firms (asargued by Opler and Titman 1994), so leverage is alsoa proxy for financial condition. Because highly lever-aged firms have incentives to increase equity capital tomitigate their financial distress, we predict that lever-age has positive effects on additional equity capital.Leverage is defined as the book value of debt dividedby the book value of assets at the beginning of fis-cal year t. Cash holdings are a proxy for liquidity. Wepredict that firms with higher liquidity are less risky,so the effects on additional registered equity are posi-tive. Cash holdings are defined as cash holdings at the
beginning of fiscal year t, normalized by total assets atthe beginning of fiscal year t.
Operating incomes are a proxy for financial con-dition. High operating incomes suggest that firmshave sufficiently high cash flows, which leads to lowcredit constraints. In addition, operating incomes area proxy for firm profitability. If firms with low cashflows face credit constraints on bank loans and useadditional equity capital, then the coefficient of oper-ating incomes will be negative. On the other hand, ifprofitable firms can use additional equity, the coeffi-cient of operating incomes will be positive. Operatingincome is defined as operating income in fiscal yeart, normalized by total assets at the beginning of fiscalyear t. Tangible fixed assets are variables representingthe amount of collateral assets. Firms with high col-lateral assets are less risky and have easier access toloans. Therefore, tangible fixed assets are a proxy offinancial condition and the availability of loans withtraditional or nontraditional institutions. The coeffi-cient of tangible fixed assets is thus negative foradditional equity capital if equity capital and bankloans are substitutes. Tangible fixed assets are definedas tangible fixed assets at the beginning of fiscal yeart, normalized by total assets at the beginning of fiscalyear t.
As a proxy for productivity, we employ simplerevenue-based total factor productivity (TFP).17 Weassume a Cobb–Douglas production function and esti-mate the following equation, dividing the sample by15 industries and 17 years:
ln(Yi,j,t
) = αj,t +βkj,t ln
(Ki,j,t
)+βlj,t ln
(Li,j,t
)+ωi,j,t (2)
where Y is 1 plus a firm’s revenue at fiscal year t, Kis 1 plus a firm’s tangible fixed assets at the beginningof fiscal year t, L is 1 plus a firm’s number of employ-ees at fiscal year t, and ωi,j,t is the error term of firmi of industry j at year t.18 We estimate (2) for all 15industries and 17 years to control for industry and yearheterogeneity, which means 255 industry/year estima-tions of Eq. 2. We define TFP as the residual for firm
17Because of a lack of panel data, it is difficult to control endo-geneity caused by an omitted variable (such as a productivityshock).18We add 1 to all variables because some firms’ tangible fixedassets and/or number of employees is zero. We remove obser-vations whose revenue is zero, but we also add 1 to revenue tomaintain consistency with the definitions of the other variables.
1076
SME policies as a barrier to growth of SMEs
Fig. 1 Pre-trend of Additional Capital Stock Dummy for Treatment1 and Controls. Note: This figure shows the mean of additionalcapital stock dummy before the policy year
i in industry j of year t (ωi,j,t ) calculated using αj,t ,βl
j,t , and βkj,t . We expect that the coefficient of TFP
can be either positive or negative. Firms with high pro-ductivity can issue more equity. If this is supported,the sign of TFP is positive. On the other hand, firmswith low productivity cannot acquire enough creditfrom banks, so they issue equity to finance investmentopportunities. If this is supported, the sign is negative.
4.2 Effects of changes in requirements in the SMEbasic act
4.2.1 Hypothesis
In Japan, SMEs can access the various SME policieslisted in Table 1. If the opportunity cost of access-ing these policies is significant, we predict that firmsensure that they remain within the SME requirementsunder the SME Basic Act so that they can utilize theseSME policies, which is a disincentive for firm growth.If a firm’s capital stock is close to the cap specified inthe SME Basic Act, they do not have an incentive to
increase their capital stock. As a result, firms do notincrease their equity capital.
To investigate this research question, we focus onthe effects of the change in 1999 in the definition ofSMEs in the SME Basic Act. As noted above, thischange resulted in the capital stock cap requirementbeing relaxed from 100 million to 300 million yenfor all industries, with the exception of the whole-sale, retail, and service industries. Following the 1999revision, firms with capital stock of around 100 mil-lion were able to increase their equity capital butretain their SME status under the requirements ofthe new Act. If the cap acts as an effective con-straint on increases in capital stock, then firms wouldincrease their capital stock after the change in theSME Basic Act. These effects would be magnifiedif a firm’s capital stock was close to the pre-1999cap. As the cap of 100 million yen does not apply tothe wholesale, retail, and service industries, we canregard the firms in these industries as a control group.We employ the difference-in-differences approach andexamine whether treatment group firms increased
1077
D. Tsuruta
Fig. 2 Pre-trend of Additional Capital Stock Dummy in Treatment2 and Controls. Note: This figure shows the mean of additionalcapital stock dummy before the policy year
their capital stock after the revision of the SME BasicAct.
Similarly, we test our hypothesis using the whole-sale, retail, and service industries as a treatment groupand other industries as a control group. As notedabove, the capital stock cap rose from 30 millionto 100 million for the wholesale industry with thechanges to the SME Basic Act. If the cap is an effec-tive constraint, firms with a capital stock of close to 30million yen in the wholesale industry would not haveincreased their capital stock before the revision of theSME Basic Act. Furthermore, this effect is significantfor the treatment group, that is, firms in the whole-sale industry. The effect of the gap between a firm’scapital stock and the cap of 30 million yen before therevision is weaker after the revision of the SME BasicAct in the treatment group. Finally, we investigate thecase of the retail and service industries. In this case,the cap was 10 million yen before the revision of theSME Basic Act, and so we focus on the gap betweena firm’s capital stock and the cap of 10 million yen.
4.2.2 Equation
To test our hypothesis, we estimate the followingequation:
�Capital Stocki,t = β1T reatmenti × Policyt
+ β2T reatmenti
× Capital Stock Gapi,t
+ β3T reatmenti
× Capital Stock Gapi,t
× Policyt
+ β4Capital Stock Gapi,t
+ Xi,tβ5 + ιt + κi + λi,t (3)
where λi,t is the error term of firm i in fiscal year t,κi is industry fixed effects of 45 industries, and ιi isyear fixed effects from FY1991 to FY2007. The def-inition of �Capital Stocki,t is the same as in Eq. 1.X includes leverage at the beginning of fiscal year t,tangible fixed assets at the beginning of fiscal year t,
1078
SME policies as a barrier to growth of SMEs
Fig. 3 Pre-trend of Additional Capital Stock Dummy for in Treatment3 and Controls. Note: This figure shows the mean of additionalcapital stock dummy before the policy year
operating incomes in fiscal year t, and firm size at thebeginning of fiscal year t. Firm size is controlled bythe natural logarithm of total assets at the beginningof fiscal year t. Larger firms are less risky and can usemore capital. Therefore, we predict that the coefficientof firm size is positive.
We use three types of treatment dummy. The firsttreatment dummy has a value of one if firms do notbelong to the wholesale, retail, or service industries,and a value of zero otherwise. This dummy focuseson the change in definition 1, shown in Table 2 (here-after, Treatment1). The second treatment dummy has avalue of one if firms belong to the wholesale industry,and a value of zero otherwise. This dummy focuseson the change in definition 2, shown in Table 2 (here-after, Treatment2). The third treatment dummy has avalue of one if firms belong to the retail and serviceindustries, and a value of zero otherwise. This dummyfocuses on the changes in definitions 3 and 4, shownin Table 2 (hereafter, Treatment3).
In addition, we add variables indicating the capi-tal stock gap (Capital Stock Gap in Eq. 3), whichis defined as the natural logarithm of the cap on cap-ital stock in the SME Basic Act before 1999 minusa firm’s capital stock at the beginning of fiscal yeart. We define three types of capital stock gap, depend-ing on the cap on capital stock. Capital Stock Gap1is defined as the natural logarithm of 100 millionyen minus a firm’s capital stock at the beginningof fiscal year t. Capital Stock Gap2 is definedas the natural logarithm of 30 million yen minus afirm’s capital stock at the beginning of fiscal year t.Capital Stock Gap3 is defined as the natural loga-rithm of 10 million yen minus a firm’s capital stock atthe beginning of fiscal year t. If firms do not increasetheir capital stock so that they can continue to satisfythe SME requirements, the treated firms with a smallercapital stock gap increase their lower capital stock.
If treated firms increase their capital stock afterthe change in the definitions of SMEs in 1999, the
Note: This table shows the summary statistics of the variables used in the econometric analysis
coefficient of T reatmenti × Policyt is positive.Because treated firms with capital stock close to thecap under the definition of SMEs have less incen-tive to increase their capital stock, we predict that thecoefficient of T reatmenti × Capital Stock Gapi,t
is negative. In addition, if the incentive to increasecapital stock is weakened by the cap set in thedefinition of SMEs, firms will increase their capi-tal stock after the revision of the Act relaxed thecap. Therefore, because the effects of T reatmenti ×Capital Stock Gapi,t are smaller after the change inthe definition of SMEs, we predict that the coefficientof T reatmenti × Capital Stock Gapi,t × Policyt
is negative. The coefficient of Capital Stock Gapi,t
controls the effects of Capital Stock Gapi,t on�Capital Stocki,t for both the treated and controlgroups. We control the effects of T reatmenti byindustry fixed effects and those of Policyt by yearfixed effects.
In Figs. 1, 2, and 3, we show the pre-trend of theadditional capital stock dummy, divided by treatmentand control groups. These figures indicate that thetrends in the additional capital stock are similar among
treatment and control groups, suggesting that firms ina different sector are a valid counterfactual.
4.3 Consequences of changes in capital stock
4.3.1 Hypothesis
As a consequence of increasing capital stock andacquiring additional equity capital, firms can increasetheir inventory and/or capital investment, whichinduces asset growth of firms. As these actionsenhance firm value, we predict that the change inthe SME capital stock requirements will lead to firmgrowth.19 If this prediction is supported, low levels ofcapital stock, caused by the cap in the definition ofSMEs, impede firm growth by decreasing acquisitionsof equity capital. However, firms can use other finan-cial sources to finance investment opportunities, such
19The consequence of additional equity capital is not only firmgrowth. Ou and Haynes (2006) argue that firms avoid default-ing on loans as additional equity capital mitigates liquidityshortages.
1080
SME policies as a barrier to growth of SMEs
as bank loans. In this case, the effects of additionalcapital stock on firm growth are not significant.
In addition, we investigate the effects of equityissues on debt finance. If capital stock constraints aresevere for small businesses, they will use other finan-cial sources, including bank loans. As a result, weassume that, prior to the changes in the SME BasicAct, firms would have used debt rather than equitybeyond the level that was optimal. If this is accu-rate, the coefficient of additional equity on total debtgrowth will be negative. On the other hand, if therelationship between equity issues and debt financeis complementary, the coefficient of additional equitywill be positive. The reason for this positive relation-ship is that firms with equity issues would becomemore creditworthy, and the supply of bank loanswould increase.
4.3.2 Equation
To investigate the consequence of additional equitycapital, we estimate the following regression:
Growthi,t = γ1�Capital Stocki,t
+ Yi,t γ2 + μt + νi + ξi,t (4)
�Capital Stock∗i,t = Zi,tω1 + Xi,tω2 + øt
+ πi + τi,t
�Capital Stocki,t
= 1 if �Capital Stock∗i,t > 0
�Capital Stocki,t
= 0 otherwise (5)
where xii,t ∼ N(0, σ 2), τi,t ∼ N(0, σ 2), andCov(ξi,t , τi,t ) = ρ �= 0. νi and πi are industry fixedeffects of 48 industries and μt and øt are year fixedeffects from FY1991 to FY2007.
We use several proxies as the dependent variable:asset growth, debt growth, tangible fixed asset growth,and inventory growth. Asset growth is defined as theannual change in total assets from the beginning to theend of fiscal year t, which is normalized by total assetsat the beginning of fiscal year t. Debt growth is definedas the annual change in total debts from the begin-ning to the end of fiscal year t, which is normalizedby total assets at the beginning of fiscal year t. Tangi-ble fixed asset growth is defined as the annual changein tangible fixed assets from the beginning to the endof fiscal year t, which is normalized by total assets
at the beginning of fiscal year t. Inventory growth isdefined as the annual change in inventories from thebeginning to the end of fiscal year t, which is normal-ized by total assets at the beginning of fiscal year t.We use the additional capital stock dummy as a proxyof �Capital Stocki,t . Yi,t includes cash holdings atthe beginning of fiscal year t, leverage at the beginningof fiscal year t, tangible fixed assets at the beginningof fiscal year t, operating income in fiscal year t, TFPin fiscal year t, and firm size at the beginning of fis-cal year t. As we argued above, �Capital Stocki,t isdetermined by many variables, which include the levelof a firm’s capital stock and the change in the defini-tion of SMEs. Therefore, because �Capital Stocki,t
is a nonrandom variable, we should control for possi-ble endogeneity. We use a treatment effects model tomitigate any endogeneity bias. We estimate the param-eter vectors using a maximum-likelihood method. InEq. 5, we employ variables (Xi,t ) in Eqs. 1 and3. In addition, we employ several types of Zi,t :Capital Stock Dummy
ji,t in Eq. 1, and T reatmenti×
Policyt , T reatmenti × Capital Stock Gapi,t ,T reatmenti × Capital Stock Gapi,t × Policyt , andCapital Stock Gapi,t in Eq. 3.
5 Estimation results
5.1 Cap on capital stock
Table 4 shows summary statistics for each of thevariables. As we omit outliers, the minimum and max-imum of each variable are not extreme values. Table 3shows the mean values of each dummy. Table 5 showsthe estimation results for Eq. 1 using the additionalcapital stock dummy and the amount of �capitalstock. As the additional capital stock dummy is abinary variable, we employ a maximum-likelihoodprobit model. Furthermore, as Table 4 shows, theamount of �capital stock has a lower limit of zero.Therefore, we employ a tobit model. In column (1),we show the estimation results using the additionalcapital stock dummy. The benchmark is observationswith 10 million yen or less of capital stock, so thateach estimated coefficient shows the magnitude ofadditional capital stock compared with firms with 10million yen or less of capital stock. The estimatedcoefficient for the capital stock dummy from 20M to60M is negative and statistically significant at the 1%
1081
D. Tsuruta
Table5
Est
imat
ion
resu
ltsfo
ref
fect
son
addi
tiona
lcap
itals
tock
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Dep
ende
ntV
aria
ble
�C
apita
lSto
ck
Prox
yof
Dep
ende
ntV
aria
ble
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Indu
stry
All
Man
ufac
turi
ng,e
tc.
Who
lesa
leR
etai
l&Se
rvic
e
Cap
italS
tock
20M
−0.0
183∗
∗∗−6
1.66
28∗∗
∗−0
.017
9∗∗∗
−68.
3610
∗∗∗
−0.0
220∗
∗∗−9
0.54
95∗∗
∗−0
.015
8∗∗∗
−40.
1270
∗∗∗
(0.0
01)
(5.0
48)
(0.0
01)
(5.8
51)
(0.0
03)
(17.
582)
(0.0
03)
(11.
128)
Cap
italS
tock
30M
−0.0
173∗
∗∗−5
7.56
52∗∗
∗−0
.018
1∗∗∗
−69.
6226
∗∗∗
−0.0
179∗
∗∗−6
8.95
39∗∗
∗−0
.009
5∗∗∗
−20.
9314
∗
(0.0
01)
(5.7
88)
(0.0
01)
(6.8
74)
(0.0
03)
(18.
205)
(0.0
04)
(12.
049)
Cap
italS
tock
40M
−0.0
102∗
∗∗−2
5.28
39∗∗
∗−0
.013
6∗∗∗
−47.
1639
∗∗∗
−0.0
056
−9.2
092
0.00
7431
.519
2∗∗
(0.0
02)
(7.7
29)
(0.0
02)
(9.2
18)
(0.0
06)
(23.
326)
(0.0
06)
(15.
991)
Cap
italS
tock
50M
−0.0
181∗
∗∗−5
5.82
39∗∗
∗−0
.019
0∗∗∗
−72.
2466
∗∗∗
−0.0
217∗
∗∗−8
3.91
79∗∗
∗−0
.008
3∗∗
−4.4
182
(0.0
01)
(6.2
91)
(0.0
01)
(7.4
60)
(0.0
04)
(22.
044)
(0.0
04)
(12.
655)
Cap
italS
tock
60M
−0.0
072∗
∗∗−9
.803
5−0
.012
1∗∗∗
−32.
7226
∗∗∗
−0.0
034
2.58
400.
0206
∗∗65
.613
7∗∗∗
(0.0
03)
(10.
027)
(0.0
02)
(12.
047)
(0.0
09)
(33.
938)
(0.0
09)
(19.
936)
Cap
italS
tock
70M
0.00
4529
.622
2∗∗∗
−0.0
038
1.19
780.
0328
∗∗10
6.06
33∗∗
∗0.
0304
∗∗∗
83.8
426∗
∗∗
(0.0
04)
(11.
383)
(0.0
04)
(13.
594)
(0.0
15)
(33.
872)
(0.0
12)
(23.
397)
Cap
italS
tock
80M
−0.0
096∗
∗∗−1
7.48
81∗
−0.0
123∗
∗∗−3
6.00
15∗∗
∗−0
.009
0−2
2.55
600.
0097
45.3
297∗
∗
(0.0
02)
(10.
242)
(0.0
02)
(11.
847)
(0.0
08)
(35.
163)
(0.0
08)
(20.
861)
Cap
italS
tock
90M
−0.0
036
9.13
59−0
.008
0∗∗∗
−11.
6267
−0.0
089
−19.
1311
0.01
94∗
69.4
875∗
∗∗
(0.0
03)
(11.
089)
(0.0
03)
(12.
683)
(0.0
09)
(41.
031)
(0.0
10)
(22.
985)
Cap
italS
tock
100M
−0.0
185∗
∗∗−3
7.72
57∗∗
∗−0
.019
8∗∗∗
−53.
1276
∗∗∗
−0.0
182∗
∗∗−5
2.01
05∗∗
∗−0
.008
2∗∗
−0.4
487
(0.0
01)
(6.7
68)
(0.0
01)
(8.6
61)
(0.0
03)
(19.
364)
(0.0
03)
(12.
134)
Cap
italS
tock
110M
0.00
6644
.434
0∗∗∗
−0.0
059
−3.5
505
−0.0
012
19.1
923
0.06
04∗∗
∗14
8.06
00∗∗
∗
(0.0
05)
(14.
045)
(0.0
05)
(19.
381)
(0.0
10)
(36.
609)
(0.0
16)
(25.
002)
Cap
italS
tock
120M
0.00
70∗
49.5
794∗
∗∗−0
.001
230
.516
3∗0.
0091
54.2
567∗
0.03
72∗∗
∗97
.721
0∗∗∗
(0.0
04)
(11.
967)
(0.0
04)
(16.
372)
(0.0
10)
(30.
468)
(0.0
11)
(19.
924)
Cap
italS
tock
130M
0.02
48∗∗
∗92
.133
6∗∗∗
0.01
27∗
64.1
917∗
∗∗0.
0165
66.4
129∗
0.06
65∗∗
∗15
8.64
98∗∗
∗
(0.0
06)
(14.
334)
(0.0
07)
(18.
762)
(0.0
14)
(36.
155)
(0.0
17)
(26.
819)
Cap
italS
tock
140M
0.01
19∗∗
55.5
432∗
∗∗0.
0056
36.2
874∗
−0.0
019
12.3
301
0.05
75∗∗
∗13
5.00
42∗∗
∗
(0.0
06)
(15.
437)
(0.0
06)
(19.
550)
(0.0
12)
(45.
200)
(0.0
17)
(27.
361)
1082
SME policies as a barrier to growth of SMEs
Table5
(con
tinue
d)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Dep
ende
ntV
aria
ble
�C
apita
lSto
ck
Prox
yof
Dep
ende
ntV
aria
ble
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Indu
stry
All
Man
ufac
turi
ng,e
tc.
Who
lesa
leR
etai
l&Se
rvic
e
Cap
italS
tock
150M
−0.0
027
27.2
770∗
∗−0
.006
3∗∗
4.62
33−0
.015
0∗∗∗
−28.
9554
0.02
59∗∗
∗10
2.42
67∗∗
∗
(0.0
03)
(10.
946)
(0.0
03)
(13.
816)
(0.0
06)
(30.
893)
(0.0
09)
(20.
665)
Cap
italS
tock
200M
0.00
66∗∗
∗57
.792
0∗∗∗
−0.0
015
27.2
011∗
∗∗0.
0066
54.4
778∗
∗∗0.
0338
∗∗∗
118.
2297
∗∗∗
(0.0
02)
(5.8
58)
(0.0
02)
(7.4
62)
(0.0
05)
(16.
064)
(0.0
05)
(11.
124)
Cap
italS
tock
300M
0.00
98∗∗
∗70
.737
0∗∗∗
0.00
0335
.774
4∗∗∗
0.01
44∗∗
∗84
.766
0∗∗∗
0.04
05∗∗
∗13
1.96
56∗∗
∗
(0.0
02)
(4.9
16)
(0.0
02)
(6.1
79)
(0.0
04)
(14.
425)
(0.0
04)
(9.3
53)
Cap
italS
tock
400M
0.01
64∗∗
∗10
1.14
03∗∗
∗0.
0058
∗∗∗
68.7
817∗
∗∗0.
0247
∗∗∗
115.
9624
∗∗∗
0.04
80∗∗
∗15
1.98
98∗∗
∗
(0.0
02)
(5.6
67)
(0.0
02)
(7.0
85)
(0.0
05)
(15.
392)
(0.0
05)
(10.
887)
Cap
italS
tock
500M
−0.0
001
63.4
194∗
∗∗−0
.005
6∗∗∗
33.3
738∗
∗∗0.
0116
∗∗∗
81.7
591∗
∗∗0.
0217
∗∗∗
107.
3872
∗∗∗
(0.0
01)
(5.3
47)
(0.0
01)
(6.6
50)
(0.0
04)
(15.
197)
(0.0
04)
(10.
063)
Cas
hH
oldi
ngs
0.01
46∗∗
∗51
.423
5∗∗∗
0.01
07∗∗
∗44
.459
9∗∗∗
0.01
39∗
41.5
097
0.01
56∗∗
∗48
.469
1∗∗∗
(0.0
02)
(7.3
73)
(0.0
03)
(9.3
21)
(0.0
07)
(27.
684)
(0.0
04)
(12.
130)
Tang
ible
Fixe
dA
sset
s−0
.009
1∗∗∗
−35.
7821
∗∗∗
−0.0
060∗
∗∗−2
2.76
33∗∗
∗−0
.014
0∗∗
−60.
7544
∗∗∗
−0.0
157∗
∗∗−5
2.32
02∗∗
∗
(0.0
02)
(5.7
67)
(0.0
02)
(7.3
17)
(0.0
06)
(22.
654)
(0.0
03)
(9.7
62)
Lev
erag
e0.
0138
∗∗∗
50.1
999∗
∗∗0.
0137
∗∗∗
52.8
053∗
∗∗0.
0158
∗∗∗
58.8
247∗
∗∗0.
0133
∗∗∗
40.2
845∗
∗∗
(0.0
01)
(2.8
39)
(0.0
01)
(3.6
76)
(0.0
03)
(9.8
64)
(0.0
02)
(4.6
71)
Ope
ratin
gIn
com
e−0
.005
3−1
9.91
58∗
−0.0
134∗
∗∗−5
1.14
52∗∗
∗0.
0313
∗∗∗
97.1
137∗
∗−0
.011
1∗−2
9.79
09
(0.0
03)
(10.
946)
(0.0
04)
(13.
494)
(0.0
12)
(43.
318)
(0.0
06)
(18.
148)
TFP
0.00
26∗∗
∗7.
5599
∗∗∗
0.00
47∗∗
∗15
.685
0∗∗∗
−0.0
017∗
∗−7
.141
3∗∗
0.00
12∗
1.86
52
(0.0
00)
(1.1
85)
(0.0
00)
(1.5
77)
(0.0
01)
(2.9
68)
(0.0
01)
(2.1
99)
1083
D. Tsuruta
Table5
(con
tinue
d)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Dep
ende
ntV
aria
ble
�C
apita
lSto
ck
Prox
yof
Dep
ende
ntV
aria
ble
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Dum
my
Am
ount
Indu
stry
All
Man
ufac
turi
ng,e
tc.
Who
lesa
leR
etai
l&Se
rvic
e
Yea
rFi
xed
Eff
ects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Indu
stry
Fixe
dE
ffec
tsY
esY
esY
esY
esY
esY
esY
esY
es
Obs
erva
tions
304,
342
304,
342
202,
260
202,
170
37,6
2037
,593
81,6
3181
,576
Log
-lik
elih
ood
−53,
764
−133
,825
−33,
207
−80,
807
−6,6
90−1
6,40
9−1
6,12
0−4
0,71
2
Not
e:T
his
tabl
epr
esen
tses
timat
esof
prob
itan
dto
bit
regr
essi
ons
with
the
addi
tiona
lca
pita
lst
ock
dum
my
and
the
amou
ntof
addi
tiona
lca
pita
lst
ock
asth
ede
pend
ent
vari
able
s.T
head
ditio
nal
capi
tal
stoc
kdu
mm
y(s
how
nas
“Dum
my”
inth
eta
ble)
isa
dum
my
vari
able
that
has
ava
lue
ofon
eif
capi
tal
stoc
kis
larg
erat
the
end
offi
scal
year
tth
anit
isat
the
begi
nnin
gof
fisc
alye
art.
�ca
pita
lst
ock
(sho
wn
as“A
mou
nt”
inth
eta
ble)
isth
edi
ffer
ence
inca
pita
lst
ock
atth
een
dof
fisc
alye
art
from
that
atth
ebe
ginn
ing
offi
scal
year
t.D
efin
ition
sof
capi
tal
stoc
kdu
mm
ies
are
show
nin
Tabl
e3.
Lev
erag
eis
defi
ned
asth
ebo
okva
lue
ofde
btdi
vide
dby
the
book
valu
eof
asse
tsat
the
begi
nnin
gof
fisc
alye
art.
Tang
ible
fixe
das
sets
are
defi
ned
asth
era
tioof
afi
rm’s
tang
ible
fixe
das
sets
toto
tal
asse
tsat
the
begi
nnin
gof
fisc
alye
art.
Cas
hho
ldin
gsar
ede
fine
das
the
ratio
ofa
firm
’sca
shho
ldin
gsto
tota
lass
ets
atth
ebe
ginn
ing
offi
scal
year
t.O
pera
ting
inco
mes
are
defi
ned
asop
erat
ing
inco
me
infi
scal
year
t,no
rmal
ized
byto
tala
sset
sat
the
begi
nnin
gof
fisc
alye
art.
Firm
size
isde
fine
das
the
natu
rall
ogar
ithm
ofto
tala
sset
sat
the
begi
nnin
gof
fisc
alye
art.
Seve
nye
ardu
mm
ies
from
FY19
92to
FY20
07an
d43
indu
stry
dum
mie
sar
ein
clud
ed.
The
refe
renc
eye
aris
FY19
91.T
hem
argi
nale
ffec
tsof
each
vari
able
are
show
nin
this
tabl
e.E
stim
ated
robu
stst
anda
rder
rors
are
show
nin
pare
nthe
ses.
The
sym
bols
∗ ,∗∗
,and
∗∗∗
deno
tesi
gnif
ican
ceat
the
10%
,5%
,and
1%le
vels
1084
SME policies as a barrier to growth of SMEs
Fig. 4 Magnitude of Estimated Marginal Effects of CapitalStock Dummies on Additional Capital Stock Dummy, by Indus-try. Note: This figure shows the estimated marginal effects ofcapital stock dummies on additional capital stock dummy using
the probit estimation. The estimated result for all firms is shownin column (1) of Table 5. Other results are not reported in anytable
level, suggesting that firms with 60 million yen orless of capital stock are less likely to increase theircapital stock. The estimated coefficients of the capi-tal stock dummy for 80M and 100M are also negativeand statistically significant at the 1% level. Focusingon magnitude, the estimated negative coefficient of thecapital stock 100M is the highest, although the manyestimated coefficients for over 100M are positive andstatistically significant. Column (2) shows the estima-tion results of the tobit regression using the amount of�capital stock as the dependent variable. The signs ofthe estimation results are almost the same as those incolumn (1). In terms of magnitude, although the esti-mated negative coefficient of capital stock 100M isnot the highest, it remains high compared with similarlevel capital stock dummies.
Columns (3)–(8) show the estimation results,divided by three groups of industries. Columns (3)
and (4) show the estimation results for manufacturing,etc., columns (5) and (6) show those for wholesale,and columns (7) and (8) show those for retail and ser-vice. The estimation results of capital stock 100M aresimilar between manufacturing, etc. and wholesale. Ifthe cap of capital stock requirement of the old SMEBasic Act for the wholesale industry has significanteffects, the coefficient of capital stock 30M is negativefor only the wholesale industry. However, the mag-nitudes of the estimated coefficients of capital stock30M for wholesale are similar to those for manufac-turing. Focusing on retail and service, we see that theestimated coefficient for capital stock 100M is neg-ative and statistically significant in column (7), butnot significant in column (8). However, the estimatedcoefficients of capital stock 110M or over and thosebetween 60M and 90M are positive and statisticallysignificant at the 1% level. Similar to the results in
1085
D. Tsuruta
Fig. 5 Magnitude of Estimated Coefficients of Capital StockDummies on Additional Capital Stock, by Industry. Note: Thisfigure shows the estimated coefficient of capital stock dummies
on additional capital stock using the tobit estimation. The esti-mated result for all firms is shown in column (2) of Table 5.Other results are not reported in any table
other columns, the magnitude of the estimated coeffi-cient of capital stock 100M is also smaller, comparedwith the similar size of capital stock (column 8).
To compare the estimated coefficients of the capitalstock dummies, Fig. 4 (using the additional capital stockdummy as the dependent variable) and Fig. 5 (usingthe amount of �capital stock as the dependent variable)show bar graphs of the estimated coefficients of the capi-tal stock dummies. We show the estimated coefficientsof the capital stock dummies using all firms in Table 5.In addition, we show the estimated coefficients of thecapital stock dummies for the three separate categoriesof industries in Figs. 4 and 5. Figures 4 and 5 show thatthe impacts of almost all capital stock dummies arenegative for the 50M or less capital stock dummies.This implies that firms with 10 million yen or lessof capital stock are very small firms. As Berger andUdell (1998) argue, they rely on insider equity becauseexternal finance is less available than it is for larger
firms, because of information asymmetry. On the otherhand, firms with between 10 and 50 million yen ofcapital stock can use external finance, so their esti-mated coefficients are negative compared with firmswith 10 million or less of capital stock. Focusing onthe over 60M capital stock dummies, we see that theestimated coefficients are positive or nearly zero, butthen, for the 100M capital stock dummies, they arenegative. In the area of the graph showing the over100M capital stock dummies, almost all the estimatedcoefficients are positive because these firms are cred-itworthy and can use external equity. Moreover, thegraph suggests that there is a large gap between firmswith 100 million yen of capital stock and those withover 100 million yen of capital stock.
In sum, these results imply that firms with 100 mil-lion yen of capital stock or less increase capital stockless than larger firms, whereas these effects are smallin firms with over 100 million yen of capital stock. As
1086
SME policies as a barrier to growth of SMEs
Table 6 Estimation results for the effects of changing the definitions of SMEs on the manufacturing industry
(1) (2) (3) (4)
Dependent Variable �Capital Stock
Proxy of Dependent Variable Dummy Amount Dummy Amount
Operating Income −0.0040 −12.9832∗∗∗ −0.0041 −13.0891∗∗∗
(0.003) (4.866) (0.003) (4.869)
TFP 0.0015∗∗∗ 2.2751∗∗∗ 0.0015∗∗∗ 2.2664∗∗∗
(0.000) (0.833) (0.000) (0.833)
Firm Size 0.0009∗∗∗ 2.0544∗∗∗ 0.0009∗∗∗ 2.0604∗∗∗
(0.000) (0.217) (0.000) (0.217)
Year Fixed Effects Yes Yes Yes Yes
Industry Fixed Effects Yes Yes Yes Yes
Observations 210,439 210,439 210,439 210,439
Log-likelihood −33470 −75229 −33460 −75224
Note: This table presents estimates of probit and tobit regressions with the additional capital stock dummy and the amount of additionalcapital stock as the dependent variables. The additional capital stock dummy (shown as “Dummy” in the table) is a dummy variablethat has a value of one if capital stock is larger at the end of fiscal year t than at the beginning of fiscal year t. Additional capital stock(shown as “Amount” in the table) is the difference in capital stock at the end of fiscal year t from that at the beginning of fiscal yeart. Treatment1 has a value of one if firms do not belong to the wholesale, retail, or service industries, and a value of zero otherwise, inorder to focus on the change in Definition 1, shown in Table 2. Policy has a value of one if the year is after FY2000 in columns (1)and (2), and after FY1999 in columns (3) and (4). Capital Stock Gap1 is defined as the natural logarithm of 100 million yen minus afirm’s capital stock at the beginning of fiscal year t. Definitions of other independent variables are shown in the note to Table 5. Themarginal effects of each variable are shown in this table. Estimated robust standard errors are shown in parentheses. The symbols ∗,∗∗, and ∗∗∗ denote significance at the 10%, 5%, and 1% levels
the cap in capital stock for SMEs is set at 100 millionyen in the Corporation Tax Act and SME Basic Act(for manufacturing and other industries before 1999),these negative effects on additional capital stock arecaused by the incentive to retain SME status. Table 5shows that the estimated coefficients of cash holdings,leverage, and TFP (apart from column 6) are positive,
whereas those of tangible fixed assets and operatingincome are negative. All estimated coefficients arestatistically significant at the 1% level. These resultssuggest that highly leveraged firms increase equitycapital to mitigate the cost of high leverage. The neg-ative coefficients for tangible fixed assets suggest thatfirms with high collateral assets can increase bank
1087
D. Tsuruta
Table 7 Estimation results for the effects of changing the definitions of SMEs on the wholesale industry
(1) (2) (3) (4)
Dependent Variable �Capital Stock
Proxy of Dependent Variable Dummy Amount Dummy Amount
Note: This table presents estimates of probit and tobit regressions with the additional capital stock dummy and the amount of additionalcapital stock as the dependent variables. The additional capital stock dummy (shown as “Dummy” in the table) is a dummy variablewith a value of one if the capital stock is larger at the end of fiscal year t than at the beginning of fiscal year t. Additional capital stock(shown as “Amount” in the table) is the difference in capital stock at the end of fiscal year t from that at the beginning of fiscal year t.Treatment2 has a value of one if firms belong to the wholesale industry and a value of zero otherwise, which enables us to focus onthe change in Definition 2, shown in Table 2. Policy has a value of one if the year is after FY2000 in columns (1) and (2), and afterFY1999 in columns (3) and (4). Capital Stock Gap2 is defined as the natural logarithm of 30 million yen minus a firm’s capital stockat the beginning of fiscal year t. Definitions of other independent variables are shown in the note to Table 5. The marginal effects ofeach variable are shown in this table. Estimated robust standard errors are shown in parentheses. The symbols ∗, ∗∗, and ∗∗∗ denotesignificance at the 10%, 5%, and 1% levels
loans, so they increase their additional equity less thando firms with fewer collateral assets.
5.2 Changes in requirements under the SME basic act
In the previous subsection, we show that firms increasecapital stock less if a firm’s capital stock is 100 mil-lion yen or less. We interpret this as indicating that
firms increase their capital stock less to remain withinthe SME requirements and retain their SME status.However, the results could also be interpreted as indi-cating that it is difficult for firms with 100 millionyen or less of capital stock to increase their capi-tal stock because they are SMEs that face seriousinformation asymmetry with investors. Therefore, weconduct another test focusing on an exogenous event,
1088
SME policies as a barrier to growth of SMEs
Table 8 Estimation results for the effects of changing the definitions of SMEs on the retail and service industries
(1) (2) (3) (4)
Dependent Variable �Capital Stock
Proxy of Dependent Variable Dummy Amount Dummy Amount
Operating Income −0.0147∗∗∗ −8.5804∗∗∗ −0.0147∗∗∗ −8.5828∗∗∗
(0.003) (2.728) (0.003) (2.729)
TFP 0.0098∗∗∗ 5.1052∗∗∗ 0.0098∗∗∗ 5.1050∗∗∗
(0.001) (1.481) (0.001) (1.481)
Firm Size 0.0136∗∗∗ 8.9122∗∗∗ 0.0136∗∗∗ 8.9156∗∗∗
(0.000) (1.344) (0.000) (1.345)
Year Fixed Effects Yes Yes Yes Yes
Industry Fixed Effects Yes Yes Yes Yes
Observations 106,634 106,634 106,634 106,634
Log-likelihood −14366 −37599 −14360 −37596
Note: This table presents estimates of probit and tobit regressions with the additional capital stock dummy and the amount of additionalcapital stock as the dependent variables. Additional capital stock dummy (shown as “Dummy” in the table) is a dummy variable witha value of one if capital stock is larger at the end of fiscal year t than at the beginning of fiscal year t. The additional capital stock(shown as “Amount” in the table) is the difference in capital stock at the end of fiscal year t from that at the beginning of fiscal yeart. Treatment3 has a value of one if firms belong to the retail and service industries and a value of zero otherwise, which enables us tofocus on the change in Definition 3, shown in Table 2. Policy has a value of one if the year is after FY2000 in columns (1) and (2), andafter FY1999 in columns (3) and (4). Capital Stock Gap2 is defined as the natural logarithm of 10 million yen minus a firm’s capitalstock at the beginning of fiscal year t. Definitions of other independent variables are shown in the note to Table 5. The marginal effectsof each variable are shown in this table. Estimated robust standard errors are shown in parentheses. The symbols ∗, ∗∗, and ∗∗∗ denotesignificance at the 10%, 5%, and 1% levels
which is the change in the definition of SMEs inthe SME Basic Act. If SMEs limit their capital stockto remain within the SME requirements, rather thanbeing because of information asymmetry, they willincrease their capital stock after the relaxation of theseconstraints following revision of the SME Basic Act.
Table 6 shows the estimation results for treatment1.We limit observations to firms with 100 million orless of capital stock at the beginning of the fiscalyear. Columns (1) and (2) show the estimation resultsfor policy, which has a value of one if the year isafter FY2000. To check robustness, we also show the
1089
D. Tsuruta
Table 9 Estimated results of treatment effects regression for additional capital stock on asset growth
(1) (2) (3) (4)
Additional Capital Stock Dummy 0.1215∗∗∗ 0.0878∗∗∗ 0.0616∗∗∗ 0.0425∗∗∗
Note: This table provides the estimates of the treatment effects model with the additional capital stock dummy and total asset growthas the dependent variables. Definitions of all variables are in the notes accompanying Tables 5, 6, 7, and 8. The symbols ∗, ∗∗, and ∗∗∗denote significance at the 10%, 5%, and 1% levels
results for the policy variable that has a value of oneif the year is after FY1999. Column (1) shows theestimation results of the probit estimation using theadditional capital stock dummy as the dependent vari-able. The estimated coefficient of treatment1 × policyis positive and statistically significant at the 1% level,implying that treated firms increase their capital stockmore after the change in the cap on capital stock thatoccurred under the SME Basic Act. The estimatedcoefficient of treatment1 × capital stock gap1 is pos-itive and statistically significant at the 1% level. Thissuggests that treated firms are less likely to increasecapital stock if their capital stock is close to the capbefore the change in the SME Basic Act. On the otherhand, the estimated coefficient of treatment1 × cap-ital stock gap1 × policy is negative and statisticallysignificant at the 1% level. Furthermore, the estimatedmarginal effect of treatment1 × capital stock gap1 is0.0080, whereas that of treatment1 × capital stockgap1 × policy is –0.0115, suggesting that the positive
effects of the distance to the cap under the SME BasicAct before 1999 are insignificant after the relaxationof the cap. In sum, these results support our hypoth-esis that SMEs had a disincentive to graduate fromSME status and increase capital stock after the changein the definitions of SMEs. The estimated coefficientof capital stock gap1 is positive and statistically sig-nificant, suggesting that all firms with a smaller gapbetween their capital stock and the cap increased cap-ital stock less. Recall that under the Corporation TaxAct, the cap for SMEs is 100 million yen of capitalstock, which is the same as the cap under the SMEBasic Act for treatment1 before 1999. The positivecoefficient of capital stock gap1 suggests that firmsdid not have an incentive to increase their capital stockif it is close to the cap under the Corporation TaxAct.
Column (2) shows the estimation results using theamount of �capital stock as the dependent variable.The estimated coefficients of treatment1 × policy and
1090
SME policies as a barrier to growth of SMEs
Table 10 Estimated results of treatment effects regression for effects of additional capital stock on debt growth
(1) (2) (3) (4)
Additional Capital Stock Dummy 0.0139∗∗∗ −0.0188∗∗∗ −0.0479∗∗∗ −0.0625∗∗∗
Note: This table provides the estimates of the treatment effects model with the additional capital stock dummy and total debt growthas the dependent variables. Definitions of all variables are in the notes accompanying Tables 5, 6, 7, and 8. The symbols ∗, ∗∗, and ∗∗∗denote significance at the 10%, 5%, and 1% levels
treatment1 × capital stock gap1 are positive and sta-tistically significant at the 1% level. The estimatedcoefficient of treatment1 × capital stock gap1 × pol-icy is negative and statistically significant at the 1%level. These results are similar to those in column (1),implying that firms increased capital stock more afterthe change in the definitions of SMEs. This effect islarger if a firm’s capital stock is close to the cap oncapital stock. Column (2) also shows the estimatedcoefficient capital stock gap1, which is not statisticallysignificant. This suggests that the result for capitalstock gap1 is not robust. Columns (3) and (4) showthe estimation results using the policy variable thathas a value of one if the year is after FY1999. Theestimation results of treatment1 × policy, treatment1× capital stock gap1, and treatment1 × capital stockgap1 × policy are similar to those in columns (1) and(2). The results of the estimated coefficients for con-trol variables are similar to those in Table 5, apart fromthose for tangible fixed assets.
Table 7 shows the estimation results using treat-ment2. Definitions of the dependent and control vari-ables in each column are the same as those in Table 6.We limit observations to firms with capital stock of 30million yen or less at the beginning of the fiscal year,which satisfy the requirements for capital stock underthe SME Basic Act before 1999. Although the levelsof the cap for treatment1 and treatment2 are differentunder the SME Basic Act, the estimation results aresimilar to those in Table 6. The estimated coefficientsof treatment2 × policy and treatment2 × capital stockgap2 are positive and those of treatment2 × capitalstock gap2 × policy are negative, and all are statisti-cally significant at the 1% level. These results suggestthat treated firms increased their capital stock lessbefore the relaxation of the cap. In particular, theseeffects are larger for treated firms with capital stockthat is close to the cap. After relaxing the cap in 1999,this effect was weakened. This implies that the cap setin the definitions of SMEs under the SME Basic Act
1091
D. Tsuruta
Table 11 Estimated results of treatment effects regression for effects of additional capital stock on tangible fixed asset growth
(1) (2) (3) (4)
Additional Capital Stock Dummy 0.0204∗∗∗ 0.0137∗∗∗ 0.0093∗∗∗ 0.0121∗∗∗
Note: This table provides the estimates of the treatment effects model with the additional capital stock dummy and tangible fixedasset growth as the dependent variables. Definitions of all variables are in the notes accompanying Tables 5, 6, 7, and 8. The symbols∗, ∗∗, and ∗∗∗ denote significance at the 10%, 5%, and 1% levels
is a significant constraint on firms’ additional equitycapital.
Table 8 shows the estimation results using treat-ment3. We limit observations to firms that satisfy therequirement for capital stock under the SME BasicAct before 1999 (firms with 10 million yen or less ofcapital stock at the beginning of the fiscal year). Thedefinitions of the dependent and control variables arethe same as those in Table 6. Similarly to the resultsin Tables 6 and 7, the estimated coefficients of treat-ment3 × policy and treatment3 × capital stock gap3are positive and those of treatment3 × capital stockgap3 × policy are negative. These coefficients arestatistically significant at the 1% level. These resultssuggest that the estimated results for the treatment andcapital stock gap are robust.
The subsample of control firms includes somefirms that have newly become SMEs after the pol-icy change. Even if we use only a subsample of
firms excluding these firms, the estimation results aresimilar to those in Tables 6, 7, and 8.
5.3 Effects of additional equity on growth
5.3.1 Asset growth
Table 9 shows the estimation results for Eq. 4 usingasset growth as the dependent variable. Equation 5 incolumn (1) is estimated using variables in column (1)of Table 5; that in column (2) is estimated using vari-ables in column (1) in Table 6; that in column (3) isestimated using variables in column (1) in Table 7;and that in column (4) is estimated using variables incolumn (1) in Table 8. The estimated coefficients inEq. 5 are similar to the estimated results of each probitmodel.
The estimated results in Table 9 show that theestimated coefficients of the additional capital stock
1092
SME policies as a barrier to growth of SMEs
Table 12 Estimated results of treatment effects regression for effects of additional capital stock on inventory growth
(1) (2) (3) (4)
Growth Growth Growth Growth
Additional Capital Stock Dummy 0.0060∗∗∗ 0.0030∗∗∗ −0.0021 −0.0015
Note: This table provides the estimates of the treatment effects model with the additional capital stock dummy and inventory growthas the dependent variables. Definitions of all variables are in the notes accompanying Tables 5, 6, 7, and 8. The symbols ∗, ∗∗, and ∗∗∗denote significance at the 10%, 5%, and 1% levels
dummy are positive and statistically significant atthe 1% level. These results are robust because weobtain similar results if we employ different variablesin Eq. 5. The estimated ρ is positive and statisti-cally significant at the 1% level; the assumption ofcorr(νi, πi) �= 0 is therefore supported.
The estimated coefficients of cash holdings are pos-itive and statistically significant at the 1% level. Theseresults suggest that firms with high liquidity increasedfirm size more. The estimated coefficients of lever-age are negative and statistically significant at the1% level. Because highly leveraged firms are gener-ally financially distressed, the performance of firms islower if leverage is high. The coefficients of tangibleassets are negative and statistically significant at the10% level in column (1). Tangible assets are a proxyfor collateral assets. We predict that the effects of tan-gible assets are positive on firm performance becausecollateral assets mitigate credit constraints for smallbusinesses. However, this prediction is not supported
by the results of Table 9. The estimated coefficientsof operating income are positive and statistically sig-nificant at the 1% level. Because profitable firms havemore good investment opportunities, they increasetheir assets more. The estimated coefficients of TFPare all positive and statistically significant at the 1%level, suggesting that firms with high productivitygrow faster. The estimated coefficients of firm sizeare negative and statistically significant at the 1%level.
5.3.2 Debt growth
Table 10 shows the estimation results for Eq. 4 usingdebt growth as the dependent variable. The variablesin the first-stage equation in each column are the sameas those in Table 9. The estimated coefficients of theadditional capital stock dummy are positive in column(1) and negative in columns (2)–(4). The coefficientsare all statistically significant at the 1% level. The
1093
D. Tsuruta
magnitude of the coefficients increases as the columnnumbers become smaller, whereas firm size (proxiedby the level of capital stock) becomes smaller as thecolumn numbers become larger. This indicates that theeffects of equity issues on debt finance are positivefor larger firms but negative for smaller firms. Thisimplies that the relaxation of the capital stock con-straint results in lower leverage for smaller firms. Forlarger firms, equity issues and debt finance are com-plements. In addition, our estimation results show thatfirms with high volatility and leverage decrease debtsmore after the relaxation of the capital stock require-ments, allowing them to adjust their capital structure(not shown in the table).
5.3.3 Tangible fixed asset and inventory growth
Table 11 shows the estimation results for Eq. 4 usingtangible fixed asset growth as the dependent variable.In columns (1)–(4), the estimated coefficients of theadditional capital stock dummy are positive and sta-tistically significant at the 1% level. These suggestthat firms achieve capital investment by increased cap-ital stock. Therefore, if firms near to the cap on thecapital stock of the SME Basic Act decrease equityissue, firms also decrease their capital investment.Table 12 shows the estimation results using inven-tory growth as the dependent variable. In columns (1)and (2), the estimated coefficients of additional capitalstock are positive and statistically significant at the 1%level. However, focusing on columns (3) and (4), thesecoefficients are not statistically significant. Theseresults suggest that larger firms increase inventoryinvestment by equity issue, but these results are notrobust.
5.4 Magnitude of policy effects on capital stockand firm growth
5.4.1 Additional capital stock and firm growth
As we show in the previous subsection, firms areunlikely to increase capital stock if the level of theircapital stock is near the cap of the requirement ofan SME. As capital stock increases enhance the assetgrowth of firms, we conclude that the capital stockrequirements of an SME are a significant constraintfor firm growth. However, we do not show the mag-nitude of the distortion caused by the policy. To show
the policy effects on �Capital Stock, we estimate thefollowing equation.
Policy�CapitalStocki,t = �Capital Stock
Policy
i,t
− �Capital StockNon−policy
i,t (6)
�Capital StockPolicyi,t is estimated by the regres-
sion results of probit models for Eq. 3 (shown inTables 6, 7, and 8), which is the actual trend ofincrease in the capital stock of SMEs. To estimatethe counterfactual of the policy, we should infer theincrease in capital stock if the requirements of SMEsare not changed. �Capital Stock
Non−policyi,t is coun-
terfactual for the increase in the capital stock. This isestimated by the regression results of probit modelsfor Eq. 3 where Policyt = 0. Policy
�Capital Stocki,t
is the gap between additional capital stock with andwithout the change in the requirement.
In addition, we estimate the magnitude of the capon the requirement of an SME in the SME Basic Acton the asset growth of firms. To investigate this issue,we estimate the asset growth with and without thepolicy change as follows.
GrowthPolicy
i,t = γ1 �Capital StockPolicy
i,t
+ Yi,t γ2 + μt + νi (7)
GrowthNon−policy
i,t = γ1 �Capital StockNon−policy
i,t
+ Yi,t γ2 + μt + νi (8)
From these equations, we estimate PolicyGrowthi,t
for each observation as follows.
PolicyGrowthi,t = Growth
Policy
i,t − GrowthNon−policy
i,t (9)
= γ1
(�Capital Stock
Policy
i,t
− �Capital StockNon−policy
i,t
)(10)
γi , i = 1, 2, μt , and νi are obtained from the estima-tion results in Table 9. We show the average values ofPolicy
�CapitalStocki,t and PolicyGrowth
i,t for each yearas the magnitude of the policy effects.
Table 13 shows the average policy effects on addi-tional capital stock, defined by Eq. 6. Columns (1)–(3)show the policy effects for treatment1. We divide thesample into three groups, firms with a capital stock of80M yen or less, over 80M to 90M yen, and over 90Mto 100M yen. In the group of 80M yen or less, wecannot find any large and positive policy effects. On
1094
SME policies as a barrier to growth of SMEs
Table13
Est
imat
edpo
licy
effe
cts
onad
ditio
nalc
apita
lsto
ck
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Tre
atm
ent1
Tre
atm
ent2
Tre
atm
ent3
Cap
italS
tock
≤80M
80M
−−90
M90
M−−
100M
≤10M
10M
−−20
M20
M−−
30M
<10
M=1
0M
1999
E(
�
Capit
alSto
ckP
oli
cy
i,t
)2.
048
1.04
70.
448
1.79
01.
442
0.47
31.
239
0.21
9
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
048
1.04
70.
448
1.79
01.
442
0.47
31.
239
0.21
9
E( P
oli
cy�
Capit
alS
tock
i,t
)0.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
0
2000
E(
�
Capit
alSto
ckP
oli
cy
i,t
)1.
884
1.98
42.
065
1.30
01.
769
1.57
60.
828
1.15
8
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
216
1.12
00.
475
1.71
81.
372
0.37
71.
349
0.31
5
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.331
0.86
41.
591
−0.4
170.
397
1.19
9−0
.521
0.84
3
2001
E(
�
Capit
alSto
ckP
oli
cy
i,t
)1.
841
1.91
42.
042
1.15
71.
568
1.40
40.
871
1.17
2
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
160
1.09
00.
488
1.52
01.
191
0.34
91.
425
0.26
7
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.319
0.82
31.
554
−0.3
630.
377
1.05
5−0
.554
0.90
5
2002
E(
�
Capit
alSto
ckP
oli
cy
i,t
)1.
490
1.58
61.
638
1.01
41.
389
1.15
20.
801
1.25
7
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)1.
753
0.89
60.
363
1.34
91.
012
0.24
71.
366
0.29
8
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.263
0.69
01.
275
−0.3
350.
377
0.90
6−0
.565
0.95
9
2003
E(
�
Capit
alSto
ckP
oli
cy
i,t
)1.
804
1.98
71.
963
1.30
91.
687
1.50
71.
099
1.49
2
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
115
1.13
60.
472
1.72
01.
264
0.33
51.
802
0.38
3
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.311
0.85
11.
491
−0.4
120.
423
1.17
2−0
.703
1.10
9
1095
D. Tsuruta
Table13
(con
tinue
d)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Tre
atm
ent1
Tre
atm
ent2
Tre
atm
ent3
Cap
italS
tock
≤80M
80M
−90M
90M
−100
M≤1
0M10
M−2
0M20
M−3
0M<
10M
=10M
2004
E(
�
Capit
alSto
ckP
oli
cy
i,t
)2.
062
2.40
02.
654
1.30
31.
743
1.55
11.
395
1.86
4
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
393
1.42
30.
677
1.72
41.
325
0.36
52.
187
0.47
6
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.331
0.97
71.
977
−0.4
210.
418
1.18
7−0
.792
1.38
8
2005
E(
�
Capit
alSto
ckP
oli
cy
i,t
)2.
145
2.37
22.
629
1.51
72.
133
1.85
02.
355
2.51
1
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
490
1.38
90.
664
1.99
01.
621
0.41
93.
440
0.65
2
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.345
0.98
41.
964
−0.4
730.
512
1.43
1−1
.085
1.85
9
2006
E(
�
Capit
alSto
ckP
oli
cy
i,t
)2.
126
2.17
92.
587
1.32
01.
729
1.53
91.
869
2.23
0
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
465
1.26
50.
653
1.73
21.
315
0.37
02.
768
0.59
4
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.338
0.91
41.
934
−0.4
120.
414
1.16
9−0
.898
1.63
6
2007
E(
�
Capit
alSto
ckP
oli
cy
i,t
)1.
998
2.09
52.
386
1.39
31.
832
1.69
11.
915
1.92
9
E(
�
Capit
alSto
ckN
on−p
oli
cy
i,t
)2.
322
1.21
30.
604
1.83
61.
410
0.45
72.
794
0.55
2
E( P
oli
cy�
Capit
alS
tock
i,t
)−0
.324
0.88
21.
781
−0.4
430.
421
1.23
5−0
.879
1.37
7
Not
e:T
his
tabl
epr
ovid
esth
ees
timat
esof
polic
yef
fect
sfo
rth
eav
erag
ead
ditio
nalc
apita
lsto
ckdu
mm
y,de
fine
dby
equa
tion
(6).
The
num
bers
are
expr
esse
din
perc
enta
gete
rms
1096
SME policies as a barrier to growth of SMEs
Table14
Est
imat
edpo
licy
effe
cts
ongr
owth
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Tre
atm
ent1
Tre
atm
ent2
Tre
atm
ent3
Cap
italS
tock
≤80M
80M
-90M
90M
-100
M≤1
0M10
M-2
0M20
M-3
0M<
10M
=10M
1999
E(
G
row
thP
oli
cy
i,t
)0.
608
−0.6
92−0
.629
0.10
00.
674
−0.2
040.
110
0.27
0
E(
G
row
thN
on−p
oli
cy
i,t
)0.
608
−0.6
92−0
.629
0.10
00.
674
−0.2
040.
110
0.27
0
E( P
oli
cyG
row
thi,t
)0.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
0
2000
E(
G
row
thP
oli
cy
i,t
)1.
575
0.40
00.
419
0.82
81.
454
0.59
00.
539
0.62
1
E(
G
row
thN
on−p
oli
cy
i,t
)1.
604
0.32
40.
279
0.85
41.
430
0.51
70.
561
0.58
6
E( P
oli
cyG
row
thi,t
)−0
.029
0.07
60.
140
−0.0
260.
024
0.07
4−0
.022
0.03
6
2001
E(
G
row
thP
oli
cy
i,t
)−0
.210
−1.5
00−1
.346
−0.4
33−0
.335
−0.5
93−0
.582
−0.3
30
E(
G
row
thN
on−p
oli
cy
i,t
)−0
.182
−1.5
72−1
.483
−0.4
11−0
.358
−0.6
58−0
.558
−0.3
69
E( P
oli
cyG
row
thi,t
)−0
.028
0.07
20.
136
−0.0
220.
023
0.06
5−0
.024
0.03
8
2002
E(
G
row
thP
oli
cy
i,t
)−0
.650
−1.9
07−1
.484
−1.3
61−0
.748
−1.4
73−1
.389
−1.5
31
E(
G
row
thN
on−p
oli
cy
i,t
)−0
.626
−1.9
67−1
.596
−1.3
40−0
.771
−1.5
28−1
.365
−1.5
71
E( P
oli
cyG
row
thi,t
)−0
.023
0.06
10.
112
−0.0
210.
023
0.05
6−0
.024
0.04
1
2003
E(
G
row
thP
oli
cy
i,t
)1.
061
−0.0
41−0
.156
0.23
40.
042
−0.1
050.
322
1.16
9
E(
G
row
thN
on−p
oli
cy
i,t
)1.
088
−0.1
15−0
.286
0.25
90.
016
−0.1
770.
352
1.12
2
E( P
oli
cyG
row
thi,t
)−0
.027
0.07
50.
131
−0.0
250.
026
0.07
2−0
.030
0.04
7
1097
D. Tsuruta
Table14
(con
tinue
d)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Tre
atm
ent1
Tre
atm
ent2
Tre
atm
ent3
Cap
italS
tock
≤80M
80M
–90M
90M
–100
M≤1
0M10
M–2
0M20
M–3
0M<
10M
=10M
2004
E(
G
row
thP
oli
cy
i,t
)2.
679
1.73
22.
015
1.10
31.
811
1.15
72.
400
1.92
2
E(
G
row
thN
on−p
oli
cy
i,t
)2.
709
1.64
71.
842
1.12
91.
785
1.08
42.
433
1.86
3
E( P
oli
cyG
row
thi,t
)−0
.029
0.08
60.
174
−0.0
260.
026
0.07
3−0
.034
0.05
9
2005
E(
G
row
thP
oli
cy
i,t
)2.
967
2.19
52.
646
1.24
82.
038
2.06
42.
188
2.28
7
E(
G
row
thN
on−p
oli
cy
i,t
)2.
998
2.10
92.
474
1.27
72.
007
1.97
62.
234
2.20
8
E( P
oli
cyG
row
thi,t
)−0
.030
0.08
60.
172
−0.0
290.
032
0.08
8−0
.046
0.07
9
2006
E(
G
row
thP
oli
cy
i,t
)3.
410
2.23
02.
663
1.95
82.
781
1.67
33.
393
3.22
1
E(
G
row
thN
on−p
oli
cy
i,t
)3.
440
2.14
92.
493
1.98
32.
755
1.60
13.
431
3.15
2
E( P
oli
cyG
row
thi,t
)−0
.030
0.08
00.
170
−0.0
250.
026
0.07
2−0
.038
0.06
9
2007
E(
G
row
thP
oli
cy
i,t
)2.
456
1.03
71.
583
1.73
02.
701
1.50
03.
585
2.52
7
E(
G
row
thN
on−p
oli
cy
i,t
)2.
484
0.96
01.
427
1.75
72.
675
1.42
43.
623
2.46
8
E( P
oli
cyG
row
thi,t
)−0
.028
0.07
70.
156
−0.0
270.
026
0.07
6−0
.037
0.05
8
Not
e:T
his
tabl
epr
ovid
esth
ees
timat
esof
polic
yef
fect
sfo
rth
eav
erag
eas
setg
row
th,d
efin
edby
equa
tion
(10)
.The
num
bers
are
expr
esse
din
perc
enta
gete
rms
1098
SME policies as a barrier to growth of SMEs
Fig. 6 Distribution of Capital Stock of Treatment1, by Kernel Density Estimation. Note: This figure shows the distribution ofestimated capital stock for treatment1 after the policy change
the other hand, focusing on firms with capital stockof over 90M to 100M yen, we see that the averageeffects are between 1.275 and 1.977%. These firmsfaced the cap of capital stock requirement of the oldSME Basic Act, so they were more likely to increasetheir capital stock after the policy change. In otherwords, the requirements of the SME Basic Act reducethe probability of increases in capital stock by 1.696%on average, while the actual average rate of capitalstock growth is 1.885%. Columns (4)–(6) show theestimation results for treatment2. Focusing on firmswith a capital stock of 20–30M yen in treatment2, wesee that the magnitude of the policy effects is smaller,between 0.906% and 1.431%, and 1.169% on aver-age. On the other hand, the policy effects for firmswith a capital stock of 10M or less are negative, butthe magnitude is small. Columns (7) and (8) show theestimation results for treatment3. The policy effects ofadditional capital stock for firms with 10M yen of cap-ital stock are between 0.843 and 1.859%. However, ifwe focus on the observations with capital stock lessthan 10M yen, these effects are negative. In sum, the
average policy effects on additional capital stock areeconomically significant for firms close to the cap ofthe requirement of SMEs.
Table 14 shows the policy effects on firm growth,defined by Eq. 10. The format of this table is the sameas that of Table 13. Columns (1)–(3) show the pol-icy effects for treatment1. Focusing on the group forcapital stock of over 90M to 100M, the policy effectson asset growth are between 0.112 and 0.174%, and0.149% on average. These effects seem to be smallbecause the numbers are around 0.15%. However, theactual average asset growth rate is 1.913% from 2000to 2007, so the impact on the growth is not small.20 Onthe other hand, the policy effects for firms with capi-tal stock of 80M yen or less are very small, between–0.023% and –0.030%. Focusing on the estimationresults for treatment2 in columns (4)–(6), we see thatthe policy effects for firms with capital stock of over20M to 30M yen are around 0.1%, which are smaller
20The low asset growth rate was caused by the financial crisisof the late 1990s.
1099
D. Tsuruta
Fig. 7 Distribution of Capital Stock of Treatment2, by Kernel Density Estimation. Note: This figure shows the distribution ofestimated capital stock for treatment2 after the policy change
than those for treatment1. In the other groups, the esti-mated effects are negative or very small. Similarly,the effects for treatment3 (shown in columns 7 and8) are between 0.039 and 0.079% even if the sampleis limited to firms with a capital stock equaling 10Myen. In sum, the policy effects on growth are positivefor firms near to the cap of the requirement of SMEs.The effects are larger for treatment1, which includeslarger-sized SMEs in terms of size of capital stock.
Tables 13 and 14 show that the effects on growthare small although those on the capital stock areeconomically significant. The reason is that firmscan substitute equity capital for debt for financinginvestment opportunities. If the substitution effects aresmall, the effects on firm growth are larger than theeffects estimated by our model.
5.4.2 Distribution of capital stock
Figures 6, 7, and 8 show the distributions of capi-tal stock after the policy change, estimated by kernel
density estimation. The value of capital stock is cal-culated as actual capital stock in the beginning of thefiscal year plus the estimated �capital stock using thetobit model shown in column (2) of Tables 6, 7, and8. Figure 6 shows the distribution for treatment1. Theobservations in this figure are limited to firms with acapital stock of 50M to 100M yen at the beginningof the fiscal year. We show the distributions with andwithout policy changes (counterfactual). The distribu-tion peaks at 100M yen of capital stock, which is thecap in the old SME Basic Act and Corporate Tax Act.Comparing the distribution with and without policychanges, the distribution is right-shifted if the policy ischanged. However, the peak of 100M yen is still large.The reason is that the cap of the Corporate Tax Act isstill 100M yen of capital stock. Figures 7 and 8 showthe distributions of capital stock with and without pol-icy changes. To focus on the cap on the requirement ofSMEs, the observations are limited to firms with capi-tal stock of over 10M to 30M yen in Fig. 7 and those of10M yen or less in Fig. 8. The distributions are right-
1100
SME policies as a barrier to growth of SMEs
Fig. 8 Distribution of Capital Stock of Treatment3, by Kernel Density Estimation. Note: This figure shows the distribution ofestimated capital stock for treatment3 after the policy change
shifted near the cap of the requirement, suggesting thatthe requirement of SMEs is a cause of the distorteddistribution. However, the magnitudes are not signif-icant because the policy effects on capital stock aresmaller than those for treatment1.
These findings are consistent with the notion ofBerger and Udell (1998) that shows sources of financefor SMEs divided by firm size. Small-sized firms usefinancial institution loans, so the equity issue is not themain source of finance.21 Therefore, the constraintson capital stock for treatment2 and treatment3 havesmaller impacts, as shown in Figs. 7 and 8. On theother hand, Berger and Udell (1998) argue that publicequity is a main source of finance for medium firms,so the constraint on capital stock for treatment1 hasa significant impact on the distribution, as shown inFig. 6.
21Berger and Udell (1998) argue that venture capital is also asource of finance for small firms, but in Japan, venture capitalis not common for SMEs.
5.5 Additional test
5.5.1 Policy dummy
The coefficients of the interaction variable for thetreatment dummies for FY1999 and FY2000 and thepolicy variables are positive and statistically signifi-cant in all tables. However, we do not test whetherthese positive effects exist for policy dummy variablesfor other fiscal years. To test the other years aroundthe changing of the SME Basic Act, we reestimate (3)including policy dummies for 5 fiscal years before andafter the pseudo policy year. If the years around thechange in the SME Basic Act have positive effects, theestimation results support our hypothesis for policyyear dummies close to FY1999.
Table 15 shows the estimation results using policydummies from FY1996 to FY2002. Panel A showsthe estimation results using treatment1, Panel B showsthose using treatment2, and Panel C shows those usingtreatment3. In all panels, the magnitude of coefficients
1101
D. Tsuruta
Table15
Est
imat
ion
resu
ltsfo
rth
eef
fect
sof
polic
ydu
mm
ies
for
FY19
96to
FY20
02
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Dep
ende
ntV
aria
ble
�C
apita
lSto
ck(D
umm
y)
Peri
odof
Polic
yFY
1996
FY19
97FY
1998
FY19
99FY
2000
FY20
01FY
2002
Pane
lA
Tre
atm
ent1
×Po
licy
−0.0
048
0.11
41∗∗
∗0.
1558
∗∗∗
0.14
48∗∗
∗0.
1254
∗∗∗
0.05
94∗∗
∗−0
.001
9
(0.0
08)
(0.0
16)
(0.0
20)
(0.0
20)
(0.0
18)
(0.0
11)
(0.0
04)
Tre
atm
ent1
×C
apita
lSto
ckG
ap1
0.00
40∗∗
∗0.
0126
∗∗∗
0.01
43∗∗
∗0.
0141
∗∗∗
0.01
32∗∗
∗0.
0092
∗∗∗
0.00
30∗∗
∗
(0.0
01)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
01)
(0.0
01)
Tre
atm
ent1
×C
apita
lSto
ckG
ap1
0.00
08−0
.018
4∗∗∗
−0.0
232∗
∗∗−0
.021
8∗∗∗
−0.0
194∗
∗∗−0
.011
2∗∗∗
−0.0
004
×Po
licy
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
01)
(0.0
01)
Cap
italS
tock
Gap
10.
0062
∗∗∗
0.00
61∗∗
∗0.
0059
∗∗∗
0.00
53∗∗
∗0.
0041
∗∗∗
0.00
19∗∗
∗0.
0004
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
00)
Yea
rFi
xed
Eff
ects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Indu
stry
Fixe
dE
ffec
tsY
esY
esY
esY
esY
esY
esY
es
Obs
erva
tions
140,
217
137,
151
134,
771
133,
229
132,
260
131,
311
130,
205
Log
−lik
elih
ood
−26,
561
−24,
668
−23,
212
−22,
045
−21,
059
−18,
035
−13,
629
Pane
lB
Tre
atm
ent2
×Po
licy
0.03
630.
5014
∗∗∗
0.61
78∗∗
∗0.
3578
∗∗∗
0.26
71∗∗
∗0.
0730
∗−0
.007
2
(0.0
38)
(0.1
38)
(0.1
34)
(0.1
14)
(0.1
00)
(0.0
42)
(0.0
06)
Tre
atm
ent2
×C
apita
lSto
ckG
ap2
0.00
400.
0279
∗∗∗
0.03
02∗∗
∗0.
0183
∗∗∗
0.01
57∗∗
∗0.
0072
∗−0
.000
7
(0.0
06)
(0.0
09)
(0.0
09)
(0.0
07)
(0.0
06)
(0.0
04)
(0.0
02)
Tre
atm
ent2
×C
apita
lSto
ckG
ap2
−0.0
093
−0.0
515∗
∗∗−0
.058
3∗∗∗
−0.0
399∗
∗∗−0
.033
0∗∗∗
−0.0
140∗
∗∗0.
0020
×Po
licy
(0.0
08)
(0.0
09)
(0.0
09)
(0.0
07)
(0.0
07)
(0.0
05)
(0.0
03)
Cap
italS
tock
Gap
20.
0285
∗∗∗
0.02
58∗∗
∗0.
0220
∗∗∗
0.01
84∗∗
∗0.
0155
∗∗∗
0.00
83∗∗
∗0.
0023
∗∗∗
(0.0
02)
(0.0
02)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
Yea
rFi
xed
Eff
ects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Indu
stry
Fixe
dE
ffec
tsY
esY
esY
esY
esY
esY
esY
es
Obs
erva
tions
103,
315
100,
857
98,8
6697
,220
95,8
9094
,414
92,7
63
Log
−lik
elih
ood
−20,
383
−18,
973
−17,
796
−16,
768
−15,
804
−12,
834
−8,8
00
1102
SME policies as a barrier to growth of SMEs
Table15
(con
tinue
d)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Dep
ende
ntV
aria
ble
�C
apita
lSto
ck(D
umm
y)
Peri
odof
Polic
yFY
1996
FY19
97FY
1998
FY19
99FY
2000
FY20
01FY
2002
Pane
lC
Tre
atm
ent3
×Po
licy
−0.0
105∗
∗∗0.
0444
∗∗∗
0.06
66∗∗
∗0.
0678
∗∗∗
0.06
83∗∗
∗0.
0410
∗∗∗
0.00
62∗
(0.0
04)
(0.0
09)
(0.0
11)
(0.0
11)
(0.0
11)
(0.0
08)
(0.0
03)
Tre
atm
ent3
×C
apita
lSto
ckG
ap3
−0.0
015
0.00
65∗∗
∗0.
0057
∗∗∗
0.00
52∗∗
∗0.
0059
∗∗∗
0.00
46∗∗
∗0.
0004
(0.0
02)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
(0.0
01)
Tre
atm
ent3
×C
apita
lSto
ckG
ap3
0.00
65∗∗
∗−0
.019
0∗∗∗
−0.0
225∗
∗∗−0
.022
3∗∗∗
−0.0
211∗
∗∗−0
.013
9∗∗∗
−0.0
020∗
×Po
licy
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
01)
(0.0
01)
Cap
italS
tock
Gap
30.
0833
∗∗∗
0.07
50∗∗
∗0.
0665
∗∗∗
0.05
90∗∗
∗0.
0528
∗∗∗
0.03
91∗∗
∗0.
0237
∗∗∗
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
02)
(0.0
01)
(0.0
01)
(0.0
01)
Yea
rFi
xed
Eff
ects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Indu
stry
Fixe
dE
ffec
tsY
esY
esY
esY
esY
esY
esY
es
Obs
erva
tions
72,3
4171
,000
70,0
5768
,995
68,5
6667
,625
66,1
19
Log
−lik
elih
ood
−11,
799
−10,
804
−10,
083
−9,6
07−9
,237
−7,3
85−4
,878
Not
e:T
his
tabl
epr
esen
tses
timat
esof
prob
itan
dto
bit
regr
essi
ons
with
the
addi
tiona
lca
pita
lst
ock
dum
my
asth
ede
pend
ent
vari
able
.The
addi
tiona
lca
pita
lst
ock
dum
my
(sho
wn
as“D
umm
y”in
the
tabl
e)is
adu
mm
yva
riab
leth
atha
sa
valu
eof
one
ifre
gist
ered
capi
tal
isla
rger
atth
een
dof
fisc
alye
art
than
atth
ebe
ginn
ing
offi
scal
year
t.D
efin
ition
sof
inde
pend
ent
vari
able
sar
esh
own
inth
eno
teto
Tabl
es5,
6,7,
and
8.T
hem
argi
nal
effe
cts
ofea
chva
riab
lear
esh
own
inth
ista
ble.
Est
imat
edro
bust
stan
dard
erro
rsar
esh
own
inpa
rent
hese
s.T
hesy
mbo
ls∗ ,
∗∗,a
nd∗∗
∗ den
ote
sign
ific
ance
atth
e10
%,5
%,a
nd1%
leve
ls
1103
D. Tsuruta
Table16
Est
imat
ion
resu
ltsfo
rth
eef
fect
sus
ing
nont
reat
edfi
rms
only
Man
ufac
turi
ngW
hole
sale
Ret
aila
ndSe
rvic
e
Tre
atm
ent1
Tre
atm
ent1
Tre
atm
ent2
Tre
atm
ent2
Tre
atm
ent3
Tre
atm
ent3
×C
apita
lSto
ckG
ap×
Cap
italS
tock
Gap
×C
apita
lSto
ckG
ap×
Cap
italS
tock
Gap
×C
apita
lSto
ckG
ap×
Cap
italS
tock
Gap
×Po
licy
×Po
licy
×Po
licy
Pseu
doC
apdF
/dx
S.E
.dF
/dx
S.E
.dF
/dx
S.E
.dF
/dx
S.E
.dF
/dx
S.E
.dF
/dx
S.E
.
350M
−0.0
189∗
∗∗(0
.003
)−0
.012
6∗∗
(0.0
06)
−0.0
244∗
∗∗(0
.003
)0.
0168
∗(0
.009
)−0
.021
8∗∗∗
(0.0
03)
−0.0
029
(0.0
06)
360M
−0.0
135∗
∗∗(0
.003
)−0
.011
9∗∗
(0.0
05)
−0.0
197∗
∗∗(0
.002
)0.
0250
∗∗∗
(0.0
09)
−0.0
180∗
∗∗(0
.003
)0.
0005
(0.0
06)
370M
−0.0
114∗
∗∗(0
.003
)−0
.011
1∗∗
(0.0
04)
−0.0
169∗
∗∗(0
.002
)0.
0154
∗(0
.008
)−0
.016
8∗∗∗
(0.0
02)
0.00
62(0
.006
)
380M
−0.0
163∗
∗∗(0
.003
)0.
0015
(0.0
04)
−0.0
173∗
∗∗(0
.002
)0.
0090
(0.0
06)
−0.0
150∗
∗∗(0
.002
)−0
.005
0(0
.006
)
390M
−0.0
192∗
∗∗(0
.003
)0.
0036
(0.0
04)
−0.0
161∗
∗∗(0
.002
)−0
.002
6(0
.006
)−0
.015
2∗∗∗
(0.0
02)
−0.0
061
(0.0
05)
400M
−0.0
216∗
∗∗(0
.003
)0.
0078
∗(0
.004
)−0
.017
1∗∗∗
(0.0
02)
−0.0
006
(0.0
06)
−0.0
126∗
∗∗(0
.002
)−0
.015
7∗∗∗
(0.0
05)
410M
0.00
01(0
.002
)−0
.002
3(0
.003
)0.
0013
(0.0
01)
−0.0
079∗
(0.0
04)
0.00
06(0
.001
)0.
0023
(0.0
03)
420M
−0.0
041∗
(0.0
02)
−0.0
037
(0.0
03)
−0.0
025
(0.0
02)
−0.0
067
(0.0
05)
−0.0
029
(0.0
02)
0.00
22(0
.004
)
430M
−0.0
088∗
∗∗(0
.002
)−0
.002
2(0
.004
)−0
.007
0∗∗∗
(0.0
02)
−0.0
052
(0.0
05)
−0.0
059∗
∗∗(0
.002
)−0
.001
5(0
.005
)
440M
−0.0
123∗
∗∗(0
.002
)−0
.001
5(0
.004
)−0
.011
6∗∗∗
(0.0
02)
0.00
20(0
.005
)−0
.009
3∗∗∗
(0.0
02)
−0.0
021
(0.0
05)
450M
−0.0
122∗
∗∗(0
.002
)−0
.001
4(0
.003
)−0
.010
8∗∗∗
(0.0
02)
−0.0
065
(0.0
06)
−0.0
103∗
∗∗(0
.002
)−0
.003
4(0
.004
)
460M
−0.0
004
(0.0
02)
0.00
01(0
.003
)0.
0014
(0.0
01)
−0.0
077∗
∗(0
.004
)0.
0006
(0.0
01)
−0.0
028
(0.0
03)
470M
−0.0
016
(0.0
02)
0.00
17(0
.003
)−0
.000
2(0
.001
)−0
.001
3(0
.004
)0.
0010
(0.0
02)
−0.0
039
(0.0
04)
480M
−0.0
016
(0.0
02)
0.00
10(0
.003
)−0
.000
7(0
.001
)−0
.001
1(0
.004
)−0
.000
4(0
.002
)−0
.003
0(0
.004
)
490M
0.00
63∗∗
∗(0
.001
)−0
.004
1∗(0
.002
)0.
0046
∗∗∗
(0.0
01)
−0.0
015
(0.0
04)
0.00
33∗∗
∗(0
.001
)0.
0040
(0.0
03)
500M
0.00
80∗∗
∗(0
.001
)−0
.002
0(0
.002
)0.
0071
∗∗∗
(0.0
01)
−0.0
008
(0.0
03)
0.00
61∗∗
∗(0
.001
)0.
0011
(0.0
02)
Not
e:T
his
tabl
epr
esen
tses
timat
esof
prob
itre
gres
sion
sw
ithth
ead
ditio
nalc
apita
lsto
ckdu
mm
yas
the
depe
nden
tvar
iabl
e.T
head
ditio
nalc
apita
lsto
ckdu
mm
yis
adu
mm
yva
riab
leth
atha
sa
valu
eof
one
ifca
pita
lsto
ckis
larg
erat
the
end
offi
scal
year
ttha
nat
the
begi
nnin
gof
fisc
alye
art.
We
use
the
sam
eco
ntro
lvar
iabl
esas
thos
ein
Tabl
es5,
6,7,
and
8.W
ees
timat
e16
equa
tions
for
each
indu
stry
.Col
umn
(1)
show
sth
eps
eudo
cap
onca
pita
lsto
ck.T
hem
argi
nale
ffec
tsof
each
vari
able
are
show
nin
each
cell.
Est
imat
edro
bust
stan
dard
erro
rsar
esh
own
inpa
rent
hese
s.T
hesy
mbo
ls∗ ,
∗∗,a
nd∗∗
∗ den
ote
sign
ific
ance
atth
e10
%,5
%,a
nd1%
leve
ls
1104
SME policies as a barrier to growth of SMEs
of treatment × policy, treatment × capital stock gap,and treatment × capital stock gap × policy is largearound FY1999. However, focusing on FY1996 andFY2002, we see that those estimated coefficients arestatistically insignificant or their magnitude is small.In sum, only the estimation results for the years aroundFY1999 support our hypothesis.
5.5.2 Estimations of pseudo cap on capital stock
In the previous section, we showed that firms whosecapital stock is close to the cap of the SME BasicAct do not increase the capital stock. To show therobustness of the estimation results, we reestimatethe effects of the pseudo cap on capital stock usingobservations of only nontreated firms. As we argued,the firms whose capital stock is under the cap ofthe old SME Basic Act are unaffected by the pol-icy change. When the capital stock is between 100and 300 million yen, firms are also affected by thepolicy change because they become an SME in thenew SME Basic Act. Therefore, firms with capitalstock of between over 300 and 500 million yen arenot affected by any policy changes of the SME BasicAct. Using the subsample of firms with capital stockof between over 300 and 500 million yen, we esti-mate (3) using several pseudo levels of the capitalstock as a proxy of the cap. We use various pseudocaps on the capital stock shown in Table 16, insteadof the actual cap of the capital stock in the old SMEBasic Act. As shown in Table 16, all the estimatedcoefficients of T reatmenti × Capital Stock Gapi,t
and T reatmenti × Capital Stock Gapi,t × Policyt
are inconsistent with the hypothesis, while the esti-mation results for the real cap in the SME BasicAct support the hypothesis. Therefore, the hypothe-sis in this paper is not supported by the untreatedfirms.
6 Conclusion
In this paper, we investigated whether firms have adisincentive to graduate from being SMEs to becomelarge firms. To test this hypothesis, we employed twoempirical strategies. First, we showed that firms with100 million yen of capital stock were less likely toincrease their capital stock. As SMEs are definedunder the Corporation Tax Act and the SME Basic Act
as firms with capital stock of 100 million yen or less,such firms had an incentive to meet the SME require-ments and retain their SME status. Second, we showedthat, after the relaxation of the definitions of SMEsunder the SME Basic Act, firms were more likely toincrease their capital stock. This effect is larger if afirm’s capital stock is close to the cap set in the SMEdefinition. This implies that the capital stock require-ment is an effective constraint on the accumulation ofadditional equity capital.
We also showed that additional capital stock hadpositive effects on firm growth (in terms of the growthrate of a firm’s total assets). As the requirements forcapital stock in the definitions of SMEs had neg-ative effects on additional capital stock, the SMErequirements impeded firm growth for SMEs.
Our study has important implications for SME poli-cies. As noted, SME policies can be important formitigating market failure. However, the menu of SMEpolicies adopted in Japan impedes firm growth. Gov-ernments should therefore be cautious about imple-menting an excessive range of policies to supportSMEs.
References
Acs, Z., & Szerb, L. (2007). Entrepreneurship, economicgrowth and public policy. Small Business Economics, 28(2),109–122. https://doi.org/10.1007/s11187-006-9012-3.
Arrow, K. (1962). Economic welfare and the allocation ofresources for invention. In The Rate and Direction ofInventive Activity: Economic and Social Factors. NationalBureau of Economic Research, pp. 609–626.
Baumol, W.J. (1962). On the theory of expansion of the firm.The American Economic Review, 52(5), 1078–1087.
Berger, A.N., & Udell, G.F. (1998). The economics of smallbusiness finance: The roles of private equity and debt mar-kets in the financial growth cycle. Journal of Banking& Finance, 22, 613–673. https://doi.org/10.1016/S0378-4266(98)00038-7.
Bernini, C., & Pellegrini, G. (2011). How are growthand productivity in private firms affected by publicsubsidy? Evidence from a regional policy. RegionalScience and Urban Economics, 41(3), 253–265.https://doi.org/10.1016/j.regsciurbeco.2011.01.005.
Cantner, U., & Kosters, S. (2012). Picking the winner?Empirical evidence on the targeting of r&d subsidiesto start-ups. Small Business Economics, 39(4), 921–936.https://doi.org/10.1007/s11187-011-9340-9.
Cerqua, A., & Pellegrini, G. (2014). Do subsidies to privatecapital boost firms’ growth? A multiple regression discon-tinuity design approach. Journal of Public Economics, 109,114–126. https://doi.org/10.1016/j.jpubeco.2013.11.005.
Chaurey, R. (2017). Location-based tax incentives: evidencefrom India. Journal of Public Economics, 156, 101–120.https://doi.org/10.1016/j.jpubeco.2016.08.013.
Cin, B.C., Kim, Y.J., Vonortas, N.S. (2017). The impact of pub-lic R&D subsidy on small firm productivity: evidence fromKorean SMEs. Small Business Economics, 48(2), 345–360.https://doi.org/10.1007/s11187-016-9786-x.
Cowling, M. (2016). You can lead a firm to r&d but can youmake it innovate? UK evidence from SMEs. Small Busi-ness Economics, 46(4), 565–577. https://doi.org/10.1007/s11187-016-9704-2.
Craig, B.R., Jackson, W.E., Thomson, J.B. (2007). Smallfirm finance, credit rationing, and the impact of SBA-guaranteed lending on local economic growth. Jour-nal of Small Business Management, 45(1), 116–132.https://doi.org/10.1111/j.1540-627X.2007.00202.x.
Czarnitzki, D., & Delanote, J. (2015). R&d policies for youngSMEs: Input and output effects. Small Business Economics,45(3), 465–485. https://doi.org/10.1007/s11187-015-9661-1.
Dawson, C., de Meza, D., Henley, A., Arabsheibani, G.R.(2014). Entrepreneurship: Cause and consequence of finan-cial optimism. Journal of Economics & Management Strat-egy, 23(4), 717–742. https://doi.org/10.1111/jems.12076.
De Meza, D. (2002). Overlending? The Economic Journal,112(477), F17–F31. https://doi.org/10.1111/1468-0297.00681.
Foreman-Peck, J. (2013). Effectiveness and efficiency of SMEinnovation policy. Small Business Economics, 41(1), 55–70. https://doi.org/10.1007/s11187-012-9426-z.
Garcıa-Santana, M., & Pijoan-Mas, J. (2014). The reserva-tion laws in India and the misallocation of productionfactors. Journal of Monetary Economics, 66, 193–209.https://doi.org/10.1016/j.jmoneco.2014.04.017.
Garicano, L., Lelarge, C., Reenen, J.V. (2016). Firm size dis-tortions and the productivity distribution: evidence fromFrance. American Economic Review, 106(11), 3439–79.https://doi.org/10.1257/aer.20130232.
Gourio, F., & Roys, N. (2014). Size-dependent regulations, firmsize distribution, and reallocation. Quantitative Economics,5(2), 377–416. https://doi.org/10.3982/QE338.
Guner, N., Ventura, G., Xu, Y. (2008). Macroeconomic impli-cations of size-dependent policies. Review of EconomicDynamics, 11(4), 721–744. https://doi.org/10.1016/j.red.2008.01.005.
Guner, N., Ventura, G., Yi, X. (2006). How costly are restric-tions on size? Japan and the World Economy, 18(3),302–320. https://doi.org/10.1016/j.japwor.2004.11.002.
Hanson, A., & Rohlin, S. (2011). Do location-based taxincentives attract new business establishments? Journal ofRegional Science, 51(3), 427–449. https://doi.org/10.1111/j.1467-9787.2010.00704.x.
Hosono, K., Takizawa, M., Tsuru, K. (2017). Size-dependentpolicy and firm growth. Discussion papers 17070, ResearchInstitute of Economy, Trade and Industry (RIETI).
Ipinnaiye, O., Dineen, D., Lenihan, H. (2017). Driversof SME performance: a holistic and multivariateapproach. Small Business Economics, 48(4), 883–911.https://doi.org/10.1007/s11187-016-9819-5.
Kang, J.W., & Heshmati, A. (2008). Effect of credit guaranteepolicy on survival and performance of SMEs in Repub-lic of Korea. Small Business Economics, 31(4), 445–462.https://doi.org/10.1007/s11187-007-9049-y.
Koga, T. (2005). R&d subsidy and self-financed r&d: the caseof Japanese high-technology start-ups. Small Business Eco-nomics, 24(1), 53–62. https://doi.org/10.1007/s11187-005-3096-z.
Lenihan, H. (2011). Enterprise Policy evaluation: is there a‘new’ way of doing it? Evaluation and Program Planning,34(4), 323–332. https://doi.org/10.1016/j.evalprogplan.2011.03.006.
Lokshin, B., & Mohnen, P. (2012). How effective arelevel-based R&D tax credits? Evidence from theNetherlands. Applied Economics, 44(12), 1527–1538.https://doi.org/10.1080/00036846.2010.543083.
Mankiw, N.G. (1986). The allocation of credit and financial col-lapse. Quarterly Journal of Economics, 101(3), 455–470.https://doi.org/10.2307/1885692.
Nakata, T. (2013). Part XI amendment of the small and medium-sized enterprise basic act. In Nakata, T. (Ed.) History ofjapan’s trade and industry policy (12): Small- and medium-sized enterprises (in Japanese). Keizai Sangyo Chosakai,Tokyo, pp. 1207-1256.
OECD (2013). SME And entrepreneurship financing: therole of credit guarantee schemes and mutual guaranteesocieties in supporting finance for small and medium-sized enterprises, Final Report, OECD iLibrary, Paris.https://doi.org/10.1787/f493861e-en.
OECD. (2016). Japan: Boosting growth and well-being in anageing society. Paris: OECD iLibrary. https://doi.org/10.1787/9789264256507-en.
Oh, I., Lee, J.-D., Heshmati, A., Choi, G.-G. (2009). Eval-uation of credit guarantee policy using propensity scorematching. Small Business Economics, 33(3), 335–351.https://doi.org/10.1007/s11187-008-9102-5.
Opler, T.C., & Titman, S. (1994). Financial distress and cor-porate performance. Journal of Finance, 49(3), 1015–1040.https://doi.org/10.1111/j.1540-6261.1994.tb00086.x.
Ou, C., & Haynes, G.W. (2006). Acquisition of additionalequity capital by small firms – findings from the nationalsurvey of small business finances. Small Business Eco-nomics, 27(2), 157–168. https://doi.org/10.1007/s11187-006-0009-8.
Saito, K., & Tsuruta, D. (2018). Information asymmetry insmall and medium enterprise credit guarantee schemes:Evidence from Japan. Applied Economics, 50(22), 2469–2485. https://doi.org/10.1080/00036846.2017.1400651.
Storey, D.J. (1994). Understanding the small business sector.London: Thomson Learning.
Storey, D.J. (2008). Entrepreneurship and SME policy,World Entrepreneurship Forum 2008 Edition. (avail-able at: http://www.world-entrepreneurship-forum.com/Publications/Articles).
Uesugi, I., Sakai, K., Yamashiro, G.M. (2010). The effective-ness of public credit guarantees in the Japanese loan mar-ket. Journal of the Japanese and International Economies,24(4), 457–480. https://doi.org/10.1016/j.jjie.2010.08.001.