Financing Preferences and Firm Growth in a Small Open Economy

Transcription

1 Financing Preferences and Firm Growh in a Small Open Economy By Winson Moore, Roland Craigwell and Chanelle Maxwell 1 Research Deparmen, Cenral Bank of Barbados, PO Box 1016, Bridgeown, Barbados ABSTRACT. This paper uilises daa on all he lised companies in Barbados spanning FINANCING PREFERENCES AND FIRM GROWTH IN A SMALL OPEN ECONOMY he period 1990 o 2000 o examine he hypohesis ha a firm s financing preferences has no influence on growh. Conrary o he previous lieraure, a model of firm growh is augmened wih financing preference variables and esimaed using insrumenal variable By Winson Moore, Roland Craigwell and Chanelle Maxwell Cenral Bank of Barbados panel daa echniques. The resuls from his sudy sugges ha he use of inernal and exernal funds is inversely relaed o firm growh, while equiy finance has a posiive impac. Moreover, he srengh of hese relaionships is enhanced for younger, smaller enerprises. Proposed Running Headline: Financing Preferences and Firm Growh 3 Sepember Moore: Senior Economis, Research Deparmen, Cenral Bank of Barbados, P.O. Box 1016, Bridgeown. Phone , Fax , wrmoore@cenralbank.org.bb Craigwell: Chief Economis, Research Deparmen, Cenral Bank of Barbados, P.O. Box 1016, Bridgeown. Phone , Fax , rccraigwell@cenralbank.org.bb Maxwell: Economis, Research Deparmen, Cenral Bank of Barbados, P.O. Box 1016, Bridgeown. Phone , Fax , cmaxwell@cenralbank.org.bb 23

2 Financing Preferences and Firm Growh in a Small Open Economy Insead of developing a heoreical model of invesmen behaviour, which is cusomary in he empirical lieraure, his sudy specifies and esimaes a model of firm growh. Then he impac of financing decisions is examined by augmening he model 1. Inroducion wih financing decision variables. In addiion, raher han spliing he sample, he paper uses ineracion erms o deermine wheher he imporance of he financing decision In 1958, Franco Modigliani and Meron Miller developed he hypohesis ha a firm s financial srucure will no affec is marke value in fricionless capial markes. This resul implies ha only echnological and produc marke opporuniies should influence a firm s real variables, and by exension, is prospecs for growh. However, a number of empirical sudies have found evidence conrary o Modigliani and Miller s irrelevance resul. Indeed, i is repored ha when cash flow is added o a heoreically derived variables varies across firms depending on heir age, size and cash flows. To he auhors knowledge, he enire approach adoped is one of he firs conduced in he area. The resuls agree wih hose auhors who refue he Modiliani-Miller hypohesis and sugges ha firms financing decisions do impac on he growh prospecs of publicly raded companies in Barbados. More imporanly, however, i indicaes ha he use of inernal and exernal funds are inversely relaed o firm growh, while employing equiy invesmen equaion, his variable ends o improve he empirical fi of he model (see finance has a posiive influence on growh. Addiionally, he impac of financing Hubbard, 1998, for a survey of his lieraure). This paper advances he above-menioned lieraure, by using insrumenal variable economeric echniques o es he hypohesis of wheher a firm s financing preferences on firm growh increases as firm size falls. These resuls are robus o a series of sensiiviy ess, and herefore seem o imply ha firms should be careful of being overly relian on inernal finance. preferences in a small open economy, influences is fuure growh prospecs. In The remainder of he paper is as follows. Secion II develops he empirical paricular, his sudy considers he imporance of exernal financing, inernal financing and equiy finance on firm growh in Barbados. One implicaion of his paper is ha if a model. Secion III discusses he daa and examines he characerisics of he Barbadian corporae secor, while Secion IV presens he economeric resuls. Secion V concludes. link beween a firm s financing decision and growh is found, i could provide evidence in favour of he credi view of he moneary ransmission mechanism, which argues ha moneary policy firs impacs on borrowers financial posiion (for example, cash flows) and hen on invesmen (see Brunner and Melzer, 1968; Bernanke and Blinder, 1992). 24

3 2. Empirical Model 2.1 Theory example, he choice beween deb and equiy) should no impac on real variables. In his insance, invesmen is only influenced by Tobin s q, which is a proxy for he increase in he value of he firm ha would resul from an addiional uni of invesmen, and is hus, The saring poin of he analysis is o specify a model of firm growh, and hen augmen his equaion wih financing variables o idenify wheher hey significanly affec growh. If one assumes ha firms maximise he expeced discouned value of dividends ( D ), hen he firm s iner-emporal maximisaion problem is as follows: subjec o: V ( K, ) max D E s 1D s (1) { I s } s 0 s 1 D ( K, ) C( I, K ) I (2) K 1 (1 ) K I (3) where represens he curren ime period, s are incremenal periods o, K is he capial sock, is a produciviy shock, E [.] is he expecaions operaor, represens he discoun facor, is he profi funcion, C is he adjusmen cos funcion, I is invesmen and is he rae of depreciaion for capial. equaion: where Hubbard (1998) shows ha one can derive he following reduced form invesmen * I q 0 (4) 0 1 i, 1 * I is he firm s invesmen in fixed capial (as a fracion of he curren fixed capial inpu) and q i is Tobin s q raio. Equaion (4) is quie similar o he resuls derived by Modigliani and Miller (1958), and shows ha a firm s financial policy (for posiively relaed o invesmen. form: Assuming he firm s log-linearised Cobb-Douglas producion funcion is of he y l ( 1 ) k (5) i i i Then aking he firs difference of Equaion (5) and subsiuing Equaion (4) while noing ha k * i I i i, one obains he following growh equaion: y q (6) * * * 0 1 li 2 i Equaion (6) indicaes ha firm growh should be posiively associaed wih changes in he amoun of labour inpu uilised and Tobin s q. This equaion can be augmened wih he erm, F i, which capures he financing decisions of firms, ha is: y F (7) i * * * * 0 1 li 2qi 3 i * Thus, if he firm s financial policy has no impac on growh hen 0 in Equaion (7). The firms examined can be divided ino various sub-samples according o some measure of financial preferences (see Hoshi, Kashyap and Scharfsein, 1991; Fazzari, Hubbard and Peerson, 1985; Whied, 1992), o es wheher he above-proposed relaionship differs beween sub-samples. This approach can lead o degrees of freedom problems and, alernaively, he financing variables can be ineraced wih he variables of ineres, for example, size. In his case, Equaion (7) becomes: * * * * * y l q F F Size (8) i 0 1 i 2 i 3 i 4 i 3 25

4 * Thus, 4 ells he researcher wheher he choice of he ype of finance used for 3. Daa invesmen varies beween firms of differing sizes. The same procedure can be done for age and a proxy for financing consrains. This paper uses published financial saemen daa for all Barbadian firms raded on he local sock exchange over he period 1990 o 2000 o examine corporae growh. The 2.2 Economeric Approach firms in he daabase accoun for 8% of oal privae secor employmen in Barbados, and generae $3.1 billion in sales, or approximaely 60% of gross domesic produc a marke Equaion (8) is usually esimaed using panel daa echniques (ha is, he pooled, fixed effecs or random effecs models). However, many of he variables in he equaion are likely o be simulaneously deermined wih he dependen variable or he oher explanaory variables. For example, hose firms wih relaively faser raes of growh are likely o have greaer cash and may find i more advanageous o employ inernal finance o purchase new invesmens. Similarly, he greaer use of labour inpus, ceeris paribus, should lead o an increase in oupu. However, an expansion in oupu could also cause he firm o uilise more labour inpus. In hese insances, he parameer esimaes from he usual panel daa echniques are inconsisen. Therefore, his sudy uilises insrumenal variable mehods (hree-sage leas squares). The presence of heeroskedasic residuals could be a problem for Equaion (8), given ha he sudy employs company-level daa, as a resul, Whie heeroskedasic consisen sandard errors are uilised in all regressions. prices. Firms included in he daabase are drawn from virually every major indusry in he domesic economy (disribuion; finance, insurance and real esae; manufacuring; ranspor, sorage and communicaions; elecriciy, gas and waer, and; ourism). One shorcoming of he daabase, however, is ha i does no include a large number of small enerprises. Tables 1 and 2 presen aggregae saisics on corporae performance in Barbados over he period 1990 o The ables show ha firm growh is highly pro-cyclical, wih he only period of decline recorded in 1992 when Barbados was in he mids of an economic recession. As a resul, average firm growh over he period was esimaed a 6.1%. The bes performing firms were hose in he finance, insurance and real esae indusry, wih an average annual rae of expansion of 14.7%. Firms engaged in disribuion, ranspor, sorage and communicaions and ourism aciviies also regisered credible raes of growh of beween 4 and 6% per year. Unsurprisingly, hose enerprises in manufacuring increased, on average, by less han one percen per year over he sample period. Table 1 also provides saisics on he oal invesmen as a percen of ne asses. I reveals ha firms inves on average 17% of ne asses in any given year. Eniies in 26

5 disribuion (17.9%) and ranspor, sorage and communicaions (31.4%) had he greaes invesmen o asse raios over he period, while he invesmen o asse raio for The pervasive use of inernal financing also has ramificaions for firm sabiliy, since financing invesmens hrough inernal sources, reduces he resources available o manufacuring esablishmens was 16.4%, probably indicaing why produc obsolescence buffer agains shocks arising from economic downurns. The uilisaion of inernal and rade liberalisaion has forced many of hese enerprises ou of operaion (Lewis- Bynoe, Griffih and Moore, 2002). Mos of he financing for hese new invesmen aciviies came primarily from company (inernal) sources. Beween 1990 and 2000, inernal financing represened, on average, 12.3% of ne asses, four percen greaer han he raio for exernal financing and significanly higher han ha obained from equiy finance, which accouned for less han one percen of ne asses. Also noeworhy is ha inernal financing was quie significan during he period of economic decline (beween 1990 and 1993), when i represened 15.9% of ne asses, almos wice he amoun obained from exernal sources. In conras, during he pos recession period, inernal financing declined o 11% of ne asses while exernal finance rose marginally. Equiy financing was he leas used source of financing invesmen aciviies during he review period, wih he highes raio o ne asses of 3.7% recorded in This laer finding is surprising given ha domesic ineres rae spreads end o be wide and exceed hose in mos Caribbean and developed economies (Moore and Craigwell, 2002). However, if firms benefi from relaionship lending his could explain why hey chose his source of financing over equiy. Parial evidence of his phenomenon is repored by Moore and Craigwell (2003), who argue ha ineres raes on large loans (over $500,000) end o be almos wo percenage poins lower han hose obained on smaller loans. In addiion, he relaively undeveloped capial marke and he bias of he ax sysem owards deb finance are imporan facors as well. finance is especially prevalen in manufacuring enerprises, where inernal finance usually represens 20.7% of oal asses in any given year. Addiionally, firms engaged in ranspor, sorage and communicaions (19.5%) and finance, insurance and real esae (12.9%) had relaively high inernal finance o asse raios. On he oher hand, firms in disribuion end o depend primarily on loans and overdrafs o finance heir aciviies, wih exernal finance averaging 10.3% of ne asses. Tourism (13.9%) and ranspor, sorage and communicaions (17.6%) enerprises also had relaively high exernal financing o equiy raios. Only firms in he finance, insurance and real esae indusry had any significan levels of equiy finance, wih average equiy o asse raios of 6.6% repored, while all he oher indusries recorded equiy finance o asse raios of less han 1%. As a preliminary analysis of he possible impac ha firms financing decisions can have on growh, Figure 1 plos he rae of expansion for hose firms which depend o a large exen on inernal finance and exernal finance. The figure shows ha here does no appear o exis any significan difference in he rae of growh beween he wo groups of firms. Over he sample period considered, he average rae of increase for hose firms which have a relaively greaer reliance on inernal finance was approximaely 6% per year, or wo percenage poins higher han hose ha depended primarily on exernal finance. This difference, however, seems o have occurred since hose firms which depend on exernal deb o finance a larger proporion of heir 27

6 invesmen purchases had a somewha greaer variance in heir rae of expansion. Thus, significanly influence he growh prospecs of Barbadian firms. The coefficien he hypohesis ha firm s financing decisions influences firm growh seems o have been rejeced by a graphical analysis of he daa. This simple graphical analysis, however, does no allow for differences in he fuure profiabiliy beween firms as suggesed by he model presened in Secion 2. In addiion, an analysis of share equiy could no be underaken since firms did no issue equiy in every year for he sample period considered. Therefore, more formal saisical echniques han hose repored above are esimaes reveal ha hose firms, wih a greaer preference for using inernal finance and deb, experience lower raes of growh han heir peers. The negaive coefficien on he deb variable could reflec a number of facors. For insance, one can argue ha deb causes he firm o uilise a fixed proporion of is earnings every period o repay he ineres and principle on he deb. This reduces he abiliy of he firm o absorb he impac of shocks on he demand for he enerprise s goods or services (see Bernanke and employed in he following secion. Campbell, 1988; Canor, 1990; Freidman, 1990). Moreover, if he reurn on he invesmen is no greaer han he rae of ineres, hen he overall value of he firm, and by exension is rae of growh, could decline over ime. 4. Economeric Findings 4.1 Baseline Resuls The difference in he size of he coefficien esimaes show ha he preference for financing invesmen purchases wih reained earnings had he greaes negaive influence on firm growh. More specifically, he sudy finds ha for every dollar in reained Column 2 of Table 3 presens he basic growh model for firms in Barbados. Alhough earnings he firm uses o finance invesmen purchases, i looses hireen cens in erms of he sudy focuses on he insrumenal variable resuls, due o possible simulaneiy possible asse growh. As in he case of deb, he large negaive impac of firm problems, OLS esimaes are also provided for comparison purposes. In accordance wih a priori reasoning, he basic growh model shows ha growh is posiively relaed o changes in he labour inpus used and Tobin s q raio. I is noeworhy ha he signs and magniudes of hese wo explanaory variables did no change appreciable in any of he hree regressions esimaed. The final column in he Table augmens he basic growh model wih variables, preferences for inernal finance could imply ha financial insiuions have more resricive sandards on financing invesmen projecs han firms. Thus, hose projecs he bank refuses o finance eiher have o be abandoned or financed from inernal sources. I is noed ha his argumen runs conrary o he financing hierarchy framework for businesses esed by Oliner and Rudebusch (1996), which argues ha firms end o finance mos invesmen from reained earnings given he cos relaed o exernal finance, which capure he financing preferences of firms in he sample. Conrary o he bu could hold for fringe projecs. However, one could hink abou projecs, which are Modigliani and Miller irrelevance resul, he financial preference variables do highly valued by he firm, or is execuives, bu does no have a high enough rae of 28

7 reurn o be financed by conracing deb (for example, a more luxurious car for he chief execuive officer). In conras, he coefficien on he equiy variable is he only one of he finance preference indicaors ha is posiively relaed o firm growh. Moreover, he coefficien esimaes sugges ha for every dollar in new invesmen financed hrough equiy, he firm s asses rise by one dollar and hiry-eigh cens. This large and posiive coefficien on he equiy variable implies ha firms end o only finance hose projecs wih a relaively high reurn on invesmen. In order o furher invesigae he impac of firm s financing decisions on growh, he auhors inerac an age dummy wih he finance variables. The coefficien esimaes provide some ineresing conclusions. Firs, all he age ineracion erms are significan, and herefore sugges ha age has a significan influence on he financing preferencesgrowh nexus. Second, he decision o use reained earnings and/or deb has a greaer negaive impac on he growh prospecs for younger firms relaive o heir older peers. However, younger firms ha issue equiy o finance invesmen purchases have greaer growh prospecs han older eniies. Third, he negaive effec of exernal finance on growh is larger han ha for young firms using inernal financing sources. This oucome can also be due in par o young firms depending more on deb in heir early sages of 4.2 Non-Linear Resuls growh, since in he iniial years of business developmen cash flows are usually negaive. This finding jusifies he use of special financing schemes for sar-ups as a means of As menioned in Secion II here exiss a large amoun of lieraure, which indicaes ha developing a corporae culure. he financing prospecs of small firms differ from hose of large eniies. As a resul, his paper ineracs a size dummy wih he financing variables o es he impac on corporae growh beween differen sizes of firms (see Table 4). Column 3 of he able reveals ha 4.3 Sensiiviy Analysis he ineracion erm wih exernal finance is posiive and significanly relaed o firm growh. This finding seems o reflec he long held view ha large firms end o benefi from financial economies of scale, boh in erms of access o finance and ineres raes. In addiion, he magniude of he coefficien suggess ha for every dollar invesed (financed hrough exernal sources), larger firms earn abou hiry cens more han heir smaller counerpars. This secion of he sudy discusses he robusness of he key resuls repored above. Table 5 summarises he ess and o conserve space, only he coefficien esimaes on he relevan financing variables are repored. Evans (1987) and Hall (1987) boh found ha firm growh should be specified as a funcion of firm size and age. They base his posure, for he mos par, on Jovanovic s (1982) learning model of firm growh. The framework assumes ha firms have an iniial disribuion of efficiency, which is 29

8 unknown o hem before hey ener he indusry. However, afer enering he indusry firms obain informaion abou heir cos hrough producion, wih he mos efficien firms more likely o survive and grow, and he inefficien enerprises having a greaer probabiliy of failure. The model, herefore, yields he predicion ha small firms should have higher bu more variable raes of growh and herefore size should be inversely relaed o growh. In conras, he learning-by-doing lieraure proposes ha firm growh should be posiively relaed o size. This concep was firs elucidaed by Wrigh (1936), who pus forward he idea ha he average cos of producion decreases wih he pas accumulaion of oupu. Since small firms accumulaed pas oupu is lower han ha for larger enerprises, one should expec ha large corporaions should expand a a faser rae han smaller eniies. One would also anicipae ha larger firms are able o benefi from economies of scale in producion and finance, since large corporaions, given heir superior asse base, are more likely o benefi from reduced borrowing cos. In conras o he wo heories above, Gibra s law argues ha plan size should be independen of learning by doing lieraure bu more imporanly, he main conclusions regarding he impac of financing preferences are unchanged. Jovanovic s (1982) heory of firm growh also proposes ha age and growh should be inversely relaed based on he assumpion ha firms learn heir efficiencies over ime. As a resul, he oupu of corporaions of various ages should converge in he long run. In addiion, since older eniies are probably closer o convergence, heir raes of economic expansion are likely o be lower. The augmened baseline economeric equaions reveal ha age is an insignifican deerminan of he firm growh, and shows ha he signs and magniudes of he financing variables are unchanged. Finally, he growh model is augmened wih he variable average cash as a raio o oal asses. This variable is employed o allow for he effec of financial consrains on firm growh and invesmen (see Devereux and Schianarelli, 1989). The inclusion of his variable would also allow for he influence of financial consrains on firms financing preferences. The inferences derived from he signs and significance of he financing variables is unchanged. growh. This hypohesis is based on he assumpion ha he disribuion of plan sizes is log normal, as a resul, in he growh model size should be an insignifican deerminan of growh. 5. Conclusions A number of definiions of he size variable are examined: (1) size demarcaed using a dummy variable, where large firms are defined as hose employing more han 500 persons; (2) size measured by he naural logarihm of he oal number of employees, and; (3) size proxied by he naural logarihm of oal asses (column 2 of Table 5). In all insances, size is posiively relaed o firm growh in Barbados in accordance wih he Convenional economic heory posulaes ha firms financing choices should have no influence on real variables such as invesmen or growh. This paper examines his hypohesis by esing he significance of financing variables in a model of corporae growh using a panel of Barbadian enerprises over he period 1990 o The 30

9 economeric resuls sugges ha financing preferences do have a significan influence on References he growh prospecs of corporaions over he shor-erm. The findings indicae ha he use of inernal and exernal finance is inversely relaed o growh, while equiy finance is posiively relaed o growh. However, he negaive influence of exernal finance on growh is reduced for larger enerprises. This finding could be reflecive of financial economies of scale or relaionship lending, a characerisic, which may be prevalen in he Barbadian financial sysem (Moore and Craigwell, 2003). However, he negaive impac of inernal and exernal financing is magnified for smaller enerprises, while he use of equiy financing by small firms is posiively relaed o growh. A series of sensiiviy ess confirm he robusness of hese resuls. The findings ariculaed in his sudy have a number of policy implicaions. Firs, he axaion sysem in Barbados should be reformed o eliminae he bias agains uilising equiy finance. Currenly, ineres coss relaed o conracing deb can be wrien-off for ax purposes, however, he cos of issuing new shares does no benefi from similar exempions. Thus, he accouning cos of issuing new equiy is greaer. The finding ha he use of reained earnings o finance invesmen has he greaes negaive impac on growh sugges ha policymakers need o coninue o focus on encouraging firms, especially sar-ups, o uilise oher forms of invesmen finance. Finally, effors o develop he local and regional sock exchanges should be encouraged. Balagi, B.H., 2001, Economeric Analysis of Panel Daa, England: John Wiley and Sons Ld. Bernanke, B.S. and J.Y. Campbell, 1988, Is There a Corporae Deb Crisis? Brookings Papers on Economic Aciviy 1, Bernanke, B.S. and A.S. Blinder, 1992, Federal Funds Rae and he Channels of Moneary Transmission, American Economic Review 82, Brunner, K. and A.H. Melzer, 1968, Liquidiy Traps for Money, Bank Credi, and Ineres Raes, Journal of Poliical Economy 76, Canor, R., 1990, Effecs of Leverage on Corporae Invesmen and Hiring Decisions, Federal Reserve Bank of New York Quarerly Review (Summer), Devereux, M. and F. Schianarelli, 1989, Invesmen, Financial Facors and Cash Flows: Evidence from UK Panel Daa, NBER Working Paper No. 3116, MA. Evans, D.S., 1987, Tess of Alernaive Theories of Firm Growh, Journal of Poliical Economy 95, Fazzari, S.M., R.G. Hubbard and B.C. Peerson, 1988, Financing Consrains and Corporae Invesmen, Brookings Paper on Economic Aciviy 1, Freidman, B., 1990, Implicaions of Increasing Corporae Indebedness for Moneary Policy, Occasional Papers Nos. 29, Group of Thiry, New York. Hall, B.H., 1987, The Relaionship Beween Firm Size and Firm Growh in he U.S. Manufacuring Secor, Journal of Indusrial Economics 35,

10 Hausman, J.A., 1978, Specificaion Tess in Economerics, Economerica 46, Hoshi, T., Kashyap, A. and D. Scharfsein, 1991, Corporae Srucure, Liquidiy, and Invesmen: Evidence from Japanese Panel Daa, Quarerly Journal of Whied, T.M., 1992, Deb, Liquidiy Consrains, and Corporae Invesmen: Evidence from Panel Daa, Journal of Finance 47, Wrigh, T.P., 1936, Facors Affecing he Cos of Airplanes, Journal of Aerospace Science 3, Economics 106, Hubbard, G.R., 1998, Capial-Marke Imperfecions and Invesmen, Journal of Economic Lieraure 36, Jovanovic, B., 1982, Selecion and Evoluion of Indusry, Economerica 50, Lewis-Bynoe, D., Griffih, J. and W. Moore, 2002, Trade Liberalisaion and he Manufacuring Secor: The Case of a Small Developing Counry, Conemporary Economic Policy 20, Mercer, L.J. and W.D. Morgan, 1972, Inernal Funds and Auomobile Indusry Invesmen: An Evaluaion of he Selzer Hypohesis, Journal of Economic Hisory 32, Modigliani, F. and M.H. Miller, 1958, The Cos of Capial, Corporaion Finance, and he Theory of Invesmen, American Economic Review 48, Moore, W. and R. Craigwell, 2002, Marke Power and Ineres Rae Spreads in he Caribbean, Inernaional Review of Applied Economics 16, Moore, W. and R. Craigwell, 2003, The Relaionship Beween Commercial Banks Ineres Raes and Loan Sizes: Evidence from a Small Open Economy, Applied Financial Economics 13, Oliner, S.D. and G. Rudebusch, 1996, Is There a Broad Credi Channel for Moneary Policy? Federal Reserve Bank San Francisco Economic Review 1,

11 TABLE 1: Corporae Performance in Barbados TABLE 2: Descripion of Variables Growh in Nomina l GDP (%) Growh in Compan y Ne Asses (%) Averag e Age (years ) Growh in Employmen (%) Toal Invesmen (% of Ne Asses) Exernal Financin g (% of Ne Asses) New Share Issues (% of Ne Asses ) Inernal Financin g (% of Ne Asses) Variable Descripion Mean S. Obs. Dev. Growh Logarihm of ne asses in period less he logarihm of ne asses in period -1 l Logarihm of he oal number of employees q Logarihm of he value of equiy plus oal deb, in -1, divided by he book value of he firm s ne asses in -1 Infin Toal value of invesmen finance from inernal sources divided by oal asses Sharefin Toal value of invesmen finance from new equiy divided by oal asses Exfin Toal value of invesmen finance from exernal sources divided by oal asses Large Logarihm of ne asses Age Dummy variable which akes he value 1 for firms which are older han he sample mean and zero oherwise

12 TABLE 3: Basic Regression Resuls Dependen variable: Growh OLS IV IV l (0.054)* (0.014)** (0.011)** q (0.021) (0.003)** (0.003)** Exfin (0.015)** Infin (0.025)** Sharefin (0.093)** Inercep (0.010)** (0.001)** (0.002)** R Observaions Noe: (1) Whie sandard errors are repored in parenheses. (2) **, * indicaes significance a he 1, 5 percen levels, respecively. TABLE 4: Non-Linear Regression Resuls Dependen variable: Growh IV IV IV l (0.011)** (0.039)** (0.011)** q (0.003)** (0.011) (0.004)** Exfin (0.015)** (0.074) (0.023) Infin (0.025)** (0.092) (0.031)** Sharefin (0.093)** (0.359)** (0.490) Exfin Large (0.129)* Infin Large (0.165) Sharefin Large (0.923) Exfin Young (0.042)** Infin Young (0.036)* Sharefin Young (0.518)* Inercep (0.002)** (0.011)** (0.004)** R Observaions Noe: (1) Whie sandard errors are repored in parenheses. (2) **, * indicaes significance a he 1, 5 percen levels, respecively. 34

13 Dependen variable: Growh TABLE 5: Sensiiviy Analysis IV wih size as an explanaory variable IV wih age as an explanaory variable Exfin (0.018)** (0.017)** Infin (0.024)** (0.024)** Sharefin (0.146)** (0.125)** Noe: (1) Whie sandard errors are repored in parenheses. (2) **, * indicaes significance a he 1, 5 percen levels, respecively. IV wih cashflow as an explanaory variable (0.015)** (0.022)** (0.128)** % FIGURE 1: Firm Growh (Disaggregaed by Major Source of Invesmen Finance) Inernal Finance Exernal Finance 35