How Do Firm Investments Respond To Debt And Aggregate Uncertainty? Study Case In Malaysia

Transcription

1 PROSIDING PERKEM VIII, JILID 1 (013) ISSN: 31-96X How Do Firm Invesmens Respond To Deb And Aggregae Uncerainy? Sudy Case In Malaysia Riayai Ahmad School of Economics Faculy of Economics and Managemen The Naional Universiy of Malaysia riayai@ukm.my ABSTRACT The paper idenifies he response of firms invesmen o he marke ineres rae uncerainy and deb holding in Malaysia as one of developing counry. The sensiiviy of firms invesmen o he ineracion beween aggregae uncerainy and deb holding is also emphasized. This paper also examines he heerogeneiy beween high- and low-indebed firms groups. The findings discover ha firms invesmen responds negaively o he aggregae uncerainy and deb holding. However, he effec of he ineracion beween aggregae uncerainy and deb holding on firms invesmen was no significan; hese resuls were quie homogenous for high- and low-indebed firms in Malaysia. INTRODUCTION The impac of deb and uncerainy on firm invesmen aracs a lo of aenions from researchers sudying in his issue. However, here is a lile aenion is given o sudy he linkage beween he win effecs of deb and uncerainy on firm invesmen.thus, he objecive of his paper is o invesigae he role of uncerainy and deb holding on firm invesmen behaviour. In paricular, his sudy analyses he effec of deb and macroeconomic uncerainy (he marke ineres raes uncerainy) on firm invesmen. This sudy also ries o idenify he cross-effec of macroeconomic uncerainy and deb holding on firm invesmen. In oher words, his sudy aemps o assess he specific channels ha affec firm invesmen in Malaysia. The final objecive in his paper is o idenify he heerogeneous effec for wo groups firms in Malaysia namely high- and low-indebed firms. Theory suggess here is a posiive relaionship beween deb and invesmen which called he signalling hypohesis as inroduced by Ross (1977). Myers (1977) inroduced he pecking order heory of financing. This hypohesis saes ha here are hree levels for financing a firm s new invesmen, which are inernal funds, o issue deb and as a las resor o issue equiy. This implies ha coss of deb financing for invesmen are higher han hose of inernal funds. Consequenly, i affecs firm financing decisions o finance heir invesmen. Furhermore, if he firm issues deb, hen he deb payback commimen suggess a lower level of liquidiy. In imperfec capial markes, a firm ha has a lower liquidiy will face higher coss of exernal capial, which discourages invesmen.thus, i shows a negaive impac of deb financing on firm invesmen, which conradics he signaling hypohesis. So i can be said ha he relaion beween deb and firm invesmen is ambiguous. There are sudies ha focus on he relaionship beween uncerainy and invesmen heoreically and empirically. Uncerainy can be divided ino wo caegories which are macroeconomics (aggregae) uncerainy and idiosyncraic uncerainy. Examples for macroeconomic uncerainy are uncerainy in inflaion raes, exchange raes and marke ineres raes, while idiosyncraic uncerainy is proxied by he uncerainy in he produciviy, cos of producion and so on. Theory also idenifies several channels or facors how uncerainy may affec firms invesmen. They include he risk aiude of firms owards risk; risk averse or risk akers, and financing consrains ha may arise from asymmeries beween borrowers and lenders. Generally, he effec of uncerainy on invesmen shows an ambiguous relaion. The radiional lieraure assumes ha he invesmen is reversible which suggess a posiive effec of uncerainy on invesmen as explained in Richard (197). Recen lieraure suggess ha invesmen is irreversible and firms have he opion o wai o inves for example in Dixi and Pindyck (1994). The irreversibiliy and he opion o wai could affec invesmen negaively. However, many sudies on his opic focus more on he developed counries raher han developing counries. For examples, Aivazian e al. (005), Byrne and Davis (004), Bo and Serken (00) and Baum (010) sudy he relaion beween uncerainy and invesmen in developed counries. Bo and Serken (00) sudy his issue for he Neherlands, Byrne and Davis (004) examine he same Persidangan Kebangsaan Ekonomi Malaysia ke VIII (PERKEM VIII) Dasar Awam Dalam Era Transformasi Ekonomi: Cabaran dan Halauju Johor Bahru, 7 9 Jun 013

2 494 Riayai Ahmad issue in he Unied Saes and Baum (010) focuses on he Unied Kingdom. There are a few sudies ha focus on he developing counry, for example, Aizenman and Marion (1999). Their sudy focuses on he macro-level daa and he main objecive is o examine he relaionship beween various volailiy measures and privae invesmen in 46 developing counries. In view of limied empirical evidence for developing counries especially a firm level, his paper will examine he aspec of invesmen and uncerainy problems in developing counries using a firm-level daase. Specifically, his paper sudies he impac of deb holdings and uncerainy on firm invesmen behaviour in Malaysia as one developing counry. Focusing on a developing counry is imporan o idenify he behaviour of firm invesmen in his group, as i may have differen behaviour wih firm invesmen in developed counries. The moivaion for his sudy is based on he argumen ha many firms are financed by equiy and deb. When uncerainy happens in he economy i will affec he invesmen decision direcly. Furhermore, he uncerainy also has an impac on he financial srucure of he firm in real erms. For example, uncerainy in he nominal ineres raes because of he high volailiy in he inflaion raes will lead o a higher ineres raes burden. Higher ineres raes burden will lower he firm invesmen, however, a he same ime i also lowers he real value of deb. This gives he firm an incenive o inves as long as he reducion of he real value of deb exceeds he increase in he ineres burden. Firms wih a high leverage may experience a posiive cross-effec of deb holdings and he ineres raes volailiy. While, for firms holding a lower amoun of deb, he benefis from he reducion of he real deb probably are oo low o cover he increase in ineres paymens. To achieve he objecives in his paper, hree hypoheses have been made. The firs hypohesis saes here is a posiive relaionship beween firm invesmen and deb, while here is a negaive relaion beween macroeconomic uncerainy and firm invesmen. I is also hypohesized ha he cross effec of deb and uncerainy on firm invesmen is posiive for high-indebed firms and negaive for low-indebed firms. Consequenly, here exiss a heerogeneous effec of he ineracion beween hese wo variables o firm invesmen for boh groups of firm. This sudy conribues o he empirical evidence in he opic of firm invesmen, deb and uncerainy relaionship paricularly for developing counries. Furhermore, based on he auhor s knowledge, here is no sudy focusing on relaionship beween firm invesmen and cross-effec of uncerainy and deb holdings in developing counries. Specifically, his sudy exens he lieraure in he firm invesmen issue in Malaysia as one small open economy. The findings in his paper can be used as a guideline for he policy makers in Malaysia o make any decisions relaed wih he uncerainy; paricularly uncerainy in aggregae level and firm invesmen. This papersars wih he inroducion and some lieraure review. I is followed by a discussion abou firm invesmen and uncerainy briefly. The model specificaion and esimaion procedure are also emphasized before he discussion abou he findings and conclusion. LITERATURE REVIEW The impac of deb and uncerainy on firms invesmen has been sudied by many researchers. However, many of hese sudies have been focused more on he developed counries such as in he Unied Saes, Unied Kingdom and Neherlands as can be seen in Driver and Moreon (1991), Bo and Serken (00), Byrne and Davis (004) and Baum (010). Theoreically, here are several hypoheses or heories ha can be linked o he relaionship beween deb and firm invesmen. Based on he signaling hypohesis as inroduced by Ross (1977), here is a posiive impac of deb on firm invesmen. The signaling hypohesis saes ha managers are issuing deb because hey are opimisic abou fuure produciviy of firm. Therefore deb issuing signals fuure profiabiliy and i encourages firms o inves more. Myers (1977) inroduced he pecking order heory of financing. This hypohesis saes ha here are hree levels for financing a firm s new invesmen which are inernal funds, o issue deb and as a las resor o issue equiy. This implies ha coss of deb financing for invesmen are higher han inernal funds and i affecs firm invesmen decisions. Besides, if he firm issues deb, hen he deb payback commimen suggess a lower level of liquidiy. In imperfec capial markes, a firm ha has a lower liquidiy will face higher coss of exernal capial, which discourages invesmen. I indicaes a negaive relaionship beween deb holdings and firm invesmen. I can be deduced ha he relaion beween deb and invesmen as ambiguous. There is also he issue of he relaion beween uncerainy and firm invesmen. Early models of a posiive linkage beween invesmen and uncerainy rely on he assumpion ha invesmen is reversible as in Richard (197). Based on his assumpion, as he new informaion is available, he exisence of uncerainy ha affecs marginal produciviy of capial would increase he opimal sock 494

3 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII and also invesmen. I indicaes a posiive relaion beween uncerainy and invesmen. On he oher hand, here is an argumen which saes ha here exiss a negaive relaion beween uncerainy and invesmen as hey assume invesmen as irreversible and here exiss he opion for waiing for examples in Caballero (1991) and Dixi and Pindyck (1994). I can be deduced ha he relaion beween uncerainy and invesmen is also ambiguous. Caballero (1991) proves ha uncerainy has a negaive impac on invesmen if irreversibiliy is assumed in combinaion wih decreasing reurns o scale or imperfec compeiion. A posiive relaion exiss beween uncerainy and invesmens when he firm is assumed o have consan reurn o scale and in he perfec compeiion. I can be summarized ha here are several channels how uncerainy affecs he invesmen which includes he risk aiude agains uncerainy and he non-lineariy of echnology as explained in Driver and Moreon (1991) and Caballero (1991). I is worh emphasizing ha uncerainy and invesmen also has an ambiguous relaionship. I is also worh noing ha here are wo ypes of uncerainy ha are faced by firms o do he invesmen. The firs is macroeconomic or aggregae uncerainy which includes uncerainy in he inflaion raes, he marke ineres raes; for examples reasury bills or governmen bond and exchange raes. Beaudry e al. (001) explained ha macroeconomic uncerainy plays a vial role o affec firm invesmen decisions. They argue ha he sabiliy in inflaion as well as ineres raes improves he efficiency of allocaion of resources. Finally, i allows invesmen o be more effecively channelled o he projecs wih he highes reurns because he bes invesmen opporuniies are more easily idenified. Driver and Moreon (1991), Byrne and Davis (004) and Rashid (011) also find ha macroeconomic uncerainy affecs invesmen negaively in he Unied Kingdom and he Unied Saes. The second is microeconomic or idiosyncraic uncerainy which affecs he firm invesmen hrough he inpu decisions. Ghosal and Loungani (000) examine he impac of profi uncerainy on invesmen using firm daa from he Unied Sae of America. The findings show ha here exiss a negaive impac of profi uncerainy on firm and a negaive impac is subsanially greaer in indusries dominaed by small firms. Bo and Serken (00) and Bo (007) also provide evidence ha firms invesmen is sensiive o idiosyncraic uncerainy. Besides, he sudy in he issue of uncerainy and firm invesmen will be more ineresing when he researcher ries o find he linkage beween he uncerainy and he financing resources for he firms, for example, in Bo and Serken (00). They idenify he ineracion effec beween idiosyncraic uncerainies; measured by he uncerainy in firm ineres raes and deb holdings on firm invesmen. They find ha firm invesmen responds negaively o he cross effec beween idiosyncraic uncerainy and deb holding paricularly for high leverage firms. In fac, many previous sudies have been mosly resriced o he Unied Saes, Unied Kingdom and oher developed counries. Based on he auhor s knowledge, he sudy on his issue for developing counries is sill limied. For example, in macroeconomic or aggregae level, Aizenman and Marion (1999) examine he linkage beween uncerainy and invesmen using aggregae daa for developing counries. They find ha he effec of uncerainy on invesmen migh be more significan in developing counries han in developed marke economy. The macroeconomic volailiy may be higher because producion and rade are less diversified. Moreover, less developed financial markes limi individual agens opporuniy for insuring agains idiosyncraic risk. Besides, since incomplee markes in developing counries may make invesmen less easily reversible, he effec of uncerainy on invesmen may be more marked han in developed counries. In microeconomic level, on he oher hand, Driffield and Pal (001) sudy he behavior of corporae invesmen and financial consrain in four Eas Asian counries which are Indonesia, Korea, Malaysia and Thailand paricularly during he crisis period. Using a daase for he 1990s, his sudy finds ha a large number of firms in he sudy depend on cash flow o finance heir invesmen. Based on he lieraures discussed above, his paper aemps o exen he lieraures on his issue by invesigaing he behaviour of Malaysian lised firms as response o deb and uncerainy. Uncerainy in his paper will be focused on he macroeconomic level insead of he disaggregae level. The moivaion for his paper has been discussed in he inroducion. FIRM S INVESTMENTAND MACROECONOMIC UNCERTAINTY Firm invesmen plays a vial role as a deerminan of aggregae oupu growh. Consequenly, many researchers have been focused on he deerminans of invesmen paricularly he impac of financial facors and uncerainy in he economic environmen. Richard (197) explains ha here exiss a posiive relaion beween firm invesmen and uncerainy paricularly wih he assumpion ha he invesmen is reversible. Moreover, Richard (197) assume ha he firms as risk neural and here is

4 496 Riayai Ahmad uncerainy in each period including he curren period. However, recen lieraure inroduces he concep of irreversibiliy of and he possibiliy o delay he invesmen decisions. Based on his concep, uncerainy affecs firm invesmen negaively for examples in Dixi and Pindyck (1994) and Caballero (1991). Thus, he focus on his paper is he impac of macroeconomic uncerainy paricularly he marke ineres raes on firm invesmen. Besides, he join impac beween macroeconomic uncerainy and deb on firm invesmen will also be focused on. This paper follows Bo and Serken (00) who assumed ha in firm invesmen behavior, he financial srucure of he firm is relevan o he impac of he ineres raes uncerainy on firm invesmen. In he heory, i has been explained ha high inflaion implies volailiy of inflaion and i leads o he uncerainy in he inflaion. As a resul, i also affecs he nominal rae of ineres where a higherineres raes will lead o a higher ineres raes burden. However, he financial srucure of he firm in real erms relies also on he nominal ineres raes and he inflaion volailiy. Inflaion reduces he real value of deb, bu deb holders are compensaed for his by an inflaion premium in he nominal ineres raes hey charge he firm. These higher nominal borrowing coss resul in lower ne income for he firm. The decline in ne income, however, is offse by he decrease in he real value of nominal liabiliies. To keep he real value of is deb consan, he firm will increase is nominal borrowing in he presence of inflaion. The firm will face he rade-off beween he increase in he coss of deb financing and he decrease in he real value of deb. If he magniude of he decrease in he real value of deb is larger han ha in he increase in he ineres paymens, deb capial gain occurs and he firm will inves more. The deb capial gain in he presence of inflaion is more likely o be experienced by firms ha have a higher level of deb. When he firm has a lower level of deb, he increase in he ineres paymens in he presence of inflaion will be higher han he decrease in he real value of deb, which means ha he inernal funds available for invesmen are decreased, leading o a negaive relaionship beween deb and invesmen holding oher hings unchanged. As high inflaion happens, i leads o a higher ineres raes. In oher words, he firm has an incenive o inves as long as he reducion of he real value of deb exceeds he increase in he ineres burden. MODEL SPECIFICATION AND ESTIMATION PROCEDURE The model specificaion in his paper is augmened version of firm invesmen model under uncerainy as in Bo and Serken (00). I can be shown as follows: I K i, I CFLOW DEBT SALE i, K i, 1 K i, K i, 5 IRU i i, (1) Where irepresens he firm invesmen and i is measured by he capial expendiure, K is capial sock for firms which is he ne firm fixed asses excludes depreciaion. However, i includes propery, plan and equipmen. CFLOW shows he cash flow for firms and i is defined as operaing income plus depreciaion. The depreciaion includes oal depreciaion, amorizaion and depleion. CFLOW indicaes he reliance of firms on he inernal sources for funding heir invesmen. DEBT is deb holdings or borrowing of he firms and i is measured by he oal deb. Then, SALE represens he growh of he firm sales while, IRU is he aggregae uncerainy. The aggregae uncerainy has been focused on he marke ineres raes uncerainy and i is measured by he lending rae uncerainy. Uncerainy in he ineres raes is measured using a GARCH model. Besides, he error erm in equaion (1) is also assumed o follow wo way error componens disurbances wih i is a firm specific effec and is a ime specific effec, while i, is he remainder sochasic disurbance erm ha is assumed o be independen and idenically disribued wih mean zero and variance re-wrien as i i i. The error erm can be, while i and presen he firms in he sample and ime, respecively. Based on he equaion (1), 1,, 3 and 4 are expeced o have value bigger han zero. In oher words, he coefficien values for lagged dependen variable, cash flow, deb and growh sales are posiive o influence firms invesmen. On he oher hand, 5 or he coefficiens value of aggregae uncerainy is expeced o affec firms invesmen negaively. Then, equaion (1) can be exended as in equaion () o es empirically he response of firm invesmen on he join impac beween he marke 496

5 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII ineres raes uncerainy and deb holding. I has been done by following Bo and Serken (00) who derive he invesmen model which has been affeced by he ineracion beween ineres rae uncerainy and deb holding; DEBTU IRU DEBT. I K i, I CFLOW DEBT SALE i, K i, 1 K i, K i, 5 IRU 6DEBTU i i, () From he Equaion (), DEBTU represens he win effecs of he marke ineres raes uncerainy and deb holding on firm invesmen. I is expeced ha he coefficien value for his ineracion is a posiive for high-indebed firms while, a negaive sign is prediced for low-indebed firms. The daa used in his paper covers all lised firms from Bursa Malaysia, Malaysia. Daa for he firms have been colleced from Worldscope Full Company Repors in Thompson which can be downloaded from he Daasream daabase ha cover he years 199 o 009. The daa for macroeconomic uncerainy which is he lending rae has been colleced from he Inernaional Financial Saisics (IFS) websie. The lending rae daa covers he period 1990 o 009 on a monhly basis. Afer checking he daa paricularly he availabiliy issue of required daa, he sample in his paper covers 508 firms. However, afer deecing he ouliers using DFITS es, he sample in his paper is only 496 firms excluding firms from financial secor. 1 This paper uses GARCH model o measure aggregae uncerainy in he economy. Themarkes ineres raes will be used (lending rae in Malaysia) which covers he period 1990:1 o 009:1 in monhly basis. Firsly, he exisences of uni roo in he series will be esed using he Augmened Dickey Fuller (ADF) es and Phillips and Perron (PP) es. The null hypohesis ha here exiss a uni roo indicaes he lending rae is nonsaionary, while he rejecion of he null hypohesis shows ha he lending rae series as saionary. To proceed o a GARCH model, he lending rae series have o rejec he null hypohesis which means he series is saionary. Nex, his paper esimaes he aggregae uncerainy using he GARCH model. The GARCH model is inroduced by Bollerslev (1986) who exends he ARCH model developed by Engle (198) o le condiional variance depend on is own lags as well as lags of he squared error. Equaion (3.3) shows he mean equaion for he regression model wih IR is he naural logarihm for firs difference of he lending rae. The lending rae has been calculaed as IR LogIR LogIR1. DU is he dummy variable wih value of 1 indicaes during he financial crisis beween 1997:7 and 1998:9, while value of 0 represens oher periods (no in he financial crisis) and is he error erm. IR 0 1IR 1 DU (3) (4) Equaion (4) indicaes he condiional variance equaion ha has been assumed o follow an auoregressive GARCH (1,1) process. From equaion (4), i can be seen ha dependson 1 ; he lag of he squared residual from he mean equaion and i indicaes news abou volailiy from he previous period (ARCH erm), and also 1 ; he las period s forecas variance (GARCH erm). The firs erm in parenheses in he GARCH (1,1) model refers o he presence of a firs-order auoregressive GARCH erm. The second erm in parenheses refers o a firs-order moving average ARCH erm. Then, his paper proceeds o he nex sep for measuring he marke ineres raes uncerainy. Firs, his paper obains he series of he condiional variance of he marke ineres raes wih monhly observaions. In order o mach hese wih he annual invesmen daa a hand, his paper uses he median of he disribuion of he condiional variance over each 1-monh period as he proxy for he uncerainy of he marke ineres raes for ha year. 1 Based on he DFITS es, 1 firms have been found as ouliers wih 10 of hem are from high-indebed firms group and of hem are from low-indebed firms group. Appendix I explains more deails he heory of DFITS es for deecing he ouliers.

6 498 Riayai Ahmad Generalized Mehod of Momens Esimaor The panel daa se ha has been used in his paper consiss of many or large firms over shor ime period. Besides, i shows ha here exiss he unobservable firm effec which indicaes he heeroscedaiciy across firms ha may be correlaed wih he explanaory variables. Furhermore, here is also he possibiliy ha some of he explanaory variables such as deb o be weak exogenous or endogenous variable. This paper is also augmened he model ha has been derived by Bo and Serken (00) by including he lagged dependen variable as one of he regressor. Thus, i implies ha here is correlaion beween he explanaory variable and he error erm. Based on hose characerisics, his paper esimaes he specificaion model using GMM esimaor as i conrols for simulaneiy bias. Arellano and Bond (1991) propose he firs differencegmm esimaor where his esimaor uses he lagged level of he explanaory variables as insrumens for firs difference variable.the firs difference GMM esimaor is asympoically valid under he assumpion ha he error erm is no serially correlaed and he lagged explanaory variables are weakly exogenous. The firs difference GMM esimaor can be shown as follows: I K I CFLOW DEBT SALE i, i, K i, 1 K i, K i, 5 IRU i i (5) In equaion (5), he operaor indicaes he firs difference of he variables in he equaion. Then, following Arellano and Bond (1991), his paper ses he following momen condiions: I E. i, i, 1 0 for s ; 3,..., T (6) K i, s ECFLOW i, s. i, i, 1 0 for s ; 3,..., T (7) EDEBT i, s. i, i, 0 for s ; 3,..., T (8) ESALE i, s. i, i, 1 0 for s ; 3,..., T (9) EIRU i, s. i, i, 1 0 for s ; 3,..., T (10) The usage of lagged level of explanaory variables as insrumens for addressing he possible I simulaneiy bias of esimaes of he regressor coefficiens and he correlaion beween and K i1 i. However, his paper uses he sysem GMM esimaor ha ha combines he difference equaion (5) and he level equaion (1) as proposed by Arellano and Bover (1995). Moreover, Alonso- Borrego and Arellano (1999) and Blundell and Bond (1998) give addiional evidence by presening ha his esimaor (he sysem GMM) is able o reduce biases and imprecision associaed wih difference esimaor. They also show ha firs difference GMM esimaor suffers wih weak insrumen problem. I can be happen when he regressors are persisen over ime. Besides, if he dependen variable follows a random walk which makes he firs lag a poor insrumen for is difference i conribues o a poenial problem where under cerain condiions he variance of he esimaes may increase asympoically and creae considerable bias. The problem of persisen over ime and random walk explanaory variable happen eiher as he auoregressive parameer approaches uniy or as he variance of he individual effecs increases relaive o he variance of he idiosyncraic error i hen lagged levels of hese variables are weak insrumens for he regression equaion in i, differences. Based on he above argumens, his paper ses he addiional addiional momen condiions for he regression in levels as follows: j 498

7 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII I I E. 0 i i for s 1 K is K is1 (3.11) ECFLOW i s CFLOW i s1. i i 0 for s 1 (3.1) EDEBT i s DEBT i s1. i i 0 for s 1 (3.13) ESALE i s SALE i s1. i i 0 for s 1 (3.14) EIRU i s IRU i s1. i i 0 for s 1 (3.15) The momens condiions presened in Equaion (3.6) o Equaion (3.15) have o be used in he sysem GMM esimaor.an addiional necessary condiion for he efficiency of he sysem GMM esimaor is ha, even if he unobserved counry specific effec is correlaed wih he regressors in levels, i is no correlaed wih heir differences. The condiion also means ha he deviaions of he iniial values of he independen variables from heir long run values are no sysemaically relaed o he counry specific effec. However, he sysem GMM faces he problem of proliferaion in he number of insrumens as discussed in Roodman (009b). I means here are oo many insrumens in he model which lead o misleadingin he some of he asympoic resuls abou he esimaors and relaed ess. However, here is lile guidance in he lieraure o deermine how many insrumens could be considered as oo many. Roodman (009b) recommends ha he number of insrumens should no exceed he number of group (in his paper is firms). Moreover, Calderon e al. (00) and Roodman (009b) propose wo mehods o conrol he number of insrumens in he GMM esimaor. Firs, by using only cerain lags of insrumens insead of all available lags, and second, by collapsing he block of he insrumen marix o reduce he dimensionaliy of he insrumenal variable marix. The GMM esimaors are ypically applied in one-sep and wo-sep varians as explained in Arellano and Bond (1991). The one-sep esimaors use weighing marices ha are independen of esimaed parameers, while he wo-sep GMM esimaor uses he opimal weighing marices in which he momen condiions are weighed by a consisen esimae of heir covariance marix. This makes he wo-sep esimaor asympoically more efficien han he one-sep esimaor. However, Bond (00) poins ou ha he resuls from one-sep esimaor are more favorable han wo-sep resuls. His argumen, based on simulaion analysis, i shows ha he wo-sep GMM esimaion is less efficien when he asympoic sandard error ends o be oo small or he asympoic -raio ends o be oo big. Windmeijer (005) also shows ha he wo-sep GMM esimaion wih numerous insrumens can lead o biased sandard errors and parameer esimaes. However, his paper applies one-sep sysem GMM o esimae he specificaion model. The unbiased, consisen and efficien resuls of GMM esimaor depend on he appropriae adopion of insrumen. On oher words, he GMM resuls depend on he validiy of he insrumens and he model is correcly specified. For ha purpose, wo ess are used in his paper o idenify he validiy of he insrumens adoped in he model. The firs is he Sargan or Hansen es of overidenifying resricions which ess he overall validiy of he insrumens. I is worh noing ha a special feaure of GMM esimaion is ha he number of momen condiions increases wih T. Therefore, a Sargan or Hansen es is performed o es he over-idenificaion resricions. These are esed by analyzing he sample analogue of he momen condiions used in he esimaion process. The es saisic is asympoically disribued as a chi-square wih degrees of freedom equal o he number of insrumens less he number of parameers. A failure o rejec he null hypohesis indicaes ha he insrumens are valid and he model is correcly specified. In oher words, he momen or orhogonaliy condiions hold o accep he null hypohesis. The second es is he serial correlaion es which examines he hypohesis of no second-order serial correlaion for he error erm in he firs difference equaion. Balagi (005) argues ha his es is very imporan because i idenifies he consisency of he GMM esimaor such ha E 0. i i THE EMPIRICAL RESULTS AND ANALYSIS

8 500 Riayai Ahmad In his secion, his paper repors he esimaion resuls of augmened Bo and Serken (00) invesmen model under uncerainy. Firsly, his paper presens he resul for uni roo es for GARCH model. I followed by he main resuls ha have been esimaed using one-sep sysem GMM. The one-sep sysem GMM resuls cover he resuls for he whole sample and followed by wo groups namely, highand low-indebed firms. Uni Roo and GARCH Resuls Table 1 repors he uni roo es for he marke ineres raes; lending rae and i indicaes ha he marke ineres raes rejec he null hypohesis in he level. I indicaes he saionary of marke ineres raes eiher using he ADF or PP ess. Nex, Table reveals he resul for GARCH (1,1) model for he marke ineres raes. In he esimaion of GARCH (1,1) model, his paper pu he dummy variables during he shock period (DU). DU is equal o 1 during he financial crisis while DU = 0 for he oher periods. Based on he resul, i can be seen ha all componens of auoregressive and moving average are significan. Then, his paper proceeds o he nex sep for measuring he marke ineres raes uncerainy as discussed above. GMM Resuls: Firm Invesmen, Deb And Marke Ineres Raes Uncerainy As shown in Table 3 in Panel A, all explanaory variables are significan in affecing firms invesmen in Malaysia wih he signs as expeced excep for deb holdings which has a negaive sign. Sales growh, on he oher hand, gives insignifican impac on firms invesmen. Specifically, i can be seen ha lagged dependen variable (las period s invesmen) gives a significan impac on curren invesmen wih a posiive sign. I indicaes ha las period invesmen deermines curren period invesmen significanly. This suppors previous sudies such as Baum e al. (010). Besides, he coefficien value for cash flow also appears as a posiive and significan. The resul is consisen wih he firm financing and invesmen heory where cash flow plays a vial role as inernal funds for firms o finance heir invesmen. I can be said as 1 percen increase in cash flow will lead o increase in firms invesmen of percen. Furhermore, he small value of coefficien for cash flow indicaes ha firms in Malaysia rely also on he exernal funds o finance heir invesmen. The resul is consisen wih previous sudies such as Sean (006). However, he coefficien value for deb appears as negaive and saisically significan. This finding conradics he expecaion in his paper and i also does no suppor he signalling hypohesis as inroduced by Ross (1977). However, his resul is also no surprising as Bo (007) argues ha here is non-linear relaion beween deb and firm invesmen. I means ha afer a cerain poin he excess level of deb will affec firms invesmen negaively. Lang (1996) also finds ha firms deb affecs he invesmen negaively. As expeced, he aggregae uncerainy which is he marke ineres raes uncerainy affecs firms invesmen negaively and saisically significan. This resul is consisen wih previous sudies such as Bo and Serken (00) and Rashid (011). The negaive effec of marke ineres raes uncerainy on firm invesmen indicaes ha he firms in Malaysia very cauious wih he uncerainy in he ineres raes and hey are no inves more when he uncerainy in he marke ineres raes happen. The resuls in Table 3 (Panel A) are suppored by wo specificaion ess which are he Arellano-Bond for second order serial correlaion and he Sargan/Hansen of over idenificaion ess. The p-value for he second order serial correlaion es indicaes ha he failure o rejec he null hypoheses. I suggess here is no second order serial correlaion problem in he esimaion model. The p-value for he over idenificaion es also shows he failure o rejec he null hypohesis. I indicaes ha he insrumens used in he model are valid and correcly specified. I can be concluded ha he esimaion resuls for he whole sample are srongly suppored by he wo diagnosic ess;he AR() es and he Sargan/Hansen of overindenifying es. Nex, Table 3 in Panel B, on he oher hand, repors he resul for high-indebed firms in Malaysia. I can be seen ha he lagged dependen variable also gives a posiive effec on curren firm invesmen. The cash flow also affecs firms invesmen posiively and saisically significan. Furhermore, he small value of coefficien for cash flow is also consisen wih he resul for he whole sample. I indicaes he high-indebed firms rely on exernal funding which are deb and equiy. Ineresingly, he effec of deb on firms invesmen in his group appears as a negaive and significan. I shows ha he excess level of deb influence he firms invesmen negaively. The marke ineres raes uncerainy gives negaive impac on firms invesmen and i is saisically significan. However, he resul for sales growh remains unchanged wih a posiive sign bu saisically insignifican. The resuls in Panel B are also suppored by wo specificaion ess namely he AR() es and he 500

9 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII Sargan/Hansen es. The AR() es shows he failure of he esimaion model o rejec he null hypohesis which indicaes here is no second order serial correlaion problem in he model. Nex, he Sargan/ Hansen es of over-idenifying resricions also indicaes ha he momen condiions hold in he GMM esimaor such ha he insrumens used in he model are valid. In addiion, Panel C in Table 3 reveals he resul for low-indebed firms in Malaysia. Based on he resuls, i shows ha all explanaory variables play imporan roles o encourage firms o inves more excep for he deb. From he resul, i indicaes ha las period s firm invesmen plays a vial role o deermine curren invesmen wih a posiive sign and saisically significan. I is followed by he cash flow which gives a significan impac on firms invesmen wih a posiive sign. Nex, sales growh also shows a significan effec on firm invesmen, also wih a posiive sign. The marke ineres raes uncerainy appears o give a negaive impac on firm invesmen and saisically significan. The resuls in Panel C are also suppored by he AR() es and he Sargan/Hansen es for over-idenifying of he insrumens used in he model. Boh ess indicae he accepance of he null hypoheses which sugges ha he model is correcly specified and he insrumens used are valid. The Cross Effec Of The Marke Ineres Raes And Deb On Firm Invesmen In Table 4, he main focus in his able is he resul for he cross effec beween he marke ineres raes uncerainy and deb on firm invesmen for he whole sample and wo main groups; namely high- and low-indebed firms. I is worh noing ha he cross-effec of deb and he ineres raes volailiy can be shown from wo channels. Firs, an increase in volailiy will increase he ineres raes burden. Secondly, he higher marke ineres raes volailiy which leads o he uncerainy in he marke ineres raes will likely decrease he real value of deb holdings. Thus, i encourages firm o do more invesmen. According o he resuls in Panel A, i can be seen ha all explanaory variables which are lagged dependen variable, he cash flow, he deb and he marke ineres raes uncerainy affec he firm invesmen significanly and consisen wih he previous resuls. Specifically, he lagged dependen variable and he cash flow affec he firm s invesmen posiively while, he deb and he marke ineres raes uncerainy give negaive effecs on firm invesmen. The deb holding remains o conradic he hypohesis in his paper. However, he resul for sales growh does no significanly affec firm invesmen in Malaysia. Besides, he join effec beween he marke ineres raes uncerainy and deb (DEBTU) is posiive. The posiive impac of he join impac of hese wo variables indicaes ha he firms in Malaysia do more invesmen when hey face he ineres raes uncerainy which means he ineres raes burden is no excess he reducion in he real value of deb. However, he resul is no saisically significan. Panel B in Table 4 reveals he resuls for high-indebed firms in Malaysia. The resuls are consisen wih he resuls in Panel A (in he same able). I can be explained ha he lagged dependen variable, he cash flow, deb and he marke ineres raes uncerainy give significan impac on firm invesmen in Malaysia. The lagged dependen variable and he cash flow give a posiive impac on firm invesmen and here are saisically significan. On he oher hand, he deb and he marke ineres raes uncerainy affec he firm invesmen negaively and saisically significan. The resuls for he sales growh and cross effec beween he marke ineres raes uncerainy and deb holding are insignifican o deermine he firm invesmen in Malaysia. Nex, Panel C in Table 4 presens he resuls for low-indebed firms in Malaysia. The resuls indicae ha all explanaory variables give significan impacs on firm invesmen in Malaysia excep for he cross effec beween he marke ineres raes uncerainy and deb holding. In oher words, i can be explained ha he own effec of deb and he marke ineres raes uncerainy are negaive in influencing firm invesmen and are saisically significan a leas a he 10 percen level. The resul for sales growh indicaes he significan impac on firm invesmen a leas a 10 percen as well. Oher explanaory variables which are lagged dependen variable and he cash flow consisenly give significan impacs on firm invesmen in Malaysia. However, he cross effec beween deb and he marke ineres raes uncerainy shows an insignifican impac on firm invesmen. All resuls in Table 4 are suppored by wo specificaion ess o idenify he validiy of he insrumens adoped in he models. Firs, he AR() es indicaes he failure o rejec he null hypohesis which means he consisen esimaes such ha E i i 0 for all panels. Second, he Sargan/Hansen es for over-idenifying resricion also shows ha he momen condiions hold in he GMM esimaor which indicaes ha he insrumens used in he models are valid. I is worh emphasizing ha he resuls in Table 4 indicae ha he join impac beween he marke ineres raes uncerainy and deb holding is insignifican o deermine he firm invesmen in Malaysia for all panels. I migh be he firms in Malaysia more sensiive wih he firm specific ineres raes o deermine heir invesmen. On he oher hand, he oher explanaory variables indicae he

10 50 Riayai Ahmad imporan influence on firm invesmen. Ineresingly, he sales growh is only significan o deermine firm invesmen in low-indebed firm group while, he resul for high-indebed firm and he whole sample is insignifican. Besides, all coefficiens for he cash flow variable indicae he value less han one and relaively low. I also shows ha he firms invesmen relies on he inernal funds o finance heir invesmen. The deb holding gives a negaive effec on firm invesmen and i conradics he hypohesis in his paper as well as he signalling hypohesis. As expeced as well, he marke ineres raes uncerainy affecs he firm invesmen negaively. This finding is consisen wih previous sudies such as Bo and Serken (00), Baum e al. (010), Rashid (011) and Driver and Moreon (1991) who found ha invesmen responds negaively o macroeconomic uncerainy. CONCLUSION The ambiguous effec of deb and uncerainy on firm invesmen has been largely documened previously and empirically has been focused on he developed counries. The main objecive in his paper is o examine he relaion beween uncerainy, deb and firm invesmen in developing counries specifically in Malaysia. Macroeconomic uncerainy has been focused on his sudy and i is proxied by he marke ineres raes uncerainy. Besides, he focus is also given on he join impac beween he marke ineres raes uncerainy and deb holding on firm invesmen. Finally, his sudy is also carried ou o idenify he heerogeneous response of wo firm groups; namely high and low-indebed firms The resuls indicae ha macroeconomic uncerainy affecs firm invesmen negaively, wheher i is proxied by he marke ineres raes uncerainy. The consisen resuls hold in all esimaions, boh for he whole sample or by spliing he sample ino wo groups; namely high- or lowindebed firms. These findings are also consisen wih previous sudies in developed counries such as in Bo and Serken (00), Byrne and Davis (004), Beaudry e al. (001) and David and Moreon (1991). Furhermore, a negaive relaion beween macroeconomic uncerainy and firm invesmen suppors he recen heory of invesmen under uncerainy such as in Caballero (1991) and Dixi and Pindyck (1994). The resuls for deb holding, on he oher hand, conradic he hypohesis in his paper. This paper hypohesized ha here exiss a posiive relaion beween deb and firm invesmen. However, he findings show ha a negaive relaion beween hem. These findings do no suppor he signalling hypohesis as proposed by Ross (1977). I is also no surprising as Bo (007) argues ha he relaion beween deb and firm invesmen is non-linear. I means, he deb affecs firm invesmen posiively a he firs sage. However, if he deb exceeds a cerain value, i affecs firm invesmen negaively. These resuls are also consisen wih Lang e al. (1996) who sudy he effec of deb on firm invesmen in developed counries. The final focus in his paper is on he join impac beween macroeconomic uncerainy and deb holding on firm invesmen. The resuls show ha he cross-effec beween hese wo variables on firm invesmen as insignifican. I indicaes he own effec of hese wo variables is more imporan han he ineracion effec. Ineresingly, he resuls show ha he impac of macroeconomic uncerainy and deb on firm invesmen is quie similar for boh groups. Boh groups respond negaively o he macroeconomic uncerainy as well as o he deb. The join effec of macroeconomic uncerainy and deb on firm invesmen is also no significan for hese wo groups. I indicaes here are no heerogeneous resuls beween hese wo groups. I is no surprising as all firms in he sample have been colleced from lised companies. The heerogeneiy is expecing when he sample consiss of he firms from lised and unlised companies. I can be concluded ha firm invesmen is known o play a vial role o foser he aggregae oupu growh in he counry. Thus, i is imporan o sudy he deerminans of firm invesmen paricularly invesmen under uncerainy. The findings in his paper have implicaions o firms as invesors and also o he policy makers. For firms, hey have o aler wih he uncerainy in he economy as uncerainy mos probably affec heir invesmen negaively. I leads o decrease heir reurn and profis as well as heir posiion in he business. For governmen and he policy makers, hey should pay more aenion on he firm invesmen behaviour paricularly under uncerainy. The sabiliy in he counry which includes he sabiliy in he marke ineres raes lead o he efficiency of allocaion of resources. Thus, i allows he invesmen o be more effecively channelled o he projecs wih he highes reurns. I is only happens when he bes invesmen opporuniies are easily idenified in he sable economic environmen. Appendix I 50

11 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII Deecing he ouliers All daa in his sudy are colleced from Worldscope Full Company Repors in Thompson which can be downloaded from Daasream daabase. The daa covers he period from 1990 o 009 which cover lised firms in Malaysia. Afer filering he daa paricularly he availabiliy issue for he required daa, his paper uses unbalanced panel daa as he sample wih he minimum period is 5 years. The firs analysis in his paper is o deec he exisence of poenial ouliers in he sample. This has been done using he DFITS es as proposed by Belsley e al. (1980). This saisic idenifies observaions wih a high combinaion of (saisical) leverageand residual. The saisic is given by DFITS r i 1 h i h i / where r i is sudenized residual given by r i s ( i) e 1 h refers o he roo mean squared error s of he regression equaion wih he ih observaion removed, and h is he leverage saisic. Belsley e al. (1980) sugges ha a cuoff value of DFITS i k n indicaes highly influenial observaions. On oher words, an observaion is considered as an oulier if he absolue DFITS saisic is greaer han variables and n is he number of firms. k / n i wih s i, where k depics he number of explanaory REFERENCES ABEL, A. B. & EBERLY, J. C The effecs of irreversibiliy and uncerainy on capial accumulaion. Journal of Moneary Economics, 44, AIVAZIAN, V. A., GE, Y. & QIU, J The impac of leverage on firm invesmen: Canadian evidence. Journal of Corporae Finance, 11, AIZENMAN, J. & MARION, N. P Macroeconomic uncerainy and privae invesmen. Economics Leers, 41, ARELLANO, M. & BOND, S Some Tess of Specificaion for Panel Daa: Mone Carlo Evidence and an Applicaion o Employmen Equaions. The Review of Economic Sudies, 58, ARELLANO, M. & BOVER, O Anoher look a he insrumenal variable esimaion of errorcomponens models. Journal of Economerics, 68, BAUM, C. F., CAGLAYAN, M. & TALAVERA, O On he invesmen sensiiviy of deb under uncerainy. Economics Leers, 106, 5-7. BEAUDRY, P., CAGLAYAN, M. & SCHIANTARELLI, F Moneary Insabiliy, he Predicabiliy of Prices, and he Allocaion of Invesmen: An Empirical Invesigaion Using U.K. Panel Daa. The American Economic Review, 91, BLUNDELL, R. & BOND, S Iniial condiions and momen resricions in dynamic panel daa models. Journal of Economerics, 87, BO, H Nonlinear Effecs of Deb on Invesmen: Evidence from Duch Lised Firms. The European Journal of Finance, 13, BO, H. & STERKEN, E. 00. Volailiy of he ineres raes, deb and firm invesmen: Duch evidence. Journal of Corporae Finance, 8, BOLLERSLEV, T A Condiionally Heeroskedasic Time Series Model for Speculaive Prices and Raes of Reurn. The Review of Economics and Saisics, 69, BYRNE, J. P. & DAVIS, E. P Permanen and emporary inflaion uncerainy and invesmen in he Unied Saes. Economics Leers, 85, CABALLERO, R. J On he Sign of he Invesmen-Uncerainy Relaionship. The American Economic Review, 81, CABALLERO, R. J. & PINDYCK, R. S Uncerainy, Invesmen, and Indusry Evoluion. Inernaional Economic Review, 37, CALDERON, C. A., CHONG, A. & LOAYZA, N. V. 00. Deerminans of curren accoun deficis in developing counries. The BE Journal of Macroeconomics,, CARRUTH, A., DICKERSON, A. & HENLEY, A Wha do We Know Abou Invesmen Under Uncerainy? Journal of Economic Surveys, 14, DIXIT, A. & PINDYCK, R Invesmen under Uncerainy, Princeon Universiy Press.

12 504 Riayai Ahmad DRIFFIELD, N. & PAL, S The Eas Asian crisis and financing corporae invesmen: is here a cause for concern? Journal of Asian Economics, 1, DRIVER, C. & MORETON, D The Influence of Uncerainy on UK Manufacuring Invesmen. The Economic Journal, 101, ENGLE, R. F Auoregressive Condiional Heeroscedasiciy wih Esimaes of he Variance of Unied Kingdom Inflaion. Economerica, 50, GHOSAL, V. & LOUNGANI, P The Differenial Impac of Uncerainy on Invesmen in Small and Large Businesses. Review of Economics and Saisics, 8, LANG, L., OFEK, E. & STULZ, R Leverage, invesmen, and firm growh. Journal of Financial Economics, 40, 3-9. MYERS, S. C Deerminans of corporae borrowing. Journal of Financial Economics, 5, PINDYCK, R Capial risk and models of invesmen behaviour. Sloan School of Managemen. RASHID, A How does privae firms' invesmen respond o uncerainy? some evidence from he Unied Kingdom. The Journal of Risk Finance, 1, RICHARD, H The effecs of price and cos uncerainy on invesmen. Journal of Economic Theory, 5, ROODMAN, D. 009b. A Noe on he Theme of Too Many Insrumens. Oxford Bullein of Economics and Saisics, 71, ROSS, S. A The Deerminaion of Financial Srucure: The Incenive-Signalling Approach. The Bell Journal of Economics, 8, SEAN, C Inernaional corporae invesmen and he relaionships beween financial consrain measures. Journal of Banking & Finance, 30, TABLE 1:UniRoo Tes Resul Variable ADF PhillipsPerron level 1s Dif. level 1s Dif IR *** *** *** *** The null hypohesis is H o: presence of he uni roo and H a :saionariy of he series *** indicaes ha he rejecion of he null hypohesis for presence he uni roo and i is significan a 1 percen level. TABLE 1: The Resul For A GARCH (1,1) Model For The Marke Ineres Raes Coefficien Sd. Error Prob. IR(-1) *** DU *** C ** Variance Equaion C 4.16E E *** *** *** *** and ** indicae he significance of he variables a 1 and 5 percens, respecively. 504

13 Prosiding Persidangan Kebangsaan Ekonomi Malaysia Ke VIII TABLE Error! No ex of specified syle in documen.: Firms Invesmen Model Wih The Marke Ineres Rae Uncerainy Panel A Panel B Panel C Explanaory variables Whole High-Indebed Low-Indebed coefficien robus s.er P-value coefficien robus s.er P-value coefficien robus s.er P-value CONSTANT LDV CFLOW DEBT SALE IRU AR() p-value Hansen es p-value No. of observaions No. of firms Noe:The resuls in Table 3.3 have been obained by esimaing he model specificaion as in equaion (3.1). The insrumens have been assumed as follows: lagged dependen variable (las period invesmen o capial raio), CFLOW and DEBT as endogenous variables, while SALE and IRU as predeermined variables. The esimaion is carried ou by conrolling he number of lag of insrumens and collapsing he insrumen marix as proposed by Roodman (009b) and Calderon e al. (00). Year dummies are no repored in order o save space.