E069-10-TBR-Vol 2 Issue 1 0310university2.taylors.edu.my/tbr/uploaded/2012_vol2_issue1.pdf · 2020. 6. 2. · A Contemporary Business Journal Vol. 2 Issue 1 February 2012 A Publication

A Contemporary Business JournalVol. 2 Issue 1 February 2012

A Publication of Taylor’s Business School, Taylor’s University Sdn Bhd

E069-10-TBR-Vol 2 Issue 1-0 Contents_0210.pmd 03-Oct-2012, 9:31 AM1

MEMBERSMEMBERSMEMBERSMEMBERSMEMBERS

Emeritus Professor PatrickEmeritus Professor PatrickEmeritus Professor PatrickEmeritus Professor PatrickEmeritus Professor PatrickHutchinsonHutchinsonHutchinsonHutchinsonHutchinsonUniversity of New England (UNE)Email: [email protected]

PrPrPrPrProfofofofofessor Dressor Dressor Dressor Dressor Dr..... Mansor H. Mansor H. Mansor H. Mansor H. Mansor H. Ibr Ibr Ibr Ibr IbrahimahimahimahimahimProfessor of Finance & EconometricsDepartmentINCEIF The Global University of IslamicFinanceEmail: [email protected]

PrPrPrPrProfofofofofessor Meressor Meressor Meressor Meressor Mervyn Levyn Levyn Levyn Levyn LewiswiswiswiswisUniversity of South AustraliaEmail: [email protected]

Associate Professor Sanjay GoelAssociate Professor Sanjay GoelAssociate Professor Sanjay GoelAssociate Professor Sanjay GoelAssociate Professor Sanjay GoelUniversity of MinnesotaEmail: [email protected]

Associate Professor Puja PadhiAssociate Professor Puja PadhiAssociate Professor Puja PadhiAssociate Professor Puja PadhiAssociate Professor Puja PadhiIndian Institute of Technology, BombayEmail: [email protected]

Dr Peter NichollsDr Peter NichollsDr Peter NichollsDr Peter NichollsDr Peter NichollsUniversity of West EnglandEmail: [email protected]

Dr CaDr CaDr CaDr CaDr Catherthertherthertheryn Khoo-Layn Khoo-Layn Khoo-Layn Khoo-Layn Khoo-LattimorttimorttimorttimorttimoreeeeeTaylor’s University, MalaysiaEmail:[email protected]

MANAGING EDITORMANAGING EDITORMANAGING EDITORMANAGING EDITORMANAGING EDITOR

Sheela Devi D SundarasenSheela Devi D SundarasenSheela Devi D SundarasenSheela Devi D SundarasenSheela Devi D SundarasenAssociate Dean (Research & Development)Taylor’s Business SchoolTaylor’s University, MalaysiaLakeside CampusEmail:[email protected]

EDITEDITEDITEDITEDITORIAL BOORIAL BOORIAL BOORIAL BOORIAL BOARD MEMBERSARD MEMBERSARD MEMBERSARD MEMBERSARD MEMBERS

DisclaimerThe selection and presentation of materials and the opinions expressed are the sole responsibility of theauthor(s) concerned. Statements made by authors do not imply endorsement or agreement by theEditor-in-Chief, the Editorial Board or Taylor’s University Sdn. Bhd.

Printed byAslita Sdn Bhd20, Jalan 4/10B, Spring Crest Industrial Park, 68100 Kuala Lumpur.

EDITOR-IN-CHIEFEDITOR-IN-CHIEFEDITOR-IN-CHIEFEDITOR-IN-CHIEFEDITOR-IN-CHIEF

Dr Ong Fon SimDr Ong Fon SimDr Ong Fon SimDr Ong Fon SimDr Ong Fon SimProfessor and Head of Department (Marketing)

Taylor’s UniversityEmail: [email protected]


ContentsContentsContentsContentsContents

DeterDeterDeterDeterDeterminants ofminants ofminants ofminants ofminants of the R the R the R the R the Relaelaelaelaelationship betwtionship betwtionship betwtionship betwtionship between Fireen Fireen Fireen Fireen Firm Lem Lem Lem Lem Levvvvverererereraaaaagggggeeeee 11111and Cash Flow: Evidence from Malaysian Companiesand Cash Flow: Evidence from Malaysian Companiesand Cash Flow: Evidence from Malaysian Companiesand Cash Flow: Evidence from Malaysian Companiesand Cash Flow: Evidence from Malaysian CompaniesMarina Mustapha & Ng Huey Chyi

Financial InteFinancial InteFinancial InteFinancial InteFinancial Integggggrrrrraaaaation and Intertion and Intertion and Intertion and Intertion and Internananananational Cational Cational Cational Cational Capital Mobility:pital Mobility:pital Mobility:pital Mobility:pital Mobility: 1717171717Evidence from ASEANEvidence from ASEANEvidence from ASEANEvidence from ASEANEvidence from ASEANMuzafar Shah Habibullah, Sarinder Kumari & Baharom, A.H.

FirFirFirFirFirm Heterm Heterm Heterm Heterm Heterooooogggggeneity and eneity and eneity and eneity and eneity and TTTTTrrrrrade Barade Barade Barade Barade Barrierrierrierrierriers – s – s – s – s – AnalAnalAnalAnalAnalysis ofysis ofysis ofysis ofysis of India’ India’ India’ India’ India’s s s s s TTTTTrrrrradeadeadeadeade 3333333333Archana Srivastava, Bikash Ranjan Mishra & Somesh Kumar Mathur

Measuring SerMeasuring SerMeasuring SerMeasuring SerMeasuring Service Quality in Luxurvice Quality in Luxurvice Quality in Luxurvice Quality in Luxurvice Quality in Luxury Hotels in Malay Hotels in Malay Hotels in Malay Hotels in Malay Hotels in Malaysiaysiaysiaysiaysia 4747474747Abaeian, V. & Khong, K. W.

Using Conjoint Analysis to Establish Consumer PreferencesUsing Conjoint Analysis to Establish Consumer PreferencesUsing Conjoint Analysis to Establish Consumer PreferencesUsing Conjoint Analysis to Establish Consumer PreferencesUsing Conjoint Analysis to Establish Consumer Preferences 6161616161for Prawns in Malaysiafor Prawns in Malaysiafor Prawns in Malaysiafor Prawns in Malaysiafor Prawns in MalaysiaAhmad Hanis Izani Abdul Hadi, Mad Nasir Shamsudin,Alias Radam & Jinap Selamat

Adoption ofAdoption ofAdoption ofAdoption ofAdoption of Islamic Banking Pr Islamic Banking Pr Islamic Banking Pr Islamic Banking Pr Islamic Banking Products and Seroducts and Seroducts and Seroducts and Seroducts and Services amongvices amongvices amongvices amongvices among 7373737373Non-Muslim Consumers in MalaysiaNon-Muslim Consumers in MalaysiaNon-Muslim Consumers in MalaysiaNon-Muslim Consumers in MalaysiaNon-Muslim Consumers in MalaysiaLiew C. S. & Leong K.W.

A Contemporary Business JournalVol. 2 Issue 1 February 2012


Determinants of the Relationship between Firm Leverage and Cash Flow

Taylor’s Business Review Vol. 2 Issue 1 February 2012 1

DeterDeterDeterDeterDeterminants ofminants ofminants ofminants ofminants of the R the R the R the R the Relaelaelaelaelationship betwtionship betwtionship betwtionship betwtionship between Fireen Fireen Fireen Fireen FirmmmmmLeverage and Cash Flow: Evidence from MalaysianLeverage and Cash Flow: Evidence from MalaysianLeverage and Cash Flow: Evidence from MalaysianLeverage and Cash Flow: Evidence from MalaysianLeverage and Cash Flow: Evidence from Malaysian

CompaniesCompaniesCompaniesCompaniesCompanies

Marina Mustapha* & Ng Huey Chyi**Taylor’s University, Malaysia

AbstrAbstrAbstrAbstrAbstract:act:act:act:act: A firm’s investment activities are believed to be affected by its leverage andcash flow relationship. Previous empirical literature has so far identified a positiveleverage-cash flow relationship under the signaling theory and a negative leverage-cash flow relationship under the pecking order theory. This paper seeks to examinethe determinants of firm leverage-cash flow relationship among companies inMalaysia. Theories have been reviewed to identify influential factors and furthertested on 100 FBMKLCI multi-sector companies in the Malaysian bourse. Firms’cash flow, liquidity, profitability, tangibility and growth represent the independentvariables in the research model. Our study demonstrates that there is no significantrelationship between leverage and cash flow. This might be due to different financingstyles in developing countries, the effect of ease in accessing external funds with lowcost, or firms’ financial flexibility to issue new debts rather than being dependent oninternal funds. While firms’ tangibility and dividend payout does not affect firms’leverage level, firms’ investment suggests a significant positive relationship withleverage only within the small firm size group. Firms’ profitability, liquidity andgrowth are also key predictors for leverage. Our results further support the peckingorder theory, where negative liquidity-leverage relationship suggests firms with highliquid assets prefer to use internally generated funds to finance their investmentactivities.

KKKKKeeeeey wy wy wy wy wororororordsdsdsdsds: Firm leverage, cash flow, liquidity, profitability, tangibility, MalaysiancompaniesJEL classification: G3

Vol 2 Issue 1 February 2012pp. 1-15

ISSN: 2232-0172

A Contemporary Business Journal

* Marina Mustapha, Taylor’s Business School, Taylor’s University, Lakeside Campus, 47500 Subang JayaSelangor, Malaysia. Email: [email protected]

**Ng Huey Chyi, Taylor’s Business School, Taylor’s University, Lakeside Campus, 47500 Subang JayaSelangor, Malaysia. Email: [email protected]

E069-10-TBR-Vol 2 Issue 1-1 Determinants Firm.pmd 03-Oct-2012, 9:47 AM1

Marina Mustapha & Ng Huey Chyi

Taylor’s Business Review Vol. 2 Issue 1 February 20122

1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION

The non-universal theories of capital structure have received much attention in thefields of corporate finance and financial economics for decades. Empirical evidencefrom studies on capital structure cannot be generalised with the reason being theconditional factors that are attached to each specific theory and presumptions ofperfect capital markets. Common conditional capital structure theories include trade-off, pecking order, free cash flow and signaling theories. The trade-off theoriesemphasise debt-tax relationship against the cost of financing, while the free cashflow theory signals high agency costs that lead to firms’ over-investment behaviour.The increase in value is achieved through their cash flow that surpasses positiveinvestment opportunities. Firms’ precariously take a high risk by engaging themselvesin high debt levels. Subsequently, the other two aspects of capital structure areassociated with asymmetric information that will explain the firms’ leverage-cashflow relationship. Though, the directions of the relationship contradict one another,yet both are well-supported by vast literature and are believed to be further reconciledusing contemporaneous and inter-temporal elements (Shenoy & Koch, 1996). Thepositive relationship is supported by the signaling theory that accounts for leverageto event changes analysis, whereas, the negative relationship is supported by thepecking theory that is based on cross-sectional analysis.

Regardless of the diverse empirical evidence, capital structure studies primarilyseek to explain firms’ financial tactics, as well as the financial decisions on investmentactivities. Hence, financing matters for most corporations and their investmentbehaviour is dependent upon the availability of internal funds and leverage levels.In addition, explicit transaction costs that affects leverage (Strebulaev, 2007;Shivdasani & Stefanescu, 2010; Faulkender, Flannery, Hankins & Smith, 2012)warrant firms to have leverage targets (Altinkilic & Hansen, 2000; Leary & Roberts,2005). In relation to this, a significant number of studies further reveal that firms’leverage adjustment costs are also influenced by other factors that demand accessto capital markets. Firms would raise external funds to finance promising investmentsthrough debt or equity issuances, and generating cash beyond positive investmentopportunities. In contrast, leverage can be adjusted to repay debts and pay dividends.Hence, there appears to be joint effects of adjustment costs and cash flows onleverage adjustments, which can be integrated with adjustment timing (Faulkender,Flannery, Hankins & Smith, 2012). Changes in market conditions also affect leverageadjustments, where a high market-to-book value means a decline in next year’samount of debt, without any significant equity changes (Frank & Goyal, 2004).

Other independent debt-related factors that have also been included byresearchers in their capital structure model, apart from cash flow are liquidity,profitability, tangibility and growth. Most capital structure research concentrates onlarge public, non-financial firms with access to global markets. Indeed, Marsh (1982)




shows that larger firms are more likely to issue more debt. These companies areexposed to both internal and external funds, as well as, the timing flexibility andlower costs attached to adjust their capital structure for investments. Nevertheless,superficial knowledge and understanding of relevant issues are still prevalent.Furthermore, complications and arguments are inevitable due to variation ofrelationships, such as, between debt and equity financing, or even debt and cashflow, within similar industries. Besides, the presence of variation over time is alsoevidenced, though with constant influential financial tactics.

2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW

Firms’ capital structure analyses raise issues on a firm’s incentive to target specificleverage ratios. An adjustment cost emerges even with a slow adjustment speed(Leary & Roberts, 2005; Strebulaev, 2007; Faulkender et al., 2012). The incidenceof leverage adjustment occurs at times of high adjustment benefits, as well as lowadjustment cost. Regardless of positive or negative operating cash flows, largefirms are more aggressive in changing their capital ratios. According to Faulkenderet al. (2012), costs of assessing external capital significantly affect leverage. Specifically,firms with high cash flows and high leverage deviations take on larger capital structureadjustment, as compared to other firms with an almost zero cash flow realisation.Likewise, unconstrained firms adjust faster than those constrained firms within anunder-leveraged condition and vice versa within an over-leveraged condition. Thus,fluctuations in adjustment benefits and costs are the consequence of changespertaining to the leverage gap, cash flow, investment opportunities, access to externalfinancing, profitability and market conditions that might change stock prices(Hovakimian, Opler & Titman, 2001).

Debates on financing corporate investment gave rise to the exploration andimportance of leverage-cash flow relationship studies, with asymmetric informationas the underlying influential factor. The above leverage-cash flow empirical evidenceis derived through two famous strands of literatures, known as the signaling andpecking order theories. However, there is hardly any research work that givesemphasis on both models simultaneously. Event studies on signaling theory, whichreflects inter-temporal aspects (Shenoy & Koch, 1996), act upon market expectationsof future cash flow as the heart of research, while pecking order theories focusprimarily on contemporaneous aspects of the cross-sectional studies. Theysuccessfully reconcile both contradictory theories in a single research model. Theirresearch results are similar with Harris & Raviv (1991), as well as Ravid & Sarig(1991) that suggest high current leverage is related to high future cash flow. Managerswith asymmetric information tend to make current financial decisions on investmentactivities by predicting the investment future payoffs. Earlier studies, however, donot support this hypothesis (Cornett & Travlos, 1989; Copeland & Lee, 1991).




However, pecking order behaviour generally indicates a negative leverage-cash-flow relationship (Ross, 1977 cited in Shenoy and Koch, 1996; Harris & Raviv,1991). Managers here are being more skeptical in that they depend on availablecash flow to support their current financing needs. Only when cash flow isinsufficient, would they proceed with alternative measures by issuing debt or equity.Initially, issuing equity measures were not so popular among established firms dueto asymmetric information being ignored (Baskin, 1989). Firms will only borrowwhen their investment needs are greater than the expected inelastic supply of retainedearnings. Even with asymmetric information incidence, equity financing is still limitedbecause this type of financing is contingent upon the unpredictable future value ofthe security. Consequently, debts become firms’ foremost source of funding. Withits preference ranking of financing sources, pecking order theory has been knownamong some researchers as a good descriptor of corporate or financing behaviour(Shyam-Sunder & Myers, 1999; Frank & Goyal, 2003), though others criticise thisview as being based on either an undersimplified or oversimplified version oftheir statistical models (Jung, Kim & Stulz, 1996; Fama & French, 2005). Leary &Roberts (2010) extended similar research where their finding converges towardsthe tradeoff theory when a firm’s debt is allowed to vary with variables of mixedtheories in their model. They further claim that pecking order behaviour ismore incentive-conflict driven, rather than information-asymmetry driven.

Over and above these findings, the pecking order theory does not providesufficient support for capital structure policies in Europe (Gaud, Hoesli & Bender,2007). Apart from the operating and market performance, the capital structure isbeing affected by national environments. Firms are considerably in good positionsince they are not in their maximum debt boundary, and face minor suffering evenif they are below the target leverage level. Internal financing is preferable overexternal financing to avoid conflicts, while paying higher dividends are observedto be a better measure than reducing debts, for profitable firms. The sameresearchers further demonstrate that investment projects with good future prospectsare funded through equity issuance. Equally, firms will issue debt and increasedividends when projects’ returns are unpredictable. Other empirical studies thattake into account measures of firms’ profitability or earnings (Vogt, 1992), debt,capital structure and dividends (Jensen, Solberg & Zorn, 1992) show comparablenegative relationships despite using different methodologies.

Research in UK done by Bennett & Donnelly (1993) used both short-termdebt and long-term debt in relation to firm size, profitability and asset structure.However, they found that size and profitability are significant in determining long-term debt but not the short-term debt. A study done by Singhania & Seth (2010)on USA, Germany, Japan, Italy and India report that firm growth and size havesignificant positive relationship with a firm’s debt. Similarly, Baskin (1989) showed




that leverage is positively related to past growth, while negatively related to pastprofits. High paying dividend firms in the past also seem to borrow more.Additionally, large firms have lesser financial distress problems and their previousyear profits are negatively related to debts. In support of a similar view is the workdone by Rajan & Zingales (1995) and Wald (1999). Their results indicate the presenceof a significant negative relation between firm’s profitability and their debt ratio inUSA, UK and Japan. A profitable firm is associated with high growth andinvestment opportunities. A spare capacity of borrowing power exists as the firmcan issue additional debts to invest. Ho, Lam & Sami (2004), in accord with Singhania& Seth (2010) also reveal that growth companies in Hong Kong have lower debtequity ratios. On top of these, debt-liquidity and debt-interest coverage ratiorelationships exhibit a similar trend.

Last, but not least, the effect of asset liquidity on leverage has been widelyresearched and debated for many years. According to a study by Sibilkov (2009)on public companies in the US, which results are in accord with Shleifer & Vishny(1992), higher asset liquidity increases leverage. Stronger asset liquidity-leverage isstronger for firms with smaller fixed assets-debt ratio and those with high probabilityof defaults. In addition, a similar positive relationship is found between asset liquidityand secured debt. Further analysis, conversely, reveals a curvilinear asset liquidity-unsecured debt relationship. However, the effect is dependent upon managers’discretion in disposing such assets (Myers & Rajan, 1998). Hence, higher asset liquidityis attached to cheaper sales of assets and value diversion from bondholders. Thistype of managerial control leads to a reduction in agency problems. Adjustmentsof capital structure are then claimed to be irregular since hefty leverage adjustmentcosts force firms to diverge from their targeted leverage ratios (Welch, 2004;Strebulaev, 2007). The expected costs of distress are considerably economicallysizeable and substantial as compared to leverage adjustment costs and the benefitsof debt. This finding might well-explain the reason why bankrupt firms incur lowleverage ratio (Ju, Parrino, Poteshman & Weisbach, 2005).

3. METHOD3. METHOD3. METHOD3. METHOD3. METHOD

We extracted our data for this study from the Malaysian bourse. Our samplesconsist of the top 20 and bottom 20 performers of the FTSE 100-index. Firmsfrom financial and REIT industries, however, were excluded due to substantialgovernment intervention. These 40 firms are large firms from various industriesand they have been purposely selected because large firms typically have moreinvestment opportunities and higher ability to issue additional debts or equity tofinance investment activities, as compared to small firms. Thus, the small firmswere excluded to avoid any ambiguous results. In our study, we also focused on




the effect of firm size on leverage and cash flow relationship between and amongfirms of the two performance categories. All data over the period of 2005-2007were extracted from the respective firms’ annual reports. Only reports with completedata were included. Specifically, data before the 2008 global crisis were used toensure the consistency of analysis. Our final sample consisted of a balanced panelof 19 companies and 57 observations.

This paper seeks to analyse the relationship between firms’ leverage and cashflow of the firms. We had also included factors such as liquidity, tangibility,profitability and growth into our regression as follows:

Dit = b1 Iit + b2 CFit + b3 CSi,t-1 + b4 Qi,t-1 + b5 Pit + b6 TANGit + b7 CRit + b8 DIVit + + eit

Dit denotes debt ratio which is the firm’s leverage level; b1 Iit denotes firm’sinvestment for the year; b2 CFit denotes cash flow and b3 CSi,t-1 denotes cash stock atbeginning of the year representing the internal cash available in the firms. b4 Qi,t-1denotes Tobin’s Q which represents firm’s growth and investment opportunitiesof the firm within the year; b5 Pit denotes firm’s profitability; b6 TANGit denotestangibility; b7 CRit denotes current ratio; b8 DIVit denotes dividend payout ratio;and eit denotes the error term.

There are two possibilities where firms with high leverage will have lowerinvestment due to lack of funds. On the other hand, firms may issue more debtsto finance additional investment activities which would bring higher returns. Thus,investment can be an important predictor here. Cash flow and cash stock representthe internal cash available in the firm and it can be used for any of the firm’sactivities. It is believed that Tobin’s Q represents a firm’s investment opportunityand future growth. High Tobin’s Q affects the firm’s issuing of new debts foradditional investment activities.

The firm’s liquidity will be proxied by current ratio which shows the firm’sability to pay back its debts. In the traditional trade-off theory, it is believed thatfirms with high current ratio should increase their leverage due to their low risk ofdefault. On the other hand, pecking order theory argues a negative relationshipbetween liquidity and leverage as firms with high liquid assets will prefer to useinternally generated funds to finance their investment activities. Firms may pay outdividends to shareholders when they have free cash flow. Thus, the internal cashflow is not available anymore for firms’ other activities like investment and firmsneed to obtain additional debts. Firm size is not included in this regression, becausewe will run the regression analysis based on market capitalisation, sales value andtotal assets size. The computation of the variables is shown in Table 1.




4. DISCUSSION4. DISCUSSION4. DISCUSSION4. DISCUSSION4. DISCUSSION

Table 2 shows the sample sizes and descriptive information on the firms’ data ofthe nine variables. The sample mean debt ratio is equal to 0.36 (s.d. = 0.19); samplemean investment is equal to 0.10 (s.d. = 0.25); the sample mean cash flow is equalto 0.20 (s.d. = 0.22); the sample mean current ratio is equal to 3.35 (s.d. = 3.30); thesample mean Tobin’s Q is equal to 3.26 (s.d. = 2.87) and the sample mean dividendpayout ratio is equal to 0.43 (s.d. = 0.31). Current ratio has the highest variability ofdata distribution while profitability has the lowest variability of data distribution.

Table 3 shows the correlation coefficient of all dependent and independentvariables. There is a strong negative relationship between leverage and current ratioof -0.517 with a p-value of 0.000. There is another strong positive relationshipbetween investment and cash flow with a correlation coefficient of 0.657 and a p-value of 0.000 < 0.01. Cash stock has many significant correlations with the othervariables such as profitability, current ratio and dividend payout ratio. Profitability

Variables Definition

Debt ratio (Dit ) Ratio of total liability of the year to net assets of the yearInvestment (Iit ) Ratio of change in net fixed assets of the year to net fixed

assets beginning of the yearCash flow (CFit ) Ratio of net income plus depreciation, amortised intangibles

and deferred taxes less dividends, to net fixed assetsbeginning of the year

Cashstock (CSi,t-1) Ratio of total cash and cash equivalent beginning of the year tonet fixed assets beginning of the year

Tobin’s Q (Qi,t-1) Ratio of firm market value to firm book valueProfitability (Pit) Ratio of net profit before taxes to total assets of the yearTangibility (TANGit ) Ratio of net fixed assets to total assets of the yearCurrent ratio (CRit ) Ratio of total current assets to total current liability of the yearDividend (DIVit ) Ratio of dividends to total net income of the year

TTTTTaaaaabbbbble 1.le 1.le 1.le 1.le 1. Study variables

Debt Investment Cash Cash Tobin’s Q Profitability Tangibility Current Dividendratio flow stock ratio payout

ratio

0.36 0.10 0.20 0.38 3.27 0.12 0.61 3.35 0.43(0.19) (0.25) (0.22) (0.42) (2.87) (0.08) (0.21) (3.30) (0.31)

TTTTTaaaaabbbbble 2.le 2.le 2.le 2.le 2. Descriptive statistics of the means values and standard deviations (s.d.)

(Standard deviations in parentheses)




also has strong correlations with current ratio, Tobin’s Q and dividend payoutratio. Since there were several strong correlations among the independent variables,multicollinearity among the other seven independent variables were tested in theregression analysis.

Table 4 shows the relationship between the firms’ leverage and the eightindependent variables. With the first step multicollinearity test, the VIF values arelower than 5, which are within the range of 1.503 to 3.587. Thus, there is nocollinearity among the independent variables and none of the independent variableswill be dropped from the regression model. The regression model has a p-value

Variables Debt Invest- Cash Cash Tobin’s Q Profitability Tangibility Current ratio ment flow stock ratio

Debt ratio 1Investment 0.067 1Cash flow -0.186 0.657* 1Cash stock -0.330**-0.173 -0.013 1Tobin Q -0.183 -0.048 -0.019 0.146 1Profitability -0.176 -0.315** 0.166 0.369* 0.401* 1Tangibility 0.250 0.183 -0.002 -0.744 -0.109 -0.314** 1Current -0.517* -0.294** 0.213 0.470* 0.033 0.528* -0.418* 1ratioDividend -0.219 -0.397* -0.279** 0.593* 0.410* 0.397* -0.358* 0.253payoutratio

TTTTTaaaaabbbbble 3.le 3.le 3.le 3.le 3. Pearson correlation coefficient matrix

* Correlation is significant at 0.01 level** Correlation is significant at 0.05 level

Variables Coefficient (b) Standard error t-value p-value

Constant 0.520 0.128 4.075 0Investment 0.086 0.165 0.520 0.606Cash flow -0.190 0.174 -1.092 0.280Cash stock -0.067 0.097 -0.685 0.497Tobin’s Q -0.017 0.009 -1.888 0.065Profitability 0.881 0.420 2.096 0.041Tangibility -0.045 0.160 -0.283 0.778Current ratio -0.032 0.010 -3.316 0.002Dividend payout ratio -0.037 0.107 -0.343 0.733

TTTTTaaaaabbbbble 4.le 4.le 4.le 4.le 4. Investment regressions of the sample firms with all key independent variables

Note: R2 = 0.370; Adjusted R2 = 0.265; F-statistic = 3.527; p-value = 0.003




of 0.003 ( 0.10). Our study focuses on the relationship betweenleverage and cash flow. However, there is no significant relationship between leveragewith cash flow and cash stock. The coefficient of cash flow shows a negativeimpact on leverage with only -0.190. Tobin’s Q and profitability show significantp-value at 0.10 and 0.05 levels respectively. Profitability has a strong positivecorrelation while Tobin’s Q has a very low negative correlation with leverage.

The current ratio has very low negative coefficients and a p-value of 0.002 at0.01 significant level. This significant negative relationship supports the peckingorder theory. Both tangibility and dividend payout ratio do not significantly affectthe leverage levels. This means firms are not paying dividends by issuing debts.Our finding is inconsistent with earlier research findings as the cash flow has nosignificant relationship with leverage. We find only Tobin’s Q, profitability andcurrent ratio have significant effect on leverage levels.

We also tested the regression model without investment, tangibility and dividendratio since they had among the highest p-values in previous analysis. This was toretest the relationship between leverage with firms’ cash flow and other significantvariables. Table 5 shows that the regression model’s explanatory power of R2 hasslightly decreased to 0.363 and appears to be unlikely to explain better variations inleverage. Cash flow and cash stock still have a p-value > 0.10; Tobin’s Q, profitabilityand current ratio are all significant in the model. Thus, we conclude that cash flowhas no effect on the leverage explanatory power in the model.

Previous research reveals that firm size has a significant effect on firm’sleverage as large firms may have higher accessibility to external funds. On theother hand, small firm size does not offer much opportunity to issue debts as suchfirms have a higher default rate and low assets to back up the debts.

Variables Coefficient (b) Standard error t-value p-value

Constant 0.484 0.044 11.028 0.000Cash flow -0.101 0.098 -1.030 0.308Cash stock -0.059 0.058 -1.013 0.316Tobin’s Q -0.018 0.008 -2.143 0.037Profitability 0.761 0.367 2.073 0.043Current ratio -0.033 0.008 -3.972 0.000

TTTTTaaaaabbbbble 5.le 5.le 5.le 5.le 5. Investment regressions of the sample firms without investment, tangibility anddividend payout ratio as independent variable

Note: R2 = 0.363; Adjusted R2 = 0.300; F-statistic = 5.807; p-value = 0.000




Many studies have included firm size as the control variable in the model.There are several measurements for firm size. Here, we used three differentmeasurements to proxy firm size: (1) market value, (2) sales value, and (3) totalassets. Since our sample size had a clear differentiation of top and bottom twentyfirms from the top 100 list, we segmented the firms into small and large firm sizeswith these three different measurements and ran the regression analysis separately.Thus, any difference between the coefficient for the independent variables acrossthe small and large firm size groups will show the effect of firm size on leverage.

Table 6 provides information on leverage regression based on firm’s marketvalue. For the small firms’ market value group, there is a significant relationshipbetween the firm’s investment and firm’s leverage level. R2 value of the model ishigh at 0.472 and p-value at the significant level of 0.05. All other independentvariables do not have significant coefficients. When testing on the large firms sizegroup, there was multicollinearity and investment (VIF = 9.402 > 5) had beendropped from the model. The new model without investment was significant forthe large firm group at 0.10 level and R2 of 0.497. Tobin’s Q was significant at 0.10level while current ratio had a negative coefficient of -0.066 with a significant p-value of less than 0.01. Subsequently, all other independent variables were notsignificantly related with firm’s leverage.

Results of investment regression based on firm’s lagged sales size are shown inTable 7. Cash stock was taken out from the model due to multicollinearity. Themodel fitted well for the small firm size group, where p-value and R2 were at 0.526 and0.015 (p-value < 0.05) respectively. Similarly, investment had a significant positiverelationship with leverage at 0.05 level. Investment had been dropped from thelarge firm size group model due to multicollinearity. Though the model wassignificant (p-value = 0.020 < 0.05) with an R2 of 0.526, no independent variablehad a significant relationship with the firm’s leverage levels.

Table 8 shows the investment regression based on the firm’s total assets size.Tangibility and dividend payout ratio had been dropped from the model. As in thecase of the previous results, the model fit for the small firm size group (p-value <0.05, R2 = 0.472) and none of the independent variables had a significant relationshipwith firm’s leverage except for investment. For the large firms total assets group,again investment had been dropped and the model showed a good fit at R2 of0.470 (p-value



Sam

ple

info

rmat

ion

Regr

essio

n es

timat

esG

roup

Mea

n fir

m s

ize

NIn

vestm

ent

Cas

h C

ash

Tobi

n’s Q

Prof

itabi

lityT

angi

bilit

yC

urre

nt D

ivid

end

R2

p-va

lue

flo

w st

ock

rat

iopa

yout

ratio

Small

RM 1

,943

.2 m

330.

490*

**0.

098

-0.0

490

0.0

450.

111

-0.0

08 0

.031

0.47

20.

029*

*La

rge

RM15

,365

.5 m

24-

-0.0

96-0

.202

-0.0

27**

1.2

01-0

.008

-0.0

66*

-0.0

820.

497

0.08

4***

TTTT T aaaa abbbb b l

e 6.

le 6

.le

6.

le 6

.le

6. I

nves

tmen

t reg

ress

ions

for f

irms b

ased

on

firm

s’ m

arke

t valu

e

* Si

gnifi

cant

at 0

.01

leve

l, **

Sig

nific

ant

at 0

.05

leve

l, **

* Si

gnifi

cant

at 0

.10

leve

l

Sam

ple

info

rmat

ion

Regr

essio

n es

timat

esG

roup

Mea

n fir

m s

ize

NIn

vestm

ent

Cas

hCa

shTo

bin’s

QPr

ofita

bilit

yTa

ngib

ility

Cur

rent

D

ivid

end

R2

p-va

lue

flow

stock

rat

iopa

yout

ratio

Small

RM 8

57.5

m29

0.5

54**

0.08

9

--0

.018

0.15

70.

176

-0.0

030.

025

0.52

60.

015*

*La

rge

RM 9

,155

.9 m

28

-

-0.1

28-0

.039

-0.0

150.

549

0.26

0-0

.069

-0.0

040.

526

0.02

0**


e 7.

le 7

.le

7.

le 7

.le

7. I

nves

tmen

t reg

ress

ions

for f

irms b

ased

on

firm

s’ lag

ged

sales

* Si

gnifi

cant

at 0

.01

leve

l, **

Sig

nific

ant a

t 0.0

5 le

vel,

***

Sign

ifica

nt a

t 0.1

0 le

vel

Sam

ple

info

rmat

ion

Regr

essio

n es

timat

esG

roup

Mea

n fir

m si

zeN

Inve

stmen

t C

ash

Cas

hTo

bin’s

QPr

ofita

bilit

yTan

gibi

lity

Cur

rent

D

ivid

end

R2

p-va

lue

flow

stoc

k

ratio

payo

ut r

atio

Small

RM 2

,010

.7 m

29 0

.394

***

-0.02

5-0

.079

-0

.003

0

.678

--0

.018

-0.

472

0.019

**La

rge

RM 1

7,99

1 m

28

-

-0.10

1-0

.087

-0

.031

**

0.2

28

0.0

63-0

.054*

*

0.0

160.

470

0.049

**


e 8.

le 8

.le

8.

le 8

.le

8. I

nves

tmen

t reg

ress

ions

for f

irms

base

d on

firm

s’ to

tal a

sset

s

* Si

gnifi

cant

at 0

.01

leve

l, **

Sig

nific

ant a

t 0.0

5 le

vel,

***

Sign

ifica

nt a

t 0.1

0 le

vel




We further ran the mean comparison test on those significant independentvariables, namely, investment, Tobin’s Q, profitability and current ratio forsmall and large firm size groups based on the three firm size measurements.Table 9 displays the t-values to test the hypothesis that there is no difference inindependent variables from two different firm size groups. A hierarchicaldifference in the coefficients was found where investment for the large firmsize group was not significantly different from the small firm size group. Thecurrent ratio showed significant results (p-value < 0.10) in two different firmsize measurements. Hence, there were significant differences between the smalland large groups. This supports the regression results earlier where current ratiowas significantly related to leverage in large firm size group rather than smallfirm size group. Both Tobin’s Q and profitability showed mixed results of nodifferences between the two firm size groups under the three measurements.

In brief, the relationships between leverage and independent variables havemixed results in small and large firm size groups. Investment is significantly relatedin small firm size group while both Tobin’s Q and current ratio are significantlyrelated in the large firm size group. However, cash flow, cash stock, tangibility andprofitability do not show any significant relationship with leverage either in smallor large firm size groups.

Investment Tobin’s Q Profitability Current ratio

Firm market value measure of firm sizesmall large firm size -1.13 (0.26) -2.15 (0.04)** 0.11 (0.92) 1.43 (0.16)

Firm sales measure of firm sizesmall large firm size -0.80 (0.44) -1.18 (0.25) 0.87 (0.39) 1.71 (0.09)***

Firm total assets measure of firm sizesmall large firm size -1.01 (0.32) -0.42 (0.67) 2.77 (0.01)* 1.68 (0.10)***

TTTTTaaaaabbbbble 9.le 9.le 9.le 9.le 9. t-values of the independent variables in the small and large firm size group.

(p-value in parentheses), * Significant at 0.01 level, ** Significant at 0.05 level,*** Significant at 0.10 level.

5. CONCLUSION5. CONCLUSION5. CONCLUSION5. CONCLUSION5. CONCLUSION

The positive leverage-cash flow relationship was supported by the signaling theorythat accounts for leverage to event changes analysis, whereas, the negativerelationshipwas supported by the pecking theory based on cross-sectional analysis.However, our study concludes that there is no significant relationship betweenleverage and cash flow. This may be due to different financing styles in developingcountries or the effect of the ease in accessing external funds at low cost. Moreover,




financial flexibility of firms gives a space capacity for them to issue new debtsrather than depend on internal funds. Subsequently, firms’ tangibility anddividend payout did not have an effect on their leverage level, regardless of sizegroup. The investment activities of firms have a significant positive relationshipwith leverage within the small firm size group but not within the large firm sizegroup. This is because larger firms are capable of financing investment activitieswith available internal funds. Small firms, in contrast, need to issue new debtsto fund any new investment activities. The profitability, liquidity and growthof firms are also key predictors for leverage. Our results support pecking ordertheory, where negative liquidity-leverage relationship suggests firms with highliquid assets prefer to use internally generated funds to finance their investmentactivities.

Future research may consider a larger sample size (e.g. all firms listed on theMalaysian bourse FTSE 100-index). This would enhance better comparisons amonga wider range of various industries, as well as, confirm all previous research results.Accuracy of findings would also be improved. It is also believed that there existsdifferent leverage-cash flow relationships in different time periods. Thus, researchcan be extended based on different time periods - before, during and after theglobal/financial crisis. This comparison may help to forecast the possibility of acrisis in future. Additionally, financial flexibility can be included as a predictor forthe study, as this variable has a consistent tendency to affect availability of internalfunds and cost of external funds for investment activities. In summary, resultsfrom this study can be of high value to researchers, especially, those who studydeveloping countries. Moreover, this research makes a significant contribution inthat it could be used as a platform for management, as well as governments ofdeveloping countries to focus on their investment efforts for sustainability andgrowth.

ReferencesReferencesReferencesReferencesReferences

Altinkilic, O., & Hansen, S.S. (2000). Are there economies of scale in underwriterfees? Evidence of rising external financing costs. Review of Financial Studies, 13,191–218.

Baskin, J. (1989). An empirical investigation of the pecking order hypothesis. FinancialManagement, Spring, 26–35.

Bennett, M., & Donnelly, R. (1993). The determinants of capital structure: someUK evidence. British Accounting Review, 25(1), 43–59.

Copeland, T., & Lee, W. (1991). Exchange offers and stock swaps-new evidence.Financial Management, 20, 34–48.

Cornett, M., & Travlos, N. (1989). Information effects associated with debt-for-equity and equity-for-debt exchange offers. Journal of Finance, 44, 451–468.




Fama, E., & French, K. (2005). Financing decisions: Who issues stock? Journal ofFinancial Economics, 76, 549–582.

Faulkender, M., Flannery, M.J., Hankins, K.W., & Smith, J.M. (2012). Cash flowsand leverage adjustments. Journal of Financial Economics, 103, 632–646.

Frank, M., & Goyal, V. (2003). Testing the pecking order theory of capital structure.Journal of Financial Economics, 67, 217–248.

Frank, M.Z., & Goyal, V.K. (2004). The effect of market conditions on capitalstructure adjustment. Finance Research Letters, 1, 47–55.

Gaud, P., Hoesli, M., & Bender, A. (2007). Debt-equity choice in Europe. InternationalReview of Financial Analysis, 16, 201–222.

Harris, M. & Raviv, A. (1991). The theory of capital structure. Journal of Finance, 46,297–356.

Ho, S.S.M, Lam, K.C.K., & Sami, H. (2004). The investment opportunity set, director,ownership and corporate policies: evidence from emerging markets. Journal ofCorporate Finance, 10, 383 – 408.

Hovakimian, A., Opler, T., & Titman, S. (2001). The debt–equity choice. Journal ofFinancial and Quantitative Analysis, 36, 1–24.

Jensen, G., Solberg, D., & Zorn, T. (1992). Simultaneous determination of insiderownership, debt, and dividend policies. Journal of Financial and QuantitativeAnalysis, 27, 247–263.

Ju, N., Parrino, R., Poteshman, A.M., & Weisbach, M.S. (2005). Horses and rabbits?Optimal dynamic capital structure from shareholder and manager perspectives.Working Paper, University of Illinois.

Jung, K., Kim, Y., & Stulz, R. (1996). Timing, investment opportunities, managerialdiscretion and the security issue decision. Journal of Financial Economics, 42, 159–185.

Leary, M., & Roberts, M. (2005). Do firms rebalance their capital structures? Journalof Finance, 60, 2575–2619.

Leary, M.T., & Roberts, M.R. (2010). The pecking order,debt capacity,and informationasymmetry. Journal of Financial Economics, 95, 332–355.

Marsh, P. (1982). The choice between equity and debt: An empirical study. Journalof Finance, 37, 121–144.

Myers, S., & Rajan, R. (1998). The paradox of liquidity. Quarterly Journal of Economics,113, 733–771.

Rajan, R., & Zingales, L. (1995). What do we know about capital structure? Someevidence from international data. Journal of Finance, 50, 1421–1460.

Ravid, S., & Sarig, O. (1991). Financial signaling by committing to cash outflows.Journal of Financial and Quantitative Analysis, 26, 165–180.

Shenoy, C., & Koch, P. D. (1996). The firm leverage-cash flow relationship. Journalof Empirical Finance, 2, 307-331.




Shivdasani, A., & Stefanescu, I. (2010). How do pensions affect capital structuredecisions? Review of Financial Studies, 23, 1287–1323.

Shleifer, A., & R. Vishny. (1992). Liquidation values and debt capacity: a marketequilibrium approach. Journal of Finance, 47, 1343–1366.

Shyam-Sunder, L., & Myers, S. (1999). Testing static trade-off against pecking ordermodels of capital structure. Journal of Financial Economics, 51, 219–244.

Sibilkov, V. (2009). Asset liquidity and capital structure. Journal of Financial andQuantitative Analysis, 44(5), 1173–1196.

Singhania, M., & Seth, A. (2010). Financial leverage and investment opportunitiesin India: an empirical study. International Research Journal of Finance andEconomics, 40, 215–226.

Strebulaev, I. (2007). Do tests of capital structure theory mean what they say?Journal of Finance, 62, 1747–1787.

Vogt, S. (1992). The role of internal financial sources in the firm financing andinvestment decisions. Working Paper 146, Center for the Study of AmericanBusiness, Washington University.

Wald, J.K. (1999). How firm characteristics affect capital structure: an internationalcomparison. Journal of Financial Research, 22(2), 161–167.

Welch, I. (2004). Capital structure and stock returns. Journal of Political Economy, 112(1), 106–131.


Financial Integration and International Capital Mobility: Evidence from ASEAN

Taylor’s Business Review Vol. 2. Issue 1. 2012 17

Financial InteFinancial InteFinancial InteFinancial InteFinancial Integggggrrrrraaaaation and Intertion and Intertion and Intertion and Intertion and Internananananational Cational Cational Cational Cational Capital Mobility:pital Mobility:pital Mobility:pital Mobility:pital Mobility:Evidence from ASEANEvidence from ASEANEvidence from ASEANEvidence from ASEANEvidence from ASEAN

Muzafar Shah Habibullah*Universiti Putra Malaysia

Sarinder Kumari**PEMANDU Unit, Prime Ministers Department

Baharom, A.H.***Taylors University

Abstract: Abstract: Abstract: Abstract: Abstract: This study was conducted to explore the linkages between savings andinvestment and to further test whether there is any evidence of relationship betweenfinancial integration and international capital mobility. The empirical model to testthe capital mobility hypothesis applied here employs the panel data approach to thebasic regression model used by Feldstein & Horioka (1980). The study coversASEAN5 and ASEAN+3 (with China, Korea and Japan included). The findings ofthis study corroborate with other studies that savings and investment are cointegratedand this can be interpreted as a manifestation of the inter-temporal budget constraintrather than evidence of low capital mobility. The long-run equilibrium betweensavings and investment is in line with the inter-temporal budget constraints of anopen economy as current account deficits cannot be sustained indefinitely. Therelatively high degree of capital mobility when China, Korea and Japan are includedin the ASEAN5 sample suggests that there is great potential for integration in thefinancial markets in these eight economies in the future

Key wordsKey wordsKey wordsKey wordsKey words: ASEAN, savings, investment, cointegration, financial integrationJEL classification: E02, E22, E44

Vol 2. Issue 1. 2012pp. 17-31

ISSN: 2232-0172

A Contemporary Business Journal

* Muzafar Shah Habibullah, Department of Economics Faculty of Economics and Management,Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia.

** Sarinder Kumari, PEMANDU Unit, Prime Ministers Department, Putrajaya, Malaysia.*** Baharom, A.H., Taylors Business School, Taylors University, Lakeside Campus, No 1 Jalan Taylor’s

47500 Subang Jaya, Selangor, Malaysia. Email: [email protected]

E069-10-TBR-Vol 2 Issue 1-2 Financial Integrationin_0210.pmd 03-Oct-2012, 10:20 AM17

Muzafar Shah Habibullah, Sarinder Kumari & Baharom, A.H.

Taylor’s Business Review Vol. 2. Issue 1. 201218

1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION1. INTRODUCTION

The general move towards international financial liberalisation by countries hasbeen motivated by the fact that integration of financial markets allows for moreefficient use and allocation of investment funds over time. Financial integrationand capital mobility are closely related and often used interchangeably as theintegration of financial markets involves the deregulation of national markets andthe liberalisation of international flows. Therefore, financial markets integration isconducive to capital mobility or increases the potential of capital flows and it canbe implied that capital mobility is a sufficient condition for financial integration orthat a high level of capital mobility indicates a high level of market integration(Moosa, 1996).

While the gains from financial integration and the associated capital mobilityare quite clear, the measurement criteria for financial integration have proven to bemore difficult. This paper focuses on the saving-investment criterion highlightedby Feldstein & Horioka (1980) as a way of measuring international capital mobility.Feldstein & Horioka ran a cross-sectional regression on 16 industrial countries andfound a high correlation between savings and investment. This was interpreted asindicating low capital mobility because if there is high capital mobility, there is no apriori reason for savings and investment to be correlated across countries as savingsin each country would respond to the worldwide opportunities for investment,while the worldwide pool of capital would finance investments in each country.However, if capital mobility were restricted, then there would be a wedge betweenthe cost of domestic and foreign savings and the incremental savings would tendto be invested domestically. Therefore a positive close-to-one correlation betweenthe savings rate and investment would be suggestive of imperfect capital mobilityand in the extreme case of zero capital mobility, savings and investment would beperfectly correlated.

The finding of a high correlation between saving and investment by Feldstein& Horioka (1980) implying imperfect capital mobility has jolted the widespreadanalytical approach by researchers that international capital mobility is nearly perfect.If international capital mobility does not require a close correlation between savingsand investment, it is difficult to rationalise the Feldstein-Horioka finding –corroborated in many similar studies – that savings and investment are in facthighly correlated. Referred to as the Feldstein-Horioka puzzle, this raised considerabledoubt on whether national markets for physical capital are highly integrated. Despitethis, the savings-investment correlation proposed by Feldstein & Horioka (1980)has been widely accepted as a measure of financial integration and capital mobility.

However, the Feldstein-Horioka finding has been subject to both econometricsand theoretical criticisms. The cross-sectional regression not only ignores the time-series and non-stationary properties of investment and savings in each country that




could lead to spurious regressions but also could be subject to simultaneitybias (Dooley, Frankel & Mathieson, 1987; Miller, 1988; Sinn, 1992). There havealso been difficulties in interpreting the results of tests based on the Feldstein&Horioka method, as the regression equations are not explicitly derived fromtheory. There is therefore no information on the size of the regression coefficientunder the null hypothesis of perfect capital mobility. Indeed there are manycases in which savings and investment are correlated even under perfect capitalmarkets (Obstfeld, 1993).

Due to these criticisms, alternative approaches have been employed to studythe saving-investment relationship. Cointegration techniques of analysis pioneeredby Miller (1988) have been considered as they take into account the time-series andnon-stationarity properties of saving and investment in each country. Thecointegration analysis essentially extends the Feldstein-Horioka interpretation byarguing that cointegration between saving and investment implies that capital ishighly immobile internationally. Absence of cointegration between savings andinvestment meansthat capital mobility is high. Results from various cointegrationtests (Miller, 1988; Barkoulas, Filizetkin & Murphy 1996; Bajo-Rubio, 1998) showeither no cointegration between savings and investment or decreasing regressioncoefficients over time. However recently it has been argued that savings andinvestment tend to be cointegrated variables (Coakley, Kulasi & Smith, 1996; Coakley& Kulasi, 1997; Coiteux & Olivier, 2000). The cointegration between savings andinvestment has been taken to imply inter-temporal budget constraints rather thanevidence of low capital mobility.

However, although the cointegration analysis isable to address the issue ofnon-stationarity of investment and savings, the conventional unit root andcointegration tests have been found to have low testing power and which couldtherefore result in erroneous results (Coakley et al., 1996; Coakley & Kulasi, 1997;Oh, Kim, Kim & Ahn, 1999).

To take into account the econometric problems of the Feldstein-Horiokaapproach, this paper employs the panel data approach to analyse international capitalmobility. Panel data analysis has been found to improve the statistical power of theconventional unit root and cointegration tests by pooling data and increasing thenumber of observations (Coakley et al., 1996; Coakley & Kulasi, 1997; Nagayasu,1998; Oh et al., 1999).

2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW2. LITERATURE REVIEW

The Feldstein-Horioka finding of a high correlation between investment rates andtheir national saving rates has been confirmed by many subsequent studies using avariety of techniques encompassing both cross-section and time-series regressionsover different time periods for both industrialised and developing countries (e.g.




Feldstein, 1983; Penati & Dooley, 1984; Dooley et al., 1987; Vos, 1988; Tesar, 1991). Infact Feldstein (1983) finds no evidence that this correlation has weakened over time.

However, there is little agreement on explanations of this apparent empiricalregularity of a high correlation between savings and investment. Economists arenot willing to accept the inference that financial markets are not highly integratedand argue that the results do not necessarily imply low capital mobility in theinternational economy (Westphal, 1983; Finn, 1990; Baxter & Crucini, 1993; Coakleyet al., 1996; Coakley & Kulasi, 1997).

A number of hypotheses have been used to rationalise the Feldstein-Horiokaregression result without concluding that capital is immobile. For example, a numberof studies suggest that the savings-investment correlation increases with countrysize due to its effects on the world interest rate (Murphy, 1984; Dooley et al., 1987;Baxter & Crucini, 1993; Krol, 1996). Another alternative proposition is that savingsand investment can be correlated even in the presence of capital mobility ifexogenous variables like productivity shocks or non-traded consumption goodsaffect both savings and investment (Wong, 1990; Tesar, 1991).

The inter-temporal approach to the current account or the solvency constraintarguments have also been used to explain the relationship between savings andinvestment as both a short run and long run phenomenon (Sachs, 1982; Finn, 1990;Sinn, 1992; Baxter & Crucini, 1993; Coakley et al., 1996; Coakley & Kulasi, 1997;Jansen, 2000). According to the inter-temporal approach to the current account,domestic savings and investment should be perfectly correlated in the long run,since current account balances (i.e. differences between savings and investment)should add up to zero. Due to the inter-temporal budget constraints in an openeconomy, current account deficits/surpluses cannot be sustained indefinitely. In thelong run therefore investments cannot deviate too much from savings. This impliesthat savings and investment should keep a one-to-one relation in the steady state; inother words, savings and investment are cointegrated variables. On the other hand,in the short run the size and sign of the correlation between savings and investmentwould depend on the structure of the economy, as well as on the nature of theshocks. So while small positive, zero or negative correlations would indicate asignificant degree of capital mobility, high positive correlations would not necessarilymean capital mobility being small.

One main criticism of the Feldstein-Horioka finding is that the national savingsand investment are both endogenous variables that respond to common factors.For example, one version of the endogeneity critique that results in investment andsavings being correlated has been attributed to government policy wherebygovernments react systematically to current account imbalances so as to offsetthese imbalances (Tobin, 1983; Bayoumi, 1990). Although endogeneity of savingsarises especially in time-series analysis it may also arise in cross-section analysis.




One way to deal with the cyclical endogeneity to remove the effects of the businesscycle is to use averaged rates of saving and investment over a period of time as ina cross-sectional regression (Feldstein & Horioka, 1980; Bayoumi, 1990; Hussein,1998)

There are essentially two ways that the relationship between savings-investment have been analysed. The first is to use time-averaged or cross-sectionaldata (Feldstein & Horioka, 1980; Feldstein, 1983; Bayoumi, 1990, Wong, 1990;)and the second is to use time-series data (Sinn, 1992). However, the use of cross-sectional data to primarily remove the business cycle effects has been a subjectof contention. For example, Sinn (1992) and Krol (1996) argue that cross-sectionaldata biases the results against capital mobility. Firstly, the averaging of dataover five or ten years may not be sufficient to establish a long-run relationshipbetween savings and investment. Furthermore, capital mobility is not just along run issue. Secondly, averaging data can introduce an empirical problem ofoffsetting changes in investment and savings over time suggesting a relationshipbetween investment and savings when none may exist. For instance, the inter-temporal budget constraints imply that there is a high correlation betweensaving and investment in the long run (Sachs, 1982; Sinn, 1992). Therefore thesavings and investment ratios averaged over sufficiently long periods must beclose despite capital mobility. This implies that the correlation between savingsand investment should be higher in the long run, compared to the short-runalthough this may not always be the case as shown by Sarno & Taylor (1998).

To overcome the shortcomings of the cross-sectional method, time-seriesestimation has been advocated. One important aspect of the time series estimationis that the same correlation coefficient is not imposed across different countries(Sinn, 1992). Therefore the time series estimates have the advantage as the countriesthat are more open and integrated with the world capital markets are not mergedtogether with economies that are more closed in a single regression. Modern theoryof inter-temporal economics also shows that even with financial integration, acountry may experience net capital inflows and outflows at different points in time(Genberg & Swoboda, 1992). This variability would not be captured by a cross-sectional analysis that is based on time-averaged data and would therefore result ina bias against the capital mobility hypothesis. A re-estimation of the Feldstein-Horioka model using annual data shows not only lower correlation coefficientsbut also considerable variability between savings and investment (Sinn, 1992).

Despite its apparent advantages, there have been concerns that the time seriesestimates may be subject to greater bias than cross-section regressions because ofthe problems of identification and estimation induced by simultaneity bias or theendogeneity of both savings and investment. For instance, the time-series estimatesdo not take into account business cycle effects on both savings and investment(Krol, 1996; Oh et al., 1999).




As a result there is now growing interest in panel data methodology that utilisesinformation from both the time-series and cross-sectional variation in the data. Itdoes not require the averaging of data and there is no loss of information associatedwith it while at the same time allowing for specific country effects such as countrysize to be taken into account.

Krol (1996) employed the panel data approach to re-estimate the saving-investment regression as he attributed the high saving-investment coefficient obtainedin the Feldstein-Horioka study (1980) to the problem of the estimation technique.It is argued that although the cross-sectional data employed by Feldstein-Horiokatakes into account time-effects like the business cycle, it does not take into accountthe unobserved or unmeasured country effects like country size. On the otherhand, time-series data does not take into account the business cycle effects. A paneldata approach would allow for both these effects to be taken into account as itcontrols for business cycle effects without averaging the data or any loss ofinformation associated with it and it also takes into account the country effects.Using the panel approach, a considerably smaller correlation between savings andinvestment than previous estimates is found. Country effects are found to havemore significant effects than international business cycle effects in the saving-investment analysis. The findings of Krol are corroborated by a number of studiesadopting a panel data approach (Vamvakidis & Wacziarg; 1998; Corbin, 2001; Ohet al.; 1999; Coakley et al., 1996).

One common criticism raised in the literature is the low testing power of theconventional cointegration tests. Coakley & Kulasi (1997) employ a variety ofcointegration tests to analyse the relationship between savings and investment. Besidesthe conventional ADF cointegration tests, a panel unit root test using the t-barstatistic based on the average ADF proposed by Im, Pesaran & Shin(1995) is alsoemployed as it has higher power than the conventional time-series tests. The resultsof the panel cointegration show that there is cointegration between saving andinvestment in all the 11 countries studies as compared to only 8 countries by theconventional tests. Oh et al. (1999) use conventional and panel unit root andcointegration tests and confirm the low testing power of conventional cointegrationtests as they fail to detect cointegration between savings and investment in contrastto the finding of the panel cointegration tests.

3. METHOD3. METHOD3. METHOD3. METHOD3. METHOD

The empirical model to test the capital mobility hypothesis applied here employsthe panel data approach to the basic regression model used by Feldstein & Horioka(1980) to analyse the relationship between savings and investment. This approachutilises information from both time series and cross-sectional variation in thedata and allows for the use of advances made in estimating models of individual




behaviour over time without having to aggregate time-series data. It does notrequire the averaging of data and there is no loss of information associatedwith it.

The regression model that is employed in this study is as follows:

ititiit YSYI εβα ++= )/()/( (1)

where I is the gross domestic investment, S is the gross national savings, Y is thegross domestic product, i is the country index, α is the constant term and β is thesaving-investment coefficient, ε is the error term and t is the observation year.

In this regression, β is subjected to the equality restriction and is constrained tohave the same value for all countries and does not vary across countries. However,to capture the country specific effects, the equality condition is not imposed on αand it is allowed to differ across countries. These constraints define the fixed-effects model (Krol, 1996; Coiteux & Olivier, 2000).

Under the assumption of high capital mobility, β should be closer to zero thanto one. The null hypothesis that is tested is β = 0 against the alternative hypothesis ofβ = 1. If β is not found to be significantly different from zero, then the null hypothesiscannot be rejected and it can be inferred that capital is mobile internationally.However, if the null hypothesis is rejected, it can be inferred that capital is notperfectly mobile.

The conventional unit root and cointegration tests used in time-series datahave been found to be limited due to their low testing power especially for panelsof moderate size where the existing test procedures may not be computationallyfeasible or sufficiently powerful. This may therefore lead to erroneous results. Thepanel data approach employs the panel unit root tests and panel cointegration teststhat have been found to improve the power of these tests. In this respect, the panelunit root tests developed by Im et al. (1997: IPS hereafter) and the cointegrationtests by Pedroni (1995) and Kao & Chiang (1998) are used.

To determine the stationarity of the individual variables, the conventional unitroot tests of Augmented Dickey-Fuller (ADF) (Dickey & Fuller, 1981) and Phillips-Peron (PP) (Phillips & Peron, 1988) are conducted. However, as these tests havelow testing power, this study employs a panel unit root test proposed by IPS totest the null hypothesis of unit root (that is, non-stationarity). This test is substantiallymore powerful than the individual time-series ADF or PP tests.

The IPS test is based on the average of the statistics obtained from individualtests. In other words, in the IPS procedure an ADF equation is estimated separatelyfor each individual country and this allows for differing parameter values, varianceand even different lag lengths. The null hypothesis for the presence of unit root isHo : βi = 0 for all i against the alternative H1: βi < 0 for i = 1, 2,…., N1; βi = 0 for i = N1+1, N1 +2,……N. The alternative hypothesis allows for βi to differ across groups.




The IPS procedure looks at the stochastic process generated by the first-orderautoregressive process:

Yit = (1 - φi )μi + φiYi ,t-1 + εit (2)

where i = 1, 2,…, N; t = 1, 2, …, T and the initial values Yt,0 are given. Thefollowing equation is estimated:

ΔYi t = αi + βiYi ,t -1 + εit (3)

where αi = (1 - φi )μi , βi = (1 - φi ), and ΔYit = Yi ,t - Yi ,t - 1 . The errors are assumed to beserially uncorrelated.

IPS proposed two test statistics to test the null hypothesis of the unit root test.These are the Lagrange multiplier or the LM-bar statistic and the t-bar statistic. Inthis study, only the t-bar test statistic is used as in the absence of autocorrelation, thet-bar test tends to perform better than the LM-bar test with finite samples. The t-bar test statistic can be calculated as follows:

( ){ })0|(

0|

==−

=iT

ITNTi

tVar

tEtN

ββΓ Γt (4)

where t TN is the cross-sectional average of the standard individual ADF unit root

t-statistics, such that t TN = ∑=

N

iiTt

N 1

1. The terms E (tT | β = 0) and Var (tT | β = 0) are

the respective finite common mean and variance of tiT for i = 1, 2,…, N, obtainedunder the null hypothesis of βi = 0. The test statistic tiT is the individual t-statistic fortesting the null hypothesis of the unit root (βi = 0) in the standard individual ADFand its critical values are reported by Im et al. (1997).

Cointegration on the individual time-series data is conducted using the Engle-Granger (1987) cointegration test. However, the conventional cointegration teststend to suffer from unacceptably low power especially when applied to series ofmoderate length. Hence, the panel cointegration tests will be employed. Panelcointegration tests generally allow for selective pooling of information regardingcommon long-run relationships from across the panel while allowing the associatedshort run dynamics and fixed effects to be heterogenous across different membersof the panel. Two cointegration tests proposed by Kao, Chiang & Chen (1999)and Pedroni (1995) are employed to test whether a cointegration relationship existsbetween savings and investment. For this study, these tests apply to homogenouspanels whereby β is assumed to be the same across all individuals. In these tests thenull hypothesis of no cointegration in the variables in each member of the panel istested.

E069-10-TBR-Vol 2 Issue 1-2 Financial Integrationin_0210.pmd 03-Oct-2012, 3:16 PM24



Kao (1998) proposes two types of panel cointegration tests, namely the Dickey-Fuller (DF) and augmented Dickey-Fuller (ADF). The DF-type tests are calculatedfrom the estimated residuals as follows:

êi ,t = γêi ,t -1 + v (5)

where êi ,t is the estimated residual from the estimated equation. To test the null ofno cointegration, (Ho: γ = 1), four DF-type tests are constructed as follows:

1. DF γ =( )

2.10

31ˆ NTN +−λ (6)

2. DF t = Nt 875.125.1 +γ (7)

3. )ˆ/ˆ2.7(3

ˆ/ˆ3()1ˆ(40

4

22*

νν

ννγ

σσσσγ

+

+−=

NTNDF (8)

4. )ˆ10/ˆ3(ˆ2/ˆ(

ˆ2/ˆ6(20

2220

0*

νννν

ννγ

σσσσ

σσ

+

+=

NtDFt (9)

where =σ νˆ 2 ∑ ∑−∑ −με ε 1p and σ νˆ 20 = ΩΩΩ −− 1εμεμ . and are based on strictexogeneity of the regressors with respect to the errors in the equation while andare based on endogenous regressors. The ADF test is calculated from the followingregression:

ti

P

jjtiititi eee ,

1,1,, ˆˆˆ νϑγ +Δ+= ∑

=−− (10)

The ADF test statistic is constructed as follows:

ADF= ( )

( ) ( )σσσσσσ

νννν

νν

ˆˆˆˆˆˆ

2

0

222

0

0

1032

26

+

+ Nt ADF(11)

where tADF is the t-statistic of γ in equation (6). The distributions of all these statisticsare normal N(0, 1).

Pedroni (1995) constructed a panel autoregressive coefficient estimator γ̂,TN

based on a pooled Phillips and Perron-type test as follows:

( )⎟⎠

⎞⎜⎝

⎛

−Δ=−

∑∑

∑∑

= =−

= =−

N

i

T

tti

N

i

T

tititi

TN

e

ee

1 2

2

1,

1 2,1,

,

ˆ

ˆˆˆ1ˆ

λγ (12)

where λ̂ i is the scalar equivalent to the correlation matrix and corrects for anycorrelation effect. The two test statistics proposed by Pedroni (1995) are:




1. PC1 = ( )

)1,0(2

1ˆ, N

NTTN ⇒

−γ (13)

2. PC2 = (14)

The cointegration relationship in the panel data is estimated using Kao & Chiang’s(1999) dynamic OLS (DOLS) and fully modified (FMOLS) estimators. The FMOLSestimator does not in general improve on the OLS estimator while the DOLSestimator is better than both the bias-corrected OLS and FMOLS estimators. TheOLS estimator of β is

(15)

where and ⎟⎠

⎞⎜⎝

⎛= ∑=

T

ttii YY

1, . The FM estimator is obtained by correcting

for endogeneity and serial correlation to the OLS estimator in (15) above and isgiven as follows:

( )( ) ( ) ⎥⎦

⎤⎢⎣

⎡⎟⎠

⎞⎜⎝

⎛ −−⎥⎦

⎤⎢⎣

⎡ −−= ∑ Δ∑∑∑=

++

=

−

= =

N

iti

T

ttiti

N

iiti

T

ttiFM TXXXXXX Y

1,

1,,

1

1

'

,1

, ˆˆˆ εμβ (16)

where Δ̂ εμ is the kernel estimates of Δεμ . The DOLS estimator can be obtainedfrom the following regression:

∑−=

+ +Δ++=2

1,,

',,

q

qjtijtiijtiiti XcXY νβα (17)

The DOLS (p,q) where p and q denote the number of leads and lags respectivelyare estimated. To test for panel cointegration, the residual based ADF statisticproposed by Kao and Chiang (1999) to test the null of no cointegration is employed.If there is a unit root, then there is no cointegration in the model.

This study uses saving and investment data of the ASEAN5+3 economiesnamely; Indonesia, Malaysia, Philippines, Singapore and Thailand, China, Japan andKorea. The sample period extends from 1970 to 1999.

4. RESULTS AND DISCUSSION4. RESULTS AND DISCUSSION4. RESULTS AND DISCUSSION4. RESULTS AND DISCUSSION4. RESULTS AND DISCUSSION

In general, the results of the panel unit tests based on the IPS procedure presentedin Table 1 suggest that the individual unit root tests fail to reject the null hypothesisof non-stationarity of the investment rate and saving rate.

The cointegration analysis is carried out for both Kao and Chiang (1999) andPedroni (1995) approaches and the results are presented in Table 2. Bothcointegration test rejected the null hypothesis of no cointegration, These findingscorroborate other studies in the literature and confirm that saving and investment

⎥⎦

⎤⎢⎣

⎡⎟⎠

⎞⎜⎝

⎛ Δ−−⎥⎦

⎤⎢⎣

⎡ −−= ∑ ∑∑∑= =

++−

= =

N

i

T

tutiiti

N

i

N

itiitiOLS TyXXXXXX

1 1,,

1

1 1

',,

ˆˆ)())((ˆ εβ

( )∑=

=T

ttii XTX

1,/1

( )( ))1,0(

2

11 ˆ, N

TNTTN ⇒

−− γ

E069-10-TBR-Vol 2 Issue 1-2 Financial Integrationin_0210.pmd 03-Oct-2012, 3:34 PM26



The presence of cointegration is consistent and complies with the argumentsthat the inter-temporal budget constraints are always operative in an open-economyas current account deficits cannot be sustained indefinitely (Jansen, 1997; 2000). Assuch regardless of the degree of capital mobility the inter-temporal budgetconstraints are powerful enough to drive the results indicating the presence of acointegration relationship.

Sample country Investment rate Saving rate

No time trend With time trend No time trend With time trend

ASEAN5 -0.31 1.63 -1.01 0.39(0.38) (0.05) (0.15) (0.35)

ASEAN5+3 -2.15 -1.85 -1.67 -1.07(0.02) (0.03) (0.05) (0.14)

TTTTTaaaaabbbbble 1.le 1.le 1.le 1.le 1. Results of IPS (t-bar) panel unit root tests

Note: Figures in brackets are p-values.

ASEAN5 ASEAN5+3Kao (1999)DFρ -12.61 -3.69

(0.00) (0.00)DFt -7.23 -2.12

(0.00) (0.00)DF * -23.00 -8.46

(0.00) (0.00)DFt -6.01 -2.70

(0.00) (0.00)ADF -2.22 -4.02

(0.00) (0.00)PedroniPC1 -48.00 -15.35

(0.00) (0.00)PC2 -47.22 -15.10

(0.00) (0.00)

*

ρ

TTTTTaaaaabbbbble 2.le 2.le 2.le 2.le 2. Results of panel cointegration tests

Note: Figures in brackets are p-values.

are cointegrated in the long run for all the countries under consideration (Jansen,1997; Anoruo, 2001).




Our point of interest is to evaluate the magnitude of β, the saving retentioncoefficient for the sample of ASEAN5 and ASEAN+3. As shown in Table 3, theDOLS estimate of β for ASEAN5 registered a value of 0.89, and when China,Korea and Japan are included in the sample, the estimated β is reduced to 0.58.Both β ’s are significantly different from zero at the 5 % level.

A comparison of capital mobility between the two sub-regional samples isnot easy. The magnitude of β cannot be used to measure degree of capital mobility.The β estimate of 0.26 for the ASEAN 5 region does not necessarily indicate alower degree of capital mobility than the estimate of zero. This is because theestimate may reflect different types of shocks that the countries may experienceunder capital mobility. By the same token, it is impossible to associate a zerocoefficient with the theoretical benchmark of perfect capital mobility. In fact evenunder perfect capital mobility the β is not expected to be zero (Feldstein & Horioka,1980).

However the results can be used to make qualitative comparisons pertaining tocapital mobility in the region. A comparison of β estimates on a regional basisshows that the ASEAN5 rejects the null hypothesis of perfect mobility indicatingthat capital mobility in these regions is low. However, the results for the ASEAN+3show otherwise. Based on these results it can be concluded that while capital mobilityis low in the ASEAN5 region as a whole, the capital mobility is relatively higher inthe ASEAN+3 region.

5. CONCLUSION5. CONCLUSION5. CONCLUSION5. CONCLUSION5. CONCLUSION

The findings of this study corroborate other studies that savings and investmentare cointegrated and this can be interpreted as a manifestation of the inter-temporalbudget constraint rather than evidence of low capital mobility (see for example:Coakley et al., 1996; Coakley & Kulasi, 1997; Coiteux & Oliver, 2000). The long-run equilibrium between savings and investment is in line with the inter-temporalbudget constraints of an open economy as current account deficits cannot besustained indefinitely. In this line of research, Lau & Baharumshah (2002) have also

β (saving retention coefficient) R2/Adjusted R2

ASEAN5 0.89 0.55/(7.26) -0.72

ASEAN5+3 0.58 0.29/(6.53) 0.03

TTTTTaaaaabbbbble 3.le 3.le 3.le 3.le 3. Results of estimated saving retention coefficient from DOLS estimates

Note: Figures in brackets are t-statistics.




provided evidence that most of the major Asian economies do not violate theinter-temporal budget constraints.

One explanation for the relatively high degree of capital mobility when China,Korea and Japan are included in the ASEAN5 sample is that there is great potentialof integration in the financial markets in these eight economies in the future. Gradualelimination of restrictions on capital inflow (FDI) so as to take advantage of thesubstantial benefits from higher investments, faster growth and higher livingstandards in these Asian countries are ongoing process. One important conclusionof this study is that low savings-investment retention suggests that Asian countrieshas already satisfied one of the criterion of the Optimum Currency Area (OCA)put forward by Mundell (1961).

ReferencesReferencesReferencesReferencesReferences

Anoruo, E. (2001). saving-investment connection: evidence from the ASEANcountries. American Economist, 45(1), 46-53.

Bajo-Rubio, O. (1998). The saving-investment correlation revisited: the case ofSpain, 1964-1994. Applied Economics Letters, 5, 769-772.

Barkoulas, J, Filizetkin, A., & Murphy, R. (1996). Time series evidence on the saving-investment relationship. Applied Economics Letters, 3, 77-80.

Baxter, M., & Crucini, M. J. (1993). Explaining saving-investment correlations. TheAmerican Economic Review, 83(3), 416-436.

Bayoumi, T. (1990). Saving-investment correlations. immobile capital, governmentpolicy or endogenous behavior. IMF Staff Papers, 37(2), 360-387.

Coakley, J., & Kulasi, F. (1997). Cointegration of long span saving and investment.Economics Letters, 54(1), 1-6.

Coakley, J., Kulasi, F., & Smith, R. (1996). Current account solvency and the Feldstein-Horioka puzzle. The Economic Journal 106, 620-627.

Coiteux, M. & Olivier, S. (2000). The saving retention coefficient in the long runand in the short run: evidence from panel data. Journal of International Money andFinance, 19: 535-548.

Corbin, A. (2001). Country specific effect in the Feldstein-Horioka paradox: apanel data analysis. Economics Letters, 72, 297-302.

Dickey, D.A., & Fuller, W.A. (1981). Distribution of the estimators for autoregressivetime series with a unit root. Econometrica, 49, 1057–72.

Dooley, M., Frankel, J., & Mathieson, D. J. (1987). International capital mobilty:What do saving-investment correlations tell us? IMF Staff Papers, 34, 503-530.

Engle, R.F., & C.W.J. Granger. (1987). Cointegration and error correction:representation, estimation, and testing. Econometrica, 55, 251–76.

Feldstein, M. (1983). Domestic savings and international capital movements in thelong run and the short run. European Economic Review, 21, 120-151.




Feldstein, M., & Horioka, C. (1980). Domestic Saving and International CapitalFlows. Economic Journal, 90, 314-329.

Finn, M.G.

Documents

E069-10-TBR-Vol 2 Issue 1 0310university2.taylors.edu.my/tbr/uploaded/2012_vol2_issue1.pdf · 2020. 6. 2. · A Contemporary Business Journal Vol. 2 Issue 1 February 2012 A Publication