Determinants of Corp Investment

This shift in approach can be attributed to the seminal work by Modigliani and Miller (1958). ... 1 The title of this paper is taken from one of such ...

0 downloads 122 Views 76KB Size
Determinants of Corporate Investment

Amado P. Saquido 2 October 2003

Determinants of Corporate Investment by Amado P. Saquido

I. Introduction Early applied research stressed the significance of financing constraints in business investment.1 Since the mid-1960’s, however, most applied work isolated real firm decisions from purely financial factors. This shift in approach can be attributed to the seminal work by Modigliani and Miller (1958). This demonstrated the irrelevance of financial structure and financial policy for real investment decisions under certain conditions. Thus, a firm’s financial structure would not affect its market value in frictionless capital markets. If their assumptions were satisfied, “real” firm decisions (e.g. fixed investment) motivated by maximization of shareholders’ claims would be independent of financial factors such as liquidity, leverage, or dividend payments. Applied to capital investment, this boils down to the neoclassical theory of investment in which the firm’s choice of optimal capital stock could be determined irrespective of financial factors. The q-theory approach, pioneered by Tobin (1969) and extended to models of investment assuming convex costs of adjusting the capital by Hayashi (1982), offers another formulation of the neoclassical model. Investment opportunities could be summarized by the market valuation of the firm’s capital stock, and, under certain assumptions, the ratio of the market value of the capital stock to its replacement cost is the basic variable explaining investment demand.2 The results of empirical studies in other countries using the q-theory of investment are mixed, with majority on the favorable side.3 Studies using Philippine data are scant. Aquino (2000) found that investment rate had no significant relationship with q. It was instead related to revenue growth, and, to a lesser extent, to the debt-equity ratio and the fixed capital intensity of the firm’s operations. He did not find a significant relationship either between investment rate and cashflow. On the other hand, imperfections in the capital market imply that internal finance is less costly than external finance.4 This further implies that investment at the firm level should be sensitive to profitability, i.e. the changes in the net worth of the firm, given constant investment opportunities. Cashflow as a proxy for changes in firm wealth—thus, the ability to source from internal funds—should then be relevant to investment. While many 1

The title of this paper is taken from one of such works, i.e. Grunfeld, Y, The Determinants of Corporate Investment, unpublished Ph.D. thesis, Department of Economics, University of Chicago, 1958; quoted in Greene (1997) and other econometrics textbooks which use his data in explaining models for panel data.

2

Hubbard (1998) gives a detailed explanation of the foregoing and of the subsequent developments in empirical work in this area. 3

Aquino (2002) includes a survey of these studies and summarizes their results.

4

See Romer (2001) and Hubbard (1998) for the theoretical underpinnings and analysis of the issues involved in empirically testing this hypothesis.

1

early studies found such a relationship, it is not clear whether it is due to changes in investment opportunities or to capital market imperfections.

II. Objectives of the Study This study aims to investigate two main determinants of investment at the firm level, namely investment opportunities as captured by Tobin’s q, and cashflow as a proxy for changes in net worth, using Philippine data. It follows the approach of Fazzari, Hubbard and Petersen (FHP) (1988) who studied the impact of financing constraints on the sensitivity of investment to cashflow. They controlled for investment opportunities using Tobin’s q, and used the dividend rate to classify firms into being financially constrained or not. This allowed them a two-fold conclusion, that cashflow affects investment and that this is due imperfections of the capital market. A vast empirical literature followed, summarized in Hubbard (1998), focusing on various factors that determine the relationship between investment and cashflow, with supportive findings. Some recent examples are Hansen (1999) for agency costs and Liu and Qi (2002) using an information model. An opposing minority is not lacking, notably that of Kaplan and Zingales (1995). Erickson and Whited (2000) even argue that when measurement errors in q are accounted for, the relationship between investment and cashflow disappear. This paper, however, will not tackle the issues in the preceding paragraph, as it will be limited to the first of the conclusions of FHP (1988). Confirming such a relationship would reject the purely neoclassical theory and cannot but hint at the presence of imperfections in the capital market. The other variables, including those that Aquino (2000) found to have some significance, will also be investigated namely: Debt-to-equity ratio. A high debt-equity ratio may put a restraint on investments (a negative relationship) as it could put a limit on borrowing capacity to finance investments particularly if existing shareholders are unable or unwilling to issue new shares as this may dilute their control over the corporation. The reverse may also be true, that increased investments may result in higher debt-equity ratio (a positive relationship). Fixed Capital Intensity. The ratio of Fixed Assets over Total Assets reflects the fixed capital intensity of the firm’s operations. Setup costs related to high fixed capital expenditures may put some restraint on additional investments (a negative relationship).

2

Revenue Growth Rate. This is analogous to the acceleration principle in macroeconomics. Consistent increases in revenues are supposed to generate confidence and expectations of growth and additional investments. Firm Size. Yu (2003) found firm size as a significant determinant of capital structure. The size of the firm could have a bearing on how much access it has to capital markets. Larger firms may be more diversified, enjoy easier access to capital markets, receive higher credit ratings for their debt issues, and pay lower interest rates on borrowed funds. This may then be a financial constraint that could affect investment. Growth Rate of Gross National Product. The growth of GNP in real terms reflects the general state of the economy. An expansionary economy implies more investment opportunities and should thus translate into more investment at the firm level. The confidence it generates, i.e. expectations of the future are positive, should have the same effect.

III. Data Sources and Coverage Corporate data for the study was obtained mainly from Technistock, Inc. Where lacking, information was completed using the Philippine Corporate Handbook, August 2001 and October 2002 editions.5 Macroeconomic data came from the Philippine Institute of Development Studies website. To have more robust conclusions the study covers panel data from 1989-2002 of all currently listed firms.6 The Philippine Stock Exchange website had 233 firms listed as of August 2003. Excluding those in the financial sector and others using the criteria7 followed by Yu (2003) leaves a total of 145 companies. Many firms either listed only years after 1989 or may simply did not have data earlier than 1996. For a few companies, where not all required data fields were available from Technistock, figures prior to 1996 5

Darwin Yu graciously provided encoded data from these volumes.

6

Baltagi (1995) cites the benefits of using panel data: a) It controls for individual heterogeneity; b) it gives more informative data, more variability, less collinearity among the variables, more degrees of freedom and more efficiency; c) it is better able to study the dynamics of adjustment; and d) better able to identify and measure the effects that are simply not detectable in pure cross-sections or pure time-series data. The limitations lie mainly in assembling the panel data: problems in collection and selectivity, and the usually short time-series duration of available information. The experience in this research confirmed this: assembling a decent panel required much time and patience. 7

Excluded were those considered dormant (no or minimal revenues in the last five years either because they were in the pre-operating stage, shut down, or undergoing reorganization); those with average growth rates of over 1000% (indicating a change of focus of business purpose); those with negative equity (thus giving no meaningful debt equity ratio); and those with less than three years of operation (data considered not enough to even out fluctuations in the business cycle). These criteria were adapted and their application reviewed to cover the longer period.

3

(i.e. which could not be filled up by the Corporate Handbooks’ 1996-2001) had to be discarded. In general, 2002 data were scanty. Thus, the resulting data set was an unbalanced panel. The Appendix lists the firms included in the sample.

Definition of variables The variables used in the regression models are defined as follows, and indicated in boldface. Investment growth INV =

It K  K t 1  t K t 1 K t 1

Kt = Fixed Assets + Other Long-term Assets = Total Assets – Current Assets In the absence of data on annual investment, the change in the firms’ capital stock Kt was used. q-proxy Q=

TD  MC TD  SE

TD = Total Debt = Short-term debt + Long-term debt MC = Total Market Capitalization SE = Total Shareholders’ Equity This proxy for q was formulated and used by Aquino (2002).8

Cash Flow CF =

CFt K t 1

CFt = After Tax Net Income + Depreciation & Amortization The firms’ cashflow is scaled using the previous year’s amount of capital stock. 8

Tobin’s formulation under certain assumptions is Q = Market Value / Replacement Cost. The difficulty in estimating the denominator given the lack of financial data motivated the formulation of this surrogate measure. Aquino (2002) justifies its acceptability by its very high correlation with values of Tobin’s q obtained using a more accurate procedure.

4

Leverage9 DE =

TD SE

Revenue Growth REVt  REVt 1 REVt 1

REVG =

REVt = Total Revenue Fixed Capital Intensity10 FATA =

FixedAssets TotalAssets

Firm Size LNREV = ln Rev

Macroeconomic factor GNPR = real annual growth rate of Gross National Product

Dummy variables HI = equals 1 if the company is a holding firm, zero otherwise PI = equals 1 if the company is in the property sector, zero otherwise MI = equals 1 if the company is in the mining sector, zero otherwise OI = equals 1 if the company is in the oil sector, zero otherwise

9

Yu (2003) found as more relevant the DE ratio defined in terms of TD rather than in terms of STD

10

Initially this was defined as FCI = (Fixed Assets + Other Long Term Assets)/Total Assets = K/TA, following one of the determinants for capital structure in Yu (2003). While this variable has little correlation with FATA, and if included in the regression model was also significant, it does not have a theoretical foundation in the context of investment. The calculated coefficient of FCI has a sign opposite to that of FATA. Moreover, its inclusion in the place of or in addition to FATA does not significantly alter the fit of the model.

5

Descriptive Statistics and Correlation Analysis The initial data set was severely skewed by some outliers. Following Liu and Qi (2002) the extreme upper (for INV, Q, CF, DE, and REVG) and lower (for Q and CF) 0.5 to 1% of observations were excluded.11 The descriptive statistics of the modified data set follows: Table 1 Descriptive Statistics of Dependent and Independent Variables

Mean Median Maximum Minimum Std. Dev. Skewness Observations Cross sections

INV

Q

CF

DE

REVG

FATA

LNREV

0.305 0.097 9.756 -1.000 0.824 5.064

2.215 1.095 34.903 0.164 3.371 4.359

0.132 0.081 1.896 -0.966 0.260 1.901

0.583 0.299 11.153 0.000 0.935 5.042

0.402 0.097 28.093 -1.378 1.998 8.715

0.362 0.345 0.992 0.000 0.270 0.291

20.330 20.603 25.648 8.006 2.349 -0.733

1275 145

1305 145

1240 145

1480 145

1315 145

1448 144

1475 145

Checking for pairwise correlation among the independent variables shows relatively low correlation, little danger of multicollinearity in the regression model. Furthermore, according to Baltagi (1995) panel data are more likely to give less collinearity among the variables. Table 2 Pairwise Correlation Matrix of Independent Variables

Q CF DE REVG FATA LNREV GNPR

Q

CF

DE

REVG

FATA

LNREV

GNPR

1.0000 0.2161 -0.0743 0.1173 -0.0278 -0.1439 -0.0146

0.2161 1.0000 -0.1592 0.1177 -0.0205 0.2594 0.0000

-0.0743 -0.1592 1.0000 -0.0374 0.2140 0.2036 0.0097

0.1173 0.1177 -0.0374 1.0000 -0.0569 -0.0347 0.0995

-0.0278 -0.0205 0.2140 -0.0569 1.0000 0.2439 -0.0195

-0.1439 0.2594 0.2036 -0.0347 0.2439 1.0000 0.0045

-0.0146 0.0000 0.0097 0.0995 -0.0195 0.0045 1.0000

11

Liu and Qui (2002) applied this procedure in order to eliminate truly extreme values that could distort the true relationships among the variables. In our data set, these outliers appear to be meaningless figures, e.g. Q = 228 (mean = 2.2, median = 1.1), INV = 8829.2 (mean = 0.3, median = 0.1). Removing the upper 0.5% of INV, for example, changed its skewness from 36 to 5.

6

IV. Methodology Variations of the following model were estimated using Pooled Ordinary Least Squares, and where applicable, Generalized Least Squares: INVit   E  1Qit   2 CFit   3 DEit   4 REVGit   5 FATAit   6 LNREVit   7 GNPRit   it

  H HI it   P PI it   M MI it   O OI it  (  H 1Qit   H 2 CFit   H 3 DEit   H 4 REVGit   H 5 FATAit   H 6 LNREVit ) HI it  (  P1Qit   P 2 CFit   P 3 DEit   P 4 REVGit   P 5 FATAit   P 6 LNREVit ) PI it  (  M 1Qit   M 2 CFit   M 3 DEit   M 4 REVGit   M 5 FATAit   M 6 LNREVit ) MI it  (  O1Qit   O 2 CFit   O 3 DEit   O 4 REVGit   O 5 FATAit   O 6 LNREVit )OI it

The observations are firm-year, for firm i = 1, 2, …, 145, and year t = 1990, 1991, …, 2002. The designation of the intercept as  E indicates the estimation approach for variants of the model, i.e. common intercept, fixed effects, random effects. The development of the final model went hand in hand with the results of the various regressions performed. Some variables were consistently significant in all models (using a 5% significance level unless otherwise stated) while some came in and out. The goal was to have the best fit possible (measured by the adjusted R2). In all variants, where it is applicable, the p-value of the F-statistic was 0.000000. The values of the R2 and adjusted R2 and the Durbin-Watson statistic for each variant is summarized in Table 3 and will be used as reference in the following discussion. Table 3 Summary of model development Dummy variables

Intercept type

R2

Adjusted R2

DW stat

none intercept none none

common common fixed effects random effects

0.144 0.157 0.315 0.298

0.138 0.148 0.204 0.293

1.591 1.609 1.961* 1.922*

5

slope

common

0.246

0.224

1.751

6 7 8

slope and intercept slope significant slope

common fixed effects fixed effects

0.249 0.383 0.375

0.223 0.265 0.271

1.759 2.070* 2.036*

Model 1 2 3 4

9 significant slope random effects 0.342 0.335 1.973* * Serial correlation is ruled out for these models at both the 1% and 5% significance levels; neither can be said of the others.

7

Stage 1 1. Basic equation without any dummy variables, i.e. INVit   E   1Qit   2 CFit   3 DEit   4 REVGit   5 FATAit   6 LNREVit   7 GNPRit   it

OLS using a common intercept, . Low explanatory power. 2. Added intercept dummy variables, i.e.   H HI it   P PI it   M MI it   O OI it OLS using a common intercept, . No significant improvement. 3. Basic equation without any dummy variables,12 but using OLS13 with fixed effects for the intercept. The fit significantly improved. 4. Basic equation again without any dummy variables, but using random effects14 for the intercept. The method turns into GLS. More improvement in explanatory power.

Stage 2. The heterogeneity of the sample points to the need to distinguish at least between sectors (thus the chosen dummy variables) for the model to have greater predictive power. Given insignificant effect of using intercept dummies, it is reasonable to expect variations across sectors to be captured in the coefficients15 of the regressors. This explains the formulation of the rest of the model. Thus, for a holding firm, for instance, the coefficient for Q would be  1   H 1 . 5. Basic equation plus slope dummy variables. OLS with common intercept. The resulting fit is better than (3) but worse than (4). 12

Intercept dummy variables cannot be used in the fixed effects and random effects models since the variations in the intercept across cross sections accounted for by the fixed/random effects.

13

Both the fixed effects and random effects models assume that time-independent variations across units can be captured in the differences in the constant term, i. The fixed effects model assumes these to be correlated with the regressors. A set of dummy variables, one for each unit, is included in the estimation. The resulting method is known as the least squares dummy variable (LSDV) model. This pays a sizeable penalty in terms of the number of degrees of freedom lost, i.e. the number of cross section units (cf. Greene (1997)). 14

The random effects model, on the other hand, assumes that these time-invariant unit-specific differences

i are uncorrelated with the other regressors. Feasible GLS is one of the methods that can be used in estimating this model. 15

Liu and Qui (2002) follow the same tack although their model involves only 2 independent variables and two dummy variables.

8

6. Added intercept dummies to (5). No significant difference. 7. Same as (5) but using fixed effects. Improved fit. 8. Same as (7) but using only the significant slope dummy variables from that regression, i.e. HI*CF, HI*DE, HI*FATA, PI*LNREV. This gives basically the same results but done mainly in preparation for the next variant.16 9. Same as (8) but using random effects. This is the best fit attained. The resulting model for which the results will be presented is: INVit   E  1Qit   2 CFit   3 DEit   4 REVGit   5 FATAit   6 LNREVit   7 GNPRit  (  H 2 CFit   H 3 DEit   H 5 FATAit ) HI it  (  P 6 LNREVit ) PI it   it

V. Discussion of Results Tables 4 and 5 summarize the estimation results for the Fixed Effects and Random Effects models respectively. The fixed and random effects for each firm are not shown. Both models are estimated since there is no simple rule to determine which is more appropriate for each case.17 Consistent results across the two would lend greater weight to the findings of the study. Two of the independent variables, Q and Cashflow, are highly significant, even at the .1% level, in both models. The signs of the coefficients are consistent with what theory predicts. A high value of Q, i.e. greater than 1, would mean that the value of every additional unit of capital is more than its cost. Thus, it pays to invest more. The higher the value of Q the higher the investment rate to be expected. Caution should be exercised though in taking this observed high significance as a validation of the q-theory of investment for the Philippines. The surrogate measure used for q is rather a rough one that would admit of certain adjustments. There is also the issue of marginal q being the appropriate measure instead of average q.

16

The random effects model could not be estimated with all the slope dummies included.

17

In the words of Johnston and Di Nardo (1997) “there is no simple rule to help the researcher navigate past the Scylla of fixed effects and the Charybdis of measurement error and dynamic selection” in reference to the pros and cons of choosing between the two models.

9

Nevertheless, that Q is significant suggests that the model adequately controls for changes in investment opportunities. The highly significant coefficient of cashflow means that the neoclassical theory is not sufficient to explain investment levels. This suggests the existence of capital market imperfections as FHP (1988) predicted and verified. The magnitude of the coefficient of CF (comparable to that for constrained firms in FHP (1988)) and the consistent high significance level may suffice to validate their first conclusion for the Philippines. However, it would be best to proceed to their second step and identify just what could be the causes of such imperfections, which lead to a hierarchy of financing choices, in the line of the other studies mentioned earlier. This is a subject for further study.

Debt-Equity Ratio and Firm Size are insignificant in both models. This means that they neither restrain nor facilitate investments. A high DE ratio, for example, does not necessarily mean that a firm cannot borrow more in order to finance investment. This is consistent with the theory of independence of the investment and financing decisions. However, this may beg the question since the condition of perfect capital markets is not fulfilled. In any case, the insignificance of these variables disqualifies them as possible measures of financial constraint for the subsequent studies suggested. Revenue Growth is significant at the 1% level in the model with more explanatory power but only at the 10% level in the other. The relationship is as predicted by theory and consistent with the results of Aquino (2000) though with a much smaller coefficient. Fixed Capital Intensity is more ambivalent: highly significant in one but does not even make it to the 10% level in the model with the better fit. In both cases, however, the sign of the coefficient is correct. Higher setup costs associated with high percentage of Fixed Assets could be a constraint on investment. The state of the economy as measured by Real GNP growth has a similar effect, albeit of smaller magnitude and lower significance, as Revenue Growth. The investment opportunities offered by a growing economy may not necessarily translate to what a firm can immediately take advantage of. The opportunities at hand are supposed to be captured by Q. On the other hand, if the mechanism is the optimism on the future, lagged values of GNPR may better explain current investment. We note the high and consistent significance of the dummy variable for Holding companies, specifically for the coefficient of Cashflow. The total sensitivity of the investment rate to cashflow for holding firms which sums up to 2.1 is much higher than the 0.4 for the rest. What makes the investment pattern of a holding firm different from one which is not, regardless of the sector of the industry to which it belongs (the other sectoral dummy variables are not significant)? Or the question could be how different are the financial constraints for holding companies from those which are not. In FHP (1988) higher financial constraints imply greater sensitivity of investment to cashflow. These questions may only be answered through further study.

10

Table 4 Fixed Effects Model Method: Pooled Least Squares Sample(adjusted): 1990 2002 Included observations: 13 after adjusting endpoints Number of cross-sections used: 143 Total panel (unbalanced) observations: 1076 Cross sections without valid observations dropped Variable Q CF DE REVG FATA LNREV GNPR HI*CF HI*DE HI*FATA PI*LNREV R-squared Adjusted R-squared F-statistic

Coefficient 0.0255 0.4224 -0.0030 0.0304 -0.7849 0.0174 0.0099 1.8601 0.1531 1.0049 0.1567 0.3752 0.2715 3.6185

Std. Error 0.0073 0.0942 0.0254 0.0169 0.2228 0.0216 0.0076 0.2102 0.0588 0.4083 0.0586

t-Statistic 3.4771 4.4835 -0.1201 1.8009 -3.5235 0.8043 1.3134 8.8490 2.6040 2.4609 2.6736

Durbin-Watson stat Prob(F-statistic)

Prob. 0.0005 0.0000 0.9044 0.0720 0.0004 0.4214 0.1894 0.0000 0.0094 0.0140 0.0076 2.0357 0.0000

Table 5 Random Effects Model Method: GLS (Variance Components) Sample: 1990 2002 Included observations: 13 Number of cross-sections used: 143 Total panel (unbalanced) observations: 1076 Cross sections without valid observations dropped Variable Intercept Q CF DE REVG FATA LNREV GNPR HI*CF HI*DE HI*FATA PI*LNREV R-squared Adjusted R-squared

Coefficient Std. Error t-Statistic 0.0598 0.2156 0.2776 0.0287 0.0064 4.5021 0.3991 0.0809 4.9306 0.0160 0.0210 0.7629 0.0420 0.0157 2.6747 -0.1257 0.0975 -1.2899 -0.0004 0.0105 -0.0344 0.0146 0.0073 2.0032 1.6870 0.1771 9.5285 0.0800 0.0485 1.6495 -0.0377 0.1490 -0.2531 0.0032 0.0039 0.8243 0.3421 0.3353 Durbin-Watson stat

Prob. 0.7814 0.0000 0.0000 0.4457 0.0076 0.1974 0.9726 0.0454 0.0000 0.0993 0.8003 0.4099 1.9730

11

VI. Conclusions The most significant determinants of firm level investment for Philippine listed firms were found to be a proxy for Tobin’s Q, a measure of the investment opportunities available to the firm; and Cashflow which corresponds to its capability to finance those investments from internal funds. To a lesser extent, growth factors such as the firm’s revenue growth rate, and the country’s GNP growth rate, also contribute to determining how much the firm invests. The amount of leverage a firm has and its size are both insignificant with regard to investment decisions that it makes. This may suggest that the investment and financing decisions are indeed independent. Or, as the study would rather take it to be, they are not indicative of financial constraints and thus not relevant in solving the investment question. As earlier discussed, the study admits many areas for refinement in order to validate the conclusions, and also poses questions that need further study.

12

Appendix: Companies Included in the Study

I. Overview of the Philippine Stock Exchange Listed Companies

Industrial Sector

Total Listed

Excluded

Included

86 65 15 10 28 29 233

17 22 9 4 7 29 88

69 43 6 6 21 0 145

Commercial & Industrial Commercial & Industrial -Holding Mining Oil Property Banks & Financial Services Total

II. Listed Companies Included Commercial & Industrial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

ABS AMC ATI BEL BMM CA CAT CBC CEU CIP DGTL DHC ECP EEI EURO FCC FER FEU GLO GPH GSMI ICT ILI IPO JDI JFC KPM LFM LIB LMG LSC

ABS-CBN Broadcasting Corporation Alaska Milk Corporation Asian Terminals, Inc. Belle Corporation Bogo Medellin Milling Company Concrete Aggregates Corp. Central Azucarera De Tarlac Cosmos Bottling Corporation Centro Escolar University Chemical Industries of the Phils. Digital Telecommunications Phils., Inc. Acesite (Phils.) Hotel Corp. Easycall Comm. Phils., Inc. EEI Corporation Euro-Med Laboratories Phil., Inc. Fortune Cement Corporation Leisure and Resorts World Corporation Far Eastern University, Inc. Globe Telecom, Inc. Grand Plaza Hotel Corporation Ginebra San Miguel Inc. Int'l Container Terminal Services, Inc. Interphil Laboratories, Inc. iPeople, Inc. Jardine Davies, Inc. Jollibee Foods Corporation Keppel Philippines Marine, Inc. Liberty Flour Mills, Inc. Liberty Telecoms Holdings, Inc. LMG Chemicals Corporation Lorenzo Shipping Corporation

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

MAH MB MBC MEP MER MJC MMI MON MVC NN PCOR PCP PEP PF PLTL PNC PPC PRC PTT RCM REG RFM SEVN SFI SMC SPI STN SWM TA TEL TFC UCC URC

Metro Alliance Hold'gs and Eqts. Corp. Manila Bulletin Publishing Corp. Manila Broadcasting Company Matsushita Electric Philippines Corp. Manila Electric Company Manila Jockey Club, Inc. Mariwasa Manufacturing Corp. Mondragon Int'l. (Phils.), Inc. Mabuhay Vinyl Corporation Negros Navigation Company, Inc. Petron Corporation Picop Resources, Inc. Premiere Entertainment Productions, Inc. San Miguel Pure Foods Co. Inc. Pilipino Telephone Corporation Philippine National Const. Corp. Pryce Corporation Philippine Racing Club, Inc. Philippine Telegraph and Tel. Corp. Republic Cement Corp. Republic Glass Holdings Corp. RFM Corporation Philippine Seven Corporation Swift Foods, Inc. San Miguel Corporation SPI Technologies, Inc. Steniel Manufacturing Corporation Sanitary Wares Manufacturing Corp. Trans-Asia Oil and Energy Dev't Corp. Phil. Long Distance Telephone Co. Phil. Tob. Flue Curing and Redry Corp. Union Cement Corporation Universal Robina Corporation

13

65 66 67 68 69

V VITA VVT WEB WGA

iVantage Corporation Vitarich Corporation Vivant Corporation Philweb Corporation William, Gothong and Aboitiz, Inc.

110 111 112

SPM UW WPI

Seafront Resources Corporation Uniwide Holdings, Inc. Waterfront Philippines, Inc.

BC DIZ LC MA PX UPM

Benguet Corporation Dizon Copper Silver Mines, Inc. Lepanto Consolidated Mining Co. Manila Mining Corporation Philex Mining Corporation United Paragon Mining Corporation

OPM OV PEC SINO SOC VUL

Oriental Pet. and Minerals Corp. The Philodrill Corporation PNOC Exploration Corp. Sinophil Corporation South China Resources, Inc. Vulcan Industrial and Mining Corporation

Mining Commercial & Industrial - Holding 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109

AAA AB ABA AC ACR AEV AGI ANS APM ATN ATRK BCI BHI BPC BRN BSC CAC CEI CMT DMC EVER FDC FJP FPH FPI HI ION JGS KPH MAC MC MED MEG MHC MIC MPC NXT PWR ROX SGI

Asia Amalgamated Holdings Corp. Atok-Big Wedge Company, Inc. Abacus Cons. Res. and Hold'gs, Inc. Ayala Corporation Alsons Consolidated Resources, Inc. Aboitiz Equity Ventures, Inc. Alliance Global Group, Inc. A. Soriano Corporation Alcorn Gold Resources Corp. ATN Holdings, Inc. ATR Kim Eng Financial Corporation Bacnotan Consolidated Industries, Inc. Boulevard Holdings, Inc. Benpres Holdings Corporation A Brown Company, Inc. Basic Consolidated, Inc. CADP Group Corporation Crown Equities, Inc. Southeast Asia Cement Holdings, Inc. DMCI Holdings, Inc. Ever Gotesco Resources and Hold'gs, Inc. Filinvest Development Corporation F and J Prince Holdings Corp. First Phil. Holdings Corp. Forum Pacific, Inc. House of Investments, Inc. Ionics, Inc. JG Summit Holdings, Inc. Keppel Philippines Holdings, Inc. Macroasia Corporation Marsteel Consolidated, Inc. MEDCO Holdings, Inc. Megaworld Corporation Mabuhay Holdings Corporation Multitech Investments Corp. Metro Pacific Corporation NextStage, Inc. East Asia Power Resources Corporation Roxas Holdings Solid Group, Inc.

113 114 115 116 117 118 Oil 119 120 121 122 123 124

Property 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145

ALI CDC CHI CMP CPV CYBR ELI EPHI FLI KEP KPP LND MRC PHES RLC RLT SMDC SMP SMPH UP ZIP

Ayala Land, Inc. Cityland Development Corporation Cebu Holdings, Inc. C and P Homes, Inc. Cebu Property Venture and Dev't Corp. Cyber Bay Corporation Empire East Land Holdings, Inc. Edsa Properties Holdings, Inc. Filinvest Land, Inc. Keppel Philippines Properties, Inc. Kuok Phil. Properties, Inc. Fil-Estate Land, Inc. MRC Allied Industries, Inc. Phil. Estates Corporation Robinson Land Corporation Philippine Realty and Holdings Corp. SM Development Corporation San Miguel Properties, Inc. SM Prime Holdings, Inc. Universal Rightfield Prop. Hold'gs, Inc. Zipporah Realty Holdings, Inc.

14

References Aquino, Rodolfo (2000). How Do Philippine Listed Firms Make Investment Decisions? Philippine Management Review, Vol. 8, 1-15. Aquino, Rodolfo (2002). Allocative Efficiency of the Philippine Stock Market. University of the Philippines College of Business Administration Discussion Paper. Baltagi, Badi (1995). Econometric Analysis of Panel Data. John Wiley & Sons, 1995. Erickson, Timothy and Toni Whited (2000). Measurement Error and the Relationship Between Investment and q. Journal of Political Economy. Vol 108, no. 5. Fazzari, Steven, R Glenn Hubbard and Bruce Petersen (1988). Financing Constraints and Corporate Investment. Brookings Papers on Economic Activity, 141-195. Fazzari, Steven, R Glenn Hubbard and Bruce Petersen (1996). Financing Constraints and Corporate Investment: Response to Kaplan and Zingales. Working Paper no. 5462, National Bureau of Economic Research. Greene, William (1997). Econometric Analysis. New Jersey: Prentice Hall, 1997. Hansen, Sten (1999). Agency Costs, Credit Constraints and Corporate Investment. Unpublished Paper. Hayashi, Fumio (1982). Tobin’s Marginal q and Average Q: Interpretation. Econometrica, Jan. 1982.

A Neoclassical

Hubbard, R Glenn (1998). Capital-Market Imperfections and Investment. Journal of Economic Literature 36, 193-225. Johnston, Jack and John DiNardo (1997). Econometric Methods. The McGraw-Hill Companies, Inc., 1997. Kaplan, Steven and Luigi Zingales (1995). Do Financing Constraints Explain Why Investment Is Correlated With Cash Flow? Working Paper no. 5267, National Bureau of Economic Research. Liu, Qiao and Rong Qi (2002). Information, Production, Cash Flow and Corporate Investment. Unpublished Draft. Modigliani, Franco and Merton Miller (1958). The Cost of Capital, Corporation Finance, and the Theory of Investment. American Economic Review, June 1958. Romer, David (2001). Advanced Microeconomics. McGraw-Hill Higher Education, 2001. Tobin, James (1969). A General Equilibrium Approach to Monetary Theory. Journal of Money, Credit, and Banking, Feb. 1969. Yu, Darwin (2003). Determinants of Capital Structure for Philippine Listed Firms. University of the Philippines College of Business Administration Discussion Paper.