The determinants of capital structure choice using linear models: high technology vs. traditional corporations.

ABSTRACT

This study adopts four linear models (multiple regression model variance-component model, first-order autoregressive model, and variance-component moving average model) with 10 independent variables to analyze the important determinants of capital structures of the high tech and traditional industries, respectively. The results of the study show that the determinants of capital structure of the high tech industry are different from that of the traditional industry. In four linear models, the variance-component model has the smallest root MSE in both industries. These indicate that time-series and cross-sectional variations are very important in analyzing the determinants of capital structure in Taiwanese industry. Managers can apply these results for their dynamic adjustment of capital structure in achieving optimality and maximizing firm's value.

1. INTRODUCTION

The revolution and evolution of the new economy has created non-traditional channels for doing business. Newly formed high tech corporations attract many investors and are able to raise money easily to meet their capital needs in the 1990s. As a result, the capital structure or the determinants of the capital structure of the high tech industry seems to be significantly different from that of the traditional industry.

Ever since Myers article on the determinants of corporate borrowing, literature on the determinants of capital structure has grown steadily (Myers, 1984). Titman and Wessels' article on the determinants of capital structure choice took such attributes of firms as asset structure, non-debt tax shields, growth, uniqueness, industries classification, size, earnings, volatility and profitability, but found only uniqueness was highly significant (Titman, 1988). But Harris and Raviv in their seminal article on the subject point out that the consensus among financial economists is that leverage increases with fixed costs, non-debt tax shields, investment opportunities and firm size (Harris, 1991). Leverage decreases with volatility, advertising expenditure, the probability of bankruptcy, profitability and uniqueness of the product. Moh'd, Perry, and Rimbey employ an extensive time-series and cross-sectional analysis to examine the influence of agency costs and ownership concentration on the capital structure of the firm (Moh'd, 1998). Results indicate that the distribution of equity ownership is important in explaining overall capital structure and that managers do reduce the level of debt as their own wealth is increasingly tied-to the firm.

In a more recent article, it seems that financial decisions in developing countries are somehow different (Mayer, 1990). Rajan and Zingales took asset structure, investment opportunities, firm size and profitability as the determinants of capital structure across the G-7 countries (Rajan, 1995). They found that leverage increases with asset structure and size, but decreases with growth opportunities and profitability. Again firm leverage is fairly similar across the G-7 countries. Booth, Aivazian, Demirguc-Kunt, and Maksimovic took tax rate, business risk, asset tangibility, firm size, profitability, and market-to-book ratio as determinants of capital structure across ten developing countries (Booth, 2001). They found that long-term debt ratios decrease with higher tax rates, size, and profitability, but increase with tangibility of assets. Again the influence of the market-to-book ratio and the business-risk variables tends to be subsumed within the country dummies. Moreover, in time-series tests, Shyam-Sunder and Myers show that many of the current empirical tests lack sufficient statistical power to distinguish between the models (Shyam-Sunder, 1999). As a result, recent empirical research has focused on explaining capital structure choice by using cross-sectional tests.

Our focus is on answering three quantitatively oriented questions and proposing a qualitative comment on the rise and fall of high tech dot com companies: 1. Do corporate financial leverage decisions differ significantly between high tech and traditional industries? 2. Do the determinants of capital structure differ significantly between high tech and traditional industries? 3. Is the time-series component important in resolving the conflicting results reported in prior research? The rest of the paper is organized as follows. Section 2 presents the data used in the investigation and four linear models for debt ratio. Section 3, presents a comparative study of the determinants of capital structure choices and an attempt to rationalize the observed regularities. Section 4 offers concluding remarks.

2. DATA SOURCE AND METHODOLOGY

2.1 Data Description

In this study, corporations are classified into two categories: the high tech and the traditional corporations. High tech corporations include electronics, telecommunications, computer hardware, software, networking, information systems, and other related corporations. The rest are traditional corporations such as clothing, textile, trading, agriculture, manufacturing, etc. Corporations with sound financial statements are selected to create a database and it included a total of 168 observations of public trading corporations in Taiwan from 1996 to 1999. There are 21 high tech corporations and 21 traditional corporations. All variables are compiled from the Taiwan Economic Journal (TEJ). Basic statistics are estimated to describe each variable collected and T-tests are conducted to determine if variables of high tech corporations are different from that of traditional corporations. Correlation coefficients (CORR) are estimated to help identify potential problem of multicollinearity for regression models.

As for methodology, four models are used, with the total debt ratio (DEBT) as the dependent variables, and firm size (SIZE), growth opportunities (GRTH), profitability (ROA), tangibility of assets (TANG), non-debt tax shields (NDT), dividend payments (DIV), and business risk (RISK) as independent variables of corporation's feature. In each model, there are three external macro-economic control variables: capital market factor (MK), money market factor (M2), and inflation level (PPI).

2.2 Multiple linear regression models

In order to test the relationship between capital structure and its determinants, the following multiple regression equation is proposed for the panel data.

Model 1: [DEBT.sub.it]=[a.sub.0]+[a.sub.1]L[SIZE.sub.it]+[a.sub.2] [GRTH.sub.it]+[a.sub.3][ROA.sub.it]+[a.sub.4][TANG.sub.it]+[a.sub.5]ND [T.sub.it]+[a.sub.6][DIV.sub.it]+[a.sub.7][RISK.sub.it]+[a.sub.8] [MK.sub.it]+[a.sub.9]M[2.sub.it]+[a.sub.10]PP[I.sub.it]+ [u.sub.it], i=1, ..., N; t=1, ..., T

The following notation is used to define the variables in the empirical model:

DEBT = the total book-debt / total assets;

LSIZE = In (Asset Size);

GRTH = average sales growth rate over the previous two year;

ROA = the earnings before interest and tax divided by total assets;

TANG = fixed assets / total assets;

NDT = ratio of depreciation, investment tax credit, and tax loss carry forward to total assets;

DIV = dividend payout ratio;

RISK = variance of the return on assets;

MK = rate of return of the overall stock market;

M2 = annual growth rate;

PPI = producers' price index

The estimation procedure involves two steps. In step one, each variable is normalized by subtracting its mean value and divided by its standard deviation to have zero mean value and unity variance for all variables. As a result, we will not have an intercept in our results and we can determine the relative importance of each independent variable in explaining variations of the dependent variable based on its estimated coefficient. Variance inflation factor (VIF) is estimated for each independent variable to identify causes of multicollinearity. Pending on the results of step one, model one is re-estimated in step two by deleting variables with insignificant coefficient or significant VIF value one at a time (stepwise) (VIF>20 implies [R.sub.j.sup.2] > .95, i.e. independent variable j is highly correlated with other independent variables of the model).

2.3 Variance components models

To determine if there is heteroscedasticity problem, model 1 is estimated based on a "variance-component" model in the later. The error term include three factors (firm, time, and random) and model 1 become:

Model 2: [DEBT.sub.it]=[a.sub.0]+[a.sub.1]L[SIZE.sub.it]+[a.sub.2] [GRTH.sub.it]+[a.sub.3][ROA.sub.it]+[a.sub.4][TANG.sub.it]+[a.sub.5]ND [T.sub.it]+[a.sub.6][DIV.sub.it]+[a.sub.7][RISK.sub.it]+[a.sub.8] [MK.sub.it]+[a.sub.9]M[2.sub.it]+[a.sub.10]PP[I.sub.it]+ [u.sub.it], i=1, ..., N; t=1, ..., T

where, [u.sub.it] = [a.sub.i] + [b.sub.t] + [e.sub.it] (three error components: firm, time, and random).

The error structure is similar to the common two-way random effects model with covariates (Fuller-Battese, 1974). The error components are estimated by the fitting-of-constants method, and the regression parameters are estimated with generalized least squares which resulted in unbiased and asymptotically normally distributed estimates. Independent variables with insignificant t-ratio are deleted from the model to finalize our estimates.

2.4 First-order autoregressive models

To determine if there are autocorrelation problems, the Parks method is adopted to estimate model 1 (Parks, 1967). The method assumes a first-order autoregressive error structure with contemporaneous correlation between cross sections. The covariance matrix is estimated by a two-stage procedure leading to the estimation of model regression parameters by GLS. The model and error term become:

Model 3: [DEBT.sub.it]=[a.sub.0]+[a.sub.1]L[SIZE.sub.it]+[a.sub.2] [GRTH.sub.it]+[a.sub.3][ROA.sub.it]+[a.sub.4][TANG.sub.it]+[a.sub.5]ND [T.sub.it]+[a.sub.6][DIV.sub.it]+[a.sub.7][RISK.sub.it]+[a.sub.8] [MK.sub.it]+[a.sub.9]M[2.sub.it]+[a.sub.10]PP[I.sub.it]+ [u.sub.it], i=1, ..., N; t=1, ..., T

where, [u.sub.it] = [[rho].sub.i] [u.sub.i,t-1] + [[epsilon].sub.it]

For a stable Parks' estimate, the absolute value of [[rho].sub.ii] should be less than unity.

2.5 Variance-component moving average models (VCMA)

To determine if there are heteroscedasticity and autocorrelation problems, the Da Silva [2] method is adopted to estimate the VCMA model. This method estimates the regression parameters using a two-step GLS-type estimator. The model and error term become:

Model 4: [DEBT.sub.it]=[a.sub.0]+[a.sub.1]L[SIZE.sub.it]+[a.sub.2] [GRTH.sub.it]+[a.sub.3][ROA.sub.it]+[a.sub.4][TANG.sub.it]+[a. sub.5]ND [T.sub.it]+[a.sub.6][DIV.sub.it]+[a.sub.7][RISK.sub.it]+[a.sub.8] [MK.sub.it]+[a.sub.9]M[2.sub.it]+[a.sub.10]PP[I.sub.it]+ [u.sub.it], i=1, ..., N; t=1, ..., T

where [u.sub.it] = [a.sub.i] + [b.sub.t] + [e.sub.it], (three error components) and [e.sub.it] = [][[epsilon].sub.t] [[].sub.l][[epsilon].sub.t-1] [[].sub.m][[epsilon].sub.t-m].

As the data set only includes a four year period, model four is estimated based on the first order of moving average error process, i.e. [e.sub.it] = [[alpha].sub.0] [[epsilon].sub.t] + [[alpha].sub.l][[epsilon].sub.t-1]. We compare and interpret results of all estimates to determine if more advanced models will generate better results than that of the basic multiple regression. More advanced models assume their residual values behave differently and offer more options for better estimates. Conclusions and policy implications are then drawn as valuable tools for managers in achieving optimal capital structure.

3. EMPIRICAL RESULTS

"Table 1" presents descriptive statistics of all variables and T-tests for variable difference between high tech and traditional corporations. The results indicate that the total debt ratio, firm size, and tangibility of the high tech corporations are insignificantly different from that of the traditional corporations. The growth opportunities (higher), profitability (higher), non-debt tax shield (higher), dividend policy (lower), and business risk (higher) of the high tech corporations are significantly different from that of the traditional corporations. Therefore, it can be inferred that although the capital structure measured by debt ratio of the high tech corporation is insignificantly different from that of the traditional corporations, the determinants of capital structure of the high tech corporations can be significantly different from that of the traditional corporations.

3.1 Multiple linear regression models "Table 2" presents the results of multiple regression model. The results indicate that 1) all three external macro-economic variables are insignificantly associated with the capital structure for both the high tech and traditional corporations; 2) the estimated VIF coefficients of all three macro-economic variables are high, i.e. VIF > 20 or [R.sub.j.sup.2] > .95, which would create multicollinearity to end up with inefficient estimates; and 3) the estimated root MSE are relatively high for both the high tech and the traditional corporations as all variables have been normalized. To improve the estimates, insignificant variables with high VIF were deleted one at a time (stepwise) and the results are presented in column 2 of "Table 3 and 4". Compare to the results in column 2 of "Table 2, 3, and 4" virtually have the same implications with no statistical improvement.

3.2 Variance components model

The Fuller-Battese method is used to estimate model 2 with the assumption that the error term [u.sub.it] = [a.sub.i] + [b.sub.t] + [e.sub.it]. Following the same stepwise procedure, the Fuller-Battese's variance components model was estimated. The results are presented in column 3 of "Table 3 and 4". Compared to results of the multiple regression, results of the error-component model have been improved statistically: 1) the error-component model is more efficient empirically in terms of its estimated root MSE; 2) all external macroeconomic variables become endogenous; and 3) the error component model is a more advanced model theoretically.

3.3 First-order autoregressive model

The Parks method is used to estimate model 3 with the assumption that the error term [u.sub.it] = [[rho].sub.i] [u.sub.i,t-1] + [[epsilon].sub.it], where [[rho].sub.i] is the first-order autoregressive coefficient. For a stable Parks' estimate, the absolute value of [[rho].sub.i] should be less than unity. Unfortunately, the estimated value of [[rho].sub.i] in the model is 1.057, which did not satisfy this criteria, but the RMS values are 0.2765 and 0.2939 for the high tech and the traditional corporations, respectively, are smaller than Model 1. The results are presented in column 4 of "Table 3 and 4".

3.4 Variance-component moving average model (VCMA)

The Da Silva method is used to estimate model 4 where the error term is assumed as [u.sub.it] = [a.sub.i] + [b.sub.t] + [e.sub.it] and [e.sub.it] = [[alpha].sub.0][[epsilon].sub.t] + [[alpha].sub.l] [[epsilon].sub.t-1]. Compared directly to the model 2, only the existence of a significant [[alpha].sub.l] the first-order autoregressive coefficient will make a difference to these two models. The results in column 5 of "Table 3 and 4" indicate that the estimated [[alpha].sub.l] is extremely small (0.0019 and 0.0000 for the high tech and traditional corporations, respectively). Hence, the Da Silva's model does not add any incremental improvement to the Fuller-Batese's model.

The Fuller-Batese's variance-component model has the smallest RMS which will produce a more reliable forecast in generating tools for optimal capital structure. These indicate that the time-series and cross-sectional variation in firm leverage are very important factors in model fitting. The third columns of "Table 3 and 4" summarize regression results for Model 2 as described in Section 2. However, the impact of the independent variables is not completely uniform across industries. For example, the sign on the business risk is positive for high tech corporations, but turns negative for traditional corporations. The growth opportunity variable is positive and highly significant for traditional corporations, and no effect for high tech corporations. The overall importance and signs on the coefficients for size, profitability, non-debt tax shield, and dividend payout ratio are similar in both industries.

Based on the results of the Fuller-Batese's model, each determinant of capital structure is discussed as follow:

(1) Firm size ([X.sub.1]) measured by corporation's total asset

Many previous studies (Harris & Raviv, 1991) and Booth, Aivazian, Demirguc-Kunt, & Maksimovic 2001) argued that the capital structure might be affected by firm size positively as larger firms are more able to borrow money to realize the benefits of financial leverage. The results of this study are consistent with this presumption. Both high tech and traditional corporations with larger size had higher debt ratio.

(2) Growth opportunities ([X.sub.2]) measured by revenue growth rate

Myers identified growth opportunities had significant and negative impact on capital structure based on the argument that firms with higher investment in intangible assets are to use less debt to reduce the agency costs associated with risky debt (Myers 1977). In contrary, this study found that growth opportunities had insignificant impact on capital structure for the high tech corporations and positive impact on capital structure for the traditional corporations. In combining with the results of phase I, it seemed that most high tech corporations are characterized by high growth opportunities (homogeneity) and therefore we could not separate and elicit the impact of high growth opportunities on capital structure statistically. Traditional corporations with higher growth opportunities had higher demand for capital to sustain their growth opportunities and borrowed more than their peers with lower growth opportunities.

(3) Profitability ([X.sub.3]) measured by rate of return on total asset

Myers suggested managers have a pecking order in which retained earnings represented the first choice, followed by debt financing, and then equity to meet their financial needs (Myers, 1984). If this is true, it would imply a negative relationship between profitability and the capital structure. The results of this study are consistent with previous studies and confirmed that both the high tech and traditional corporations' profitability had negative and significant impact on capital structure.

(4) Asset structure ([X.sub.4]) (collateral value) measured by total fixed asset/total asset

As higher collateral value would enable firms to borrow more, previous studies suggested that firms' collateral value had a positive relationship with their capital structure. The results of this study indicated that the relationship between firms' collateral value and capital structure is insignificant for both the high tech and traditional corporations. It seemed that firms with higher collateral value might not necessarily exercise their borrowing ability.

(5) Non-debt tax shield ([X.sub.5]) measured by total depreciation/net sales

As non-debt tax shield could lower the benefit of financial leverage, previous studies suggested a negative relationship between the non-debt tax shield and the capital structure. The results of this studies confirmed that both the high tech and traditional corporations had a negative and significant impact on capital structure.

(6) Dividend policy ([X.sub.6]) measured by cash dividend/stockholders' equity

As higher cash dividend payments reflected lower capital demand, previous studies suggested that the relationship between cash dividend and capital structure should be negative. The results of this study confirmed that both the high tech and traditional corporations had a negative relationship between cash dividend and capital structure.

(7) Business risk ([X.sub.7]) measured by variance of firm's profitability

As financial leverage would accelerate firm's profitability and vice versa, it was expected that there would be a positive relationship between capital structure and business risk, especially when business risk is measured by the variance of firm's profitability. The results of this study indicated that there is a positive and significant relationship between business risk and capital structure for the high tech corporations, but insignificant relationship for the traditional corporations. In combining with the results of phase I, it seemed that most traditional corporations are characterized by relatively low business risk (homogeneity) and therefore we could not separate and elicit the impact of business risk on capital structure statistically.

(8) Capital market factor ([X.sub.8]) measured by rate of return of the overall stock market

Refer back to Myers' argument again, "retained earnings always represented the first choice in meeting managers' financial needs," it was expected that there is a negative relationship between capital market factor and capital structure (Myers, 1984). The results of this study confirmed that both the high tech and traditional corporations had a negative and significant relationship between capital structure and rate of return of overall stock market.

(9) Money market factor ([X.sub.9]) measured by annual growth rate of M2

The increase of money supply (M2) implied lower interest rate and created incentives for managers to borrow more (positive relationship). The results of this study confirmed that there is a positive and significant relationship between M2 and capital structure for both the high tech and traditional corporations.

(10) Inflation level ([X.sub.10]) measured by producers' price index

In a similar token, higher inflation level would imply higher interest rate and hence borrowing cost (negative relationship). The results of this study confirmed that there is a negative and significant relationship between inflation level and capital structure for both the high tech and traditional corporations.

4. CONCLUSIONS

This paper partly answers the questions posed in the introduction. It offers some hope, but also some skepticism. Other then growth opportunities, variables which are relevant for explaining capital structures in the high tech corporations are also relevant in the traditional corporations.

Smith and Watts hypothesize that debt ratios of larger firms are less limited by the costs of financial distress, because they have more diversification than smaller firms (Smith, 1992). Consistent with the hypothesis, the coefficient of firm size is positive in both industries.

A consistent result in both industries is that the more profitable the firm, the lower the debt ratio. This finding is consistent with the Pecking-Order Hypothesis. It also supports the existence of significant information asymmetries. This result suggests that external financing is costly and therefore avoided by firms. However, a more direct explanation is that profitable firms have less demand for external financing, as discussed by Donaldson and Higgins (Donaldson, 1963) (Higgins, 1977). This explanation would support the argument that there are agency costs of managerial discretion.

Both industries also support for the role of non-debt tax shields in financing decisions. DeAngelo and Masulis present a model of optimal capital structure that incorporates the impact of corporate taxes, personal taxes, and non-debt-related corporate tax shields (DeAngelo, 1980). They argue that tax deductions for depreciation and investment tax credits are substitutes for the tax benefits of debt financing. As a result, firms with large non-debt tax shields relative to their expected cash flow include less debt in their capital structures.

The major difference of the determinants in both industries is business risk. The coefficient on business risk is negative for traditional and positive for high tech corporations, and traditional corporations have substantially lower ratios of business risk. This is because of the characteristic of high tech industry. Generally, high tech is the more speculative industry. More speculation is associated with high risk. It seems that companies increase their debt financing, when the business risk increase. In traditional industry, business risk is an estimate of the probability of financial distress. It notes that low business risk enhances a firms ability to issue debt.

Thus, the answer to the first two questions posed in the introduction is: both industries seem to have the same debt financing, but the determinants of capital structure of the high tech corporations are different from that of the traditional corporations. The answer to the third question is: the time-series cross-sectional error-component model generates the best fit of the data set.

Implications on the dot com companies

Regarding the qualitative aspects of capital formation within the dot companies of the 90s, we find that beginning about 1995 a mob mentality set in within the investment community. Essentially, no rational reason could be quantified for the ability of the dot corns to attract large amounts of investment capital. That is, on the surface, there seemed to be an irrational behavior within the investment community. If we mine the information deeper, it would be quite rational for the venture capitalists to fund the dot coms to the extent that they did. Examining the phenomenon of the high tech dot coms, several factors come into play. Firstly, the general economy was doing well and the allure of high tech business was irresistible to stock purchasers. The thought that much of the world business would be internet/computer orientated took root and became the glamorous hot issue of the day. Venture capitalist read the fervor and proceeded to fund startup companies in record numbers.

The initial focus of the dot coms wasn't on profit maximization but on customer base accumulation. Standard business plans with cost/profit projections essentially were disregarded because that type of thinking process was passe according to young entrepreneurs and the race was on for developing customer bases in lieu of profitability. It became rapidly apparent to the venture capitalists that there were significant returns to be made by funding pretty much any half baked dot com company and bringing it to IPO status. The stock purchasing public, in the throes of visions of wealth, purchased any new IPO and raised the capitalization of virtually any insignificant dot com company to improbable highs.

Although the stock buying public poured in capital into the dot corns in a totally irrational fashion, the venture capitalist acted in a more rational fashion. They, the venture capitalist, recognized the rush to riches and capitalized startup companies in record numbers and in record time. They also brought these companies to IPO status in a speedy fashion. Once on the open stock exchange, the public ran stock prices to record highs and at this point, acting in a rational fashion, the venture capitalist cashed in their holdings and made record profits for their capital inputs. We can see that historically the dot com market crashed in 2000 and many ordinary purchasers along with institutional buyers were left with record losses. The basis of the crash was that the original concept of a good, solid cost/profit business plan was overlooked for customer base accumulation. It didn't take long for the stock buying public to find out that there wasn't much possibility of profitability from many, if not most, of these companies. Most of the business ideas were impractical from the start so why were they capitalized so well? The only possible answer can't be quantitative but must be addressed from a mob mentality/irrationality point of view. The buying of dot com stock and subsequent high capitalization was the "in" thing to do. The only rational actor during the hay day of the dot coms was the venture capitalist.

Consequently, there is much that needs to be done, both in terms of empirical research as the quality of databases increases, and in developing theoretical models that provide a more direct link between profitability and capital structure choice.

TABLE 1. DESCRIPTIVE STATISTICS AND RESULTS OF T-TESTS

             Mean value        Std. Error        Min. value

Variable   HC (1)   TC (2)   HC (1)   TC (2)   HC (1)   TC (2)

DEBT       0.37     0.39     0.11     0.13     0.12     0.14
LSIZE      6.88     6.99     0.45     0.60     5.98     5.71
GRTH       0.35     0.10     0.35     0.16     -0.35    -0.23
ROA        0.12     0.09     0.07     0.05     -0.07    -0.01
TANG       0.26     0.32     0.17     0.17     0.04     0.04
NDT        0.08     0.05     0.09     0.04     0.001    0.001
DIV        0.21     0.49     0.48     0.59     0.00     0.00
RISK       4.45     2.41     4.67     1.21     0.23     0.27
MK         0.16              0.22              -0.22
M2         8.56              0.38              8.30
PPI        97.83             2.26              95.58

             Max. value

Variable   HC (1)   TC (2)   T-test (3)

DEBT       0.71     0.67     -1.08
LSIZE      7.66     8.01     -1.34
GRTH       1.28     0.80     5.95 *
ROA        0.35     0.23     3.20 *
TANG       0.64     0.68     -1.53
NDT        0.32     0.12     2.79 *
DIV        2.00     3.00     -3.37 *
RISK       24.45    5.91     3.88 *
MK         0.34
M2         9.20
PPI        100.14

* Significant at 5% level; (1) High tech corporations; (2) Traditional
corporations; (3) T for [H.sub.0]: [[mu].sub.1]=[[mu].sub.2] (High
tech corporation = Traditional corporation)

TABLE 2. RESULTS OF TIME-SERIES REGRESSION

(Dependent variable: capital structure measured by debt ratio)

Indep.        High tech    VIF         Traditional   VIF
Variabl

LSIZE         0.4092       2.11        0.7292        1.81
              (0.1380) *               (0.0800) *
GRTH          0.1255       1.50        0.2565        1.25
              (0.1166)                 (0.0664) *
ROA           -0.3500      1.74        -0.3220       1.84
              (0.1253) *               (0.0806)
TANG          0.2546       2.85        0.1790        2.12
              (0.1606                  (0.0865) *
NDT           -0.6569      3.90        -0.3323       2.49
              (0.1877) *               (0.0937) *
DIV           -0.1527      1.57        -0.2404       1.74
              (0.1192)                 (0.0783) *
RISK          0.3341       1.77        -0.1663       1.16
              (0.1264) *               (0.0641) *
MK            -0.7520      27.42 **    0.1563        24.26 **
              (0.4974                  (0.2925)
M2            1.8734       147.24 **   -0.4366       130.97 **
              (1.1527)                 (0.6796)
PPI           -1.9364      178.12 **   0.5262        161.62 **
              (1.2679)                 (0.7549)
Root MSE      0.8654                   0.54105
Adj.          0.3322                   0.7390
  [R.sup.2]
F-value       3.68 *                   20.95 *
Sample size   84                       84

** [R.sub.j.sup.2] > .95 (Independent variable j is highly correlated
with other independent variables) (VIF: Variance inflation factor);

* Significant at 5% level (standard error).

TABLE 3. RESULTS OF FOUR REGRESSION MODELS FOR
HIGH TECH CORPORATIONS

(Dependent variable: capital structure measured by
debt ratio)

Indep. Variable         Multi-Reg.   Var-Comp.

[X.sub.1](size)         0.3453       0.2295
                        (0.1323) *   (0.0445) *
[X.sub.2](growth)       0.0943       N/A
                        (0.1118)
[X.sub.3](ROA)          -0.2539      -0.2793
                        (0.1151) *   (0.1358) *
[X.sub.4](FA%)          0.2214       N/A
                        (0.1578) *
[X.sub.5](t-shield)     -0.6399      -0.4567
                        (0.1873) *   (0.2225) *
[X.sub.6](dividend)     -0.2203      -0.0683
                        (0.1121) *   (0.0221) *
[X.sub.7](risk)         0.3422       0.0050
                        (0.1253) *   (0.0024) *
[X.sub.8](market)       N/A          -0.0947
                                     (0.0393) *
[X.sub.9](M2)           N/A          0.2664
                                     (0.0667) *
[X.sub.10](PPI)         N/A          -0.0351
                                     (0.0086) *
RMS                     0.8693       0.06126
Adj. [R.sup.2]          0.235
Variance ([v.sub.i])                 0.0069
Variance ([e.sub.t])                 0.0000
Variance ([[epsilon].                0.0023
  sub.it])
Sample size             84           84

Indep. Variable         Autoreg.   Var-Comp-MA

[X.sub.1](size)         0.1404     0.1793
                        (0.0974)   (0.0396) *
[X.sub.2](growth)       -0.0419    N/A
                        (0.0304)
[X.sub.3](ROA)          -0.0615    -0.2271
                        (0.2735)   (0.1316)
[X.sub.4](FA%)          -0.0136    N/A
                        (0.2770)
[X.sub.5](t-shield)     -0.7067    -0.6836
                        (0.5105)   (0.2071) *
[X.sub.6](dividend)     -0.0538    -0.0663
                        (0.0261)   (0.0233) *
[X.sub.7](risk)         0.0059     0.0061
                        (0.0030)   (0.0022) *
[X.sub.8](market)       -0.0844    -0.0245
                        (0.0820)   (0.0074) *
[X.sub.9](M2)           0.0876     0.1873
                        (0.1533)   (0.0579) *
[X.sub.10](PPI)         -0.0129    -0.0716
                        (0.0207)   (0.0392)
RMS                     0.2766     1.0385
Adj. [R.sup.2]
Variance ([v.sub.i])               0.0031
Variance ([e.sub.t])               0.0000
Variance ([[epsilon].              [alpha.sub.0]
  sub.it])                           = 0.0059,
                                     [alpha.sub.1]
                                     = 0.0019
Sample size             84         84

* Significant at 5% level.

N/A: independent variable is deleted stepwise.

TABLE 4. RESULTS OF FOUR REGRESSION MODELS FOR
TRADITIONAL CORPORATIONS

(Dependent variable: capital structure measured by
debt ratio)

Indep. Variable         Multi-Reg.   Var-Comp.

[X.sub.1](size)         0.7237       0.2636
                        (0.0789) *   (0.0265)  *
[X.sub.2](growth)       0.2651       0.0747
                        (00628)  *   (0.0254)  *
[X.sub.3](ROA)          -0.3293      -0.2734
                        (0.0790) *   (0.1342)  *
[X.sub.4](FA%)          0.1806       N/A
                        (0.0852)
[X.sub.5](t-shield)     -0.3360      -0.8965
                        (0.0923) *   (0.3952)  *
[X.sub.6](dividend)     -0.2142      -0.0257
                        (0.0718) *   (0.0090)  *
[X.sub.7](risk)         -0.1510      -0.0050
                        (0.0617) *   (0.0034)
[X.sub.8](market)       N/A          -0.1103
                                     (0.02031) *
[X.sub.9](M2)           N/A          0.2027
                                     (0.0304)  *
[X.sub.10](PPI)         N/A          -0.0318
                                     (0.0044)  *
RMS                     0.5352       0.02889
Adj. [R.sup.2]          0.710
Variance ([v.sub.i])                 0.0049
Variance ([e.sub.t])                 0.0000
Variance ([[epsilon].                0.0005
  sub.it])

Sample Size             84           84

Indep. Variable         Autoreg.     Var-Comp-MA.

[X.sub.1](size)         0.0918       0.1985
                        (0.0522) *   (0.0238) *
[X.sub.2](growth)       0.1525       0.1033
                        (0.1793)     (0.0331) *
[X.sub.3](ROA)          -0.2488      -0.3786
                        (0.2351)     (0.1715) *
[X.sub.4](FA%)          -0.0218      N/A
                        (0.1499)
[X.sub.5](t-shield)     0.8613       -0.9615
                        (1.0392)     (0.3802) *
[X.sub.6](dividend)     -0.0372      -0.0309
                        (0.0291)     (0.0122) *
[X.sub.7](risk)         -0.0175      -0.0068
                        (0.0157)     (0.0045)
[X.sub.8](market)       -0.0212      -0.0209
                        (0.0327)     (0.0043) *
[X.sub.9](M2)           -0.0159      0.1344
                        (0.0627)     (0.0352) *
[X.sub.10](PPL)         -0.0006      -0.0811
                        (0.0084)     (0.0287) *
RMS                     0.2939       0.9113
Adj. [R.sup.2]
Variance ([v.sub.i])                 0.0037
Variance ([e.sub.t])                 0.0000
Variance ([[epsilon].                [alpha.sub.0] =
  sub.it])                             0.0016, [alpha.
                                       sub.1] = 0.0000
Sample Size             84           84

* Significant at 5% level.

N/A: independent variable is deleted stepwise.

REFERENCES

Booth, L., Aivazian, V., Demirguc-Kunt, A., and Maksimovic, V., "Capital structures in developing countries", Journal of Finance, Vol. LVI, No. 1,2001,87-130.

Da Silva, J. G. C., The analysis of cross-sectional time series Data, Ph.D. dissertation, Department of Statistics, North Carolina State University, 1975.

DeAngelo, H. and Masulis R., "Leverage and dividend irrelevancy under corporate and personal taxation", Journal of Finance, 35, 1980, 453-464.

Donaldson, G., "Financial goals: management vs. stockholds", Harvard Business Review, 41, 1963, 116-129.

Fuller, W. A. and Battese, G. E., "Estimation of linear models with cross-error structure", Journal of Econometrics, 2, 1974, 67-78.

Harris, M. and Raviv A., "The theory of capital structure", Journal of Finance, 46, 1991, 297-355.

Higgins, R., "How much growth can a firm afford", Financial Management, 1977, 7-16.

Mayer, C., Financial systems, corporate finance and economic development, in R. Glenn Hubbard, ed., Asymmetric Information, Corporate Finance and Investment, University of Chicago Press, Illinois, 1990.

Moh'd M. A., Perry, L.G. & Rimbey, J. N., "The impact of ownership structure on corporate debt policy: a time-series-cross-sectional analysis", The Financial Review, 1998, 85-98.

Myers, S. C., "Determinants of corporate borrowing", Journal of Financial Economics, 5, 1977, 147-175.

Myers, S., "The capital structure puzzle", Journal of Finance, 39, 1984, 575-592.

Parks, R. W., "Efficient estimation of a system of regression equations when disturbances are both serially and contemporaneously correlated", Journal of the American Statistical Association, 62, 1967, 500-509.

Rajan, R. G. and Zingales, L., "What do we know about capital structure? Some evidence from International data", Journal of Finance, 50, 1995, 1421-1460.

Shyam-Sunder, L. and Myers, S., "Testing static tradeoff against pecking order models of capital structure", Journal of Financial Economics, 51, 1999, 219-244.

Smith, C. and Watts, R., "The investment opportunity set and corporate financing, dividend and compensation policies", Journal of Financial Economics, 32, 1992, 263-292.

Titman, S. and Wessels, R., "The determinants of capital structure choice", Journal of Finance, 43, 1988, 1-19.

Hsiao-Tien Pao, earned her Ph.D. at National Chiao-Tung University, Taiwan. Currently, she is an Associate Professor at the Department of Management Science, NCTU, Taiwan.

Bohdan Pikas is a Professor at the Department of Commerce, Niagara University, USA.

Tenpao Lee is a Professor of Logistics at the Department of Commerce, Niagara University, USA.