Financing decisions: lessons from the Spanish experience.

This paper studies the determinants of financing decisions of Spanish firms. There are three main institutional features which distinguish the Spanish financial system from the American one. First, most Spanish banks are at the core of large industrial groups insomuch as they are both the main creditors and the main shareholders of members of these groups. As in Japan, banks in Spain have historically played a very important role in financing and managing companies.

Second, Spanish securities markets are still underdeveloped in four respects: 1) they have low capitalization, 2) their transactions volumes are small, 3) there are high transaction and flotation costs, and 4) there are only a few main players and they have significant market shares. Banks also play a fundamental role in these markets by owning and managing most of the investment funds, insurance companies, and stock brokerage firms.

Third, the Spanish bankruptcy law and the Spanish tax code are characterized by some features which differentiate them from their American counterparts.(1) On the one hand, in contrast to American bankruptcy law, Spanish bankruptcy law gives control of the firm to a pool of the main creditors after bankruptcy has been declared. This reduces the bargaining power of the management in negotiating the reorganization plan with creditors. On the other hand, Spanish firms have available many types of non-debt tax shields, such as different types of investment tax credits and tax incentives for hiring additional workers.

The features of the financial system of a country play an important role in defining the determinants of financing decisions taken by the firms of that country. Only by comparing the decisions taken by firms operating in different financial environments, can we fully analyze the effect of these institutional features. However, most of the research on the determinants of capital structure have been done using mainly US data. Therefore, the study of decisions taken by firms operating in other financial environments should improve our understanding of the determinants of these decisions. In particular, the study of the Spanish financial system is interesting, since different theories have argued that the three institutional features which distinguish it influence financing decisions.

Theories about asymmetric information problems have suggested that firms should be sensitive to who provides the funds, besides whether it is in the form of debt or equity. Since different fund providers have access to different information, and have different abilities to monitor firm behavior, they will be willing to pay different prices for the securities issued by a firm. This concern about the provider of funds should be especially relevant in financial systems where banks play an important role and where organized financial markets are underdeveloped.

It has been argued(2) that the existence of close and long-term relationships between banks and firms is likely to mitigate the asymmetric information and agency problems that arise when debt and equity are diffusedly held. Banks serve as corporate monitors who bear, in a more efficient way, the costs of monitoring and gathering information about firms demanding funds.(3) In underdeveloped financial markets, the flotation of new issues is more difficult and costlier, since markets are less liquid and asymmetric information problems are more severe. It can be argued that in such a situation, there is a positive role for banks to act as holding companies, where the effect of the above market imperfections are minimized for two reasons. First, banks can internally allocate cash among firms belonging to their industrial group, thus, mitigating liquidity problems. Second, banks can monitor more effectively since they have access to detailed information, thereby reducing asymmetric information and agency problems.

Banks also play a fundamental role in Spanish financial markets. Since banks control most of the investment funds and insurance companies, they have the ability to successfully place any issue in which they are interested. Therefore, the decisions about whether to use the private or public market taken by firms belonging to an industrial group should not be related to variables which proxy for their specific asymmetric information and agency problems. They should rather be related to characteristics of the industrial group to which they belong.

Expected bankruptcy costs and tax effects have been suggested as two important determinants of the debt-equity choice of US corporations. Since the Spanish bankruptcy law and the Spanish tax code are different from the American ones, one can expect that the variables which explain the choices of Spanish firms between debt and equity may be different from the variables which explain the choices of American firms.

The econometric methodology used in this paper is intended to capture the relationship between the above-mentioned institutional features of the Spanish markets and the financing decisions of Spanish firms. Most empirical studies on capital structure have tried to explain the variation in debt ratios across firms and industries. They have only been concerned with corporations' relative use of debt and equity. In this paper, I distinguish between factors which affect the choice of type of financing, debt or equity, and factors which determine whether to use the private or public market. I use a multivariate qualitative choice model similar to that used by Mackie-Mason (1990a) to analyze the financing decisions of American firms.

This paper provides substantial empirical evidence that Spanish firms have distinct preferences for different providers of funds (i.e. between using the private and public markets), even after controlling for the type of security. The choice between the different providers of funds is explained by variables which proxy for the existence of asymmetric information and agency problems. The evidence confirms the results obtained by Mackie-Mason (1990a) for American firms.

This study also offers some indirect empirical evidence about the role of banks in the Spanish financial system. I cannot reject the hypothesis that close bank monitoring does not help to mitigate the information problems faced by firms while raising funds in public financial markets. In particular, I show that for Spanish firms belonging to an industrial group led by a bank (and, therefore, likely to be subject to close bank monitoring), variables which proxy the existence of firm-specific asymmetric information and agency problems explain the choice between using private or publicly marketed funds. Moreover, firms belonging to an industrial group led by a bank seem to face the same type of firm-specific information problems as the rest of Spanish firms while raising funds in financial markets.

Finally, I show that results do not provide support for the idea that Spanish firms take their financing decisions following the financing-hierarchy logic implied by Myers' (1984) "Pecking Order Theory." This theory states the existence of a priori hierarchical preference concerning sources of funds. It also claims that firms only use less preferred sources once the most preferred sources are exhausted. However, in this paper I present evidence that supports the hypothesis that the marginal financing decisions of Spanish firms are explained by factors like tax shields, bankruptcy costs, and asymmetric information problems.

The remainder of the paper is organized as follows: Section I reviews some of the theories of capital structure decisions; Section II presents the econometric model and the explanatory variables; Section III presents the empirical results; and finally, Section IV contains concluding remarks.

I. The Capital Structure Puzzle 

In this section, I briefly review some of the major theories of capital structure decisions. Following Myers (1984), I emphasize the distinction between two approaches to explain firms' capital structure: the existence of an optimal capital structure versus the existence of a hierarchy of financing decisions.

A. The Optimal Capital Structure

This view emphasizes the existence of a debt-equity trade-off. Firms are viewed as setting a target debt-to-equity ratio and gradually moving towards the target. In choosing their "optimal" target leverage, firms balance different costs and benefits associated with debt and equity financing.

Debt and equity can be thought of as contingent claims on the firm's cash flows, with different contingencies. The literature on "optimal capital structure"' has been concerned with the study of the contingencies which are specific to each source, and which determine their respective benefits and costs as sources of financing.

In this paper, I will be concerned with three of these contingencies which are usually emphasized as the determinants of the optimal capital structure(see, e.g., Roden and Lewellen, 1995).

1. Tax Claims

The cash flows accruing to debt and equity are taxed differently by the government. Since interest is tax-deductible and dividends are not, debt financing presents a tax advantage. However, as Miller (1977) pointed out, the corporate tax advantage from debt could be offset by personal tax disadvantages. In Miller's aggregate equilibrium, taxes have no effect on the choice of security at the level of the firm, because the after-tax cost of financing with debt equals the cost of equity financing. DeAngelo and Masulis (1980) reintroduced the notion of a tax advantage from debt: additional interest commitments reduce the probability that the firm will be paying taxes and a tax-exhausted firm will lose the benefits of all types of tax shields. In terms of Miller's framework, the declining marginal value of interest tax shields implies a downward-sloping supply curve of taxable corporate bonds. This supply of corporate bonds translates into a lower equilibrium rate of return to bondholders and hence a lower after-tax cost of debt relative to equity.

2. Bankruptcy Costs

Since debt increases the likelihood of bankruptcy, the optimal capital structure involves balancing the tax advantage of debt against the direct and indirect costs associated with financial distress. The probability of encountering financial distress for a given degree of financial leverage is positively related to the variability of operating earnings. Bradley et al. (1984) demonstrate via simulation that the bankruptcy cost argument implies an inverse relation between earnings volatility and financial leverage.

3. Principal-Agent Problems 

As Jensen and Meckling (1976) suggested, limited liability firms with high debt levels may have an incentive to take on excessively risky projects. Debt may also lead to underinvestment in future opportunities because debtholders have first claim on the cash flows from the new projects. If increasing debt raises the likelihood of conflict of interests between bondholders, stockholders, and managers, then in a rational expectations framework, increasing debt raises the cost of borrowing.

B. Financial Hierarchy: The Pecking Order Theory

This theory states that firms have a hierarchical preference over their sources of financing. When issuing securities, firms prefer internal to external financing, and debt over equity. According to this school of thought, firms choose sources low on the hierarchy only when the more preferred are exhausted.(4)

The theoretical foundation for the financial hierarchy theory needs to explain two points: the reason which leads firms to have such hierarchical preferences, and the reasons which determine why sometimes firms face financing constraints that force them to use sources low in the hierarchy. The more solid theoretical explanation for the hierarchical financing preferences of firms has been based on asymmetric information problems. Myers and Majluf (1984) argued that better informed investors would be willing to pay more for new securities than would less informed investors. Thus, firms will prefer to obtain funds from the better informed investors. Therefore, firms will prefer retained earnings to external financing; or bank debt to publicly marketed bonds assuming that the banker has better access to relevant information than do bondholders.

Asymmetric information problems are not only related to the type of financing but also to the provider of funds. Thus, we should expect firms to be sensitive to both the source and type of funds. The main prediction of the theories about asymmetric information problems is that firms are more sensitive about their provider of funds (i.e. asymmetries of information) when the perceived likelihood of an asymmetric information advantage is high, or when the difference in valuation due to asymmetries of information is high. Specific variables to measure these effects will be discussed in Section II, E. In this paper, following the methodology proposed by Mackie-Mason (1990a) I study the factors which determine the capital structure of Spanish firms, emphasizing the distinction between factors that influence the type of financing and those that explain preferences for different types of funds providers (public or private funds).

II. Data and Methodology

A. Previous Studies

Standard empirical approaches to capital structure decisions test whether companies behave as if they have target debt ratios. A firm's choice of financing instrument should depend on the difference between its current and target debt-equity ratio: they should issue equity if they are above their target level and issue debt if they are below. Since target ratios are unobservable, these empirical approaches are concerned with estimating their likely determinants. There are two major difficulties with the empirical implementation of these tests.

First, since flotation costs have a large fixed component, when companies plan new issues, they have to balance the costs of deviating from their target ratio against flotation costs. High-fixed flotation costs may give rise to infrequent issues. Therefore, firms' capital structure may change less frequently than their target ratios. Therefore, more often than not, their current leverage ratios may not be their target ones. Due to these problems, in order to explain a firm's current leverage ratio we need models that specify how firms balance over time the costs of deviating from their target ratio against flotation costs. Unfortunately, the theory does not provide any good model for a firm's adjustment process towards its target leverage ratio.

The second difficulty, as Mackie-Mason (1990b) pointed out, is the simultaneity of financing and investment decisions. Any capital structure theory that does not recognize the irrelevance of capital structure implies that financing and investment decisions are joint, and that they interact with each other. Thus, a firm's optimal leverage ratio should depend on the characteristics of its set of investment opportunities and its investment decisions. To avoid these problems, we need both a good theoretical model to disaggregate investment decisions, and data about investment and operating decisions. Unfortunately, none of these are available. Further, these problems are more likely to be faced when working with data from countries where companies usually face credit rationing problems, as in the case of Spain.

Here, how companies select between financing instruments at a given point in time is the focus, rather than about explaining their current capital structure. A few other papers use a similar statistical approach, such as Marsh (1982) for British companies and Mackie-Mason (1990a) for American firms.

B. The Model Specification

The model used is similar to that used by Mackie-Mason (1990a) to show that American firms have distinct preferences for different providers of funds as well as for different types of financing. This model is an application of the nested logit model of qualitative choice behavior developed by McFadden (1981). The nested logit model has been derived from the theory of probabilistic choice. Probabilistic-choice models assume that agents are rational in the sense that they make choices that maximize their perceived utility. However, they also assume that there can be errors in this maximization problem, due to imperfect perception and optimization, as well as to the inability of agents to measure exactly all the relevant variables. They assume that agents perceive the utility level associated with their decisions as having a random component.

Capital structure theories distinguish between factors that influence the type of financing and those that explain preferences for different types of providers. Therefore, in the model, suppose that each firm may take four different financing choices: publicly marketed equity, publicly marketed debt, private equity, and private debt (bank debt and trade credits). I define an underlying latent random variable, [U.sub.i], which denotes the level of utility associated with the ith financing choice. A firm will take that financing decision whose associated level of utility has the greatest probability of being the highest utility level that can be derived from taking any financing decision.

For each type of financing decision, i, and each firm, n, I define a binary variable, [Y.sub.i,n], which take the value 1 if the nth firm takes the ith financing decision, and 0 otherwise. I also define an underlying latent variable, [U.sub.i,n], to denote the level of utility associated for the nth firm with the ith financing decision. Thus, the observed variables [Y.sub.i,n], are defined ex-post as:

[Y.sub.i,n] = 1 if [U.sub.i,n] = (Max [U.sub.1,n], [U.sub.2,n], [U.sub.3,n], [U.sub.4,n])

[Y.sub.i,n] = 0 otherwise

If I were to use the multinomial logit model (McFadden, 1981), I would write the level of utility associated with each alternative as a linear function of some attributes

[U.sub.i,n] = [X[prime].sub.n][b.sub.i] + [[Epsilon].sub.i,n] (1)

where [X.sub.n] would be the vector of attributes or determinants of the financing choices of the nth firm, [b.sub.i] the vector of parameters to be estimated, and [[Epsilon].sub.i,n] the residual error term, which captures unobserved variations in the attributes of alternatives, firm's preferences, and error in the perception and optimization by the firm. The estimation of parameter vectors, [b.sub.i], could easily be carried out using a maximum likelihood method (ML).(5)

Unfortunately, the multinomial logit model has the property commonly referred to as "independence of irrelevant alternatives." This means that the odds ratio for the ith and jth choices is [e.sup.X[prime]nbi]/[e.sup.X[prime]nbj], which is the same irrespective of the total number of choices considered. If firms are offered an expanded financing choice set, that does not change the odds ratio between the above two choices.

Nested multinomial logit models are one way of avoiding this problem. These models assume that choices are made in a sequential process. It is assumed that each alternative is described by a set of characteristics (the type of financing and the type of providers, in this model) and that at each stage of the process a characteristic is selected with a given probability distribution. McFadden (1981) showed that these nested logit models can be derived from the theory of stochastic utility maximization, in a manner similar to that of the multinomial logit model. In this paper, I estimate two different nested logit models.(6) In the first model, firms decide whether to use private or public funds, and then given the previous choice, they decide whether to choose equity or debt. In the second model, firms choose debt or equity, and then the type of provider. (See Table 1.)

C. Sample and Data

The data used in this paper have been taken from "Central de Balances del Banco de Espana." This data base is a confidential databank of annual financial accounts, including balance sheets, income statements, and sources and uses of funds statements, for almost 2,000 Spanish companies. It has been built by the Spanish Central Bank for research purposes. All companies have incentives to send their data to this data base since it is confidential, and in exchange, they receive aggregated data about their industry competitors. The data set includes 132 listed companies. Firms for which data was available for all the years of the sampling period, December 1984 to December 1988, were included. The final sample contains 82 listed firms, divided into two subsamples. The first subsample contains 47 firms which do not belong to any industrial group led by a bank. The second subsample includes 35 firms "managed" by banks. A firm is considered to belong to an industrial group managed by a bank when more than 20% of its shares are held directly or indirectly by a bank.

Table 1. Nested Logit Model

Decision Tree Nested Logit Model I

                             Publicly Marketed
Private Funds                    Funds

Private      Private       Public Debt      Public
Debt         Equity                         Equity

Decision Tree Nested Logit Model 2

Debt                            Equity

Private      Private       Public Debt      Public
Debt         Equity                         Equity

D. Dependent Variable: Marginal Financing Decision

In order to estimate the econometric model, I had to specify the dependent variable, the qualitative variable which indicates the type of financing decision taken by the firm, and the explanatory variables for these financing choices. Ideally, the sample should include as many observations as single financing decisions taken by all the firms included in the data set. For each observation, I had to classify the financing decision and determine the value of the explanatory variables corresponding to the firm making that financing decision at the precise instant at which the decision was made. Since the data set for the explanatory variables contains only annual accounting statements, I had to assume that firms make their financing decisions once per year. Thus, the sample contains 328 observations of financing decisions. But this assumption implies that any annual observation for a firm might include several "simultaneous" financial decisions.

In order to partially avoid these problems, I have adopted an approach to defining financing choices suggested by Mackie-Mason (1990a). I classify the financing decision as "public equity" if the firm issues stock, and as "public debt" if the firm issues bonds,(7) "private debt" if the firm increases its bank loans but does not make any public issue, and "private equity" in all other cases.

By defining firms' financing decisions in the above manner, I presume that firms follow a sort of a priori financing-hierarchy logic. This a priori logic is consistent with the following empirical evidence found in the sample: 1) all firms have used as a source of financing retained earnings; 2) 95% of the financing decisions involving the issue of new publicly marketed securities were made by firms that in the same year also decided to use private debt (bank loans and trade credit); and 3) only 12% of firms in the sample issue any kind of publicly marketed security.

Given the above a priori financing-hierarchy logic, the dependent variable really records the marginal financing decisions. Hence, the econometric model will determine the explanatory variables which explain companies' marginal financing choices.

The main proposition of the Pecking Order Theory is that firms have a hierarchical preference over their sources of financing. According to this school of thought, firms resort to sources low on the hierarchy, only when they face credit constraints with the more preferred sources. Thus, in the pure Pecking Order Theory only those factors associated with firms' credit and financing constraints should explain marginal financing decisions. However, if some of the alternative theories of capital structure decisions are correct, other factors unrelated to credit constraints should influence these marginal financing decisions.

E. Explanatory Variables

The explanatory variables used in the model fall into three groups. The first consists of variables which measure attributes that different theories of capital structure suggest may affect the firm's debt-equity choice. The second set of variables are selected as indicators of the presence and the magnitude of asymmetric information problems. This set of variables is expected to affect the choice between the private and public market. The third group consists of variables which measure the level of past investment and past leverage ratios. These variables are included to test whether firms behave as if they have target debt ratios, and whether they face credit constraints.

Most of the variables that will be proposed have been used previously in other empirical studies, mainly by Titman and Wessels (1988) and Mackie-Mason (19908). All explanatory variables will be measured for the year prior to the financing decision. Finally, I estimate annual dummy variables A1, A2, and A3 to capture unobservable fixed effects for each year of the sample.

1. Variables Explaining the Choice between Debt and Equity

In Section I, I identified three determinants of the debt-equity choice. Several variables will be used to proxy for these determinants.

Tax Claims. I have argued that tax deductions for depreciation, investments tax credits, and other non-debt items are substitutes for the tax benefits of debt financing. Firms with large non-debt tax shields are expected to use less debt. I use as indicators of these non-debt tax shields the ratio of depreciation expense to total assets (AMO), and a direct estimate of these tax shields (FISC), as proposed by Titman and Wessels (1988).(8) As explained in the introduction, Spanish firms have available many non-debt tax shields. The variable FISC should capture the magnitude of these non-debt tax shields.

Bankruptcy Costs. We know that the more likely the incurrence of bankruptcy costs, the less likely firms are expected to issue debt. I use the variance of annual operating income (VARI) as an indicator of bankruptcy likelihood. As explained in the introduction, Spanish bankruptcy law gives control of the firm to a pool of the main creditors after bankruptcy has been declared. This reduces the bargaining power of the management in negotiating the reorganization plan with creditors and likely mitigates the bankruptcy costs associated with reorganization processes. Therefore, it is unlikely that variables that are indicators of bankruptcy likelihood can explain the choice between debt and equity in Spanish firms.

Principal-Agent Problems. Most capital structure theories - Galai and Masulis (1976), Jensen and Meckling (1976) and Myers (1977) - argue that firms with assets that can not be used as collateral may be expected to use less debt since they face higher agency costs with debtholders and, therefore, a more costly debt. It can be argued that these firms will tend to rely more on bank loans as source of funds, since the effective monitoring role played by banks help to overcome some of these agency costs. We include the ratio of intangible assets to total assets (COLAT) as an indicator of the collateral value of the firm.(9)

2. Variables Explaining the Choice between the Private and Public Market

The theories about information asymmetric problems predict that firms are concerned about the type of providers of funds, besides the type of financing they provide. I selected as variables explaining the choice between the private and public market, several proxies for the existence and magnitude of these asymmetric information problems. The first indicator of potential asymmetric costs is the variance of the firm's operating income (VARI). I assume that the larger this variance the more difficult it will be for investors to predict future earnings based on public information.

The second variable is a dummy variable (DIV) indicating whether firms pay dividends. The literature suggests that dividends are costly signals (due to double taxation of dividends(10)) used by managers to convey private information to markets. Thus, dividend-paying firms are expected to use public-market financing more often than non-dividend-paying firms.

A dummy variable (REGUL) to control firms which belong to industries that are subject to public regulation is also included. Regulated industries are usually expected to have a more stable operating environment. Further, regulators usually collect information and make it publicly available. For these two reasons, asymmetric information problems are likely to be less severe for firms in regulated industries. Hence, these firms should have greater access to the financial markets. I also include the total assets (ACT) as a size variable. The flotation costs for issuing securities are related to firm size. Since flotation costs have an important fixed component, access to the securities markets is expected to be relatively cheaper for larger firms. Moreover, large firms are followed much more closely by investment analysts, therefore, information problems should be lower. Hence, large firms are also expected to rely more than the average firm on public market.

3. Other Explanatory Variables

As discussed in Section I, the hierarchy theory of financing decisions predicts that firms have no optimal leverage ratio. Thus, according to this theory, financing choices should be independent of the current leverage ratio. To test this hypothesis, I include the lagged leverage ratio (RDT) and the absolute value of the difference between this lagged ratio and its average value during our sample period (DRD). I use the average ratio as a proxy for the target ratio.

Further, the hierarchy theory suggests that available liquid assets and the total amount of funds needed to undertake the set of investment opportunities should be an important determinant of firms' financing decisions. Past profitability of a firm, and hence the amount of retained earnings available, and the total amount of funds needed to undertake the set of investment opportunities should be an important determinant of firms' financing decisions. To test this hypothesis, I use the variable (LIQ), defined as the ratio of last period's net cash flow plus cash equivalents on the balance sheet to total assets, as an indicator of available liquid assets, and the ratio of total investments to total assets (INVER) as an estimator of the total volume of required funds.

III. Empirical Results

The empirical results are presented in Tables 2, 3, and 4. Table 2 presents the results of the estimation of all the branches of the first nested logit model for the sample of firms which do not belong to an industrial group. The results of the estimation of the second nested logit model for the same sample of firms are shown in Table 3.

Most of the variables expected to affect either the choice of funds provider or the type of security, have for every branch the predicted signs and significant coefficients. In particular, Tables 2 and 3 show that there are variables which help to explain the choice between private and public funds in the direction predicted on the basis of asymmetric information, and which are not related to preferences between debt and equity. These results give support to the hypothesis that Spanish firms care about who provides the funds, even after controlling for the type of security. This hypothesis is also confirmed by the inclusive value of the first nested logit model. (See Table 2.) The inclusive value captures whether there are characteristics of the firm's unobservable preferences that distinguish public from private funds. Given that the value of this parameter is statistically different from one, I can reject the hypothesis that Spanish firms do not care about whether to use private or public funds. These results are similar to those obtained by Mackie-Mason (1990a) for American firms.

In the following two subsections, I discuss the signs and level of significance of each of the variables used in the econometric models.

A. Variables for Debt and Equity

In Section II, I identified four variables in order to study the determinants of the debt-equity choice. Two of these variables have the predicted signs and significant coefficients: the direct estimate of the non-debt tax shields (FISC) proposed by Titman and Wessels (1988), and the variance of annual operating income (VARI), i.e. the indicator of bankruptcy likelihood. The variable (VARI) also has a statistically significant coefficient, as predicted.

However, the results indicate that the coefficient of the ratio of depreciation to total assets (AMO) is not significant. This result is consistent with those of other empirical studies, e.g. Titman and Wessels (1988), that have failed to find tax shield effects for American firms. For the other variable, the ratio of intangible assets to total assets (COLAT), which is used as an indicator of the collateral value of the firm, I do not find empirical evidence consistent with the predicted effect on debt-equity choice. Nevertheless, the results indicate that the collateral value of the firm facilitates its access to the public markets. Very often this variable is used by external financial analysts as an indicator of a firm's financial health; so it could act as an instrument to [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] convey information to outside investors.

The results obtained for the variables FISC and AMO are consistent with the joint hypothesis that Spanish firms have available many non-debt tax shields and that the variable (FISC) should capture their magnitude. However, the significant coefficient and correct sign that are obtained for the indicator of bankruptcy likelihood do not confirm the hypothesis that Spanish firms are not subject to severe bankruptcy costs due to the special features of Spanish bankruptcy law.

The results also provide strong empirical support for the existence of target leverage ratios for Spanish firms. Thus, the variable (DRD), the absolute value of the difference between this lagged ratio and its average value during the sample period, has significant coefficients with the predicted sign for all debt-equity choices. However, the lagged leverage ratio (RDT) has statistically insignificant coefficients for debt-equity choices, but significant ones for public-private choices. This result also suggests that this ratio could be used by financial analysts as an indicator of the financial condition of the firm, and therefore it should be considered as an informational signal for securities markets.

B. Variables for Private and Public Market Funds

I specified four variables as possible indicators of the existence and magnitude of asymmetric information problems. The empirical results provide support for the predicted effects for two of them: VARI and DIV.

However, the signs of the dummy variable (REGUL) for industries that are subject to public regulation and the size variable (ACT), differ from those predicted. The dummy variable (REGUL) has statistically insignificant coefficients for all choices. On the other hand, the size effect only seems to affect the issue of new equity. One could argue that this empirical evidence is consistent with the fact that flotation costs are generally reported to be greater for equity issues than bond issues.

Finally, it is interesting to note that the ratio of last period's net cash flow plus cash equivalents to total assets, (LIQ), has nonsignificant coefficients, and that the ratio of total investment to total assets (INVER) only has significant coefficients for the choice between publicly marketed bonds and new share issues. These results do not provide support for the financing hierarchy theory, since the past profitability of a firm and the total amount of funds needed to undertake the set of investment opportunities do not appear to be the main determinants of firms' financing decisions. Thus, my results support the hypothesis that the marginal financing decisions of Spanish firms are explained by factors like tax shields, bankruptcy costs, and asymmetric information problems, which are not related to financial rationing problems.

C. Role of Banks

As explained, in Spain, many industrial groups are led by banks. These banks act both as the main creditors and as the main shareholders of member firms. It can be argued that there is a positive role for Spanish banks to act as holding companies. First, given the small size of the Spanish capital markets, banks can internally allocate cash among firms belonging to their industrial group, thus, mitigating liquidity problems. Second, banks can monitor more effectively since they have access to detailed information, thereby reducing asymmetric information and agency problems.

Since Spanish banks control most of the investment funds and insurance companies, they have the ability to successfully place any issue in which they are interested. Therefore, the decisions about whether to use the private or public market taken by firms belonging to an industrial group should not be related to variables which proxy for their specific asymmetric information and agency problems. They should rather be related to characteristics of the industrial group to which they belong.

Table 4 provides empirical evidence concerning this hypothesis. This table presents the results of the estimation of all the branches of the first nested logit model for the sample of firms which belong to an industrial group led by a bank.(11) The variables expected to affect either the choice of type of provider of funds or the type of security have the same signs and level of significance as for the sample of firms not belonging to any industrial group.(12)

Therefore, firms which belong to an industrial group also care about whether to use private and public funds. Moreover, the choice between these two financing sources is explained by factors related to their own specific asymmetric information and agency problems. These results support the hypothesis that bank monitoring does not help to overcome the asymmetric information problems faced by firms while raising funds in public financial markets.

[TABULAR DATA FOR TABLE 4 OMITTED]

IV. Conclusions

The objective of this paper has been to study the determinants of financing decisions of Spanish firms. The study of the Spanish financial system is interesting, since it has institutional features which distinguish it from the often-studied American one.

In this paper, I distinguish between factors which affect the choice of type of financing, debt or equity, and factors which determine whether to raise funds in the private or public market. My results provide substantial empirical evidence that Spanish firms care about who provides the funds (private or public market) besides whether it is in the form of debt or equity. This evidence confirms the results obtained by Mackie-Mason (1990a) for American firms.

In Spain, many industrial groups are led by banks. These banks act both as the main creditors and as the main shareholders of member firms. It can be argued that there is a positive role for Spanish banks to act as holding companies. First, given the small size of the Spanish capital markets, banks can internally allocate cash among firms belonging to their industrial group, thus, mitigating liquidity problems. Second, banks can monitor more effectively since they have access to detailed information, thereby reducing asymmetric information and agency problems. However, the results presented in this paper do not support these hypotheses. Spanish firms belonging to an industrial group led by a bank seem to face the same type of firm-specific information problems as the rest of Spanish firms when raising funds in financial markets.

The author gratefully acknowledges helpful comments from Samuel Bentolila, Jean Dermine, Pekka Hietala, Tim Opler, Gonzalo Rubio, Theo Vermaelen, and especially, from Saugata Banerjee. This paper is partially based on a research project started while the author was at CEMFI (Spain).

1 The appendix describes certain features of the Spanish income tax code.

2 See, for example, Diamond (1984) and Hull and Mollenberndt (1994).

3 For example, Hoshi, Kashyap, and Scharfstein (1990) present empirical evidence on the benefits of the monitoring role of banks for the Japanese economy.

4 An important problem of the pecking order theory is that it is unable to explain some financing decisions that we actually observe in reality. An interesting example are stock repurchases financed by issuing new debt. See Vermaelen (1981).

5 See the work of McFadden (1981) for the proofs and some alternative procedures to implement ML estimations.

6 The nested logit models used here are identical to those proposed by Mackie-Mason (1990a). The interested reader can find there a more detailed explanation of these models.

7 No firm in our sample issued stock and new bonds in the same year. Convertibles and warrants are considered public debt when they are issued and as public equity the year they are effectively covered.

8 FISC=[operating income - interest payments - (income tax payments/Spanish corporate tax rate)]/total assets

9 The larger the value of the COLAT, the lower the tangible assets of the firm.

10 The Spanish income tax code imposes a double taxation of dividends.

11 For these firms, I do not report the results of the estimation of the second nested logit model since they are similar to those in the first subsample.

12 As suggested by a referee, I estimated a pooled model with dummy variable interactions using all those explanatory variables that were significant in the two separate models. The hypothesis that the coefficients of the significant variables are the same across the two groups could not be rejected for all of the variables at the 0.05 significance level. These results are available from the author.

References

Banco de Espana, 1989. "Rentabilidad y financiacion de las empresas que facilitan informacton ala Central de Balances (1984-1987)," Boletin Economico (February).

Bradley, M., G. Jarrell, and E. Kim, 1984, "On the Existence of an Optimal Capital Structure: Theory and Evidence," Journal of Finance (July), 857-878.

DeAngelo, H. and R.W. Masulis, 1980, "Optimal Capital Structure Under Corporate and Personal Taxation," Journal of Financial Economics (January), 3-29.

Diamond, D., 1984, "Financial Intermediation and Delegated Monitoring," Review of Economic Studies (July), 393-414.

Hull, R.M. and R. Mollenbernt, 1994, "Bank Debt Reduction Announcements and Negative Signaling." Financial Management (Summer), 21-30.

Galai, D. and R.W. Masulis, 1976, "The Option Pricing Model and the Risk Factor of Stock," Journal of Financial Economics (January/February), 53-81.

Hoshi, T., A. Kashyap, and D. Scharfstein, 1990, "Bank Monitoring and Investment: Evidence from the Changing Structure of Japanese Corporate Banking Relationship," R. Glenn Hubbard, Ed., Asymmetric Information, Corporate Finance and Investment, Chicago, IL, Univ. of Chicago Press.

Jensen, M.C. and W.H. Meckling, 1976, "Theory of the Firm: Managerial Behavior, Agency Cost and Ownership Structure." Journal of Financial Economics {October), 305-360.

Kim, E.H., 1982, "Miller's Equilibrium, Shareholder Leverage Clienteles, and Optimal Capital Structure," Journal of Finance (June), 301-318.

Mackie-Mason, J., 1990a, "Do Firms Care Who Provides Their Financing," in R.G. Hubbard, Ed., Asymmetric Information, Corporate Finance and Investment, Chicago, IL. Univ. of Chicago Press.

Mackie-Mason, J., 1990b, "Do Taxes Affect Corporate Financing Decisions," Journal of Finance (December), 1471-1493.

Marsh, P.R., 1982, "The Choice Between Equity and Debt: An Empirical Study," Journal of Finance (March), 121-144.

McFadden, D., 1981, "Econometric Models of Probabilistic Choice," in C.F. Mansky and D. McFadden, Eds., Structural Analysis of Discrete Data with Econometric Applications, Cambridge, MA, MIT Press.

Miller, M. and F. Modigliani, 1958, "The Cost of Capital, Corporation Finance and the Theory of Investment," American Economic Review (June), 261-297.

Miller, M.H., 1977, "Debt and Taxes," Journal of Finance (June), 261-275.

Modigliani, F. and M. Miller, 1963, "Taxes and the Cost of Capital: A Correction," American Economic Review (June), 333-391.

Myers, S.C., 1977, "Determinants of Corporate Borrowing," Journal of Financial Economics (November), 147-176.

Myers, S.C., 1984, "The Capital Structure Puzzle," Journal of Finance (July), 572-592.

Myers, S.C. and N.S. Majluf, 1984, "Corporate Financing and Investment Decisions When Firms Have Information that Investors Do Not Have," Journal of Financial Economics (June), 187-221.

Roden, D.M. and W.G. Lewellen, 1995, "Corporate Capital Structure Decisions: Evidence from Leveraged Buyouts," Financial Management (Summer), 76-87.

Titman, S. and R. Wessels, 1988, "The Determinants of Capital Structure Choice," Journal of Finance (March), 1-18.

Vermaelen, T., 1981, "Common Stock Repurchases and Market Signaling," Journal of Financial Economics (June), 139-181.

Appendix. Selected Features of the Spanish Income Tax Code

In Spain, individuals have to pay taxes on both income and property. The income tax brackets go from 20% to 60%. Property tax is much more moderate. For individuals, income from bonds, bank accounts, other investment income (interest on debentures, dividends, rents, etc.), and realized capital gains are taxable as ordinary income. Corporations pay taxes only on income, and the tax rate is 35%. All interest and dividend income is taxable for all corporations. Interest is tax deductible.

Jesus Saa-Requejo is an Assistant Professor of Finance in the Graduate School of Business at the University of Chicago.