Theory of Capital Structure - A Review

Transcription

1 Theory of Caital Structure - A Review Stein Frydenberg Λ Aril 29, 2004 ABSTRACT This aer is a review of the central theoretical literature. The most imortant arguments for what could determine caital structure is the ecking order theory and the static trade off theory. These two theories are reviewed, but neither of them rovides a comlete descrition of the situation and why some firms refer equity and others debt under different circumstances. The aer is ended by a summary where the otion rice aradigm is roosed as a comrehensible model that can augment most artial arguments. The caital structure and cororate finance literature is filled with different models, but few, if any give a comlete icture. JEL classification: G32 Λ Sør-Trøndelag University College, Deartment of business administration, Jonsvannsvn. 82, 7004 Trondheim, Norway. stein.frydenberg@toh.hist.no. 1 Electronic coy available at: htt://ssrn.com/abstract=556631

2 The view that caital structure is literally irrelevant or that nothing matters in cororate finance, though still sometimes attributed to us,...is far from what we ever actually said about the real-world alications of our theoretical roositions. Miller (1988) I. Introduction The aer introduces the reader to two main theories of caital structure, which is the static trade-off theory, and the ecking-order theory. Underlying these theories are the assumtions of the irrelevance theorem of Miller and Modigliani. Since the irrelevance theorem is indeed a theorem, the assumtions of the theorem, has to be broken before caital structure can have any bearing on the value of the firm. If the assumtions of the irrelevance theorem are justified, the theorem follows as a necessary consequence. II. The Irrelevance Proosition In comlete and erfect caital markets, research has shown that total firm value is indeendent of its caital structure. An otimal caital structure does not exist when caital markets are erfect. Taxes and other market imerfections are essential to building or roving a ositive theory of caital structure. Changes in caital structure benefit only stockholders and then if and only if the value of the firm increases. An exroriation of wealth from the bondholders would in a rational exectations equilibrium be exected by the bondholders, and the stockholders would ultimately carry the costs of the exroriation. Miller and Modigliani (1958b) wrote the seminal article in this field of research, using an arbitrage argument. If a firm can change its market value by a ure financial oeration, the investors in the firm can take actions that relicate the resulting debt osition of the firm. These transactions would merely change the weights of a ortfolio and should, in a erfect caital market, give zero rofit. If the market were efficient enough to eliminate the rofits for the investors, any rofit for the firm would be 2 Electronic coy available at: htt://ssrn.com/abstract=556631

3 eliminated too. Modigliani and Miller in their original articles Miller and Modigliani (1958b) and Miller and Modigliani (1958a) assume several strict constraints. ffl First, caital markets are assumed to be without transaction costs and there are no bankrutcy costs. ffl All firms are in the same risk class. ffl Cororate taxes are the only government burden. ffl No growth is allowed since all cash flows are eretuities. ffl Firms issue only two tyes of claims, risk free debt and risky equity. All bonds (including any debts issued by households for the urose of carrying stocks) are assumed to yield a constant income er unit of time, and the income is regarded as certain by all traders regardless of the issuer Miller and Modigliani (1958b) ffl Information is symmetric across insider and outsider investors. ffl Managers are loyal stewards of owners and always maximize stockholders wealth. Coeland and Weston (1988) Later, others such as Stiglitz (1974) and Merton (1990) have removed the assumtion of risk class. Myers (1984) said that lifting these restrictions, one at a time, start ossible causes for the caital structure uzzle. The theoretical models of caital structure in a world in which caital markets are not erfect relates caital structure to several measurable and nonmeasurable attributes of a firm. The irrelevance roosition rovides conditions under which the caital structure of a firm is irrelevant to total firm value. Turning the irrelevance roosition around, the roosition also tells us which factors that may be the causes of cororate caital structure. The assumtions giving irrelevance as a result may cause relevance if they are broken. The question is, do they, and if so to what extent? And what if several imerfections exist simultaneously? Besides the irrelevance hyothesis of Modigliani and Miller there are several other theories relevant to caital structure. These are the asset substitution hyothesis, under-investment hyothesis, the free-cash flow hyothesis, the signaling hyothesis and roduct markets arguments. 3

4 .1. Value Additive Model of Caital Structure State reference models have been used extensively in the finance literature as a general framework for exlanation of the irrelevance hyothesis. Both Lewellen and Mauer (1988), Kraus and Litzenberger (1973), Stiglitz (1969) and Hirshleifer (1966) have used this aroach. The MM Proosition 1 Assume erfect caital markets and no taxes on cororate income. An equilibrium in the caital market requires that the value of the firm, V t, should be indeendent of the roortions of debt and equity in the firms caital structure. V t = V B;t +V E ;t (1) Let V be the total market value of the firm debt and equity. V B = market value debt V E = market value equity Define a set of ossible future states of nature, assumed to be finite and exhaustive. Given that a state of nature occurs, all relevant future events are known with certainty. I do not, however, know which state that will occur. The return to an investor deends on which state of nature actually occurs. An investor thus has a bundle of state contingent returns. A security that returns one dollar, if a certain state occurs and zero otherwise is called a rimitive security. The rice of this rimitive security is called a state-rice and is the rice today for one dollar tomorrow given that a articular state occurs. A comlex security i.e. a stock or bond will thus be comosed of several rimitive securities, one or more for each state in which the stock gives dividend. In states where the stock ays more than one dollar, say ten dollars, the stock must contain ten rimitive securities. In the states where the stock does not give dividend, for instance if bankrutcy occurs, the rimitive security for this state is not included in the 4

5 stock s bundle of rimitive securities. The value to the investor of any articular bundle is derived from the value to him of a marginal increase in his wealth in a given state. The value now of any future contingent claim is deendent on the robability that the state will occur, the investor s references for a wealth increase in this state based on a reresentative agent s utility and the time references reresented by the risk free rate of return. Three assumtions are made before I roceed with the multi-eriod state reference model. These assumtions are: Assumtion 1: The firm s investment strategies are given - in articular, although secific investment decisions are not known, the rules governing those decisions are known. These rules are consistent with maximization of the wealth of the security holder. In other words, I am assuming that financing and investment decisions are not connected. Assumtion 2: Perfect caital markets without transaction cost, informational asymmetry or bankrutcy costs. Assumtion 3: No taxes at either cororate or ersonal level. An un-leveraged firm s time t market value V t is exressed as eq. 2. According to Lewellen and Mauer (1988), the levered and un-levered firm total values coincide and the irrelevance roosition can be justified by the following statements: V t = Z Ω V (Θ)t+1dP(Θ)t+1 (2) where V (Θ)t+1 is total firm value in state Θ at time t + 1, Ω = All the ossible states of nature. Θ = State i of nature that the economy may have in time t, i = 1..n B(Θ)t+1 = Market value of debt in state at time t + 1, E(Θ)t+1 = Market value of stocks in state at time t + 1. P(Θ)t+1 = The distribution of corresonding state rices. R t+1 = Interest ayment romised to bondholders at time t + 1, which is state-indeendent. 5

6 A leveraged firms time t market value is exressed as the sum of debt and equity value. V (Θ)t = E(Θ)t + B(Θ)t (3) How the total firm value will be divided between stockholders and bondholders deend uon when there is a default on debt obligations. There will be a default and consequently bankrutcy when R t+1 E(Θ)t+1 (4) or when R t+1 + B(Θ)t V (Θ)t+1 (5) At the oint of maturity, the bondholders claim will be worth: V (Θ)B;t+1 = min[r t+1 + B(Θ)t+1;V (Θ)t+1] (6) At the oint of maturity, the equity holders claim will be worth: V (Θ)E ;t+1 = max[v (Θ)t+1 (R t+1 + B(Θ)t+1);0] (7) V (Θ)E ;t+1 = max[e(θ)t+1 R t+1 ;0] (8) The value of the equity holders claim is equivalent to a call otion on either E(Θ)t+1 with exercise rice R t+1,oronv(θ)t+1 with exercise rice R t+1 + B(Θ)t+1. In a risk neutral evaluation, all cash flows are certain, given that a secific state occurs. The time t values of equity and debt in the levered firm can then be written as: Equity : Debt: E(Θ)t = B(Θ)t = Z Z Ω Ω V E (Θ)t+1dP(Θ)t+1 (9) V B (Θ)t+1dP(Θ)t+1 (10) 6

7 Given the linearity of the oerator, I can add together equity and bond values. Due to the nature of the ay off of debt and equity investment, the total firm value can be written as max[v (Θ)t+1 (R t+1 + B(Θ)t+1);0]+min[R t+1 + B(Θ)t+1;V (Θ)t+1] (11) CASE 1 V E (Θ)t+1 > (R t+1 + B(Θ)t+1) (12) eq:(11) ) V (Θ)t+1 (13) CASE 2 V E (Θ)t+1» (R t+1 + B(Θ)t+1) (14) The value of the leveraged firm is E(Θ)t + B(Θ)t = Z Ω eq:(11) ) V (Θ)t+1 (15) V E (Θ)t+1 +V B (Θ)t+1dP(Θ)t+1 = V (Θ)t+1 (16) With the assumtions above, I have showed that the values of the unleveraged and leveraged firm are equal. The original irrelevance roosition of Miller and Modigliani were based on the notion of risk -classes, but these risk - classes are not necessary. As Stiglitz (1969) and Kraus and Litzenberger (1973) have shown, the irrelevance roosition hold in a more general time-state framework. III. The Trade-off Theory In this section I will review literature that suggest that debt has a central role in firm financing. Jensen (1986) argues that debt is an efficient means by which to reduce the agency costs associated with equity. Klaus and Litzenberger show that with the tax advantages of debt, otimal caital structure includes debt financing. Ross (1977) and Leland and Pyle (1977) argue that 7

8 debt can be valuable as a device for signaling firm value. The three main hyotheses that are used to exlain differences in caital structure between comanies are the transaction-cost hyothesis, the asymmetric information hyothesis and the tax hyothesis. According to Harris and Raviv (1991), leverage increases with fixed assets, non-debt tax shields, investment oortunities, and firm size and decreases with volatility, advertising exenditure, the robability of bankrutcy, rofitability and uniqueness of the roduct. This theory claims that a firm s otimal debt ratio is determined by a trade-off between the losses and gains of borrowing, holding the firm s assets and investment lans constant. The firm substitutes debt for equity, or equity for debt until the value of the firm is maximized. The gain of debt is rimarily the tax-shelter effect, which arises when aid interest on debt is deductible on the rofit and loss account. The costs of debt are mainly direct and indirect bankrutcy costs. The original static trade-off theory is actually a sub theory of the general theory of caital structure because there are only two assumtions that are broken here, the no tax incentive assumtion and the no bankrutcy cost assumtion. In the more general tradeoff theory several other arguments are used for why firms might try to adjust their caital structure to some target. Leverage also deends on restrictions in the debt-contracts, takeover ossibilities and the reutation of management. A negative correlation between debt and monitoring costs is roosed by Harris and Raviv (1990). Diamond (1989) suggest that vintage firms with a long history of credits will have relatively low default robability and lower agency costs using debt financing than newly established firms. A common factor for all these firm characteristics are that they are roxies meant to measure some form of costs related to a rincial-agent roblem. There may simultaneously be several rincial-agent roblems between the different classes of securities in the firm or between stockholders and managers in the firm. This multilicity of roblems can easily confuse the analyst and lend an air of incomrehensibility to the field of cororate finance. A construction of a ositive theory of debt financing, builds on arguments on the advantages and disadvantages of debt. First, debt is a factor of the ownershi structure that discilines managers. Limiting control to a few 8

9 agents that control the common stock, while the rest of the caital is raised through bond sale, can reduce agency cost of management. Second, debt is a useful signaling device, used to inform investors a message of the firm s degree of excellence. Third, debt can also reduce excessive consumtion of erquisites because creditors demand annual ayments on the outstanding loans. Debt also has its disadvantages. First, there is the roblem of agency cost of debt that includes risk substitution and under investment. Second, debt also increases bankrutcy ossibility by increasing the financial risk of the firm. I will discuss this reasoning in the following sections. A. The Theoretical Tax Incentive Taxes The hyothesis is that an increase in tax rate will increase value of firm tax-shield. The firm reduces income by deducting aid interest on debt and thereby reducing their tax liabilities. An increase in tax rates should hence increase leverage. Tax systems currently adoted by most industrial countries can be classified into classical systems and imutation systems. In the classical systems, interest ayments are deductible at the cororate level, but dividends are not. At the ersonal level dividends and interest are deductibles. Imutation systems reduce or eliminate taxation of dividends by granting a tax credit to reciients of dividends, equal to some fraction of the cororate tax aid on earnings used for dividends. The tax incentive is a art of the static trade off theory that contends that a manager balance increased financial risk and tax deductions. In Norway this is not necessarily true since the tax system is and has been neutral towards debt and equity. After the tax reform in 1992, dividends are not taxed as income for the investor, while interest income is. The firm ays the dividends out of taxed earnings while interests on debts are tax deductible in the firm s income statement. So, in Norway, dividend and interest rate taxation should be neutral regarding which financial instrument, debt or equity, firms have an incentive to issue. Norway have had deferred taxation after the tax reform of Taxation, which, owing to tax legislation results in timing differences, i.e., differences between incomes comuted for taxation uroses 9

10 and for financial accounting uroses. Unlike ermanent differences, timing differences are caable of being reversed in future eriods. In the deferred tax model, deferred tax is a longterm liability. Deending on the direction of the difference, deferred taxes could be a liability as well as an asset. Discounting the deferred taxes is not resently acceted in any accounting standards. Deferred taxes are the tax rate (28%) multilied with the difference in income between the financial rofit and loss account and the reorted rofit or loss for taxation uroses. If the deductions are higher in the tax reort, deferred taxes are ositive. 1 The exense for taxes is shown in the rofit and loss account/income statement. In the deferred tax model, the amount reresents the total tax to be aid on the accounting rofit for that year. In other models, just the amount to be aid is shown. The 1992 tax-reform The Norwegian Tax Codes was changed in the 1992 tax-reform. Consequently, this has insired us to test whether a tax incentive of debt financing exist in Norway. The yearly regressions in Frydenberg (2001), dislay that the tax variable nondebt is significant in 1991, 92 and 1999 and year 2000 for the interest carrying debt variable. According to Bøhren and Michalsen (2001) age 261, these are the years, excet for 1999, that the Norwegian tax system is not neutral! This is a new contribution, using the natural exeriment in Norway, to show that the tax system matters for the debt ratio. Before the accounting reform in 1992, taxes resented in the accounts referred to taxes ayable for the current year. After 1992, taxes include both ayable taxes and changes in deferred taxes. Changes in deferred tax occur as a result of temorary differences between book value and tax book value, and tax loss carried forward. Sjo (1996) claims that the marginal tax rate could influence the debt structure whenever the tax codes are not entirely neutral. A lower marginal tax rate should decrease leverage. With the tax-reform, dividend 1 This is common for accelerated dereciation. The dereciation ercentage is 30 % for office equiment, which is considerable more than the 1/5 linear economic dereciation in the financial statements. The 10% difference multilied with the tax rate of 28% is considered a long-term liability. Sooner or later, here already after three years, the financial linear dereciations are higher than the accelerated tax dereciations and the deferred taxes will be reduced with 28% of the difference between tax and financial dereciations. 10

11 taxes were substantially reduced. Dividends are no longer double taxed. The investor s taxes on dividends are reduced by the amount that is aid in the comanies. This change would increase the attractiveness of being a stockholder in Norwegian firms. A change in incentives towards equity is a consequence of the tax-reform. Some authors have claimed that the firms serve as a financial intermediary. Firms may not have a tax incentive to increase debt if both the investor and the firm are in a tax osition. The firm ays the taxes on the equity return, while the investor ays the tax on the debt return. If the debt investor is not in a tax osition, he may have other tax deductions and deferrals, which give him, zero taxable income, then the debt caital return is not taxed and there is an incentive to rely on debt financing. The financial intermediary argument also claims that firms may lend at more favorable terms than investors. The idea is that firms can make better use of the benefits of debt financing if they are in a tax osition. The firm benefits from a reduction of the taxes aid to the government. Interest rates aid to debt holders are deductible in the rofit and loss account and hence reduces the tax burden. For rivate debt, this is clearly an advantage. The marginal tax advantage of interest deductibility is exected to sink as leverage increases when the marginal tax rate is rogressive. The last amount of debt that is incurred does not invoke a very large tax effect since the marginal tax rate decreases as the income is reduced. For cororate debt, however, the marginal tax rate in Norway is 28% and constant for all rofit levels. The tax incentive for debt financing is therefore not marginally decreasing as the leverage increases when the cororate tax-rate is constant. The firm tax is a ure net-rofit tax and where there is no rofit, even the government tax collector has lost his claim. In his article Debt and taxes, Miller (1977) argues that the debt level is not deendent on tax rate when both ersonal and cororate taxes are considered. Thakor (1989) is also critical to the tax advantage of debt since debt was used rior to 1913 when there were no cororate taxes in the US. Preferred stock ays a fixed dividend such as debt but the dividends are not tax-deductible. The issuance of such referred stock also shows that the tax advantage of debt only artly can exlain the usage of debt. 11

12 B. Bankrutcy Incentive A bankrutcy without costs would not change the value of the firm. Stiglitz (1969) have shown that the value of a security is the same, regardless of whether a bankrutcy would occur or not under some secific conditions. There are, however, costs of financial distress. These costs are divided into direct costs and indirect costs. Bankrutcy costs are exected to increase for all levels of leverage. Bankrutcy costs are not only the direct costs of transferring the assets to the new owners, lawyer s fee and court fees, but there are also indirect costs of bankrutcy arising from the bankrutcy rocess itself. The bankrutcy trustee, as an agent of the court, has the authority to oerate the firm. Warner (1977) said that it is not clear if this agency relation give the trustee any incentive to run the firm efficiently and take decisions which are in fact value maximizing. The stockholders face the ossibility that they may lose control of the firm when a financial distress situation occurs. The otion to receive dividends if the firm oerates rofitably in the future is thereby lost for the current stockholders. Indirect costs arise when qualified emloyees that can ursue alternative oortunities quit the firm when bankrutcy is closing in. Unrest in the organization is often the result, and suliers and customers that rely on continuous business relations may lose confidence in the firm. Risk shifting Risk shifting is described in Section F but is worth mentioning here too because the gains and temtation to lay this game is strongest when the odds of default are high. If the firm robably will go under anyway, why not make a final bet that might salvage the firm. The stockholders and management is betting with the creditors money. The oint in risk shifting is how stockholders of levered firms gain when business risk increase. This oint can also be described with a Call otion. C. Dividends Payments and Claim Dilution Problem Brealey and Myers (2000) describe this as the cash in and run roblem; stockholders can take out the valuable assets, while bondholders are ket in the dark. Bonds are riced under the 12

13 assumtion that dividend olicy remains unchanged. Reducing the investments by increasing the dividends will reduce firm value, increase the risk of outstanding debt and harm the bondholders. The stockholders can ay out all the assets and leave the bondholders with an emty shell. As the firm aroaches bankrutcy and financial distress is more and more transarent for the firm s stakeholders, management may chose to lay games with the creditors. These games can be layed all the time, but stakes are higher near bankrutcy. C.1. Claim Dilution The stockholders can use the value of the assets as bait, and then switch to another strategy. Bonds are riced under the assumtion that no new debt will be issued. When the firm issues new debt with the same or higher riority than the old debt, the claim of the old bondholders will be reduced. The reason is that the robability of default on the old debt increases. The firm starts with a conservative olicy, issuing a limited amount of relatively safe debt. Then the firm suddenly switch and issue a lot more. That makes all the debt risky, imosing a caital loss on the old bondholders. Their caital loss is the stockholders gain. D. Product Market Cometition and Caital Structure In his model, Titman (1984) ostulate that a seller of a roduct enters contracts of service to a roduct. A seller who can credibly enter this kind of service contract can demand a higher rice for his outut roduct. If this service becomes too costly, he will renegade on the contract. Bankrutcy is a ossible method to renegade a service contract. A higher debt-ratio increases the ossibility of bankrutcy and consequently the mark-u - rice the seller can demand, since the customers view the debt ratio as a signal of ossible bankrutcy. Williams (1988) have also studied the caital structure roblem when there are more than one firm in the model. An agency roblem occurs where a firm has to choose between two technologies. One labor intensive and one caital intensive are the two available technologies. 13

14 If the entrereneur invests in the labor-intensive technology, he may consume some of the investment roceeds from external financing. The caital-intensive technology is assumed to give higher returns than the labor-intensive technology. Debt can then be used to control management incentives by setting the debt level so high that an investment in labor-intensive industry is not feasible. An imortant conclusion from the industrial equilibrium literature is that several financing arrangements may coexist at the same time. Large caital-intensive firms with high debt levels may coexist with smaller and marginally rofitable firms that are having lower levels of debt. Williams (1988) model could therefore exlain the large heterogeneity of debt ratios observed in the data. E. Otimal Contract and Cororate Governance Theory In this section I exlore three subjects that are often mentioned as ossible causes of caital structure decisions. First, I consider the cororate governance roblem. How should the firm be controlled? A tightly held firm where a majority or large shares of the stocks are controlled by an individual or a grou is suosed to be a more efficient managed firm than the firm controlled by entrenched self-interested managers. Second, oerating risk has to be shared by the owners of the firm and the debt-equity aradigm have shown remarkable ability to allocate risk to the eole that has the residual control rights of the firm, the equity holders. Third, does transaction cost theory have anything to do with caital structure? The cororate governance roblem Aghion and Bolton (1988) and also Zender (1991) emhasize that contracts that grant control to one class of agents exclusively may not be efficient because they fail to give the controlling agent the incentives to make the first best decisions. The intuition behind the existing financial instruments equity and debt are that they are designed so that in each state the owner of the residual control rights own the residual cash flow. If this is not the case, there is a otential conflict because owners of one of the securityclasses could exroriate wealth from owners of the other security class. The standard debt 14

15 contract (SDC) is an otimal contract and dominates all other tyes of contracts in the Costly State Verification(CSV) literature. A SDC minimizes the exected transfer of wealth from the entrereneur to the outside investor. See Allen and Winton (1995) for a comrehensive review of the literature. The defining character of the CSV literature is that that the true cash flows of the firm are only observable to the manager or entrereneur. The cash flows of the firm are only verifiable at a cost in a bankrutcy rocess. The outside investor receives only a fixed sum of money when the entrereneur income is not verified and a strictly lower ayment when income is verified. Ownershi structure could be an imortant factor exlaining the leverage structure. Firms that are tightly controlled by major stockholders will have less agency costs. Fama and Jensen (1983) argue that it makes sense to searate the financial claims into only two tyes: a relatively low risk comonent, the debt caital, and a relatively high risk comonent, the equity caital. This is an otimal form of contracting because it reduces contracting costs and it lowers the cost of risk-bearing services. Stockholders and bondholders do not have to monitor one another. Only the bondholders must monitor the stockholders and not vice-versa. This one-way monitoring will hel to reduce contracting costs. Risk sharing Members of a syndicate are eager to regulate two main elements in the artnershi. First, the ayoffs from ositive NPV-rojects must be divided, and second, risk has to be shared. The risk stems from the time varying realization of ayoffs and liability if bankrutcy occurs. In an equity-debt model, equity-owners carry almost all the risk, and receive all the ay-off if the firm reaches a certain rofit-target. On the other hand, if rofits are below target they get no ay off. The debt owners have ractically no risk and fixed ay-off structure unless losses incurred wie out the equity. Neither equity nor debt owners can lose more than what they have invested, guarantees set aside. A first best solution to this risk-sharing roblem would be to give the risk-neutral investor all the risk and divide the ay-off accordingly. No risk results in a risk free return for the debt owners. A second best solution makes the risk averse investor carrying more risk than the otimal solution. This induces an excessive risk remium that can be attributed to the non-otimal risk sharing; the less risk averse investor should accet a larger risk. 15

16 Transaction cost theory Coase (1937) suggested that the neo-classical literature of the firm is consistent with there being one huge firm in the world and also that every division of a firm could be divested to single autonomous firm. Organizations emerged because of market failure, because of asymmetric information between agents. This asymmetric information between members give organizations their reasons for existence. Williamson (1981) claimed that other reasons for existence are transaction costs and economies of scale and scoe. A general emloyment contract between the firm and the emloyee reduces the necessary details in a contract comared to contracts that secify exactly the content of the services rovided by the emloyee. Instead the firm can order the emloyee to erform tasks within a broadly defined area. The organizations are not markets and they are not individuals. The literature of firm financing, views firms as something between. Firm financing can be seen as a set of contracts between eole that cooerate towards a common goal while they are ursuing their self-interest. Financing the investments of these organizations requires a trade off between costs and benefits of security and loan contracts. The literature of cororate finance relies on the early work of Coase (1937) and Williamson (1981) that exlain the reasons for existence of organizations. The cororation has served as a major form for organizing economic activity. The ability of the cororation to survive through the last century must reflect benefits from the organizational structure that rotects the real and financial resources. The value of the cororation must be derived from costs and imerfections inherent in markets: information asymmetries, transaction costs, economies of scale and scoe, and forms of taxation. Financing is a mirror image of the real activities of an organization. Asymmetric information and transaction costs faced by agents in the organization create an oortunity to organize economic activity differently from how a market or a contractor-based activity would be organized. Instead of roject financing the firm takes on cororate debt, thereby reducing transaction costs, and construct budgets and incentive schemes for the sub-divisions of a firm. 2 2 A roject manager is allocated caital from the firm s investment budget. Alternatively, she could have raised caital in a bank or in the caital market. That would be difficult if the asset base for the roject is shared with many other rojects and ossible collateral is already occuied. The cororate debt reduces transaction costs comared to many searate roject debts to different lenders. 16

17 F. Risk Shifting (Asset Substitution) Problem The basic idea, that increases in leverage induce equity holders to ursue riskier strategies, was introduced by Jensen and Meckling (1976). Since decisions concerning dividend ayments, issuance of new debt, and investments are made by the owners, these decisions are a otential source of conflict between equity investors and debt investors. The financial structure affects the cash flow through investment decisions. Smith and Warner (1979) argues that there are four major sources of conflict. These are dividend ayments, claim dilution, asset substitution, and under-investment. Briefly, asset substitution is an incentive roblem associated with debt. Stockholders are the residual claimants to the firm cash flow. Their claim is analogous to a call otion on the firm s asset, with a strike rice equal to the face value of the debt. It is well known that the value of a call otion increases as the risk in the underlying asset increases. Bonds are, however, sold with the rosect of a certain level of risk. If stockholders increase risk beyond this initial level, they can exroriate wealth from bond-investors. Bondholders will, under rational exectations, recognize this incentive, and require a discount when they invest in debtsecurities in a firm. This discount is a cost that reduces the total value of a firm and as such is eventually born in total by the stockholders. This risk shifting incentive can be mitigated by covenants in the debt contract, legislation and by using convertible debt or straight debt with warrants. The asset substitution roblem can be described using a model that integrates models used by Green (1984) and Green and Talmor (1986). When the investment decision of the bondholders recede that of the stockholders, it is to the stockholders advantage to choose a distribution of returns with relatively more weight in the uer tail. Rothscild and Stiglitz (1970) have shown that increasing the investment in the risky roject is equivalent to increasing the mass in the tails of the distribution of total return. How likely is this theory to have any effect in ractice? Does lenders care about the ossible asset substitution when contemlating a loan to a firm? In the loan alication rocess the investment objects and future strategy is certainly a talking oint. The roblem is 17

18 that it is almost imossible to construct a contract where the borrower cannot renegade and invest in something more risky. That might not even be in the best interest of the lender either. If firm management sees a better investment roject than the one they originally lent money for, should they not invest because the loan contract has restrictions on the ossible investments? It is likely that it is better to control management incentives through stock otions and stock market monitoring than through costly restrictions in the loan contracts. The risk-shifting incentive roblem can be controlled through restrictions in the loan contract and through management incentives. But such monitoring is costly and with uncertain effects. It is therefore likely that lenders seek to insure themselves to an increase in business risk. The simlest method is to demand collateral for the loan and to demand an equity of some size in the firm before lenders grant a loan. This incentive is even addressed by legislature in Norway that demands an equity of minimum NOK. Even for a small firm, the amount is very low and the demand for ublic disclosure of financial reort is therefore more imortant. In the US, quarterly disclosures of financial reorts are the standard among listed firms, the activities of management is therefore more transarent than in Norway. The conclusion must be that the risk-shifting roblem is hyothesized to be severe, but it is emirically very difficult to measure, if not imossible. G. Under-Investment Problem The roblem is that stockholders will have to share the extra value created by their additional investments with creditors. Myers (1993) Under-investment is defined as forgoing a roject with a ositive net resent value. The argument in this section was introduced by Myers (1977) and is called the under-investment hyothesis. Highly levered comanies are more likely to ass u rofitable investment oortunities. Firms execting high future growth should therefore use a greater amount of equity financing. 18

19 I 0 is the investment in roduct develoment made at time t = 0. Due to the investment in roduct develoment, the firm holds a growth oortunity. At t =1, the firm must invest an additional I 1, to get the cash flow X (I 0 ;α; r) from the roduct develoed. The investment I 1 could for instance be machinery needed to roduce the develoed roduct. The investment otion is firm secific and cannot be sold. The firm that made the initial investment has an exclusive right to exercise the otion. If the firm chooses not to invest, the otion has zero value. This constitutes an imerfection in the market for real otions of this category. First I consider a firm that is entirely equity financed and then a firm that issues debt to finance the initial investment I 0. The intuition behind the model is not comlicated. The consequences and ractical imlications of the model, however, is a subject for discussion. The intuition is that an investor ossessing a growth otion will not share the gains of the otion with financiers if the growth otion has rior debt financing. After having aid off the debt holders to the otion, there will be too little left for the investor of the rofit that comes from imlementing the growth otion. When B 0 < V is fulfilled, the incentive not to invest in certain states are resent. Whether the otion has any value when it exires deends on the asset s future value and also on whether the firm chooses to exercise. The funds needed to make the investment I 0 in this case are raised by selling bonds with face value F. To be able to invest at t = 1, the firm must raise I 1, and does so by outlays from the current stockholders. We abstract from the signalling considerations of a security issue. The timing of events at time t = 1 is imortant. First, the return is revealed to the stockholders, the initial investment becomes certain at this oint. Second, investment I 1 is made, and then the debt matures. The firm will invest if the return cash flow is large enough to cover both the first investment which is debt financed and the second investment which is equity financed. However, since the first investment, made at time zero, is in fact sunk cost at time t = 1, the firm should make the investment if the return cash 19

20 Figure 1. The Firm s Investment Decision with Prior Debt Financing Return X in state s 6 ρ ρρ ρ ρρ ρ ρρ ρ ρρ d(s)=0 s a ρ ρρ ρ ρρ ρ ρρ ρ ρρ d(s)=1 s b ρ ρ I 1 + F I - s, State of the world flow is larger than the equity investment at time t = 1. In some unfavourable states of nature the firm will forgo a ositive NPV investment, because of the seniority of the debt. From the stockholders ersective, they would be throwing good money after bad money. This is a cost that in a rational exectations equilibrium in the caital market will be borne entirely by the stockholders. The bond holders will require a remium to finance a roduct develoment that may not result in roduction in anticiation of the shareholders not being willing to suly the firm with additional funds to buy the necessary machinery for the roduction if the return does not cover the cost of the machinery. This remium reduces the value of the firm and is a cost of debt financing comared to equity financing. Firms with growth otions would hence be reluctant to finance themselves with debt. I can illustrate this incentive not to invest in certain unfavorable states of nature in a figure. Shaded area in Figure (G) is loss of value in some states. A high F imlies a larger loss of value. Myers (1977) is often cited in the literature. This is artly due to the seminal real otion asect, but also to the fact that he describes a unique incentive roblem. Myers argument hinges on the fact that that the lender has no additional collateral excet the growth otion. If 20

21 you do not invest, you walk away from the debt on the growth otion. If the growth otion could be sold to the lender, she could invest herself as long the second investment, I 1,is ositive NPV. But, that is not ossible, the growth otion is a real-otion which cannot be sold in the market as described in the aer by Myers (1977). A contracting ossibility would therefore be that the lender and the firm slit the rofit otherwise lost, in a Nash (50/50) bargain solution, by letting the firm exercise the growth otion and finance only a art of the new investment while the lender financed the rest. As long as V(s) is larger than I 1, both arties could be made better off by sharing the investment and gains than letting the otion exire unexercised. Imagine that the debt F is just a tiny bit larger than V(s)- I 1. The creditor could then recou almost half the debt in a bargaining solution. This is also seen in real life, it is not uncommon for banks to grant further loans to a firm to salvage some of their outstanding debt. H. Free Cash Flow Problem Debt creation, without retention of the roceeds of the issue, enables managers to effectively bond their romise to ay out future cash flows. Thus debt can be an effective substitute for dividends, something not generally recognized in the cororate finance literature. Jensen (1986) The argument of the free cash flow and the role of debt to control oortunistic management is due to Jensen (1986). Debt reduces management oortunity to send excess cash flow in non-rofitable investments. Management has less control over the firm s cash flows since these cash flows have to be used to reay creditors. Jensen and Meckling (1976) have argued that managerial incentives to allocate the firm s resources to their rivate benefit are larger when the firm is mainly equity financed. The free cash flow term is the amount by which a firm s oerating cash flow exceeds what can be rofitably reinvested in its basic business and the emhasis is here on the word rofitably. Conflicts of interest between stockholders and managers over ayout olicies are esecially severe when the organization generates substantial free cash flow. So, there is a dark side to the financial slack. Too much of it may encourage managers to take it easy, exand their erks, or emire-build with cash that should 21

22 be aid back to stockholders. The roblem is to motivate managers to disgorge the cash rather than investing it below the cost of caital or wasting it in organizational inefficiencies. Jensen (1986) validates his stories by referring to the emirical literature on debt for common stock exchanges that leads to stock rice increases. This evidence is, however, also credited for the otential signalling effect of debt. Debt as a signal of high-quality firms is treated in Section IV. The evidence from the leveraged buy out and going rivate transactions is that many of the benefits in an LBO seem to be due to the control function of debt. The conclusion of this theory is that Jensen claim that by straing the management to the mast i.e. make them ay out fixed amounts of money to the investors each year, the agency cost of free-cash flow can be reduced. H.1. Profitability Profitability affects leverage in at least two directions. First, higher rofitability usually rovides more internal financing. More earnings can be ket in the firm and hence a lower level of debt. Less debt is then needed to finance already lanned investments. Secondly, debt introduces an agency cost argument. Management will refrain from the building of emires and excessive consumtion of erquisites, when large sums of money must be aid to creditors each year. Debt kees the firm slim and cost efficient. Unnecessary non-rofitable investments will be avoided because creditors demand annual ayments and claim any free cash flow. High rofitability should result in higher leverage according to the free cash flow hyothesis, but a high leverage would result in high rofitability on the basis of the ecking order hyothesis. The roblem is that leverage and rofitability are linked both ways, and that the causal direction is uncertain. This roblem could be avoided emirically if I estimate a system of two equations instead of one equation. This aroach is used in Stein Frydenberg (2001). Since causality here works both ways, it would be ossible to control these two effects by a system of equations. 22

23 IV. The Pecking Order Theory According to the ecking order theory, the firms will refer internal financing. The firms refers internal to external financing, and debt to equity if the firm issues securities. In the ure ecking order theory, the firms have no well-defined debt-to-value ratio. There is a distinction between internal and external equity. Several authors have been given credit for introducing signaling as an argument in the discussion of debt s exlanatory factors. Ross (1977), Leland and Pyle (1977) and Myers and Majluf (1984) are often quoted as the seminal articles in this branch of the literature. Myers and Majluf (1984) describes the reference like this: The firms refer internal financing, they target dividends given investment oortunities, then chose debt and finally raise external equity. The ecking order was traditionally exlained by transaction and issuing costs. Retained earnings involve few transaction costs and issuing debt incurs lower transaction costs than equity issues. Debt financing also involves a tax - reduction if the firm has a taxable rofit. Myers and Majluf (1984) invoked asymmetric information to give a theoretical exlanation for the ecking order henomena. The signaling model described in Section A leads to a ecking order concet of caital structure, where retained earnings are referred to debt and debt is referred to new equity. The signaling model showed that only low rofit tye firms would issue equity in a searating equilibrium. Rational investors foresee this and demand a discount in Initial Public Offerings (IPO). This discount is a cost of raising equity that will be borne by the internal stockholders. Debt signals to the caital market that the issuing firm is a high erformance firm. A. Asymmetric Information (Signalling) Problem Asymmetric information between old and new investors, and managers and investors incite to signaling games where the amount of debt, and the timing of new issues is viewed as a signal of the erformance of the firm. Akerlof (1970) introduced an adverse selection argument 23

24 that exlains why rices of used cars dros significantly comared to new cars. The seller of a used car will usually have more information about the true erformance of the car than the rosective buyer. The buyer s best guess would be the average erformance of cars in the market. The buyer, when offered a car, exects the erformance of the car to be below average, otherwise the car would not have been offered to the market. Consequently, the rice of used cars decline and the only cars offered for sale are the cars not well made and maintained. The buyers require a discount to comensate for the ossibility that they might urchase an Akerlof lemon. Earlier there used to be a tendency that cars built on Mondays for some obscure reason had more roblems than other cars. The focus on quality control may have ut this roblem to an end, but still some cars seem to endure longer than others. The heart of the matter is that the seller is aware of the ucoming rearations and roblems, but the buyer is not. A recent trend in Norway to overcome this roblem is to require a comlete check of the car by an authorized worksho before buying. The work - sho roduces a status reort on the technical condition of the car, which serves as a signal of erformance of the car to rosecting buyers. This signal reduces the asymmetric information in this market. Analogously, the only firms for sale, in the market for cororate control, are those with below average excellence. The firms will, according to the adverse selection argument only issue new equity when the stock is overriced. Issuing debt can be a signal to the caital market that the firm is in fact an excellent firm and that the management is not afraid to borrow money. The bankrutcy ossibility is suosedly not large enough to let extensive borrowing bring about the current management s control of the firm. The idea underlying the signaling models, is that stockholders or managers signal rivate information to the security market in order to correct the market s ercetion of excellence. 24

25 V. Summary of the Theoretical Literature Caital structure remains enigmatic Chirinko and Anuja R. Singha (2000). Although we have come along way towards a better understanding of why and how firms choose their caital structure, there are still unresolved issues. First, there is no current model that ut all the ieces of theory together in a model that might be suitable for textbook resentation and, hence, resentable to the general ublic. Thus, students when they enter the job market have no comrehensive model to relate their caital structure decisions to. Second, the emirical evidence is mixed and does not oint out a single emirical model as a good exlainer of cororate ractice. At the same time, everybody understands that borrowing too much is not good for your firm s health and not borrowing at all is a waste of recious equity. The academic community s 40 year struggle with the issue since Miller and Modigliani (1958b) have in essence found that your caital structure is a trade off between many interests and that some times you would refer to issue debt and sometimes equity. I think the root to the roblem here is exactly many interest. The caital structure being a mirror image of the real side of the balance sheet is a too comlex fabric to fit into a single model. Academics have not tried to make a single model for how firms should invest or oerate; realizing that the environments and circumstances a firm could be oerating under is endless. Instead we have several artial models for how firms should oerate and invest contingent on the environment and circumstances surrounding the firms. It is time to realize that this goes for the financing asect as well. I can only hoe to make models that suit one tye of firm well; other tyes of comanies need other aroaches. One of my contributions is to show that the theoretical models develoed seem to exlain differently if I divide the data into several generic grous of firms. This idea is exlored in Frydenberg (2001). Models invented for homogenous firms should be tested on homogenous 25