The capital structure decision for real estate firms listed at the Stockholm Stock exchange

Transcription

1 The capital structure decision for real estate firms listed at the Stockholm Stock exchange Simon Fylking and Jason Rånes Stockholm Business School Bachelor s Degree Thesis 15 HE credits Subject: Business Administration Autumn semester 2014 Supervisor: Michael Graham

2 Abstract The main purpose with our study was to investigate whether real estate firms listed at the Stockholm OMXSPI, during the time period , tended toward the trade- off theory or the pecking order theory in a direct comparison. The trade off theory proclaims that there exists an optimal capital structure, where equilibrium between tax- shield and cost of financial distress is obtained. While the pecking order focuses, not on the capital structure but in financial preferences such as; internal funds is to be used over external and debt is preferred over equity. Prior research, made by Bond and Scott (2006), found that the pecking order offers a good fit for real estate firms listed in the UK in comparison with the trade- off theory. In contrast to that, we found that the trade- off theory dominates the pecking order theory, but that the pecking order does offer a good fit for large- cap firms. Due to the financial crisis in 2008, the Basel III reform was implemented, and we investigated how the reform affected real estate firms. However, it is hard to draw any conclusions of the impact of the Basel III reform, and one reason for this is that there are more underlying factors affecting a firms debt ratio. Thus, what we could see was that mid- cap firms was more affected by the financial crises than large- cap firms, with higher fluctuations in net debt issuance, net equity issuance and net external financing. Key words; Capital structure, Trade- off theory, Pecking order theory We would like to express our gratitude s to PhD Michael Graham for his guiding and support during the writing of this thesis. 2

3 Table of contents Thesis structure 4 1. Introduction Background Real estate industry Aim and research questions Research questions 7 2. Literature review Ideas and concepts behind the capital structure when taxes and asymmetric information is present Empirical studies when the assumption regarding perfect capital markets was violated Empirical Studies of the Capital Structure Theoretical Framework Theoretical Toolbox Method Scientific Perspective Operationalization Data Sample Dependent Variables Descriptive Statistics Empirical Method Regression Models The Simple Pecking Order Model The Simple Target Adjustment Model The Conventional Leverage Regression Model Empirical presentation, Analysis and Findings The simple pecking order model The Simple Target Adjustment Model The Conventional Leverage Regression Model (Large- Cap) Discussion and Critical Reflection Limitations Conclusion 39 References 41 Appendix 43 3

4 Thesis structure This thesis is structured as follows: first a background about the conception of capital structure and the characteristics of the real estate industry. This is followed by a review of previous studies. Thirdly, we describe our data and the empirical methods applied. Last, we present and analyse our results, discuss our findings and limitations and finally we draw our conclusions. 1. Introduction The capital structure has baffled researchers for decades. The reason for this is that there exist no universally accepted theory, explaining why a firm issues specific types of securities and for what reason a firm divide securities in different setups. The question considered is if there is a unique combination of financing sources that maximize firm value. Paradoxically, Miller and Modigliani (1958) concluded that the capital structure is irrelevant to the firm in order to maximize firm value. However, Miller and Modigliani assumed that perfect capital markets exist, which: / /implied that investors and firms trade the same set of securities at competitive market prices equal to the present of their future cash flows. And there are no taxes, transaction costs, or issuance costs associated with security trading. And a firm s financing decision do not change the cash flows generated by its investments, nor do they reveal new information about them. (Berk and Demarzo 2013 p.455) The contribution of Miller and Modigliani is usually a base for research in this field and this will also be the starting point for our intended study. 1.1 Background The current debate in the field of capital structure consists mainly of two theories, the trade- off theory of optimal capital structure and the pecking- order theory of financial choices. The trade- off theory states that a firm will hold some debt in order to take advantage of tax benefits. Thus, the optimal capital structure would exist where the 4

5 equilibrium between equity and debt lies between the tax benefits and the cost of bankruptcy. The pecking- order theory has another view of the capital structure and the focus is not on an optimal but instead on the financial choice. The financial choice consists of internal or external financing where the external consists of debt and equity (such as corporate bonds and preferred equity). The pecking order states that the firm has a preselected order when financing an investment. If possible the firm will use internal funds and if this is not possible, the preferred financing would be to use debt over equity. The pecking order can be found because of the uneven knowledge between a manager at a firm and the stockholders regarding a firm investment. The only way to wipe out the information asymmetry between the manager at the firm and the stockholder is to finance the new investment with either internal financing or financing with debt. If the manager issues equity, there will exist some degree of information asymmetry, the reason for this is that the market would either see the issue of equity as a failure of lending from the banks or as a defensive investment where the investment prospects does not look so good. Prior research has questioned whether perfect capital markets really exist. For instance, Bond and Scott (2006) found empirical evidence suggesting that real estate firms prefer internal over external financing and debt over equity when issuing new securities and financing sources to finance investments. Signal differently to the market, would affect the stock prices negatively and thus decrease firm value. After the financial crises in , the Basel committee introduced a regulatory standard implying that banks needed to increase their capital adequacy requirements. The reason for this was to decrease the risk of financial bubbles. 1 As a consequence of the capital adequacy requirements, real estate firms got harder lending conditions by the banks with a lower loan- to- value of capital. Thus, part of the financing should have moved away from the banks and moved towards corporate bonds and preference shares (Fastighetstidningen 2013). 1 Bank For International Settlements Communications, (2010), Basel III: A global regulatory framework for more resilient banks and banking systems. 5

6 1.2 Real estate industry Real estate firms tend to behave in a different way than firms in other industries, thus making it an interesting market to study. 2 According to Bruggeman and Fisher (2010), the investment strategy real estate firm use, is either to increase the value of the properties and/or increase rents. The real estate industry is characterized by a very large market and is highly competitive. Real estate firms are also characterized by comprising high debt levels. According to Bruggeman and Fisher (2010), there are three reasons that can explain the high debt levels. The first reason is that real estate firms lack enough equity to purchase a property, a second reason is that real estate firms wants to diversify by purchasing more than one property and the third reason is that real estate firms wants to take advantage of tax deductibility s of the mortgage interest. The reason why we focus only on one specific industry is because prior research, mainly by Solomon (1963), found that corporations operating in the same market tended to have similar compositions of the capital structure. Thus, we would decrease the odds of misleading numbers when running our regressions if we look at a specific industry. 1.3 Aim and research questions The aim of the thesis is to document the financial behaviour of real estate firms. By comparing the pecking order- and the trade- off theory, we would get a solid foundation when documenting the financial behaviour of real estate firms. Also, by looking at the financial behaviour before and after the financial crises, we could examine the impact of the Basel III regulatory standard on real estate firm s financial behaviour. This research should especially be interested for investors investing heavily or with parts of their portfolio in listed real estate companies. 2 We will focus mainly on real estate firms oriented towards income producing properties. 6

7 The area is quite well studied regarding the pecking order and the trade- off theory, but we believe there is a knowledge gap regarding the Swedish market Research questions Bond and Scott (2006) found that the pecking order offers a good fit for real estate firms listed in the UK compared with the trade- off theory. Can we find support that the pecking order offer a good fit compared with the trade- off theory among listed real estate firms at the Stockholm OMXSPI? Has the financing behaviour regarding debt changed for managers at real estate firms listed at the OMXSPI since the financial crises in 2008, due to higher equity demands by banks subject to the Basel III reform? 2. Literature review The literature review is posted in the following way; first we investigate the ideas and concepts behind the capital structure when taxes and asymmetric information is present, secondly we look at studies when the perfect capital market assumptions was violated, thereafter we summarize studies of the capital structure and finally we draw our theoretical framework. 2.1 Ideas and concepts behind the capital structure when taxes and asymmetric information is present The main theory explaining the capital structure when there are taxes associated with security trading is the trade off- theory. As Myers (2001) wrote The trade- off theory justifies moderate debt ratios. The reason for this is that a firm will lend up to the point where the marginal value of tax shields on additional debt is just offset by the increase in the present value of possible costs of financial distress. If managers exploit the tax shield, the observed outcome would be that more profitable firms has more taxable 7

8 income to shield and that profitable firms can service more debt with less risk of financial distress according to Myers (2001). 3 However, studies regarding the determinants of the firm s debt ratio consistently find that the most profitable companies in a given industry tend to borrow the least. 4 The main theory explaining how a firm s financing decision affect its firm value is the pecking order- theory. The underlying assumption behind the pecking order- theory according to Myers (2001) is that investors do not know the true value of either the current assets or the new investment opportunity. Or, put it in Harris and Ravivs (1991) words. In these theories, firm managers or insiders are assumed to possess private information about the characteristics of the firm s return stream or investment opportunities. In one set of approaches, choice of the firm s capital structure signals to outside investors the information of insiders. This explains why most of external financing comes from debt, signalling differently to the market would affect the firm s value negatively. For example, if a firm is required to finance a new project by issuing equity, the investment may be rejected even if its net present value is positive. The reason for this is that the under- pricing when issuing equity is so big, that the new investors capture more NPV of the new project than the existing shareholders. However, by issuing debt instead of equity, much of the undervaluation will not occur. Therefore, firms will prefer debt above equity and internal financing above external financing. According to Myers and Majluf (1984), if the main goal of the manager were to maximize market value, the manager would avoid external equity financing if the manager hold better information than outside investors, assumed that the investors are rational. Optimistic manager who believe that their company are undervalued, will instead of equity chose to issue debt. Myers (2001) reason that only managers that are pessimistic would want to issue equity. The reason for this is that the investors would response more positive when firms issue debt instead of issuing equity. The pecking order theory also explains why more profitable firms borrow less. The main reason for this is that profitable firms have more internal financing available than less 3 We define financial distress as Myers as the cost of bankruptcy and the agency cost when there is question regarding a firm creditworthiness. 4 See, Myers (1984), Harris and Raviv s (1991) 8

9 profitable firms, therefore less profitable firms accumulate more debt through external financing. However, allowing firms a wider range of financing choices can void this result in some cases. Brennan and Kraus (1987) and Constantinides and Gundy (1989) concluded that the underinvestment problem could be fixed, by signalling with a richer set of financing options. For instance, Brennan and Kraus found that by combining equity issues with debt retirements, the information asymmetry could be resolved. Another solution suggested by Constantinides and Gundy would be to create an optimal contract between the management and the outsiders Empirical studies when the assumption regarding perfect capital markets was violated Prior research found evidence against perfect capital markets when asymmetric information arise, for instance Garmaise and Moskowitz (2004) found strong evidence suggesting that asymmetric information is significant in the commercial property market. Garmaise and Moskowitz investigated the role of asymmetric information in the commercial real estate market by proposing a novel and exogenous measure of information, based on the quality of the property tax assessments. The reason why they focused on the real estate industry was because information consideration is likely to be of high importance. The reason for this, according to Garmaise and Moskowits, is that the market is highly illiquid implying that the price mechanism is slow to convey information to market participants. The reason why commercial real estate is of interest is because commercial properties and vacant land are difficult for outsiders to value. In order to do a measure of information, Garmaise and Moskowitz test a no- trade implication, which show that uninformed agents will not trade with informed counterparts. According to Garmaise and Moskowitz, this should lead to limited market participation on the parts of agents, which are particularly informational disadvantage. To separate uninformed and informed buyers, Garmaise and Moskowitz look at local market conditions. The outcome of Garmaise and Moskowitz research are clear evidence of limited participation, selective offering and market segmentation. Which suggest that buyers tends to be local when information asymmetries is high, the properties age 9

10 matters and that brokers are most likely to sell to other brokers when the information asymmetry is high. Smith (1986) also found evidence against the perfect capital market assumption when he studied the process by which corporations raise debt and equity capital and the associated effects on security prices. The motivation behind the study was that capital markets play an important role in the theory of capital structure, and mentions that capital market prices provide vital signals for corporate investments decisions. In order to investigate the associated effects on security pricing, Smith derived five hypotheses. The first hypothesis is the optimal capital structure, which shows the stock price reaction due to adjustments of a firm liability structure. Second is the implied cash flow change, which puts stock price reactions in relation with announcements of future expected cash flows. Third is the unanticipated announcement, which puts stock price reactions in relation with unanticipated announcements. Fourth is the information asymmetry, which states that corporate managers has more information than the marginal purchaser of securities and therefore is more likely to issue securities when they are overpriced in the market. The fifth hypothesis is ownership changes, which implies that changes of the control rights affects the value of the firm shares. After a firm decides the security it wants to issue, it must choose among a number of methods to market it with and according to Smith, this can affect the spread. The methods that Smith mentions is the rights or underwritten offers, negotiated or competitive bid, best efforts or firm commitment contracts and shelf or traditional registration. By looking at findings from related studies, the outcome of Smith s study are that there is weak support for the optimal capital structure and the effect on stock prices. Smith found support for the cash flow change and regarding the unanticipated announcement, the evidence appears consistent with the hypothesis. Looking at the information asymmetry, evidence suggests that the information asymmetry is consistent with the hypothesis and that the information asymmetry is more severe when issuing common equity, than for issuing debt. The fifth hypothesis regarding ownership changes also suggests that it follows the hypothesis well. The reason for this according to Smith is that announcements of transactions that increase ownership concentration raise share price. And those transactions that lower ownership concentration actually lower the share prices. 10

11 Schwartz and Aronson (1967) investigated whether differences in debt ratios existed among different industries and tested whether there exist optimal capital structure in specific industries. In order to test differences in debt ratios and if optimal capital structure existed in specific industries, Schwartz and Aronson collected statistics of firm financial structure and compared the composition between industries. If they found similarities within specific industries, the outcome according to Schwartz and Aronson must be an indication of optimal capital structure. Otherwise, there should not exist a recognizable pattern and the capital structure within a specific industry should vary randomly. The outcome of Schwartz and Aronson research was that the typical financial structure of a firm within a specific industry differs significantly from the financial structure of a firm belonging to a different industry. Hovakimian et al. (2001) tested whether firms tend toward a target debt ratio when they either raise new capital or repurchase existing capital. Hovakimian et al. employed a two- stage estimation procedure. In the first step, estimation of target debt ratio is observed by regressing observed debt ratios. The second step calculates the difference between the estimated target debt and the actual ratio, which implies whether the firm raises debt or equity. Hovakimian et al. concluded that firms tend to move toward a target ratio in the long run. They also found that a firms target ratio may change over time due to changes of the firms profitability and stock price. A negative relation between profitability and debt level indicated pecking order tendencies. Flannery and Rangan (2004) studied if nonfinancial firms have long- run leverage targets and if this is the case, they wanted to investigate how quickly the firms adjusted toward them. In order to measure if firms have long- run leverage targets and if so, how quickly firms tend toward them, Flannery and Rangan used a partial adjustment model. Previous empirical models impose unjustified assumptions about the adjustment speed and the properties of target leverage and this is the motivation behind Flannery and Rangans study. The outcome of Flannery and Rangans research is that a typical firm closes about 30% of the gap between its actual and its target debt ratio each year with the partial adjustment model, this is significantly higher than results presented in earlier studies. 11

12 Graham (2000) investigated how big the tax benefits of debt are. The reason why he investigated the tax benefits of debt is because earlier research showed evidence implying that tax benefits are one of the factors that affect financing choice. The questions asked is if tax benefits of debt affect corporate financing decision and how much they add to firm value. In order to estimate the tax costs and the benefits of debt, Graham derived an interest- deduction benefit function that consisted of one variable that assumed tax rate without interest deduction and a second variable with tax rate, including tax deduction. To analyse whether firms balance the costs and benefits of debt when they make financial decisions, Graham compared the nontax factors that affect debt policy, such as information asymmetry and the expected cost of financial distress among other nontax factors to the tax benefit function. Graham found that a firm that make use of the tax benefit could increase firm value by 9.7 per cent. Graham also found that large, liquid and profitable firms with low expected distress costs use debt conservatively Empirical Studies of the Capital Structure To investigate the capital structure, financial leverage is often used as a dependent variable. Depending on industry and country, the definition of leverage may differ. Due to the fact that we focus on the real estate industry, we defined leverage as net debt divided by total asset. The reason why we took this definition was that we assumed that current liabilities would not affect the managerial decision for listed real estate firms to a great extent when facing a potential investment. Rajan and Zingales (1995) stated that the determinants of capital structure reported in the US such as tangibility, size, growth and profitability are similar in the G7 countries (I.e US, Japan, Germany, France, Italy, UK and Canada). The outcome of Rajan and Zingales research, show that the independent variables tangibility, growth, size and profitability follow a similar pattern as in the US. The reason for this is that tangibility is positive correlated with leverage in all countries, growth is negatively correlated with leverage in all countries except Italy, size is positive correlated with leverage in all 12

13 countries except Germany and profitability is negatively correlated with leverage in all countries except Germany. According to Damodaran (2001), firms using debt instead of equity yields a tax advantage, which could make managers more disciplined in their choice of investments. However by using debt instead of equity, the expected cost of bankruptcy will increase. So by balancing the benefits of debt against the cost, the trade- off is obtained. In other words, firms have a target debt level or ratio to which they slowly adjust to, which will lead to that there should be an optimal capital structure for each firm which maximises firm value. According to Myers (2001), if the theory is right, a value- maximizing firm should never pass up interest tax shields when the probability of financial distress is remotely low. According to Myers and Majluf (1984) the pecking order emphasize asymmetric information and with that indicating problems associated with external financing. Myers (2001) state that when internal cash flows are not sufficient to fund capital expenditures, firms will borrow rather than issuing equity and also that dividends are sticky, so that dividends are not used to finance capital expenditure and by that Myers state that changes in net cash, show up as changes in external financing. The pecking order also says that if internal cash flow exceeds capital investment, the financial surplus is used to pay down debt. Shyam- Sunders and Myers (1999) tested the trade- off theory against an alternative hypothesis, based on the pecking order theory. In order to conduct the research, the authors constructed a target adjusted (trade- off) model and a simple pecking order model (these models will be presented further down in our thesis). The simple pecking order model is intended to explain much more of the time- series variance of actual debt ratios, than the target adjusted model based on the trade- off theory. Thus they emphasize that the models constructed, cannot in general be precise in its predictions. What they found when implementing the models was that the pecking order is an excellent first order descriptor of corporate financing behaviour. However, there are questions regarding if the pecking order will do as well for growth companies investing 13

14 heavily in intangible assets. Also, there could be questions regarding that the models are very simple. In contrast to that, Frank and Goyal (2004) found, when testing the pecking order (using the same model), that internal financing is not sufficient to cover investment spending on average and that external financing is heavily used, which contradicts the pecking order. Along with that they also found that debt financing is not preferred over equity financing. To obtain a more solid result, Frank and Goyal tested the simple pecking order model against a modified conventional leverage model (this will also be presented later in our thesis) and the results show that the added financial deficit did not overtake the conventional leverage factors. In fact it had a very small influence on the explanatory power. Along with that they also found that the pecking order performs much worse for small firms and high- growth firms but for mid- sizes firms, the theory gains support the larger the firm is. In other words, they found that the pecking order in fact perform much better for large firms but not good enough to account for the broad patterns of capital structure among American firms, which is contradictive to Shyam- Sunders and Myers conclusion. 2.3 Theoretical Framework The previous section showed empirical studies when the perfect capital market assumptions hold by Miller and Modigliani was violated, which interfered with their conclusion that the capital structure is irrelevant when maximizing firm value. In this section we plot the theoretical toolbox Theoretical Toolbox The theoretical toolbox consists of the two theories, the trade off- and the pecking order theory. This will be described in the following subheadings Trade- Off Theory The definition of the trade- off theory is that the total value of a levered firm is equal to the value of a firm without leverage plus the present value of the tax savings from debt, less the present value of financial distress costs. 14

15 The trade- off theory explains why firms finance with debt. Balancing the cost against the benefits, leads to the idea of an optimal capital structure Pecking Order Theory The definition of the pecking order theory of capital structure is derived from Myers (2001), which implies that; firms prefer internal over external financing, dividends are sticky, if external funds are required for capital investments the firm will issue the safest security first and if internally generated cash flow exceed capital investment the firm would pay down debt, each firm s debt ratio reflects the need for external finance. 3. Method The method chapter is posted in the following way; first we discuss the scientific perspective, secondly we operationalize our research questions, thirdly we discuss our data sample and finally we discuss the applied regression models. 3.1 Scientific Perspective This thesis is of a quantitative nature, investigating firm financing behaviour. We collect secondary data from data- stream and test with the OLS regression method, test if changes in debt ratios is driven by the need for external funds or by an attempt to reach an optimal capital structure, we also test if there has been a change in the capital structure before and after the financial crises. This thesis is based upon a positivistic approach and we use a similar definition as Bryman and Bell (2005). 3.2 Operationalization In order to investigate whether listed Swedish real estate firms follow the pecking order and/or the trade off, we need models to test this empirically. The models used to test the pecking order is the simple pecking order model, the model used to test the trade off is the simple target adjustment model and to investigate whether the firms fits the pecking order or the trade off theory better, we have used the conventional leverage regression model. 15

16 In order to study the Basel III effects on the listed Swedish real estate firm s debt levels, we have studied different time periods, including prior to the financial crises, after the financial crises and for the whole period. Due to the fact that previous studies found significant differences between large and small firms, we have included tests on mid- cap firms and on large- cap firms. 3.3 Data Sample The data sample consists of yearly collected data between the years , from 15 companies. The total amount of observations is 125. Table I. The targeted real estate firms listed at the omxspi. Name Symbol Valuta ISIN Sektor ICB- kod Cap Atrium Ljungberg B LJGR B SEK SE Financials 8600 Mid- cap Fast. Balder B BALD B SEK SE Financials 8600 Mid- cap Catena CATE SEK SE Financials 8600 Mid- cap Corem Property Group CORE SEK SE Financials 8600 Mid- cap Diös Fastigheter DIOS SEK SE Financials 8600 Mid- cap Fast Partner FPAR SEK SE Financials 8600 Mid- cap HEBA B HEBA B SEK SE Financials 8600 Mid- cap Kungsleden KLED SEK SE Financials 8600 Mid- cap Klövern KLOV SEK SE Financials 8600 Mid- cap Victoria Park A VICP A SEK SE Financials 8600 Mid- cap Wihlborgs Fastigheter WIHL SEK SE Financials 8600 Mid- cap Castellum CAST SEK SE Financials 8600 Large- cap Fabege FABG SEK SE Financials 8600 Large- cap Hufvudstaden A HUFV A SEK SE Financials 8600 Large- cap Wallenstam B WALL B SEK SE Financials 8600 Large- cap The first criteria we used were to search after a specific industry that real estate firms are classified to, which in this case was Financials. In order to specify our intended target to listed real estate firms, we targeted a specified industry classification benchmark (ICB- kod). Once the sorting was finished, we had 21 companies left. After looking through the targeted companies, we concluded to remove JM due to some orientation to the construction management. We also removed Besqab, Hemfosa 16

17 Fastigheter, Platzer Fastigheter Holding and Tribona, due to the short time listed on the Nasdaq OMX Stockholm. 3.4 Dependent Variables In the simple pecking order model the dependent variable is either the change in net debt or the change in net debt divided by the net assets. In the simple target adjustment model the dependent variable is the ratio of net debt and total assets. In the conventional leverage regression model the dependent variable is the change in total debt to market capitalization ratio. 3.5 Descriptive Statistics Table II: Real estate firms cash flows (all firms) Illustration of real estate firms cash flows, each variable is divided by net assets. 5 Year N Dividend (a) 0,029 0,021 0,064 0,038 0,025 0,045 0,136 0,027 0,025 Investments (b) 0,041 0,279 0,146 0,088 0,043 0,096-0,109 0,136 0,168 WC (c) 0,238 0,001-0,022 0,041-0,108 0,052 0,242 0,042-0,039 Cash Flow (d) 0,071 0,072 0,066 0,062 0,072 0,066 0,050 0,053 0,055 DEF (a+b+c- d) 0,237 0,229 0,122 0,105-0,112 0,128 0,220 0,152 0,100 Net Debt issue (e ) 0,046 0,280 0,188 0,133 0,018 0,021 0,108 0,150 0,261 Net Equity issue (f) 0,191-0,051-0,066-0,028-0,130 0,107 0,111 0,002-0,161 Net External Finance (e+f) 0,237 0,229 0,122 0,105-0,112 0,128 0,220 0,152 0, ,49 is the average debt ratio for all companies during the period

18 All Uirms 0,400 0,300 0,200 0,100 0,000-0,100-0, Net Debt Net Equity Net External Financing 0,6000 0,4000 0,2000 0,0000-0,2000-0,4000 Mid- Cap Net debt Net equity Net External Financing 0,4 0,2 Large- Cap Net External Financing 0-0,2-0, Net Equity Net debt 3.6 Empirical Method The most common method to estimate parameters of a regression line is the ordinary least squares (OLS). There are other methods, like the maximum likelihood or method of moments (where the later can be more complicated than necessary) used to estimate parameters, but in our case we have decided to work with OLS. This because our sample size can in some periods be too small to obtain an unbiased parameter when using the maximum likelihood method. 18

19 3.7 Regression Models The models we use to test our thesis derive from Shyam- Sunders and Myers (2001), Frank and Goyal (2003) and Bond and Scott (2006). The models are widely accepted and have been used frequently to investigate capital structure among different corporations and industries The Simple Pecking Order Model To make this study possible we have to make the assumption that there are three sources of funding available to firms: retained earnings, debt and equity. Of the three, equity is the one exposed to severe information asymmetry, debt only a minor information asymmetry problem and while retained earning have no information asymmetries. In order to test the pecking order we need to aggregate some variables. We use the same variables as Frank and Goyal (2003): DIVt = Dividends in year t (Cash dividend paid- Total) It = Investment in year t (Net Cash Flow Investing, represents the net cash receipts and disbursements resulting from capital expenditures, decrease/increase from investments, disposal of fixed assets, increase in other assets and other investing activities. A positive value in this field represents an outflow (use) of funds. A negative value in this field represents an inflow (source) of funds.) Wt = Change in working capital in year t (Change in operating working capital + change in cash and cash equivalents + change in current debt) CFt = Dt = Internal cash flow in year t (Funds From Operations represents the sum of net income and all non- cash charges or credits. It is the cash flow of the company.) Net debt in year t (Net Debt represents Total Debt minus Cash. Cash represents Cash & Due from Banks for Banks, Cash for Insurance companies and Cash & Short Term Investments for all other industries.) 6 Et = Net equity in year t These variables are aggregated to form a new variable, an accounting identity: 6 We also test a dependent variable with change in net debt/net assets, we define net assets as NET ASSETS = TOTAL ASSETS TOTAL DEBT 19

20 (1) DEFit = DIVt + It + Wt - CFt = Dt + Et, = Financial deficit (surplus) With this variable, the pecking order hypothesis to be tested is: (2) Dit = α + βdefit + εit, εit =Error term It is possible that DEFit contain information that helps account for Dt, but not in the manner hypothesized by the pecking order theory. To check if this aggregation step is acceptable we run the equation on a disaggregated basis to see whether the data satisfies this step. We form the following; (3) Dt = α + βdivdivt + βiit + βwwt + βcfcft + εt An increase in any of the components of DEFit need to have the same impact on Dt under the pecking order theory, since it is DEFit itself that matters. The hypothesis for the aggregations step is therefore: βdiv = βi = βw = βcf = 1. If the hypothesis hold the aggregation in Equation (1) is justified. The pecking order hypothesis: H0: βpo = 1 H1: βpo The Simple Target Adjustment Model According to the trade- off theory, managers are always trying to find the optimal capital structure. To match the time- series nature of the pecking order specification, the trade- off model can be re- cast as a target adjustment model, where changes in the debt ratio is driven by the deviation of the current ratio from a target. 20

21 The following variables are used, which is taken from Bond and Scott (2006). D!" = D!"!! = Target debt level in year t (The target debt level is an average of the debt level through a specified period of time) Debt level in time t- 1 (4) D!" = α + β!" D!" D!"!! + ε!", εit = Error term D!" = Net debt issuance (reduction) in year t (Net debt divided by total assets) In order to do this test we have to assume that the target is constant over the period. To obtain a target debt level we use a historical debt ratio as a proxy. The trade- off hypothesis: H0: 0 < βta < 1 The reason why the β value should lie between zero and one is because the adjustment towards an optimal level will be larger than zero and the cost of the adjustment, such as the cost of bankruptcy, must be smaller than one The Conventional Leverage Regression Model The conventional leverage regression model aims to explain the relationship between the independent variables tangibility, growth, size and profitability with the dependent variable leverage. This model is proposed to describe the amount of leverage in contrast to the pecking order, which is intended to explain the changes. Table IV: The determinants of leverage and the predicted relationship between the conventional leverage variables and leverage under different specifications of the capital structure decision. Trade- off Model Pecking Order Model Tangibility + - Growth - + Size + + Profitability

22 (5) Dit = α + βt Tit + βmbt MBTit + βls LSit + βp Pit + βdefit Dit = Tit = Debt ratio (change in total debt to market capitalization ratio) Tangibility of assets (Property, plant and equipment divided by total assets) MBTit = Market- to- book ratio (Market value divided by book value) 7 LSit = Size (Natural Logarithm of Net sales or revenues) Pit = Profitability (Operating income) DEFit = Financial deficit (surplus) Equation (5) is a conventional regression run in first differences and with financing deficit as an added factor. The conventional leverage hypothesis: According to Harris and Raviv (1991), firms with scarce tangible assets in relation to the value of the firm could have problems associated with information asymmetries. This fact makes tangible assets to one of the most important variables when investigating if one can find any support for the pecking order when using the conventional leverage regression. The prediction according to Harris and Raviv (1991) is that βt < 0. Since the trade- off theory states that the optimal capital structure would exist where the equilibrium between equity and debt lies between the tax benefits and the cost of bankruptcy, the tangible assets should be able to reduce that cost and therefore firms with a considerable amount of tangible assets should thereby work with more debt (Rajan and Zingales, 1995). According to Frank and Goyal (2003) the prediction is that βt > 0. In order to measure growth, Frank and Goyal (2003) states that if a firm have a high market- to- book ratio, the firm may have greater opportunities to grow in the future and that real estate firms might increase their use of debt financing when the growth opportunities decline. The trade- off predicts that firms with growth opportunities will use a smaller amount of debt and this, according to Bond and Scott (2006), occurs because of the increased risk of bankruptcy. The common prediction for the trade- off is that βmtb < 0. 7 Market value is defined as (stock price per share*outstanding shares) and the book value is defined as (book value per share*outstanding shares) 22

23 Frank and Goyal (2003) stated that the pecking order suggests small firms with growth opportunities are likely to face information asymmetries and will therefore strive to issue debt in order to minimize the asymmetry. The common prediction for the pecking order is that βmtb > 0. According to Frank and Goyal (2003), large firms are commonly known to be more diversified, have a better ranking in debt markets and by that having lower costs when borrowing. The larger firms should therefore have more debt in their capital structure. They also state that there is an inverse relation between size and the probability of bankruptcy, which is what the trade- off predicts. According to Bond and Scott (2006), the pecking order expects that larger firms, ought to use more debt. The prediction for both the trade- off and the pecking order is that βls > 0. Since the pecking order states that firms prefer internal financing over external and that firms will use profits to pay down debt. Fama and French (2002) stated that there is a negative relation between profits and leverage. The prediction for the pecking order is therefore that βp < 0. On contrary to the pecking order, the trade- off predicts that firms with high profits should have higher debt level in order to take advantage of the tax- shield. The prediction for the trade- off is therefore βp > 0. According to Frank and Goyal (2003), if the pecking order dominates the trade- off in a direct comparison, the addition of the financial deficit variable in the conventional leverage regression should capture all the effects and erase the influence of the conventional variables. 23

24 4. Empirical presentation, Analysis and Findings To test the hypothesis we form the test statistic;!!!!"#$%#&%!"#$%&$'( the estimated value and β* is the beta value under the null hypothesis. ~ N(0,1), where β is 4.1 The simple pecking order model Net debt issued, all firms. When running the first regression through the whole period ( ) and including all companies, we find weak support for the pecking order. The reason for this is that the beta value that should lie close to one was instead 0,67 and when calculating the t- statistic we obtained a value of - 6,2 which is larger than 2,576 (critical value of t- distribution, 1% significance level). Therefore we reject the null hypothesis. Running the second regression through the period ( ) and including all companies, we also find weak support for the pecking order because we get a beta value of 0,62. Even in this period we reject the null hypothesis since the test statistic obtained = - 5,97 which is larger than (critical value of the t- distribution, 1% significance level). The results after running the third regression including all companies through the period ( ), we obtained a beta coefficient of 0,945, which indicates high support for the pecking order. When calculating the test statistic we got a value of - 0,45 which is smaller than 1,994 ( critical value of t- distribution, 5% significance level) and by that we fail to reject the null hypothesis that β = 1. See appendix table VII. Net debt issued as a fraction of Net assets, all firms. When dividing the variables with net assets, the aim is to obtain a more solid result, this because the proportions of the firm s variables may have an impact on the regression coefficients. So, in the fourth regression, including all companies during the whole period ( ), we got a beta coefficient of 0,29, which is even smaller than the beta value obtained for whole period without the fraction of net assets. We reject the null hypothesis, since the test statistic obtained ( - 17,3 ) is larger than the critical value of the t- distribution and with that, we fail to find any support of the pecking order. 24

25 Running the fifth regression including all companies during the period ( ), we find support for the pecking order. The reason for this is that we fail to reject the null hypothesis since the beta coefficient of 0,69 cannot be rejected that of being equal to one on the significance level 1 percent due to the test statistic obtained stays within its limits. Also when looking at the 99 percent confidence interval, the conclusion stays the same and this because the interval include the null value. This contradicts the conclusions drawn from the same period, but without fractions of net assets, where we could not find any support for the pecking order. When running the sixth regression including all companies during the period ( ), we got a beta coefficient of 0,24. In contrast to the regression done without net assets where we found support of the pecking order, we here reject the null hypothesis and by that we find no indication of the pecking order. The reason for the big contrast or why the conclusions differ may lie in the impact that the large firms may have on the dependent variable since, as will be shown further down, that the support of the pecking order is strong among large firms. See appendix table VII. Net debt issued for Mid- cap firms Regression for the whole period ( ), we got a beta coefficient of 0,36. Since the 99 percent confidence interval does not include 1, we reject the null hypothesis. This can also be enhanced by looking at the test statistic ( - 6,98 ) which is larger than the critical value of the t- distribution. For the period ( ), we obtained a beta coefficient of 0,403. The test statistic value ( - 5,17 ) and the 99 percent confidence interval, both tells us to reject the null hypothesis. And for the last period ( ) the beta value obtained was 0,28. The test statistic ( - 4,42 ) and the confidence interval, both forced us to reject the null hypothesis. The support for the pecking order was in this case absent. See appendix table VIII. Net debt issued for mid- cap firms as a fraction of Net assets For the whole period ( ), when running the regression, we got a beta coefficient of 0,35. The test statistic ( - 11,71 ) for this beta value is larger than the critical value of the t- distribution (2.632, 1%) and we therefore reject the null. 25

26 When running the regression for the period ( ), the beta value obtained was 0,63. The 99% confidence interval includes the null value and by that, in contrast to the whole period, we fail to reject the null hypothesis. This can also be concluded by looking at the value of the test statistic ( - 2,64 ) which is smaller than the critical value (2.750, 1%). When running the regression for the final period ( ), we obtained a beta coefficient of 0,29. The null hypothesis was rejected both by looking at the test statistic ( - 12,3 ) and the confidence interval and as a consequence, no trace of the pecking order. See appendix table IX. Net debt issued for Large- cap firms When running a regression including only large- cap firms for the whole period ( ), we get a beta coefficient of 0,88. Due to both test statistic ( - 0,87 ) and confidence interval, we fail to reject the null. We can in other words find support for the pecking order for this period. For the period ( ), the beta value obtained was 0,92 and also in this period we fail to reject the null since the test statistic ( - 0,33 ) is larger than the critical value and the 99 percent confidence interval include the null value, the support for the pecking order is strong. When running the regression for the final period ( ), we got a beta coefficient of 0,82. In alliance with previous results for large firms, we fail to reject the null and this because of both the value of the test statistic ( - 1,31 ) is smaller than the critical value and the confidence interval (99 and 95 percent) contain the null value. See appendix table VIII. Net Debt Issued for Large- Cap Firms as a Fraction of Net Assets Running the regression for the whole period ( ), we got a beta coefficient of 0,96. The low test statistic ( - 0,33 ) and the confidence interval both made it impossible to reject the null. The support for the pecking order is consistent. When running the regression for the period ( ), the beta value obtained was 1,04. The test statistic ( 0,18 ) is smaller than the critical value of the t- distribution (2.977, 1%, 2.145, 5%) and the confidence interval include the null value. 26

27 For the final period ( ) when running the regression, we got a beta coefficient of 0,79. The support for the pecking order is strong even for this period is strong since the test statistic ( - 1,79 ) is smaller than the critical value (2.879, 1%, 2.101, 5%). See appendix table IX. The Aggregation Step Running the regression regarding the aggregation step, we conclude that the aggregation step is not fully justified. The reason for this is that the beta coefficient for working capital is not close to 1 and to support this conclusion we performed an F- test, which gave us a test statistic of 21,86, this is larger than the critical value of the F- distribution and the null is therefore rejected. However, the other variables are close to 1. The problem regarding the low working capital variable also arises for Nuri and Archer (2001). Table V: Test of the aggregation step. Dt = α + βdivdivt + βiit + βwc WCt+ βcfcft + εt The hypothesis to be tested is; H0: βdiv = βi = βwc = βcf = 1. The test refers to the whole period. Net Debt Issue P- Value Net debt Issue/Net Assets P- Value Intercept, α ,234 0,058 0,138 β Dividend 1,076 0,001 0,996 0,001 β Investments 0,980 0,000 1,044 0,000 β Working Capital 0,246 0,001 0,022 0,726 β Cash Flow 1,066 0,001 1,161 0,039 N R- square 0,693 0, The Simple Target Adjustment Model Net debt based on a historical average ( ) When running the regression for the whole period ( ) including all companies, we find support for the trade- off theory. The reason for this is that we got a beta coefficient of 0,79. Due to the hypothesis that the beta should be between zero and one, 27

28 we found it sufficient to look only at the confidence interval, in this case, in a 99 percent confidence interval we fail to reject the null hypothesis. See appendix table III. Net Debt Based on a Five- Year Rolling Average ( ) Running the regression for the period ( ) including all companies, we find no support for the trade- off theory. The reason for this is that we got a beta coefficient of 0,93 and in a 95 percent confidence interval we reject the null hypothesis. See appendix table III. Net Debt Based on a Five- Year Rolling Average ( ) Running the regression for the period ( ) including all companies, we find strong support for the trade- off theory. The reason for this is that we get a beta coefficient of 0,57. In both a 99- and 95- percent confidence interval, we fail to reject the null hypothesis. See appendix table III. Net debt, Mid- Cap Firms For mid- cap firms over the whole period ( ) the support of the trade- off theory is weak, and this because the confidence interval (95%) for the beta value (0,778) stretches beyond one. This result stays the same for the period ( ) where the beta value obtained was 0,912 with one difference that none of the confidence intervals (90%, 95% and 99%) stayed within the limits. For the last period ( ) we found strong support of the trade- off theory since all the confidence intervals (90%, 95% and 99%) for the beta value (0,547) stayed within the boundaries. See appendix table III. Net Debt, Large- Cap Firms For Large- Cap firms over the whole period ( ), we fail to find support for the trade- off theory is and this because the confidence intervals (95%, 99%) for the beta value (0,857) contain values larger than one. This stays the same for the first period ( ) where the beta value of 0,918 has confidence intervals that all stretches over one. In the final period, the support of the trade- off is, in contrast to previous findings, strong. The beta value of 0,538, have confidence intervals (90%, 95% and 99%) that stays within its limits. See appendix table III. 28

29 4.3 The Conventional Leverage Regression Model (Large- Cap) Since we only found strong support for the pecking order among large- cap firms we found it sufficient to do the conventional leverage regression only on large firms to see if the variable DEF has any affect or not. Table VI: The determinants of leverage and the relationship for large- cap firms for the time period prior to the financial crises, after the financial crises and for the whole period Tangibility Growth Size Profitability Financial Deficit Change in total debt to market capitalization ratio ( ) When running a regression for the whole period ( ) including only large- cap firms and taking notice of the financial deficit the following results was obtained. Regarding the tangibility variable, we conclude that it follows the trade- off theory. The reason for this is that we got a positive beta of 5,81 and with a 99 percent confidence interval that tells us that the beta coefficient is stays positive. When looking at the growth variable, we find a negative beta coefficient of - 1,25, also indicating the trade- off theory. To get support, we observe the confidence interval (99%), which only contain negative values for the coefficient. The size variable tends towards neither of the theories. The reason for this is that the size variable is negative with a beta coefficient of - 0,4. However, the 99 percent confidence interval range between - 1,33 to 0,53, which can indicate a positive relationship but be aware of the high p- value (0,24) which tells that there is a relative high probability that the coefficient true value is zero. The profitability variable tends towards the trade- off theory. The reason for this is that the profitable variable is positive with a beta coefficient of 21,16 and is positive through a 99 percent confidence interval. Looking at the financial deficit (surplus) variable, we find some evidence of the pecking order theory. The reason for this is that we get a positive beta coefficient of 2,19, and we 29

30 get a positive value through out in a 99 percent confidence interval. See appendix table X. Change in total debt to market capitalization ratio ( ) Looking at the tangibility variable, the indication is that it follows the trade- off theory. The reason for this is the positive beta coefficient of 5,73. However when looking at the 95 percent confidence interval we have both positive and negative values. Thus, we cannot conclude that it follows neither the trade- off nor the pecking order. When observing the growth variable, we get an indication that it follows the trade- off theory. The reason for this is that we get a negative beta coefficient of - 0,74. However, as for the previous variable, when looking at the 95 percent confidence interval we obtained both positive and negative values. Thus, we cannot conclude that it follows neither the trade- off nor the pecking order. The size variable through this period indicates that it follows neither the trade- off nor the pecking order. The reason for this is firstly that we get a negative beta of - 0,36 and secondly that the values are both negative and positive in a 95 percent confidence interval. The profitability variable indicates that it tends towards the trade- off theory. The reason for this is that we get a positive beta coefficient of 20,20, however we cannot conclude this because we get both positive and negative values in our 95 percent confidence interval. We also get a very high p- value, thus we cannot draw any conclusions. The financial deficit (surplus) variable indicates some evidence of the pecking order. The reason for this is that we have a positive beta coefficient of 2,85, and we get a positive value throughout the 95 percent confidence interval. We also get a very low p- value, which indicates that we can find evidence that large- cap firms during this period of time tends towards the pecking order theory. See appendix table X. Change in Total Debt to Market Capitalization Ratio ( ) For this period, when looking at the tangibility variable, we conclude that it follows the trade- off theory. The reason for this is that we obtained a positive beta value of 9,67 and with a 95 percent confidence interval that the beta coefficient stays positive. 30

31 Looking at the growth variable, the negative beta coefficient of - 1,42 we obtained, indicated the trade- off theory. We find support for this with all negative value in a 99 percent confidence interval. The size variable tends towards neither of the theories; the reason for this is that the size variable is negative with a beta coefficient of - 0,03. However, the 95 percent confidence interval range between - 1,33 to 0,53, which can indicate a positive or no relationship at all since the p- value is extremely high (0,97). The profitability variable tends towards the trade- off theory. The reason for this is that the profitable variable is positive with a beta coefficient of 12,28 and is both positive and negative through a 95 percent confidence interval also the p- value is very high (0,20). By that we fail to draw any conclusion from this. Looking at the financial deficit variable, we find some evidence of the pecking order theory. The reason for this is that we obtained a positive beta coefficient of 1,47, and we have a positive value through out in a 95 percent confidence interval. See appendix table X. 5. Discussion and Critical Reflection According to the pecking order theory, firms will want to finance investment first with internal funds, second with debt and last with equity. However, internal funds have no degree of information asymmetry, debt have minor information asymmetry and equity have sever information asymmetry. If we first look at the descriptive statistics for all firms (table II), we can see that the net external financing fluctuated between 10% to 23,7% during the period if we exclude year 2009, which had a negative net external financing at 11,2%. Clearly, we could see the affects of the financial crises during this period, with a more passive investment strategy as a consequence. If we then break out the parts that creates the net external financing, which are the net debt issuing and the net equity issuing, we could see that prior to the financial crises net debt lied between 4,6% to 28% and reduced year 2009 and 2010 with net debt ranged between 1,8% and 2,1%, after 2010 net debt ranged from 10,8% to 26,1%. Noticeably, we can see the affects of the financial crises, with low net debt issuance for the years 2009 and 2010, with numbers close to zero. According to the pecking order theory, debt should constitute the most of the net 31

32 external financing and thus meaning that net equity issuance should be close to zero. If we look at the period prior to the financial crises, we could see that year 2005 composed of 19,1% of net equity issuance, this is clearly of the mark and thus not follow the pecking order particularly well. However, during the period , the net equity issuance ranges between minus 2,8% to minus 13%, this implying that net equity was paid down during this period. During the period year 2010 and 2011, net equity issuance stretched from 10,7% to 11,1%, implying that net equity issuance was a significant financing tool during this period of time. Year 2012, the net equity issuance was close to zero and year 2013 it dropped to minus 16,1%. When looking at the relationship between the net external financing and the net debt issue for all firms, we could see that net debt does not follow net external financing that well compared to large- cap firms. When running the regression with the simple pecking order model, our empirical finding was that large- cap firm follows the pecking- order theory more closely than the middle- cap firms. This is consistent throughout the whole period ( ), the period before the financial crises ( ) and the period after the financial crises ( ). This results support previous studies. Frank and Goyal (2003) found that large firms follow the pecking order, however they didn t find support for the pecking order at all for small firms. This is in fact very similar to our results. Our findings suggest that there is no support for the pecking order regarding small firms (mid- cap firms), except prior to the financial crises ( ), when using variables as a fraction of net assets. One reason that there was such a distance between the mid- cap and large- cap firms could do to the fact that many of the mid- cap firms were small- cap firms in the beginning of our study period and evolved to mid- cap firms during the time period of our research. However, Bond and Scott (2006) found that the pecking order fits best among small firms and they assumed that this is subject to larger information asymmetries for small firms than for big firms. This is the complete opposite of our findings, where we find no evidence that the pecking order performs best among mid- cap firms. How come that our empirical findings differ so much? The rational thought explaining the differences is of course different markets, we investigated the Swedish market and Bond and Scott investigated the UK market. The time aspect of the study must also be taken as a 32

33 consideration, the reason for this is that our study investigated the years , while Bond and Scott examined the time period between the years Also, to take notice of is that we have included financial crises in our time period versus Bond and Scott has not done so. Another thing that we have reflected over is that in Bond and Scott s study, real estate firms net debt issuances follows the net external financing closely, indicating that the net equity issuance is close to zero. This is not the case in our empirical findings, where we found different fluctuations. One interesting thing with our study is that we investigated the financing behavior regarding debt subject to the Basel III reform. When evaluating the estimations of our regressions regarding the simple pecking order model, we can see that for both mid- cap firms and large- cap firms, the pecking order offers a better fit prior to the financial crises than after the financial crises. One possible reason for this could be that the Basel III regulation aimed to put pressure on the banks to increase their capital adequacy requirements, which in turn puts greater equity demands on its lenders. Thus, the real estate firms would need to increase their amount of equity to finance debt or finance directly with equity as preference shares or by corporate bonds. Noticeably, we could see that this was the case, especially during the period ( ). During this period, the net debt requirement remained the same although net equity increased. As the pecking order tells us firms prefer debt before equity and thus without higher capital adequacy requirements firms financing would follow the pecking order better. Garmaise and Moskowitz (2004) found strong evidence suggesting that asymmetric information is significant in the commercial property market. Although Garmaise and Moskowitz compared informed and uninformed buyers, we think that we can compare this with managers and stockholders. The reason for this is that managers at firms that work at a specific market in the Real estate industry would know the market conditions very well and this would imply that the manager would know what a good respectively a bad investment would be in this environment. If its a good investment, the manager would probably finance the investment with debt and would not take into consideration the increased risk, due to higher bankruptcy cost associated with the increased debt as a result of the leverage effect. However, is the manager unsure of the outcome regarding an investment, the manager would probably issue equity. The reason for this is that the 33

34 associated risk is taken by the shareholders and minimally by the managers. So a shareholder or an investor would probably feel uninformed if a manager raises cash for an investment with equity instead of debt. Looking at the period 2010 and 2011 in our descriptive statistics, the net external financing was partly raised by equity. This would have made the investors and shareholders uninformed due to the associated risk. Smith (1986) investigated the process by which corporations raise debt and equity capital and the associated effects on security pricing. Smith found evidence suggesting that the information asymmetry is consistent with the hypothesis and that the information asymmetry is more severe for common stockholder when a firm issues equity than when a firm issues debt. This reflects the security prices and is consistent with the pecking order theory, which implies that when a firm issues debt the information asymmetry is minimal in comparison when issuing equity. Comparing Smith s outcome with our results, we can see that large- cap firm s follows the pecking order better than mid- cap firms. One possible reason for this could be that large- cap firms issued mainly debt when financing investment, versus mid- cap firms that used a mixture of debt and equity. One of the reasons why large- cap firms issued mainly debt could be because large- cap firms have greater or more severe information asymmetries and the effects of not issuing debt would impact the stock price more negatively than for mid- cap firms. Other possible explanations why large- cap- firms finance investment mainly by debt, could be because they tend to have better credit rating, lower bankruptcy cost, higher age which effects the lending conditions from the banks. Thus, creating incentive for large- cap firms to lend. The trade- off theory explains why firms issue debt and why a firm strives towards an optimal debt ratio. As mentioned earlier, firms using debt instead of equity yields a tax advantage. However, by using more debt this will increase the risk of bankruptcy. Therefore, the optimal debt ratio is obtained where the benefits of debt balances the cost of bankruptcy. Running the regression for the simple target adjustment model including all companies and through the whole period, our empirical findings state that firms tend to follow the trade- off theory closely. Our finding is supported by prior research done by Bond and 34

35 Scott (2006) and Frank and Goyal (2003). One possible reason for this is that firms in the same industry tended towards an optimal debt ratio. This is supported by studies done by Schwartz and Aronson (1967). The reason why a firm in a specific industry tends towards an optimal debt ratio, according to Schwartz and Aronson, is because the typical financial structure of a firm within a specific industry tends to be similar to other firms in the same industry. Thus implying that firms strive toward an optimal capital structure. Hovakimian et al. (2001) presented results suggesting that when firms either raise or retire new capital, their choices move them toward the target capital structures suggested by the static trade- off models. The outcome of our results suggests that Swedish listed real estate firms follow the simple target adjustment model closely, indicating that the firms follow the trade- off theory and takes advantage of tax shields. However, when running the regression including all companies and through the period prior to the financial crises ( ), we reject the null hypothesis which results in that we cannot draw any conclusion that firms tended towards the trade- off theory before the financial crises. However, we can see some tendencies towards the trade- off, the reason for this is that we get a beta coefficient between 0 < β < 1. But the confidence interval stretched outside the restrictions of the null hypothesis. Flannery and Rangan (2004) used a partial adjustment model of firm leverage and they found strong evidence suggesting that nonfinancial firms identified and pursued target capital ratios. Although Flannery and Rangan investigated nonfinancial firms and we investigated financial firms, we think that the outcome of Flannery and Rangan s research indicated that firms in general tended to move relatively quickly towards target debt ratios in the long run. Which indicates that Swedish listed real estate firms tend towards a target debt level in the long run. If we connect our findings to the Basel III reform, we could see that after the financial crises the support for the trade- off increased both for the mid- cap and large- cap firms. One possible reason for this could be that due to lower loan- to- value conditions offered by banks as a result of higher capital adequacy requirements. Thus, firms tended towards maximizing the level of debt versus equity and if this were the case for the 35

36 overall firms, the trade- off would fit better. But further research is needed to draw such conclusion. When doing regressions regarding the conventional leverage regression model, we investigated only the large- cap firms. The reason for this was that the large- cap firms tended towards the pecking order more than the mid- cap firms, thus the large- cap firms is of more interest in a direct comparison with the trade- off theory. Running the regressions with the large- cap firms, we conclude that the trade- off is more present than the pecking order. The reason for this is that the outcome of the independent variables tangibility, growth, size and profitability supports the trade off predicted outcome better than the pecking order predicted outcome. But interestingly, the explanatory power (R 2 ) of the regression increased with almost 60% when adding the additional variable DEF. 8 If we first look at the independent variable tangibility, we could see a positive relationship in all time periods with statistical significance. This implying that an increased level of tangible assets has a positive effect on the leverage of a firm. The positive relationship between tangibility and leverage follow the trade off predicted outcome. According to Gaud et al. (2005) and Rajan and Zingales (1995), the reason why this relationship occurs is because an increase in tangible assets should reduce the cost of financial distress. The reduced cost of financial distress should then according to Gaud et al. and Rajan and Zingales motivate an increased level of debt. Thus, a high amount of tangible should support high debt levels well. The observed relationship is the opposite of the pecking order prediction, which predicts a negative relationship between leverage and the tangibility. Looking at the independent variable growth, we can see tendencies towards the trade off prediction, implying a negative relationship between growth and leverage. The reason for this is that we obtained a negative beta value for the variable growth for all the periods. However, the period prior to the financial crises is not statistically significant and has a considerably low beta value in comparison with the time period after the financial crises and for the whole period. This can indicate that there was a change in the financial behaviour prior to the financial crises, however we cannot draw 8 Without the financial deficit variable, we got a R 2 value of 0,37 for the whole period. 36

37 any conclusion about this due to the fact that not all time periods is statistically significant. According to the trade off theory when a firm have growth opportunities, the associated risk increases when investing in such investments, this according to Bond and Scott (2006). The increased associated risk, increases the bankruptcy cost and as a consequence the debt level will be less. Thus, implying a negative relationship between growth and leverage. However, due to the fact that the time period prior to the financial crises were not statistical significant, we cannot reject the idea that the opposite relationship could be the case. If the opposite relationship occurred, implying a positive relationship between growth and leverage, the pecking order prediction will hold. The pecking order theory states, according to Frank and Goyal (2003), that there should be a positive relationship between growth and leverage and the reason for this is that there exist information asymmetries between the manager and the stockholder. In order to wipe out the information asymmetry, the firm would want to issue debt and therefore increase its debt level. When investigating the independent variable size and its relationship with leverage, we cannot draw any conclusions at all. The reason for this is that the estimated beta value could either be positive, negative or zero during all time periods. So, either there is no relationship at all between size and leverage for real estate firms listed at the Stockholm Stock exchange, or there could exist a positive or a negative relationship. If there exist a positive relation between size and leverage, both the trade off prediction and the pecking order prediction is obtained. According to Frank and Goyal (2003), the trade off theory predicts a positive relationship between size and leverage. The reason for this is that larger firms has lower cost of bankruptcy in comparison with smaller firms and therefore could increase their levels of debt. According to Bond and Scott (2006), the pecking order expects that larger firms will employ more debt. Looking at the independent variable profitability and its relationship with leverage, we obtained a positive beta value in all time periods. However, we only got a beta value that is statistically significant when running a regression for the whole period. Although, all time periods shows tendencies toward a positive relationship, we can only draw 37

38 conclusions for the whole time period and not for the period prior and after the financial crises. A positive relationship between profitability and leverage indicates the trade- off theory. The reason why the trade off theory predicts a positive relationship between profitability and leverage is according to Frank and Goyal (2003) because more profitable firm will employ more debt in order to take advantage of tax shields. Adding the financial deficit variable, we can see that it does not erase the influence of the conventional variables, which according to Frank and Goyal (2003), indicates that the trade- off dominates the pecking order in a direct comparison. However, after the financial crises we see tendencies towards the pecking order theory for large- cap firms. The reason for this is that the effect of the conventional variables decreases when we include the financial deficit variable, indicating that the financial deficit variable captures some effect of the conventional variables. As a parallel to the discussion, Graham (2000) found that firms tended to be not enough leveraged to maximize firm value. We cannot draw any conclusions regarding this, but this is an interesting topic for further research. The reason for this is that listed real estate companies is very low leveraged compared with similar companies for example in the UK. 9 However, Graham also found that large, liquid, profitable firms with low expected distress costs use debt conservatively. This may not be fully comparable with our results due to the liquid factor, real estate firms tends to be not so liquid because of the heavy investments in tangible assets. But putting this a side and looking at the similarities, such as large, profitable firms with low expected distress costs, our empirical result suggests that large- cap firms are profitable firms with low expected distress costs and that they use debt conservatively Limitations We must highlight that we obtained quite low explanatory power when running our regression and especially when testing the pecking order. However, this may not be an issue because the low explanatory power was obtained for mid- cap companies, in which we found no statistically significant support of the pecking order. Also when dividing 9 Bond and Scott found an average debt ratio of 0,82 for listed real estate firms in UK, the time period was between Se appendix, Table XI. 38

39 with net assets, the purpose was to obtain a more solid result but we question if this is the case since when running regression on all companies divided by net assets, the explanatory power decreased. This was only the case when running on all firms, because when running regression on mid- cap and large- cap separately the outcome was the opposite. Another issue is that the intercept in our simple pecking order model should lie close to zero according to the hypothesis, this was not obtained and how this affect the results is hard to tell. Fortunately, we are not alone in obtaining this, see the empirical results of Bond and Scott (2006) and Frank and Goyal (2003). Looking at the target- adjustment model, we obtained a very low explanatory power in the cases where we accepted the hypothesis, i.e. when we found support for the trade- off theory. In contrast to that, when we rejected the hypothesis, the R 2 value was instead on a more acceptable level. Due to time limit, we used the database Data- stream in order to gather data, however to get the best possible data, annual reports and a dialog with people in the industry may be needed to verify key performance indicators. One shortcoming of our study is that we limited our research to large- cap companies, when comparing trade- off and pecking order with the conventional regression model, due to only finding support for the pecking order among large- cap companies. 6. Conclusion The main purpose with our study was to investigate whether real estate firms listed at the Stockholm Stock exchange, tended towards the trade- off theory or the pecking order theory in a direct comparison. Earlier research such as Bond and Scott (2006) found that the pecking order offers a good fit for real estate firms listed in the UK in comparison with the trade- off theory. In contrast to Bond and Scott (2006), we found that the trade- off theory offers a good fit for real estate firms listed at the Stockholm Stock exchange compared with the pecking 39

40 order. The reason for this was that the trade- off theory offered a god fit for both mid- cap and large- cap firms, while the pecking order theory only got support among large- cap firms. Also, in a direct comparison, the trade- off theory dominates the pecking order theory regarding large- cap firms. The reason for this was that the conventional variables tangibility, growth and profitability follow the trade- off prediction. However, in the regression regarding the conventional leverage model, we found tendencies towards the pecking order for large- cap firms after the financial crises. The reason for this is that the financial deficit variable captures some effect of the conventional variables and the explanatory power is higher in this period. 11 Due to the Basel III reform after the financial crises, which put pressure on banks to increase their capital adequacy requirements. Our second research question investigates whether financing behaviour regarding debt has changed for managers at real estate firms listed at the Stockholm Stock exchange since the financial crises in 2008, due to higher equity demands by banks subject to the Basel III reform. Our descriptive statistics shows that large- cap firms has an overall debt level that is lower than mid- cap firms, indicating that large- cap firm uses debt conservatively. We can also see that net debt issue follows net external financing better for large- cap firms, indicating support for the pecking order. To verify the support for the pecking order, we got the same support when running regressions of the simple pecking order model. We can also see that mid- cap firms was affected more by the financial crises than large- cap firms, with higher fluctuations in net debt issuance, net equity issuance and net external financing. However, it is hard to draw any conclusions of the impact of the Basel III reform, the reason for this is that there are more underlining factors affecting a companies debt ratio. What we can observe is that both mid- cap and large cap firms are low leveraged compared with the listed real estate firms in the UK, and with that the Basel III impact may be limited. Future research may look into factors that determine why large- cap firms follow the pecking order and why mid- cap firms don t. Factor that may be of interest in such research is if large- cap firms have greater or more severe information asymmetries than 11 The adjusted R 2 increases from 0,74 to 0,80. 40

41 mid- cap firms. One way of investigating this, may be to look at the announcement effects on security prices, when a firm issues equity or issues debt. If the change in security prices is higher for large cap- firms than for mid- cap firms, the information asymmetries is greater and more severe for large- cap firms. References Bank For International Settlements Communications. (2010), Basel III: A global regulatory framework for more resilient banks and banking systems. Berk, J. DeMarzo, P. (2013) Corporate Finance. Pearson 3rd Global Edition. Bond, S. and Scott, P. (2006) The Capital Structure Decision for Listed Real Estate Companies, p.1-33 Brennan, M. and Kraus, A. (1987) Efficient Financing under Asymmetric Information. Journal of Finance, vol.xlii, no.5, p Bruggeman, William, B. and Fisher, Jeffery, D. (2010), Real Estate Finance and Investments, McGraw- Hill, 14th edition. Bryman, A. & Bell, E. (2005) Företagsekonomiska forskningsmetoder. Malmö: Liber Constantinides, G. Grundy, B. (1989) Optimal Investment with Stock Repurchase and Financing as Signals. Oxford university press, Vol. 2, No. 4, p Damodaran, A. (2001) "Corporate Finance Theory and Practice." 2nd edition, Wiley International Fama, E. and French, K. (2002) "Testing Trade- off and Pecking Order Predictions About Dividends and Debt." Review of Financial Studies, vol.15, no.1, p.1-33 Flannery, M. and Rangan, K. (2004) "Partial Adjustment towards Target Capital Structures." AFA 2005 Philadelphia Meetings, Frank, M. and Goyal, V. (2003) "Testing the Pecking Order Theory of Capital Structure." Journal of Financial Economics, vol.67, p Garmaise, M. and Moskowitz, T. (2004) "Confronting Information Asymmetries: Evidence from Real Estate Markets." Review of Financial Studies, vol.17, no.2, p Gaud, P., Jani, E., Hoesli, M. and Bender, A. (2005) "The Capital Structure of Swiss Companies: An Empirical Analysis Using Dynamic Panel Data." European Financial Management, vol. 11, no. 1, p

42 Graham, J. (2000) "How Big Are the Tax Benefits of Debt?" Journal of Finance, vol.55, no.5, p Harris, M. and Raviv, A. (1991) "The Theory of Capital Structure." Journal of Finance, vol.46, p Hartomaa, Juha. (2013) Finansieringen flyttar bort från banken. 4 februari. flyttar- bort- fran- banken (Hämtad ) Hovakimian, A., Opler, T. and Titman, S. (2001) "The Debt- Equity Choice." Journal of Financial and Quantitative Analysis, vol.36, no.1, p.1-24 Modigliani, F. and Miller, M. (1958) "The Cost of Capital, Corporation Finance and the Theory of Investment." American Economic Review, vol.48, no.4, p Myers, S. and Majluf, N. (1984) "Corporate Financing and Investment Decisions When Firms Have Information that Investors Do Not Have." NBER Working Paper 1396 Myers, S. (2001) "Capital Structure." Journal of Economic Perspectives, vol.15, no.2, p Nuri, J. and Archer, S. (2001) Target adjustment model against pecking order model of capital structure. European Financial Meeting Association Rajan, R. and Zingales, L. (1995) "What Do We Know About Capital Structure? Some Evidence from International Data." Journal of Finance, vol.50, no.5, p Schwartz, E. and Aronson, R. (1967) "Some Surrogate Evidence in Support of the Concept of Optimal Financial Structure." Journal of Finance, vol.22, p Shyam- Sunder, L. and Myers, S. (1999) "Testing Static Trade- off Against Pecking Order Models of Capital Structure." Journal of Financial Economics, vol.51, p Smith, C. (1986) "Investment Banking and the Capital Acquisition Process." Journal of Financial Economics, vol.15, no.12, p.3-29 Solomon, E. The Theory of Financial Management (Columbia University Press, 1963) 42

43 Appendix Table III: Test of the Trade- off theory. The regression estimated with panel data covering 15 firms over 10 years is ΔD!" = α + β!" D!" D!"!! + ε!" where ΔD!" is the net amount of debt issued. The null hypothesis tested is 0 < β TA < 1. Three different specifications for D it are tested. The debt target is based on a firm historical average ( ) net debt to total assets. The debt target for the period before the financial crisis is based on a five- year rolling average debt ratio ( ). The debt target for the last period is based on a five- year rolling average debt ratio ( ). Standard errors are reported in parentheses. * Indicates significance at the 1% level and ** indicates significance at the 5% level. All companies Net Debt Net Debt Net Debt Period Intercept, α - 0,01110 (0,00577) - 0,02080** (0,00860) 0,00081 (0,00819) Β TA 0,79138* (0,07238) 0,93276* (0,09075) 0,56910* (0,12464) N R 2 0, , ,26440 Mid- Cap Net debt Net debt Net debt Period Intercept, α - 0,01 (0.08) - 0,013 (0,012) - 0,001 (0,011) Β TA 0,778** (0,09) 0,912 (0,12) 0,547* (0,143) N R 2 0,46 0,64 0,26 Large- Cap Net debt Net debt Net debt Period Intercept, α - 0,14 (0,06) - 0,029 (0,009) - 0,003 (0,005) Β TA 0,857*** (0,094) 0,918 (0,114) 0,538*** (0,216) N R 2 0,71 0,82 0,31 43

44 Table VII: Test of the Pecking order. The regression is estimated with panel data covering 15 firms over 10 year is ΔD!" = α + β!" DEF!" + ε!" where ΔD!" is the net debt issued or the net debt issued/net assets. The null hypothesis tested is, β PO = 1. * Indicates significance at the 1% level and ** Indicates significance at the 5% level. Net debt issued Net debt issued Net debt issued Net debt issued/net Assets Net debt issued/net Assets Net debt issued/net Assets Period ** ,12628* 0,12865** 0,11007* (129818) (223400) (158348) (0,0314) (0,04980) (0,03740) Intercept, α βpo 0,66684* (0,0537) 0,61713* (0,06412) 0,94489* (0, ,29296* (0,040) 0,68697* (0,11683) 0,23739* (0,04037) N R 2 0, , , , , ,32142 Table VIII: Test of the Pecking order. The regression is estimated with panel data covering 15 firms over 10 year is ΔD!" = α + β!" DEF!" + ε!" where ΔD!" is the net debt issued. The null hypothesis tested is β PO = 1. * Indicates significance at the 1% level and ** Indicates significance at the 5% level and *** Indicates significance level at the 10% level. Net debt issued Net debt issued Net debt issued Net debt issued Net debt issued Net debt issued Mid- Cap Firm s Large- Cap Firm s Period Intercept, α * (207737) ** (347238) *** (268761) (256445) (561841) (174086) Financing deficit, β PO 0,36128* (0, ,40320* (0,11542) 0,28241*** (0, ,88157* (0,13624) 0,92442* (0,22972) 0,81895* (0,13795) N R 2 0, , , , , ,

45 Table IX: Test of the Pecking order. The regression is estimated with panel data covering 15 firms over 10 year is ΔD!" = α + β!" DEF!" + ε!" where ΔD!" is the net debt issued/net assets. The null hypothesis tested is β PO = 1. * Indicates significance at the 1% level and ** Indicates significance at the 5% level and *** Indicates significance level at the 10% level. Net debt issued/ net assets Net debt issued/ net assets Mid- Cap Firm s Net debt issued/ net assets Net debt issued/ net assets Net debt issued/ net assets Large- Cap Firm s Net debt issued/ net assets Period Intercept, α 0,15892* (0,04216) 0,17741** (0,07069) 0,13136** (0,05074) 0,01455 (0,01751) 0,03305 (0,03785) 0,00922 (0,01102) Financing deficit, β PO 0,35035* (0,05548) 0,63484* (0,13817) 0,28975* (0,05773) 0,95678* (0,13117) 1,03892* (0,22088) 0,78655* (0,11907) N R 2 0, , , , , ,70796 Table X: The conventional leverage regression model. The regressions refer to large- cap firms. Standard errors are reported in parentheses. * Indicates significance at the 1% level and ** Indicates significance at the 5% level and *** Indicates significance level at the 10% level. Period Intercept, α - 0,0926-0,07 0,0517-0,15-0,0986*** (0,0701) (0,2028) (0,05157) Tangibility 5,8137* (1,7175) Growth Size - 1,2451* (0,3460) - 0,4034 (0,3375) Profitability 21,1605* (6,4957) Financial Deficit, DEF 2,1875* (0,5563) (0,08) 5,41 (2,08) - 1,48 (0,41) - 0,19 (0,4) 16,21 (7,72) 5,7259*** (2,8324) - 0,7365 (0,8406) - 0,3606 (0,5564) 20,1962 (11,810) - 2,8521** (1,1731) (0,22) 4,97 (3,39) - 1,56 (0,92) - 0,26 (0,679 16,29 (14,07) 9,6683** (3,7462) - 1,4228* (0,2690) 0,0274 (0,7146) 12,2821 (9,221) - 1,4706** (0,6353) - 0,04 (0,05) 12,98 (3,93) - 1,49 (0,30) 0,90 (0,69) 14,01 (10,44) - N R 2 0,5864 0,37 0,5517 0,29 0,8496 0,79 45

46 Table XI: Debt ratio. Debt ratio 0,65 0,55 0,45 0, Debt Level Large- Cap Debt Level Mid- Cap Debt Level All Firms 46