MASTER OF SCIENCE, 30 CREDITS, SECOND LEVEL STOCKHOLM,
SWEDEN 2020
Determinants of Capital Structure
Testing the Pecking Order Theory on the Swedish Construction Industry
Henrik Ek & Sofia Fjelkestam
ROYAL INSTITUTE OF TECHNOLOGY
DEPARTMENT OF REAL ESTATE AND CONSTRUCTION MANAGEMENT
Master of Science Thesis
Title Determinants of Capital Structure - Testing the Pecking Order Theory on the Swedish Construction Industry Authors Henrik Ek, Sofia Fjelkestam
Department Real Estate and Construction Management Master Thesis number
Supervisor Åke Gunnelin
Keywords Capital Structure, Pecking Order Theory, Construction industry, Leverage Regressions, Sweden
Abstract
Building new homes and offices are vital for the well-being in a country from both an economic sense and from a viewpoint that a growing population needs more housing.
For construction companies to be able to meet their objectives, an important issue is the capital structure choice. The discussion of capital structure and company value became one of the most contentious areas in finance in the decades following the publication of the famous paper by Modigliani and Miller. Since then, there has been a lot of work exploring the optimal capital structure for firms in different situations, however the work has been rather limited in Sweden. This study tested the pecking order theory of capital structure on publicly traded Swedish construction firms between 1995 to 2019.
The study had a focus on how well the pecking order theory can account for financial decisions made by Swedish construction firms through a series of tests. The purpose is to understand what factors are the most important, with regards to the capital structure, for listed firms in the Swedish construction industry. The result of the tests showed that the pecking order theory failed to outperform the conventional leverage model. Therefore, the study was unsuccessful in finding support for the hypothesis that the pecking order theory would be the dominant financial theory to explain capital structure decisions.
TRITA-ABE-MBT-20470
Acknowledgement
This study has been our master thesis on the master’s programme Real Estate and Construction Management, finishing our education in Civil Engineering and Urban management at Royal Institute of Technology in Stockholm. We would like to express our sincere gratitude to our supervisor Åke Gunnelin, who came with constructive criticism and guidance during the course of the work. We also want to give thanks to those who participated in interviews, Stefan Björklund from the construction firm Einar Mattsson and an anonymous respondent. Finally, we would also like to thank our seminar group that contributed with valuable comments.
Stockholm, June 2020
Henrik Ek and Sofia Fjelkestam
Masterexamensarbete
Titel Determinanter av kapitalstruktur - Testande av pecking order teorin på den svenska byggbranschen Författare Henrik Ek, Sofia Fjelkestam
Institution Fastigheter och byggande Examensarbete Master nivå
Handledare Åke Gunnelin
Nyckelord Kapitalstruktur, Pecking order teorin,
Byggbranschen, Finansiell hävstångsregression, Sverige
Sammanfattning
Att bygga nya bostäder och kontor är avgörande för välfärden i ett land ur både ekonomisk mening och ur en synvinkel att en växande befolkning behöver fler bostäder.
För att byggföretag ska kunna nå sina mål är valet av kapitalstruktur en viktig fråga.
Diskussionen om kapitalstruktur och företagsvärde blev ett av de mest debatterade områdena inom finansiell ekonomi under årtiondena efter den berömda publiceringen av Modigliani och Miller. Sedan dess har det gjorts mycket arbete med att utforska den optimala kapitalstrukturen för företag i olika situationer, däremot är forskningen något begränsat i Sverige. Denna studie testade pecking order teorin om kapitalstruktur hos börsnoterade svenska byggföretag mellan 1995 och 2019. Studien hade ett fokus på hur väl pecking order teorin kan förklara ekonomiska beslut som fattats av svenska byggföretag genom en serie av tester. Syftet är att förstå vilka faktorer som är de viktigaste med avseende på kapitalstrukturen för börsnoterade företag i den svenska byggbranschen. Resultatet av testerna visade att pecking order teorin inte överträffade den konventionella modellen av finansiell hävstång. Därmed hittades inte stöd för hypotesen att pecking order teorin skulle vara den dominerande finansiella teorin för att förklara beslut av kapitalstruktur.
TRITA-ABE-MBT-20470
Förord
Det här arbetet har varit vårt masterexamensarbete på masterprogrammet Fastigheter och byggande som avslutar vår civilingenjörsutbildning i Samhällsbyggnad på Kungliga Tekniska högskolan. Vi vill rikta ett stort tack till vår handledare Åke Gunnelin som kommit med konstruktiv kritik och vägledning under arbetets gång. Vi vill även tacka de som ställt upp på intervju, Stefan Björklund från byggföretaget Einar Mattsson och en anonym respondent. Slutligen vill vi även tacka vår seminariegrupp som bidragit med synpunkter.
Stockholm, juni 2020
Henrik Ek och Sofia Fjelkestam
Contents
1 Introduction 1
1.1 Purpose and Contribution . . . . 2
1.2 Research question . . . . 3
1.3 Hypothesis . . . . 3
1.4 Limitations . . . . 3
2 Literature review 4 3 Theoretical Framework 9 3.1 Modigliani-Miller theorem . . . . 9
3.2 Pecking order theory . . . . 9
3.3 Trade-off theory . . . . 11
3.4 Summary of expected coefficients . . . . 12
4 Methodology 13 4.1 Research strategy . . . . 13
4.2 Quantitative analysis . . . . 13
4.2.1 Quantitative method . . . . 13
4.2.1.1 Pecking order model . . . . 13
4.2.1.2 Conventional leverage regression model . . . . 14
4.2.2 Quantitative data description . . . . 16
4.2.2.1 Description of the firms . . . . 17
4.2.3 Descriptive statistics . . . . 18
4.3 Qualitative analysis . . . . 20
4.3.1 Qualitative method . . . . 20
4.3.2 Qualitative data description . . . . 21
4.4 Validity and reliability . . . . 21
4.5 Ethical considerations . . . . 22
5 Results 23 5.1 Results from quantitative analysis . . . . 23
5.1.1 Pecking order model . . . . 23
5.1.2 Conventional leverage regression model . . . . 25
5.2 Results from qualitative analysis . . . . 28
6 Discussion 29 7 Conclusion 33 7.1 Conclusions from research . . . . 33
7.2 Suggestions for further research . . . . 34
References 35
Interview subjects . . . . 38
Appendix 39
1 Introduction
Construction companies are vital for the well-being in a country from both an eco- nomic sense and from a viewpoint that a growing population needs more housing. For construction companies to be able to meet their objectives, the capital structure choice is an important issue. The optimal capital structure is the combination of debt and equity that minimises the average cost of capital and thus make construction compa- nies more profitable. A firm’s capital structure is one of its most important choices.
It impacts the firm’s overall risk perception, the ability to access funding and its cost, the expected return that investors and lenders have and how the firm is affected by microeconomic business decisions and macroeconomic downturns. There are several theories about the choice of capital structure and there has been debate on the area since Modigliani and Miller (1958) presented their groundbreaking theories on capi- tal structure and firm value in which they proclaimed that the value of a firm is not impacted by the chosen capital structure, under conditions of a perfect market. This study focuses on examining the pecking order theory of capital structure. A statistical analysis will be conducted on how well the theory can account for financial decisions made by enterprises in the Swedish construction industry. According to the pecking or- der theory, firms will use their internal financing as long as possible and thereafter turn to external financing. For external financing debt is preferred ahead of equity.
A company must make several decisions when choosing a specific capital structure which they deem optimal dependent on company preference with regards to circum- stances that are prevalent at that time. There are multiple variables that impacts the company’s decision. Previous studies have shown that profitability, firm size, growth opportunities and tangibility have an effect on leverage ratios (e.g. Rajan and Zingales, 1995; Frank and Goyal, 2003). It is common to simply state how leveraged a company is when describing its capital structure in a quick manner. Leverage is used in this way to describe the risk exposure a company has and it can easily be utilised when comparing different companies’ risk levels.
Choosing an optimal capital structure is not an easy task for a company as there are many different variations of equity and debt, and the accompanying costs of capital and risk is different for each combination. Since there is such a wide range of options avail- able it is important to consider the alternatives. This becomes even more indisputable when external forces impact the currently used structure, as could have been the case for a Swedish construction company when the banks altered the rules of the game.
The banks require that firms have sold a significant part of the planned apartments.
The ability to sell homes in the planning stage naturally dropped as the restrictions from the banking sector was implemented. This fundamental shift for the industry
can impact how the firms operate. They may need to find funding alternatives outside of the traditional banking world. It is interesting to get a better understanding on how the actors in the construction industry consider their future in terms of financing projects.
The problem of capital structure is academically interesting because of the perpetual debate regarding which theories are most applicable to real world settings. From a so- cietal perspective there is value in having a fundamental understanding on how actors in an important industry make their decisions. Failing to understand this may lead to unsuccessful interventions from the government which comes at a great cost. There- fore, this study will test the pecking order theory on Swedish construction firms, and analyse how well it can account for financial decisions, compared to more conventional methods.
1.1 Purpose and Contribution
The study tests the pecking order theory on capital structure for Swedish listed con- struction firms and is validated with an interview with representatives from two Swedish construction firms. The purpose is to understand what factors are the most impor- tant, with regards to the capital structure, for listed firms in the Swedish construction industry. Furthermore, this study will analyse how the capital structure can change in the future due to unconventional debt instruments.
This is a research question that is relevant to explore, and an area in which there is a lack of research. Our work will provide knowledge on this question which will be of value for researchers that can use the results in further studies. There are sev- eral previous studies that examine the determinants of a company’s capital structure, however, most studies have been conducted in the United states. In addition, many researchers have studied the capital structure choice across different industries. This study covers only construction firms to solely account for financing decisions made by Swedish construction firms.
1.2 Research question
1. How applicable is the pecking order theory on capital structure decisions made by publicly traded firms in the Swedish construction industry?
2. What impact will profitability, firm size, growth opportunities and tangibility have on the choice of capital structure for the firms?
3. Will the restrictions on loans from the banking sector have an impact on future capital structure choices for the firms?
1.3 Hypothesis
The pecking order theory will be applicable on Swedish construction firms, and will do a better job, compared to alternative theories, at accounting for financial decisions made by the companies.
The impact from profitability, firm size, growth opportunities and tangibility on the firms will not match traditional findings, and will follow the pecking order theory instead. This means a negative correlation with debt levels for profitability, growth opportunities and tangibility, and a positive correlation for firm size.
Restrictions on loans for home buyers will make it more difficult for construction firms to have high debt ratios.
1.4 Limitations
The data for the quantitative analysis is limited to Swedish listed construction firms due to desired geographical delimitation and data availability. The construction firms chosen for both the quantitative and qualitative analysis are construction firms with a main focus on buildings. No account has been taken of how the firms’ debts are distributed among different types of loans when doing the statistical analysis. The as- sumption has been made that managers strive to maximise firm value. The qualitative analysis covers fewer firms than originally hoped due to complications as a result of COVID-19. This has a natural effect on the quality of the qualitative analysis sec- tion. The answers are more susceptible to bias from respondents which would have been limited with input from more respondents. This impacts the ability to draw con- clusions from the empirical findings, as the conclusions can suffer in reliability. As a result of the limitations on the qualitative analysis, a higher focus has been put on the quantitative analysis instead.
2 Literature review
The discussion of capital structure and company value became one of the most con- tentious areas in finance in the decades following the publication of the famous paper by Modigliani and Miller in which they proclaimed that the value of a firm is not impacted by the chosen capital structure, under perfect market conditions (Modigliani and Miller, 1958). Since then, there has been a lot of work exploring the optimal capital structure for firms in different situations. However, the literature is lacking in research on the Swedish construction industry. Today, researchers agree that cap- ital structure has an impact on the value of any given firm. Several researchers (e.g.
Shyam-Sunder, 1991; Masulis, 1983) have shown that stock prices change when firms announce a different leverage ratio.
How to best explain capital structure decisions by firms is a topic that has been aca- demically relevant and challenging for decades. One general and popular method to account for debt levels is through regression analysis on the level of debt, dependent on four variables; size, asset tangibility, growth opportunities and profitability. Examin- ing debt levels dependent on these four independent variables is commonly referred to as the conventional method. The conventional variables have been used in academical studies for decades, and it is therefore necessary to include this model in a study that challenges its role as the main model for explaining debt levels. Important work with the conventional model has been done by e.g. Rajan and Zingales (1995), where they examined the determinants of capital structure choice. The study included firms from a broad set of countries, the United States, Japan, Germany, France, Italy, Canada and the United Kingdom. The study finds that the coefficients size and tangibility are positively correlated with debt, whilst growth opportunities and profitability have a negative relationship with debt.
The reason for why firms tend to become more leveraged when the size of the firm increases can possibly be explained by diversification. Larger firms tend to be more diversified which leads to less risk of financial distress. This enables the company to become more leveraged. (Rajan and Zingales, 1995).
Tangible assets can work as collateral, this means having a high fraction of tangible assets makes it less risky for the lender. Therefore, a firm with higher tangibility should have the opportunity to take on more debt than a firm with a smaller fraction of fixed assets to total assets. Since tangibility creates the opportunity to take on more debt, there should be a positive correlation between tangibility and leverage according to the conventional leverage model (Rajan and Zingales, 1995).
Highly leveraged firms are less likely to grasp an investment opportunity, compared to a less leveraged firm. This can be because of the higher connection to the risk of the
investment failing. A firm that predicts upcoming investment opportunities or growth opportunities will therefore be less inclined to take on debt (Myers, 1977).
Profitability is negatively correlated to leverage, as explained by Myers and Majluf (1984) since firms with bigger cash flows have the opportunity to utilise internal funds before taking on more debt.
These predictions and studies have been consistently tested for decades and still stand relevant, which speaks highly of the work made by the researchers in question and the solidity of the conventional leverage model.
One of the most prominent theories on capital structure choice and a challenger to the conventional regression model is the pecking order theory. Myers (1984) states that due to adverse selection, firms prefer internal financing to external financing. When external financing is used debt is preferable over equity. According to the pecking order theory firms do not have a target debt ratio. The adverse selection problem with equity is rooted in the idea that the equity investors knows that companies issue new shares when they believe their stock is overvalued to maximise their profitability, which is shown by Hovakiam, Opler and Titman (2001). Further evidence on market timing being an important factor for a company when making financial policy decisions is found by Baker and Wurgler (2002).
Shyam-Sunder and Myers (1999) find support for the pecking order theory in an em- pirical study covering a sample of 157 publicly traded industrial firms. The financing deficit for these firms are matched with issuance of debt. Similar conclusion is reached by Fama and French (2002) who found evidence supporting the pecking order theory, more profitable firms have less debt since they use retained earnings at first hand.
Frank and Goyal (2003) test the pecking order theory on a sample of publicly traded American firms between 1971 to 1998 and found that net equity issuance track the financing deficit more closely than net debt issuance, which contradicts the pecking order theory. This becomes even clearer when running the conventional leverage re- gression and including the financing deficit to conventional leverage factors. Results from the regression in the study is that leverage has a negative relationship with growth opportunities and profitability and is positively correlated to tangibility and firm size.
Inclusion of the financing deficit should, according to the pecking order theory, cancel out the impact from conventional variables. This did not happen in their study. The financing deficit, whilst empirically relevant, is not crucial in explaining debt issuance.
The pecking order theory was not able to sufficiently outmatch the conventional lever- age model for these firms (Frank and Goyal, 2003).
Another relevant theory is the trade-off theory which argues that a company’s ideal capital structure consists of a trade-off between leverage-related costs and tax benefits
of debt (Bradley, Jarrell and Kim, 1984). Issuing debt has a tax benefit since firms can deduct interest on taxable earnings and in that way pay less tax. Graham (2000) found that the capitalised tax benefit of debt is 9.7% of firm value. And firms can double the tax benefits by using more debt until the marginal tax benefit starts to decline.
Unsurprisingly, more debt tends to be used in situations where there is a larger tax gain by being leveraged (Fan, Twite and Titman, 2011). According to a survey study by Graham and Harvey (2001) covering 392 CFOs from different industries, firms often have a target debt ratio or have a target in mind when issuing new debt or equity.
Research has indicated that firms are most likely to adjust their capital structure in two situations. The first scenario is debt levels are above target level and a financial surplus. The alternative is at below target debt combined with financial deficit. Firms act differently at above and below target level debt. When firms find themselves at above target they will use their financial surplus to pay off debt, but in a situation where debt levels are below target a firm will pay off both equity and debt (Byoun, 2008).
What variables should be used when examining capital structure has been an ongoing debate in the academic field, and different researchers include various variables when empirically explaining the relationship between the variables and the choice of capital structure. Latter studies have further developed the conventional leverage regression model by including more independent variables to test if they are affecting the leverage, in an attempt to improve the model. Variables added are for example risk, cost of debt, ownership structure, industry type, country, labor protection laws, covenant protection, uniqueness (Billett, King and Mauer, 2007; Kayo and Kamura, 2011; Morri and Cristanziani, 2009; Serfling, 2016; Titman and Wessels, 1988).
Managers of riskier companies tend to choose a lower leverage to lower the company’s degree of risk, while cost of debt and ownership structure are not found to be impor- tant factors to explain leverage (Morri and Cristanziani, 2009). The type of industry was not found to play a significant role when explaining the capital structure in Kayo and Kamura’s study (2011). The lack of importance was also found to be applicable to what country a company is located in, which has an even lower importance. Serfling (2016) studied how the costs associated with discharging workers impact the capital structure decisions. The study is based on the assumption that firms with labor pro- tection laws have increased firing costs. By comparing firms with and without labor protection laws, Serfling found evidence that an increase in firing costs is followed by reduced debt ratios. The leverage is also affected by covenant protection. The nega- tive relationship between leverage and growth opportunities is significantly reduced by covenant protection (Billet, King and Mauer, 2007). Titman and Wessels (1988) find debt levels to be negatively related to the uniqueness of a firm’s business.
When analysing companies’ decisions over their capital structure some researchers assume that all corporations follow the same optimal strategy, as was done when ex- ploring if companies have a target capital structure (Flannery and Rangan, 2006). In contrast to this there is research on heterogeneity regarding the strategy and decision making (Arce, Cook and Kieschnik, 2015). Research that has found heterogeneity has been conducted through both surveys, but also empirical studies (Dissanaike, Lam- brecht and Saragga, 2001). This highlights the importance of not only focusing on one single company and making conclusions on other companies as well. There is evidence that short-sighted actors tend to be more conservative regarding debt. (Liu et al., 2017)
There is evidence from theoretic research that capital structure is expected to be counter-cyclical. This means that the target leverage level becomes lower in economic booms and the target rises in recessions (Korajczyk and Levy, 2003). This has been supported in further studies. However, the counter-cyclical variance in leverage is unique for less constrained firms (Hennessy and Levy, 2007). This becomes more relevant when combined with the fact that the construction industry is very susceptible to economic fluctuations during business cycles (Berman and Pfleeger, 1997).
Industries face different challenges, the construction industry is a large market and construction firms often have high debt levels. Construction firms often have periodic earnings (Kangari, 1988). Therefore, construction firms tend to have more capital as a buffer against economic fluctuations.
A study on capital structure determinants for construction companies in South Korea showed that firm size was positively correlated with leverage. A negative relationship with leverage was identified for growth, tangibility and profitability. The researchers found the construction firms to be moderately accommodating of the pecking order theory (Choi et al., 2014).
Feidakis and Rovolis (2007) conducted a study on construction firms in the European Union and found that country specific factors were relevant, but the results were still comparable over the European Union. The study identified profitability to be nega- tively correlated with leverage, and a positive relationship for firm size and tangibility.
Chiang and Cheng (2009) further emphasise the importance of tangibility to enable loan opportunities for construction firms in Hong Kong.
A publishing from 2017 disclosed a research investigation on the capital structure of construction companies in Russia from a viewpoint of the risks and opportunities for the firms. The researchers attempted to identify factors which affect the decision making, one example presented was that larger companies tend to use larger amounts of borrowed capital. There was a rather high diversity in terms of how leveraged
different firms chose to be when conducting their business (Guzikova, Plotnikova and Zubareva, 2017).
According to Achleitner, Lutz and Schraml (2009), when lending becomes more restric- tive, like in times of a recession, the demand for other financing options like mezzanine financing increases. They found that firms tend to use equity when no other financing options are available. In addition, problems caused by financial distress increases the firms’ willingness to incorporate mezzanine debt. The findings are in line with the pecking order theory, firms only turn to equity or mezzanine financing if there is no other funding option. Ryan, Ross, and Yen (2007) states that mezzanine debt can reduce the average costs of capital by using mezzanine instead of the more expensive equity finance.
Harris and Artur Raviv (1991) reviewed scientific work that had been done on the topic of capital structure. They further emphasise the importance of cohesion between the long-term business strategy a firm has and the optimal capital structure. Decisions made by a company regarding their capital structure has significant impact long-term.
It is crucial that the chosen strategy is consistent with the general long-term strategy from the firm (Barton and Gordon, 1987).
3 Theoretical Framework
3.1 Modigliani-Miller theorem
In their groundbreaking work, Modigliani and Miller (1958) states that the value of a company is not impacted by the chosen structure of debt or equity or the dividend paid out, see equation 1. They argue that the value of a company consists solely of earnings and the risk accompanied with the business.
VU = VL (1)
VU= value of an unlevered firm VL= value of a levered firm
This notion is based on the assumption of perfect market conditions. A perfect market has no taxes, is free of friction, all parties have perfect information with no asymmetry and all parties can lend and borrow with the same conditions. This perfect world does not exist, so further theories are used to explain the capital structure decision of firms.
3.2 Pecking order theory
In real life conditions, there is asymmetric information between the executives of a company and the equity investors. The general market is aware of the fact that ex- ecutives have more profound knowledge regarding the company’s risk exposure and opportunities. This is the basis for the pecking order theory, which states that there exists a preferred order of financing sources for a business, see figure 1. The important notion is that issuing new debt is preferred over new equity. Investors will be suspicious towards a company that issues new stock, the signals from the executives are that the company is overvalued or it does not possess the financial capabilities of taking on more debt. If a company is correctly valued and want to take on additional funding, choosing equity over debt will affect stock prices negatively since rational investors demands a discount of the current market price because of the information asymmetry (Myers and Majluf, 1984).
The pecking order for financing sources is in order – internal financing, new debt and finally new equity. According to the pecking order theory there is a causal relationship between higher profitability and lower debt. Internal sources of financing are above loans in the pecking order, and will therefore be used before the company borrows new funds. Firms will also increase liquid assets during highly profitable years. These funds can be used during financial downturns to make sure the company keeps using financial sources higher up in the pecking order (Myers and Majluf, 1984).
Figure 1: Pecking order theory diagram
External financing via equity financing will lower the share price since the investors will question if the share is worth its current price, which makes equity financing expensive for a firm. External financing via debt financing requires firms to provide banks with information about their credit risk. Internal financing via firm generated profits does not require any disclosure to investors and creditors.
According to Myers (1977), firms with lower leverage are more likely to take advantage of a growth opportunity. This is due to the financial costs of the potential failure of the investment, which is greater for a firm with high leverage. The relationship between growth opportunities and debt is therefore expected to be negative. The pecking order theory predicts a negative relationship between tangibility and debt levels. Harris and Raviv (1991) argues that firms with fewer tangible assets will have more problems with asymmetric information. This leads to a situation where firms with less tangible assets start accumulating more debt, and therefore higher tangibility will correlate with lower leverage ratios. This is different from the conventional idea that tangible assets work as collateral, and allows for more debt. In the conventional way of thinking tangibility is positively correlated with leverage (Frank and Goyal, 2000). Lastly, according to the pecking order theory, larger firms have a greater ability to accumulate retained profits which means less debt is needed (López-Gracia, Sogorb-Mira 2008). However, Myers (1984) explains that larger firms have less problems of information asymmetry between owners and creditors which allows larger firms to obtain cheaper debt. The pecking order theory can thus predict both positive or negative relationship between debt ratio and firm size.
3.3 Trade-off theory
In contrast to Modigliani-Miller theorem, the trade-off theory of capital structure states that the value of a firm is impacted by the choice of capital structure. The theory starts with the notion that there is an advantage to debt financing due to tax benefits.
However, there are also costs related to financing with debt, such as the costs of financial distress, including bankruptcy costs of debt (Kraus and Litzenberger, 1973).
The firm can lower their costs by increasing the amount of debt in financing up to a certain threshold. Further debt in excess of this level will decrease the value of the firm due to the high risk affiliated with high debt levels and the corresponding agency cost of debt (see figure 2). This is discussed in more detail in the next section. The optimal debt ratio is found at the threshold, where the optimal trade-off between tax benefits and cost of debt is found.
Figure 2: Trade-off theory
According to the trade-off theory, there is positive relationship between profitability and debt since more profitable firms have capacity for higher levels of debt (Fama and French, 2002). The higher the firm’s growth opportunities are, the lower the debt. Myers (1984) states that since agency costs and bankruptcy costs are greater for firms with higher growth opportunities, these firms are more likely to not choose debt as their first financing option. The trade-off theory further states that there is a positive relationship between the firm’s proportion of tangible assets and the level of debt. Tangible assets can be used as a protection against firm bankruptcy and there- fore firms with a higher level of tangible assets have easier access to finance through debt (Michaelas, Chittenden and Poutziouris, 1999). Lastly, larger firms tend to have
a higher level of debt according to the trade-off theory. Titman and Wessels (1988) explains the positive relationship with that larger firms often have greater diversifica- tion of activities hence less likelihood of bankruptcy. Smith and Stulz (1985) add that larger firms tend to have less volatile profits and are therefore more likely increase the debt level to take advantage of the tax benefits of debt.
3.4 Summary of expected coefficients
Table 1 presents a summary of the expected signs on the chosen debt ratio for each variable according to the pecking order theory and trade-off theory.
Table 1: Summary of the relationships between the determinants of capital structure Variable/Theory Pecking order theory Trade-Off Theory
Profitability - +
Growth opportunities - -
Tangibility - +
Firm size +/- +
4 Methodology
4.1 Research strategy
This thesis is a mixed methods research. It includes collecting secondary data for quantitative regressions and a qualitative research part involving interviews. By mixing both quantitative and qualitative research, the study gains in depth and width. The research has a deductive approach since the hypothesis is based on already existing theory and the research strategy is designed to test the hypothesis (Saunders, Lewis and Thornhill, 2016).
4.2 Quantitative analysis
4.2.1 Quantitative method
The quantitative study is conducted through statistical analysis that enables explo- ration of the relationships between variables in the collected secondary data. Secondary data has a time-saving and cost-efficient advantage. The data is already collected and may already have been used in previous research which can make further research eas- ier to implement. However, the secondary data could have been collected for a different purpose than desired and can therefore contain incorrect data. All data sources must be evaluated since the researcher lacks control to affect the data quality (Saunders, Lewis and Thornhill, 2016).
4.2.1.1 Pecking order model
The pecking order theory assumes that there are three sources of funding, retained earnings, debt and equity (Shyam-Sunder and Myers, 1999). When retained earnings have been spent the pecking order theory predicts that any financing deficit will be financed by issuance of debt. Issuance of equity will only be issued as a last resort.
The financing deficit variable is constructed by equation 2.
DEFi,t = DIVi,t+ Ii,t+ ∆Wi,t− Ci,t = ∆Di,t + ∆Ei,t (2) By performing the regression model seen in equation 3, the pecking order theory pre- diction that βP O = 1 can be tested (Frank and Goyal, 2003).
∆Di,t = α + βP ODEFi,t+ i,t (3)
DEFi,t = financial deficit for firm i in year t DIVi,t = cash dividend paid for firm i in year t Ii,t = investment for firm i in year t
∆Wi,t = change in working capital for firm i in year t
Ci,t = cash flow after interest and taxes for firm i in year t
∆Di,t = net debt issued or gross debt issued for firm i in year t
∆Ei,t = net equity issued for firm i in year t
∆ is the first differences
The different components in the financing deficit may have different levels of impact on the debt issued (Shyam-Sunder and Myers, 1999). To test this, the variables will be separated and the regression will also be run in a disaggregated form, see equation 4.
∆Di,t = α + βDIVDIVi,t+ βIIi,t+ βW∆Wi,t − βCCi,t+ i,t (4) The pecking order theory predicts βDIV = βI = βW = βC = 1. If the theory is applicable on Swedish construction firms, a unit increase in any of the components of deficit will have the same unit impact on the debt issued. If however, the significance is not driven by all components, but rather some of the individual ones a range of alternative coefficient patterns are possible.
4.2.1.2 Conventional leverage regression model
Applying the conventional leverage regression model (equation 5, 6 and 7), constructed as Frank and Goyal (2003), on Swedish construction firms will display if there is a significant relationship, and to what extent, between the debt ratio and the variables used in the regression.
The regression is estimated with fixed firm effects i.e. the analysis has no regard to differences in time. It is conducted in first differences meaning the time series is a series of changes from one period to the next period. The basic regression (equation 5) is also augmented with first the financing deficit (equation 6). The equation is then augmented with a lagged leverage variable (equation 7). With regards to the pecking order theory, the most interesting regression is when the financial deficit is included in equation 6. If the pecking order theory is applicable, then the inclusion of the financing deficit will cancel out the impact of the conventional variables (Frank and Goyal, 2003).
∆Di,t = α + βT∆Ti,t+ βM T B∆M T Bi,t+ βLS∆LSi,t+ βP∆Pi,t+ i,t (5)
∆Di,t = α + βT∆Ti,t+ βM T B∆M T Bi,t+ βLS∆LSi,t+ βP∆Pi,t+ βDEFDEFi,t+ i,t (6)
∆Di,t = α+βT∆Ti,t+βM T B∆M T Bi,t+βLS∆LSi,t+βP∆Pi,t+βDEFDEFi,t+βDt−1Di,t−1+i,t
(7) Di,t = ratio of total debt to market capitalisation, the firm’s leverage for firm
i in year t
Ti,t = tangibility defined as the ratio of fixed assets to total assets for firm i in year t
MTBi,t = market-to-book ratio defined as the change in ratio of market capitalisation to the total book value, proxy for the firm’s growth opportunities for firm i in year t
LSi,t = natural logarithm of constant sales, proxy for the size of the firm for firm i in year t
Pi,t = profitability defined as the ratio of operating income (EBIT) to total book value, profitability of the firm for firm i in year t
DEFi,t = financial deficit and defined as the sum of dividends, investment, change in working capital minus the cash flow after interest and taxes for firm i in year t
Di,t−1= lagged leverage defined as the leverage at the previous time period
∆ is the first differences
4.2.2 Quantitative data description
The quantitative analysis will be conducted by collecting historical data from the listed construction companies in Sweden using Thomson Reuters Eikon since historical data can be used to forecast future similar events (Simonton, 2003). The collected data is both quarterly and annual data in SEK millions.
The dataset is unbalanced panel data covering the eight construction companies seen in figure 3.
Figure 3: Summary of firms and years in dataset
The sample period is 1995-2019. However, data is not available for all selected firms during the whole sample period, hence an unbalanced dataset. The complete dataset consists of cross-sectional and time variant observations over a 25-year period. The reason for choosing this time period is due to availability of public data for these construction firms.
The collected data contains leverage and the factors that have shown to affect the leverage; tangibility, firm size, profitability and growth. The study is limited to these independent variables since previous studies have shown evidence that they are strongly correlated to firm’s capital choice (Rajan and Zingales, 1995; Frank and Goyal, 2003).
Another reason for choosing these variables is lack of data to develop proxies for other factors.
4.2.2.1 Description of the firms Besqab
Besqab is a housing developer who has chosen to concentrate on the Swedish market, on the local markets in Stockholm and Uppsala. They have a turnover of approximately SEK 1.8 billions and 120 employees. They are listed on NASDAQ Stockholm since 2014.
Bonava
Bonava is a leading housing developer in Northern Europe. Bonava has 2,100 employees with operations in Germany, Sweden, Finland, Denmark, Norway, Russia, Estonia and Latvia and a turnover of SEK 14 billion in 2018.
JM
JM is one of the Nordic region’s leading project developers of housing and residential areas with about 2500 employees. The business is focused on new production of housing in attractive locations in Sweden, Norway and Finland. They have a turnover of around SEK 16 billion and is public company listed on NASDAQ Stockholm since 1982.
NCC
NCC is a Nordic building and property development company that develops commer- cial real estate and builds housing, offices, industrial buildings and roads and other infrastructure. NCC operates in Sweden, Norway, Denmark and Finland and have a turnover of nearly SEK 58 billion and 15,500 employees.
Peab
Peab is one of the Nordic region’s leading construction and civil engineering companies with operations in Sweden, Norway and Finland. Construction is their largest busi- ness area. They have a turnover of approximately SEK 54 billion and about 14 000 employees.
Serneke Group
Serneke is operating in construction, civil engineering, project development and prop- erty management in Sweden. Serneke had a turnover of 6.7 billions in 2019 and 1100 employees.
Skanska
Skanska is one of the world’s leading construction and project development companies with operations in Sweden, Norway, Finland, Denmark, Poland, the Czech Republic, Slovakia, Hungary, Romania, the United Kingdom and the United states. Their biggest business stream is construction with revenue of approximately 85% of the total revenue.
They have been listed on NASDAQ Stockholm since 1965 and have a turnover of nearly 173 billions and 33 000 employees.
Veidekke
Veidekke is one of Scandinavia’s largest construction, civil engineering and housing development companies. They operate in Norway, Sweden and Denmark and have a turnover of 38 billions. Veidekke is listed on the Oslo Stock Exchange since 1986 but since Veidekke has a large part of its operations in Sweden, Veidekke is included in the study. The data collected is in NOK but is converted to SEK.
4.2.3 Descriptive statistics
The descriptive statistics are produced by using both quarterly and annual data. The reason for this is no quarterly data on the variables constructing the financing deficit was found.
Table 2 shows the average value for the dependent and independent variables for the pecking order model. The statistics are presented for each firm separately. Complete descriptive statistics can be found in the Appendix.
Table 2: Average funds flow and financing as a fraction of total assets, for each firm.
Annual data.
Besqab N=10
Bonava N=7
JM N=22
NCC N=25
PEAB N=22
Serneke N=7
Skanska N=23
Veidekke N=23 DIV 0.032 0.014 0.0314 0.0243 0.0189 0.0047 0.0283 0.0305 I -0.019 0.0073 0.0082 0.0121 0.0246 0.0528 0.0075 0.0522
∆W 0.049 0.0286 -0.0136 0.0058 0.0118 -0.0070 -0.0705 0.0079 C 0.078 0.0468 0.0665 0.0618 0.0685 0.0053 -0.0441 0.0834 DEF -0.017 0.0034 -0.0405 -0.0196 -0.0132 0.0452 0.0093 0.0073
∆D 0.012 0.0014 -0.0052 -0.0115 -0.0110 0.0245 -0.0092 0.0057
∆E 0.042 -0.0004 -0.0341 0.0000 -0.0018 0.0376 -0.0033 0.0003
∆D+∆E 0.054 0.0010 -0.0392 -0.0115 -0.0128 0.0621 -0.0125 0.0060
Table 3 illustrates summary of the average value for the dependent and independent variables for the pecking order model. The statistics are presented for year 2000, 2005, 2010, 2015 and 2019 separately. Complete descriptive statistics can be found in the Appendix.
Table 3: Average funds flow and financing as a fraction of total assets, for the selected years. Annual data.
2000 N= 5
2005 N=5
2010 N=6
2015 N=8
2019 N=8 DIV 0.0178 0.0407 0.0297 0.0245 0.0221 I 0.0355 0.0114 0.0380 0.0151 0.0083
∆W 0.0266 -0.0736 0.0142 0.0014 0.0401 C 0.0253 0.0745 0.0658 0.0502 0.0309 DEF 0.0545 -0.0960 0.0161 -0.0092 0.0396
∆D 0.0515 -0.0442 0.0157 -0.0197 0.0280
∆E -0.0027 -0.0141 -0.0016 -0.0031 -0.0001
∆D+∆E 0.0489 -0.0584 0.0141 -0.0228 0.0279
Table 4 illustrates summary of the average value for the dependent and independent variables for the conventional regression model. The statistics are presented for each firm separately. Complete descriptive statistics can be found in the Appendix.
Table 4: Average value for the conventional variable and deficit for each firm using quarterly data.
Besqab N=18
Bonava N=11
JM N=71
NCC N=72
PEAB N=76
Serneke N=9
Skanska N=64
Veidekke N=57
∆D -0.0018 0.0736 -0.0227 -0.197 -0.0613 0.4537 -0.262 -0.0500
∆T 0.0020 0.0006 0.0001 -0.0064 0.0026 0.0056 -0.0001 -0.0066
∆MTB -0.171 -0.239 0.0262 0.295 0.0502 -0.392 0.0650 0.108
∆LS 0.00518 0.0325 0.0307 0.0112 0.0654 0.151 0.0162 0.0852
∆P -0.0058 -0.0192 -0.0027 0.0102 -0.0021 0.0043 -0.0015 0.0013 Lagged
Leverage 0.1224 0.550 0.353 1.20 0.591 0.709 0.159 0.387
DEF -0.017
(N=10)
0.0034 (N=7)
-0.405 (N=22)
-0.0196 (N=25)
-0.0132 (N=22)
0.0452 (N=7)
0.0093 (N=23)
0.0073 (N=23)
Table 5 illustrates summary of the average value for the dependent and independent variables for the conventional regression model. The statistics are presented for year 2001, 2005, 2010, 2015 and 2019 separately. Complete descriptive statistics can be found in the Appendix.
Table 5: Average value for the conventional variable and deficit for the selected years using quarterly data
2001 N=8
2005 N=20
2010 N=20
2015 N=26
2019 N=29
∆D -0.117 -0.523 -0.376 -0.0758 0.180
∆T -0.0206 -0.0050 -0.0028 -0.0019 0.0163
∆MTB -0.0409 0.776 0.603 0.459 0.264
∆LS 0.294 0.120 -0.0596 0.0602 0.0227
∆P -0.0157 0.0226 -0.0087 -0.0088 0.0266
Lagged leverage 1.35 0.374 0.319 0.237 0.448
DEF -0.0154
(N=5)
-0.0960 (N=5)
0.0161 (N=6)
-0.0092 (N=8)
0.0396 (N=8)
4.3 Qualitative analysis
4.3.1 Qualitative method
Interviews allow researchers to access the knowledge that the respondents have and can therefore be used as a way of gathering data (Busher and James, 2014). The interviews in this research are conducted in non-real time, asynchronously, through e-mail due to restrictions on meetings as a result of COVID-19. Interview questions will be rather open which allows the respondent to expand on their answers in ways that had not been thought of beforehand. All respondents will have identical questions which is necessary to provide a foundation to discuss the answers. There are specific topics that are needed to be examined, but concurrently the respondents are allowed to add new knowledge and initiate further questions that have yet to be considered (Rabionet, 2011).
E-mail interviews allows the respondents to answer the questions at a time suitable to their schedule and gives the respondents more time to reflect on the responses which enriches the result. Furthermore, e-mail interviews are simpler to administer. The interview questions was sent to the participants and the responses were sent back in e-mail which prevents the researcher from having to transcribe (Busher and James, 2014). The disadvantage of e-mail interviews is that it can take several days or even weeks to get an answer. The worst case is not get an answer at all, since it is rather easy to ignore e-mails if they are to busy or if they have no interest in the research (Kivits, 2005). E-mail interviews can, if they are conducted long term, provide the researcher with in-depth data thanks to repeated interactions and closer reflection. However, the asynchronous approach can lead to the respondent giving a more socially expected response, rather than one that is spontaneous (Busher and James, 2014).
4.3.2 Qualitative data description
The respondents for the qualitative analysis consist of :
• Group Head of Treasury from a listed Swedish housing construction company who has chosen to be anonymous (further called Group Head of Treasury in this thesis)
• Stefan Björklund, Head Analyst from the construction company Einar Mattsson.
Interview questions can be found in Appendix.
The qualitative analysis covers fewer respondents than originally hoped due to com- plications as a result of COVID-19. This has a natural effect on the quality of the qualitative analysis section. The answers are more susceptible to bias from respon- dents which would have been limited with input from more respondents. This impacts the ability to draw conclusions from the empirical findings, as the conclusions can suffer in reliability.
4.4 Validity and reliability
When determining the quality of the research, validity and reliability are central parts when evaluating science and quantitative research. Validity regards the accuracy of the analysis of the results while reliability regards the possibility of repeating research and achieving equal results (Saunders, Lewis and Thornhill, 2016).
Reliability can be divided into internal and external reliability. Internal reliability can be achieved by having more than one researcher participating in a study to ensure that the researchers have the same opinions about the finding of the research project.
Research with high reliability is not dependent on who conducts the study (Saunders, Lewis and Thornhill, 2016).
Validity can be divided into three different parts, measurement validity, internal valid- ity and external validity. Measurement validity involves assessing the intention. Inter- nal validity is established when a causal relationship between two variables has been sufficiently demonstrated. External validity refers to the possibility of the findings to be relevant to other settings or occasions (Saunders, Lewis and Thornhill, 2016).
Validation involves verifying the data, analysis and the interpretation of these to de- termine how credible the research is. Reliability tests if the study’s results are due to the conclusions reached by the research or if the results are due to other extraneous variables. High validity and reliability are thus very important for a research since it provides information about the quality of the research. A research which has neither high validity or reliability is of low quality. The findings of the research are much more
likely to be false than other researches with high validity and reliability. The research is therefore not as likely to be considered as a knowledge contributor (Saunders, Lewis and Thornhill, 2016).
To ensure high reliability the research process is comprehensively thought through and evaluated. The quantitative method and data used have been used by several researchers in previous studies (eg. Frank and Goyal, 2003; Shyam-Sunder and My- ers, 1999; Rajan and Zingales, 1995) to further ensure high reliability. Each part of the study is described transparently to allow other researchers to replicate the study.
Regarding validity, the data has been verified by using more than one method of data collection to confirm the credibility. Furthermore, the research data was sent back to the respondents to allow them to comment and correct it to validate it.
4.5 Ethical considerations
A subject of a study gives consent to the researcher when they approve of being a part of the research. When asking for permission to gain access to information, or for par- ticipation in an interview it is important to be honest and not to put on pressure.
There are three different forms of consent – a lack of consent, inferred consent and informed consent. When receiving a negative answer, a researcher must be accepting of the response gained and be respectful towards the subject despite the potential disappointment. If the subject gives consent but is not fully understanding of the context or their rights regarding the study it is called inferred consent. If the researcher is not completely open and honest with what the information will be used for, or if it is stored and analysed without clarifying with the subject it is also a form of inferred consent. The researcher must also be honest with the purpose of the study (Saunders, Lewis and Thornhill, 2016).
To ensure that respondents could give proper consent they were informed of the nature of the study, how the information would be used, stored and analysed. They were made aware that they would be able to get in touch to take back their given consent, or if there were any other questions.
