Bachelor’s project within the Business and Economics Bachelor Programme
Examining Determinants Shaping Capital Structure: A
Study of Swedish Industrial Firms
Empirical evidence in Sweden between 2017-2023
Bachelor's thesis 15p
Authors: Hampus Fredriksson & Gustav Hansson
Supervisor: Dawei Fang
Spring 2024
Abstract:
This study examines the determinants influencing the capital structure of Swedish firms in the
industrial sector. Further, it investigates the effects of the Covid-19 pandemic on those
determinants. Empirical data from the beginning of 2017 to the end of 2023 are collected
from Capital IQ and used together with an Ordinary Least Square (OLS) regression analysis.
The study analyses key determinants such as tangibility, profitability, growth, non-debt tax
shields, and firm size. Results reveal that tangibility had the most significant impact on
leverage before the pandemic since firms holding more tangible assets resulted in higher
leverage. Further, during Covid-19, the emphasis on enduring liquidity limited tangibility’s
influence on leverage. Profitability showed a negative relationship with leverage throughout
the study, aligned with the pecking order theory. Non-debt tax shields showed a significant
negative influence on the leverage ratio, implying they can substitute the tax benefits of debt
financing. Finally, the thesis concludes that determinants like tangibility, profitability, and
non-debt tax shield are crucial for capital structure, while it is also important to account for
unexpected external economic shocks in the analysis.
Acknowledgement:
We extend our sincerest gratitude to our supervisor Dawei Fang for his invaluable
contributions to the completion of this Bachelor's thesis. His support, insightful guidance,
dedication, and inspiring knowledge about the subject have greatly enriched our research
journey. We would also like to extend a big thank you to Aineas Mallios for his invaluable
help and knowledge on how we should handle our collected data.
2
Table of Contents
1 Introduction 4
1.1 Background 4
1.2 Problem description and analysis 6
1.3 Purpose and hypothesis 6
2 Theoretical Framework and Hypothesis Development 8
2.1 Theory 8
2.1.1 Modigliani-Miller Theorem 8
2.1.2 Trade-off Theory 8
2.1.3 Pecking Order Theory 9
2.1.4 Agency Theory 10
2.2 Hypothesis Development 10
2.2.1 Profitability 10
2.2.2 Tangibility 11
2.2.3 Growth 11
2.2.4 Non-Debt Tax Shield 11
2.2.5 Revenue (Size) 12
3 Previous research 13
4 Methodology 16
4.1 Model Specification 16
4.2 Robustness Test 18
5 Data 20
5.1 Data collection 20
5.2 Cleaning Data 23
5.3 Descriptive Statistics 24
6 Empirical Results 26
6.1 First Regression Model 26
6.2 Second Regression Model 27
6.3 Third Regression Model 28
7 Discussion 28
7.1 Analysing the results 29
7.2 Limitations 32
8 Conclusion 33
References 35
3
1 Introduction
1.1 Background
The concept of "Capital structure" is more relevant recently than it has been for many years
with increased inflation, interest rates, and overall uncertainty in the world. These factors
make companies' decisions in their capital structure immensely important. Capital structure
represents the combination of debt and equity that a business uses to finance and support all
of its operations and expansion. Additionally, the theory of capital structure explains the
origins of funding and assesses the strategic approach that firms should evaluate when
acquiring assets or investing in projects. The decision-making process between utilising debt
or equity financing presents an intimidating challenge, entailing distinct requirements and
obstacles unique to various businesses across diverse industries (Mostafa and Boregowda,
2014).
The main objective of this thesis is to analyse the important determinants that influence
capital structure in publicly traded Swedish industrial companies. Furthermore, the research
also aims to examine the influence of the outbreak of Covid-19 and its effect on the
determinants of the capital structure for these firms. The investigation's data will come from
Capital IQ, a vast database that includes both qualitative and quantitative datasets. Capital
structure will be the dependent variable and a variety of potential determinants will be the
independent variables in an Ordinary Least Squares (OLS) regression analysis to determine
which factors have the greatest impact on the capital structure of firms within the Swedish
industrial sector. In our OLS regression analysis, we will incorporate the years pre-Covid-19
as a binary variable. This inclusion aims to ascertain the extent to which various determinants
are influenced by the Covid-19 outbreak. Such incorporation facilitates a thorough and
comprehensive examination of the factors under consideration.
To ascertain the optimal capital structure for our industry within the framework of our thesis,
it is essential to acknowledge that there exists a variety of theories regarding the perfect
debt-to-equity ratio for companies. The Modigliani and Miller theorem is one of these
theories, which was developed in a sequence of publications throughout the early 1960s, and
is relevant when determining the worth of a business. It states that a firm's capital structure
decisions do not affect the overall value of the firm by two clams. These claims are made in
4
support of this theory, which states that alterations to a company's dividend policy or capital
structure do not affect its market value. According to the theory, equity holders are also
assumed to be apathetic regarding the firm's financial practices. Theoretically, the theorem is
approached under the assumption of perfect market conditions, free from things like neutral
taxes, capital market frictions, unequal access to credit markets, and company financial
practices that obscure information. Miller and Modigliani also postulated that every company
in the market belonged to a "risk class" that had steady profits in all regions of the globe
(Modigliani and Miller, 1958).
The trade-off theory is another theory, which states that companies carefully balance tax
benefits with debt levels to obtain an advantage while reducing the possible expenses
connected with financial instability. Furthermore, when a project's finance requirements
exceed the available internal cash flow, the pecking order theory suggests that businesses
prefer to prioritise debt over equity. Moreover, in the context of necessitating external
financing for investment, companies typically prioritise issuing the most secure financial
instruments initially, commencing with debt securities, succeeding with hybrid instruments
such as convertible bonds, and ultimately resorting to equity issuance. According to
theoretical frameworks, the absence of a prescribed debt-equity ratio stems from the
recognition of two distinct forms of equity, internal and external. Consequently, the debt ratios
of individual firms reflect the aggregation of their external financing needs (Myers, 1984).
Fama and French (2002) explores the dynamics of capital structure in firms by analysing the
trade-off and pecking order models. The study focuses on how firm profitability, investments,
dividend payouts, and debt levels interplay in shaping the capital structure decisions of
companies. It sheds light on how more profitable firms and those with fewer investments tend
to have higher dividend payouts but may be less leveraged, contradicting traditional trade-off
model predictions. The research contributes to understanding how financial decisions within
firms impact their capital structure choices, providing valuable insights for investors and
stakeholders in assessing firm performance and risk management strategies, especially during
uncertain times.
5
1.2 Problem description and analysis
Since there are a lot of theories regarding the optimal capital structure for firms in different
industries, the question of which determinants are influencing the capital structure the most is
relevant to analyse. The selected timeframe for this thesis (2017-2023) also makes it relevant
to analyse the extent of influence Covid-19 had on those determinants of capital structure due
to crucial changes in macroeconomic factors. Therefore this problem is relevant to discuss
and analyse. In order to conclude which determinant variables affect the capital structure the
most, several theories regarding capital structure will have to be taken into consideration,
evaluated, and adapted to publicly listed industrial firms on the Swedish market between the
years 2017-2023.
C onsequently, the questions arises:
1. Which factors are the primary determinants influencing the capital structure of
publicly listed companies within the Swedish industrial sector?
2. How has Covid-19 affected the determinants of capital structure for publicly listed
companies operating within the Swedish industrial sector?
The collection of information and research projects seeks to answer these questions in
accordance with the timeframe from 2017-2023, which in turn helps to clarify the main goal
of the thesis.
1.3 Purpose and hypothesis
The purpose of this thesis is to conduct an analysis of the significant factors influencing the
capital structure of publicly traded Swedish industrial firms. This analysis will primarily focus
on examining the relationship between the determinants, tangibility, profitability, growth,
non-debt tax shield, and firm size and the Debt-Equity ratio, which we will call “leverage” in
our thesis. The study aims to undertake a comparative evaluation of the impact of these
determinants from January 1, 2017, to December 31, 2023, with the intention of finding
relationships among different determinants opposite the leverage ratio during the specified
timeframe. Additionally, the thesis will investigate the influence of the Covid-19 pandemic on
the determinants of the observed companies within the Swedish industrial sector and their
capital structures' correlation with the Debt-Equity ratio.
6
Our research contributes to the existing literature on capital structure, especially regarding
which determinants are the most influential and also includes to what extent the outbreak of
Covid-19 influenced the determinants for firms in the Swedish industrial sector. Specifically,
this thesis supports the prior literature regarding the importance of using tangible assets in
measuring leverage and also an extension of this result that is attributed to the Covid-19
affecting leverage. The results also provide support to the pecking order theory that firms
value internal funds more than external funds through debt. More to this, some aspects like
growth, non-debt tax shields, and size of the firm complicate the determination of capital
structure. In summary, our research offers some implications concerning the future prospects
of economic investment plans for investors and managers where macroeconomic
environments are becoming less predictable.
7
2 Theoretical Framework and Hypothesis Development
In this chapter, the reader will encounter theories connected to the finance sector and more
accurately capital structure. Further, the determinants used in this thesis will be presented in
Hypothesis Development with theories and previous research as the basis for our hypotheses.
The reader will also be inquired about our expected hypotheses and what we should expect
later in the results and analysis.
2.1 Theory
2.1.1 Modigliani-Miller Theorem
The Modigliani-Miller Theorem, which was discovered in 1958 by Merton Miller and Franco
Modigliani, is a fundamental theorem in corporate finance that has an immense amount of
impact on the field of finance. The theory asserts that a firm's overall value is unaffected by
its capital structure. There are two claims in the M&M theorem. The first proposal implies
that there are no taxes, transaction costs, bankruptcy costs, or asymmetry of knowledge in a
fully efficient market. In this case, a company running without leverage is worth the same as a
company operating with leverage. According to the second proposition, a company's cost of
equity and its degree of leverage are intimately associated. As a result, in order to offset the
higher risk involved with leverage, investors want a higher rate of return (Modigliani and
Miller, 1958).
2.1.2 Trade-off Theory
The static trade-off theory by Myers (1984) is based on the Modigliani-Miller Theorem. It
highlights the balance between the tax savings that arise from debt and the bankruptcy and
financial distress costs. These costs of financial distress include legal and administrative costs
of bankruptcy and distress costs, stemming from stakeholders' perceptions that the company
may not be able to sustain its operations. Achieving the optimal leverage level demands
finding a balance between the advantages derived from tax reductions and the costs associated
with debt issuance. However, there are costs associated with adjustments in the capital
structure such as transaction costs and agency costs. Firms experience a delay in adjusting
their capital structure, meaning changes do not occur immediately. Consequently, firms
cannot promptly counteract random events that push them away from their optimal position.
8
The large adjustment costs is one possible explanation to the large variation in companies’
leverage levels (Myers, 1984).
2.1.3 Pecking Order Theory
The pecking order theory is a theory presented by Stewart Myers and Nicolas Majluf in 1984.
The theory states that managers display a pattern of preferences when it comes to sources of
funds for investment opportunities. First, through the company's retained earnings followed
by external debt, and lastly external equity. The pecking order theory suggests that firms
generally prefer internal financing over external financing. However, when external financing
becomes necessary, firms tend to prioritize debt over equity. This preference arises from the
asymmetric information between company managers and the public. Due to this gap, external
investors lack important insights into the company's operations which its managers hold.
Thus, when the company issues stock, it must do so at the market value, despite managers
knowing the stock's true worth, based on insider information, is worth more. Such events can
result in companies missing out on profitable opportunities with positive net present value if
they're required to finance them with equity. Thus, opting for debt issuance over equity is
more beneficial for the company (Myers, 1984).
The picture illustrates the general managerial preferences when it comes to financing according to the pecking
order. Firstly retained earnings, secondly external debt, and lastly issuing new equity.
(Corporate Finance Institute)
9
2.1.4 Agency Theory
The agency theory, also called the principal-agent model, describes a linkage between an
agent and a principal. In such an arrangement, the agent is assigned certain responsibilities to
serve the best interests of his or her principal which constitutes a relationship referred to as an
agency. The theory suggests that the agent and principal’s interests diverge most famously,
within a corporation, there are relationships among managers and shareholders (the agents), or
between a creditor (the principal) and shareholders and managers (the agents). Sometimes
each stakeholder or stakeholder group just does what is in their own best interest, so there’s
not a lot of action on strategy.
When considering capital structure, agency theory offers a lens through which to analyse the
motivations behind firms' financing choices. The alignment of interests between shareholders
and managers, as well as the mechanisms employed to mitigate agency conflicts, profoundly
influences capital structure decisions (Grigore and Stefan-Duicu, 1976).
2.2 Hypothesis Development
2.2.1 Profitability
The pecking order theory proposes that firms prioritize their financing choices in a specific
order. Firms typically prefer internal funds over external ones. Should external financing be
necessary, the preference is to first select debt issuance, then possibly consider hybrid
securities like convertible bonds, and lastly issue new equity. This behaviour derives from the
costs associated with issuing equity, stemming from factors such as asymmetric information
or transaction costs. There is almost an agreement regarding the impact of profitability on
leverage with Rajan and Zingales (1995) and Myers and Majluf (1984) forecasting a negative
relationship in line with the pecking order theory. However, Jensen (1986) argues that there
can be both a negative and a positive relationship. He anticipates a positive relationship under
the assumption of an effective market for corporate control. On the other hand, if the market
is deemed ineffective, Jensen (1986) anticipates a negative relationship which aligns with the
other research. We predict that the relationship will be negative according to Rajan and
Zingales (1995) and Myers and Majluf (1984).
Hypothesis 1: We expect a negative correlation between profitability and leverage.
10
2.2.2 Tangibility
Long and Malitz (1985) presents that a company's leverage is influenced by tangible assets
that are related to the pecking order theory. They contend that material possessions can reduce
the agency cost of debt, which might encourage firms to make riskier investments after taking
on debt. It is anticipated that tangibility and a company's leverage will increase. Harris and
Raviv (1990) argued that leverage should increase in conjunction with the asset's liquidation
value.
Hypothesis 2: We expect tangibility will be positively related to a company’s leverage.
2.2.3 Growth
According to the pecking order theory, there is considerable uncertainty surrounding the
growth factor concerning its impact on leverage and the methodologies for its measurement.
Initially, a positive correlation between growth and leverage is anticipated due to the premise
that increased growth prospects necessitate greater capital infusion, thus fostering a
preference for external financing through debt. Further, the agency theory argues that firms
engaging in investments in assets with the potential for high growth in the future may
encounter challenges in leveraging such assets for borrowing purposes. Consequently, there is
a shift in expectation towards a negative correlation between growth and leverage (Song,
2005). Therefore we anticipate there could be both a negative and positive correlation
between growth and leverage according to both theories mentioned in the paragraph.
Hypothesis 3: We expect that Growth could potentially be both positively or negatively related
to a company’s leverage according to the previous theoretical work.
2.2.4 Non-Debt Tax Shield
Modigliani and Miller (1958) states that interest tax shields provide strong incentives for
companies to raise the amount of leverage. However, the influence of non-debt related
corporate tax shields, such as tax deductions for depreciation and investment tax credits, also
plays a crucial role in determining leverage. DeAngelo and Masulis (1980) suggest that these
non-debt tax shields can act as substitutes for the tax advantages of debt financing, thereby
impacting the tax advantage of leverage. As such, when other tax deductions, like
11
depreciation, increase, the tax advantage of leverage diminishes. Consequently, an
augmentation in non-debt tax shields is anticipated to have a negative effect on leverage.
Hypothesis 4: We expect that Non-debt Tax Shield will be negatively related to a company’s
leverage.
2.2.5 Revenue (Size)
Chakraborty (2010) states that larger companies generally disclose more information
compared to small companies and therefore reduce the risk of problems arising from
asymmetric information and should hence have more equity. When there is less asymmetric
information, companies can issue new equity that is more favourable for the company since
the investors need lower risk premiums. When they can issue more equity the leverage ratio
drops. This aligns with the pecking order that suggests larger firms will have a negative
relation to leverage. On the other hand, according to the trade-off theory, the relation between
the size of a company and the leverage ratio is expected to be positive. Deesomsak et al.
(2004) contended that the higher leverage observed in larger companies can be attributed to
reduced bankruptcy risk and costs. Additionally, they asserted that factors such as diminished
agency costs of debt, lower monitoring expenses, and enhanced access to credit markets
associated with size also positively influence leverage.
Hypothesis 5: The size of a company could be either positively or negatively related to
leverage.
12
3 Previous research
In this chapter, the reader will be introduced to already existing literature and how our thesis
contributes to the addition of previous papers within the capital structure sector. The
presented work in this chapter will serve as a basis for our work, allowing us to make
comparisons with our results and compare if we receive similar results. We also discuss
weaknesses and limitations from the previous literature and provide how our thesis
contributes to their weaknesses. Further analysis will be discussed in a later chapter of the
thesis.
This thesis aims to investigate the key variables influencing the capital structure of Swedish
industrial companies and how these variables were impacted by Covid-19. While there is
extensive literature on the general determinants of capital structure, studies focusing
specifically on Swedish companies, particularly within the industrial sector, are limited.
Similarly, research examining the effects of Covid-19 on these variables is sparse.
The foundation for capital structure theory arises from the Modigliani-Miller Theorem
released in 1958. New research and theories have been developed based on their findings with
the article by Stewart C. Myers released in 1984, being one of the most influential and
contributing papers on the subject of capital structure. It elaborates on the M&M propositions
and introduces new theories known as the pecking order theory and also the static trade-off
hypothesis, which explores the connection between debt levels and a firm's market value, as
well as the interplay between the present value of tax benefits and the present value of
financial distress costs. Myers suggests that in the absence of adjustment costs, each firm
would naturally converge to its optimal debt-to-value ratio. However, due to the presence of
adjustment costs, firms experience delays in achieving this optimal capital structure since the
changes tend to lag. These delays result in considerable variation in capital structures across
firms, as unforeseen events constantly change the optimal levels of debt.
Based on Miller, Modigliani, and Myers’ findings numerous articles and studies have been
made on the determinants of capital structure such as Frank and Goyal’s (2009) article. In this
article, they investigate the correlation between capital structure and various factors such as
profitability, firm size, growth, market-to-book assets ratio, expected inflation and industry
median leverage within a sample of American companies from 1950 to 2003. Frank and
Goyal explore different theoretical frameworks and present their findings through regression
13
analysis, demonstrating how changes in different variables affect the outcomes. Their analysis
encompasses firm-specific and industry-specific elements, incorporating macroeconomic
factors like interest rates and inflation. The conclusions drawn suggest that larger firms and
firms with a lot of tangible assets tend to exhibit higher leverage, and when the expected
inflation is higher, firms tend to increase leverage. Additionally, the research reveals a
negative correlation between leverage and profitability, indicating that firms with higher
profits tend to have lower leverage ratios. The market-to-book asset ratio has a negative but
relatively low impact on the leverage levels for the firms studied. The industry median
leverage shows a strong positive correlation with leverage, making it a relevant variable to
include in their study especially since firms active in different industries are analysed.
When it comes to studies conducted on Swedish firms, Song's (2005) research explores the
connection between different elements influencing capital structure and total debt levels of
about 6000 Swedish enterprises between 1992 and 2000. Additionally, Song notes that high
leverage is a common tendency among Swedish businesses. He uses OLS regression using
panel data, taking into account time dummies, uniqueness, sales, growth, profitability,
non-debt tax shields, tangibility, and firm size as dependent factors. Song states that non-debt
tax shields, which are determined by dividing total depreciation by total assets, should show a
negative correlation with leverage. Song argues that there is a positive relationship between
leverage and corporate size. However, his findings suggest that there is no relationship
between growth and leverage. Furthermore, he argues that there is a negative relationship
between leverage and profitability, which is consistent with the pecking order and other
research Song. However, there are some weaknesses in Song’s research. It was conducted
during relatively stable years in terms of inflation. Sweden experienced significant inflation in
1991 and 1992, according to SCB, which Song excluded from his dataset. Consequently, his
findings might differ from ours due to the differing macroeconomic conditions during our
study period. Additionally, Song did not focus on a specific industry, which Frank and Goyal
emphasised as important. Something that could lead to larger standard errors.
Edberg and Kjellander’s (2022) study investigates the impact of the Covid-19 outbreak on the
capital structure of Swedish corporations. To achieve this, they employ both panel regression
techniques with fixed and random effects estimation, as well as dynamic panel regression
utilising Arellano-Bond estimators. While the results exhibit some variability, they indicate an
overall increase in corporate leverage after the Covid-19 outbreak by 1.3%. However, the
14
average leverage level for the industrial sector decreased during the pandemic as the
coefficients for all variables became less positive or more negative in their regressions when
analysing the pandemic period. Edberg and Kjellander discovered that the relationship
between profitability and leverage is more negative in the industrial sector compared to other
industries during the pandemic. Edberg and Kjellander’s study does not examine the full
effects of the pandemic since their dataset only includes data from Q1 2017 to Q3 2021 and
therefore only encompasses the initial effects from the Covid-19 pandemic. Our dataset on the
other hand covers the entire years from 2017 and 2023, capturing more of the long-run
consequences of the pandemic.
With this in mind, we will include profitability, firm size, growth as variables inspired by
Frank and Goyal and the non-debt tax shield inspired by Song in our thesis. We chose not to
include the market-to-book assets ratio due to its relatively low effect on leverage in their
research, and since we aim to differentiate our research with existing studies. Similarly, we
exclude industry median leverage as a variable since our analysis only focuses on a single
industry while Frank and Goyal examined multiple different industries. Our research will fill
a void in the existing literature by examining some of the most relevant variables regarding
capital structure and including the most recent data available to see the entire effect from the
Covid-19 pandemic.
15
4 Methodology
In this chapter, the reader will be introduced to a model specification where we describe how
and why this model is established in alignment with our research questions. The reader will
also be presented with our regression models, the dummies used and explanations of the
variables. Lastly, we present all Robustness tests we have done in our data set.
4.1 Model Specification
To answer our frame of questions, we have performed three different regression models with
panel data. The primary regression analysis looks into the key factors influencing the capital
structure choices made by firms in the Swedish industrial sector. The purpose of the second
regression is to look at how Covid-19 has affected the determinant “Tangibility” of the
enterprises in the same industry. The objective of the final regression analyses is to examine
the impact of the Covid-19 outbreak on the “logarithm of revenue”.
In subsequent analysis, we will use two sets of dummy variables. The first set includes a
time-dummy representing years before the outbreak of the Covid-19 pandemic. Including the
years 2017, 2018, and 2019, with the time frame spanning from 2020 to 2023 as the
benchmark group. Additionally, we have categorized each company into one of three
industries within the industrial sector: Capital Goods, Commercial and Professional Services,
and Transportation. We conducted a Hausman test to see whether we should use Fixed or
Random effects and based on the results, we used a fixed effect model in our regression
analyses. This model will include these industry categories as dummy variables, without
keeping them constant, in the three specified regression models.
The regressions are as following:
Regression 1:
16
In all regressions, we use the Debt/Equity ratio as the measure of leverage. In Regression 1,
we first included growth as an independent variable, defined as the percentage growth in
revenue from the previous year. Notably, the growth in revenue for 2017 considers the data
from 2016.
The variable "profit" is defined as earnings before interest and taxes (EBIT) divided by total
assets. Tangibility, labelled as “Tang” in our regression, is calculated as the ratio of property,
plant, and equipment to total assets. The non-debt tax shield (NDTS) is calculated as the ratio
of total depreciation to total assets, similar to the method Song (2005) used in his paper.
Finally, "rev" represents the logarithm of total revenue, used to determine the size of a
company and to facilitate fair comparisons among the companies in our sample.
Regression 2:
In regression 2, we retain the same independent variables as those in regression 1. The
primary distinction in this regression model is the inclusion of an interaction term.
Specifically, we have introduced the variable “tan_precovid”, which represents the interaction
between the determinant tangibility and the years preceding the Covid-19 pandemic (2017,
2018, and 2019). The reason for involving the years before Covid-19 as an interaction term is
to acknowledge that economic circumstances in business contexts of the years in question
may differ. Which potentially impacts the role of tangibility in the context of performance. It
is crucial to note that the period before 2020 exhibits different economic conditions that may
contain specific effects in the utilisation of tangible assets. Studying this interaction enables
us to determine the manner in which the tangible assets of a firm affect the leverage ratio in
different macroeconomic conditions. By doing so, we will achieve the desired interaction
between tangible assets and Covid-19, which should reveal the effects that Covid-19 had on
tangible assets or could have in the future. It allows us to examine the relationship between
tangibility and firm performance when tangibility has the lowest sensitivity, and to better
comprehend how tangibility affects firm performance throughout different economic epochs.
17
Regression 3:
In regression 3, we have conducted a regression analysis similar to that in regression 2. In this
regression we have substituted the previous interaction term for a new one, logrev_precovid.
This term is defined as the logarithm of revenue, as explained in the second paragraph of
regression 1, and represents the logarithm of revenue for the years preceding the Covid-19
outbreak. The reason for interacting the logarithm of revenue with the years prior to Covid-19
is because of the idea that business environment and dynamics before the existence of the
pandemic are separate factors that define how revenues affect firms. Revenue is almost
always reciprocal and logarithm of revenue is applied to normalise its values and show
percentage changes most accurately. This interaction helps us to disentangle the temporal
effect of revenue because it identifies its operation within a certain period of time and
therefore enables us to better understand the workings and implications of this variable for the
firms under examination under distinct economic conditions. It helps to consider all factors
which can influence macroeconomic conditions and their reflection in performance, which
gives a deeper look at revenue’s effect over time.
4.2 Robustness Test
When doing observations within the same group, in this case, companies and industry sectors,
they are likely to be more correlated with each other than with observations from other
groups. Clustering helps to correct standard errors for this intra-group correlation and also to
capture the heterogeneity between the groups. This ensures that the estimated standard errors
are more accurate and leads to more reliable hypothesis testing and confidence intervals.
Further, we will also conduct the Hausman test in order to acknowledge if we should use
fixed effect or random effect for our regression.
In the presence of multicollinearity in the dataset, the independent variables exhibit high
correlation among themselves which results in elevated variance and standard errors. To
18
assess the dataset for multicollinearity, we will perform a VIF test utilising the formula below
(Daoud, 2017).
Table 1 provided below outlines the thresholds for VIF values. A VIF value of 1 indicates the
absence of multicollinearity in the dataset. Values ranging between 1 and 5 suggest a
moderate correlation, while values exceeding 5 indicate a high correlation among the
variables. In instances of high multicollinearity, the variables will be reassessed in order to
minimize their correlation with each other.
Table 1:
Table 1 shows different value of the VIF test and how they are correlated
(Daoud, 2017)
19
5 Data
This chapter presents the data for the Swedish publicly listed industrial companies, ranging
from January 1, 2017, to December 31, 2023. It discusses the selection of 142 companies
from an initial 194 due to data availability, the methodological decisions such as
winsorization to handle outliers, and the use of fixed effects models based on the Hausman
test. Additionally, it addresses the categorization of companies into three industries and the
reasons for not using industry dummy variables in the final regressions. Descriptive statistics,
correlation matrices, and robustness checks are provided to ensure data integrity and
comparability.
5.1 Data collection
Our sample was collected from Capital IQ, focusing on publicly listed Swedish industrial
companies. Initially, we identified 194 companies meeting these criteria. However, due to the
lack of data for some companies, we narrowed it down to 142 companies. The time frame
from January 1st, 2017, to December 31st, 2023 equals 7 years of studied firms and was the
primary focus of our analysis. The reason behind choosing this period of time is threefold.
Firstly, the aim was to conduct a study in a more present tense compared to previous studies
referenced in this thesis, ensuring relevance and credibility in our analysis. Secondly, the
chosen timeframe provided a sufficient amount of data to facilitate a comprehensive
examination, allowing for a thorough exploration of the research topic. It is worth noting that
the available data for 2024 was deliberately excluded in our analysis since we wanted to have
a clear definition of our timeframe. Finally, we wanted to examine the historical and current
effects of Covid-19 on the capital structure of Swedish publicly listed industrial firms.
Prior to conducting regressions in Stata, cross-sectional data sourced from Capital IQ was
manually transformed into balanced panel data. This conversion process was necessary in
order to run the regression models on our dataset. Detailed definitions and explanations of the
variables utilised in our regression analyses can be found in Table 1, located within the
Methodology section. We also collected data about three distinct industries within the
industrial sector, as previously mentioned. However, we concluded that the Transportation
sector was too small of a sector, with only 7 companies, compared to Capital Goods and
Commercial and Professional Services, with 31 and 104 companies respectively, to provide
any significant results for our thesis. Thus, we excluded the industry dummies in our
20
regressions. After conducting the Hausman test, we determined that the fixed effects model
was the most suitable for our study since the fixed effects model accounts for the variations
across the different industries in our regression analysis.
Table 2: Graph of Debt/Equity before Winsorization
Table 2 presents that the Debt/Equity ratio experiences considerable outliers illustrated on the
range on the y-axis, spanning from +32 to -17. The x-axis provides the number of
observations for each Debt/Equity ratio in our data set. The wide line in the middle illustrates
a box for the first quartile up until the third quartile of all data, representing 50%. Further, as a
result of this wide spread of observations, a decision was made to replace negative
Debt/Equity ratios and also positive outliers from our model. This was achieved through
winsorization, illustrated in table 5 below.
21
Table 3: Correlation Matrix before Winsorization
Table 3 presents the correlation among the variables, indicating that tangibility emerges as the
only significant variable. We can also note that tax shield and the logarithm of revenue
variable are significantly correlated with profitability respectively tangibility. Despite this, the
anticipated signs for all variables are consistent with theoretical predictions.
Table 4: Descriptive Statistics before Winsorization
In table 4 above, there is a wide range of values reflecting substantial outliers, including some
companies with notably large negative debt/equity ratios. The absence of a value for logrev
relates to Vestum AB's revenue for 2020. According to Capital IQ, their revenue for 2020 was
zero. Since the logarithm of zero is undefined, logrev has one missing value due to this
circumstance.
22
5.2 Cleaning Data
In order to ensure the integrity of our data, we conducted several robustness checks. The
results of these checks are presented in the table below. Further to address outliers in our
dataset, we applied winsorization at the one percent level on the entire panel data. This
involved replacing the extreme one percent of values at both ends of the distribution with
values corresponding to the boundary of that percentile.
Table 5: Graph of Debt/Equity after Winsorization
Table 5 displays the updated debt/equity graph following winsorization. The box illustrates
the first quartile up until the third quartile of all data, representing 50%. A notable observation
is the absence of the outliers in the dataset when comparing Table 5 to Table 2. The
previously prominent outliers at the upper percentile have been substituted with considerably
lower values, and negative ratios have been uniformly adjusted to zeroes.
23
Table 6: Correlation
In table 6 the correlation matrix after cleaning our data is provided. Prior to winsorization, the
significance between the variables was notably low, potentially indicating multicollinearity in
the dataset. However, following winsorization, all variables exhibit significance, with the
exception of a slight insignificance observed in revenue growth.
5.3 Descriptive Statistics
Below, table 7 presents the descriptive statistics after winsorization of the dependent variable
and five determinants selected to address our hypothesis. These determinants are derived from
142 publicly listed Swedish companies within the industrial sector. Our dependent variable is
Leverage (Debt-to-Equity ratio), while Tangibility, Logged Revenue, Profitability, Growth in
Revenue, and Tax Shield serve as independent variables.
Table 7: Post Winsorization
In table 7 it is worth noting that, following winsorization, the minimum and maximum values
of the Debt/Equity ratio have been adjusted to 0 and 3.36, respectively, instead of the original
24
-17 and +32. This adjustment prevents misleading results in our regressions. Additionally, we
have scaled down the extremely high maximum value of Growth through our winsorization,
to enhance the comparability of our panel data across different companies.
Table 8: Collinearity test
Table 8 provides a collinearity test in order to make our data more robust. We used the VIF
test in order to conclude if there is any collinearity, the calculated VIF in the dataset is 1.42,
meaning that there is very low multicollinearity in the dataset.
25
6 Empirical Results
In this chapter, the results of three regression models will be presented to the reader. The
coefficients and the significance of the determinants utilised in each model will be provided
and briefly discussed. Additionally, comparisons between the different regression models will
be included.
6.1 First Regression Model
In Table 8 below, the regression output for the first regression model is provided. The total
number of observations reported is 993, which includes all available data points, with one
observation missing, as previously elucidated in Table 4. The dependent variable in this
regression model is Debt/Equity, while various determinants serve as independent variables,
as presented in the methodology section.
Table 9: Regression Output First Model
The findings presented in Table 9 reveal a mix of expected significant and insignificant results
which are presented in order to support the answer of our first research question; “Which
factors are the primary determinants influencing the capital structure of publicly listed
companies within the Swedish industrial sector?”. Notably, Profitability, Tangibility, and Tax
Shield emerge as significant determinants at a 5% significance level, aligning with theoretical
expectations outlined in the thesis. Conversely, Growth and Revenue yield insignificant
results. However, the sign of each determinant corresponds with theoretical predictions as
26
previously mentioned, with all determinants having a negative effect on the debt-equity ratio
except Tangibility which the regression implies has a positive effect on the ratio.
6.2 Second Regression Model
Table 10: Regression Output Second Model
After we introduce the interaction term tan_covid in table 10, the results on the other variables
remain qualitatively the same as from the first model. The coefficient for tan_precovid is
significantly positive, implying that a firm's tangibility has a more positive effect on the firm's
debt to equity ratio pre covid than during covid years (2017, 2018 and 2019). Additionally,
the coefficient for tangibility has been reduced from 1.497 to 1.332 from the first regression to
the second one. We can also conclude that the interaction term provides a positive coefficient
at 0.657 suggesting that tangibility had a greater impact on the leverage ratio before the
pandemic. If we look at the other determinants compared to the first regression we see that
Growth, Profitability, and Tax shield now have a more negative coefficient. While the
coefficient logarithm of revenue has a more positive coefficient than in the first regression
model, it is still negative. Even if only Profitability, Tangibility, Tax Shield, and Tangibility
pre-Covid-19 are significant at a 5% level (1% level for Tangbility pre-Covid and Tax Shield),
it is still interesting to see that every coefficient aligns with previous theories. These results
also provide useful information for answering our second research question; “How has
Covid-19 affected the determinants of capital structure for publicly listed companies operating
within the Swedish industrial sector?”
27
6.3 Third Regression Model
Table 11: Regression Output Third Model
In table 11 we present the third and final regression model, which includes the interaction
term "logrev_covid" (an interaction between the logarithm of revenue and the years preceding
Covid-19), showing several notable results. Compared to the first regression model, the
logarithm of revenue remains insignificant. On the contrary, the interaction term is now
significant at the 5% level, providing a statistically significant relationship. Additionally, the
coefficient of the interaction term has shifted from negative to positive.
In terms of other variables, Profitability and Tax Shield continue to be significant at the 5%
level, according to our previous models. Tangibility, which was significant at the 5% level in
the first regression model, is now significant at the 1% level, indicating an increased level of
significance. Additionally, the coefficient for Tangibility has increased by 0.1 data points. The
coefficients for Growth, Profitability, and Tax Shield have decreased, while Growth remains
insignificant, similar to the second regression model.
An important notation, the inclusion of the interaction term reveals that the logarithm of
revenue is significant for the years before the Covid-19 outbreak. Despite these changes, the
signs of the coefficients remain consistent with those in the first and second regression
models, except for the noted change in the interaction term.
28
7 Discussion
In this chapter, the reader will be presented with an analysis of the factors influencing the
capital structure of Swedish industrial firms for the years 2017-2023, particularly the effect of
the Covid-19 pandemic. The chapter provides insights presented in the Empirical Results and
how our results are in line with previous theories and literature. Further, the reader will pre
presented with what limitations have been made in this paper.
7.1 Analysing the results
The regressions indicate that tangibility is the most important variable for Swedish industrial
firms when shaping their capital structure, both before and after the Covid-19 outbreak due to
the large coefficients. In the first regression where Covid-19 was an omitted variable, the
coefficient for tangibility showed a strong and significant but inflated relationship between
tangibility and Debt/Equity ratio. This is in line with Long and Malitz's (1985) findings that
firms with higher tangible assets tend to have higher leverage due to the lower perceived risk
by lenders.
When the interaction term for tangibility and the pre-Covid period was included in the second
regression model, the coefficient for tangibility decreased by 0.165 points. Indicating that the
coefficient was inflated since Covid-19 was omitted. During the pandemic, companies may
have faced difficulties financing their debt due to increased interest rates. To mitigate this and
avoid financial distress, companies might have needed to sell tangible assets, which have high
liquidation value, to finance their debt. The new reduced coefficient for tangibility in the
second regression demonstrates the average effect tangibility had on the debt-to-equity ratio
following the Covid-19 outbreak. The new interaction variables indicate that, before the
Covid-19 outbreak, the value was 0.657 higher than it was after the outbreak. When
controlling for the years prior to the Covid-19 outbreak, the average effect of tangibility on
the debt-to-equity ratio increased to 1.989 (from 1.332 to 0.657) as presented in table 9. This
shift can be attributed to the heightened economic uncertainty and risk aversion among
lenders, who preferred to finance firms with higher liquidation values of tangible assets. This
preference is rooted in the reduced risk associated with tangible assets, which can be easily
collateralized as Harris and Raviv (1990) stated. This also aligns with the research by Frank
29
and Goyal (2009), who concluded that tangible assets tend to increase a firm's leverage. Our
results acknowledge and confirm these similarities.
Profitability emerged as a significant determinant in all regression models, consistently
exhibiting a negative relationship with leverage. This supports the pecking order theory,
which suggests that firms prefer retained earnings over external debt. As firms become more
profitable, they rely less on debt, leading to a lower Debt/Equity ratio. This relationship held
steady even when accounting for the impact of Covid-19, indicating that the fundamental
preference for internal over external financing remained unchanged during the pandemic. As
discussed in the works of Rajan and Zingales (1995) and Majluf and Myers (1984), a negative
relationship was predicted. However, Jensen (1986) posited a positive correlation under
conditions of an efficient market for corporate control. Nevertheless, Jensen anticipated that
in an inefficient market, a negative relationship between profitability and leverage would
prevail. Consistent with our expectations, we observed a negative association with leverage,
which aligns closely with the findings of Rajan and Zingales (1995) and Majluf and Myers
(1984), contradicting Jensen's initial hypothesis.
Growth, measured by revenue growth, did not show significant results across the regression
models. This could be due to the mixed theoretical predictions about the relationship between
growth and leverage. One might argue according to the pecking order that higher growth
opportunities should increase leverage due to the need for more external financing, others
argue according to the trade-off theory that high-growth firms might avoid debt to mitigate
risk and preserve flexibility. Additionally, according to agency theory, the agent, represented
by the manager of the company, aims to invest in high future growth potential to enhance the
company's value, often overlooking the associated risks. However, firms opting to invest in
assets with substantial future growth potential may encounter opposition from shareholders
due to the heightened risk it poses to their investments. Consequently, an expected negative
correlation exists between growth and leverage. The insignificance of growth in our models
suggests that Song (2005) was correct in his findings that growth is not a suitable measure for
leverage. This notion is further supported that the relationship between growth and leverage
may be more complex and influenced by additional factors not captured in our study as
explained by the pecking order theory.
30
The non-debt tax shield, measured through depreciation, showed also a significant negative
relationship with leverage. DeAngelo and Masulis (1980) posited that non-debt tax shields,
such as those arising from depreciation, can serve as substitutes for the tax advantages
associated with debt financing. Thus, as these alternative tax deductions rise, the tax benefit
of leverage diminishes. Our findings are consistent with the proposition by DeAngelo and
Masulis (1980), as evidenced in Tables 8, 9, and 10. The significance of our results
underscores the pivotal role played by non-debt tax shields in shaping leverage.
The third regression model, which introduced the interaction term between the logarithm of
revenue and the pre-Covid period, highlighted a significant shift in the impact of revenue on
leverage. Prior to the introduction of the interaction term, revenue size (logged revenue) did
not significantly influence the Debt/Equity ratio. However, the coefficient was still negative
as Chakraborty’s (2010) prediction of a negative relationship was right. This negative
relationship may be due to the strict accounting standards in Sweden, particularly for large
companies. The strict accounting standards for large firms leads to less asymmetric
information between the companies and investors which increases the level of equity. This
aligns with the pecking order, suggesting a negative relationship between leverage and firm
size.
When controlling for the pre-Covid era, the relationship became statistically significant, with
the coefficient increasing by 0.012 data points. In the absence of significance prior to
controlling for the pre-Covid dummy variable, interpretation of the results was limited,
beyond noting the negative coefficients, which contradicted previous literature and theories.
Nonetheless, with the inclusion of the pre Covid-19 dummy as control variable, it becomes
evident that there exists a significant and positive relationship between firm revenue (size)
and leverage. This suggests that larger firms were better positioned to secure debt financing
before the pandemic, which aligns with the prediction of the trade-off theory. This could be
due to larger firms' perceived stability making them more attractive to lenders which is in line
with the findings of Deesomsak et al. (2004).
There appears to be a low level of multicollinearity, as evidenced by the variance inflation
factor (VIF) values being below 10. The mean VIF is approximately 1.4, which is considered
excellent. The low R-squared value indicates a weak linear relationship between the
dependent variable and the independent variables. Typically, an R-squared value below 0.3 is
31
considered unacceptable. The R-squared value is not particularly significant in this context, as
it is generally low when using panel data. If we were conducting a time-series analysis, a high
R-squared value would be more important. There may be no definitive answer to the optimal
debt-to-equity ratio, especially when considering the delays in adjustments according to the
static trade-off model. Additionally, companies may have varying risk preferences and
operational risks, leading to different levels of risk aversion and, consequently, different
leverage.
7.2 Limitations
As per our database, Capital IQ, certain data entries were not reported, which required the
removal of several companies with missing data. Thus, the number of analysed firms was
reduced from 194 to 142. Additionally, we used balanced panel data and therefore Winsorized
our dataset in Stata to address outliers, replacing the bottom and top 1% of the data. We opted
for Winsorization instead of trimming since we still wanted to include the outliers in the
dataset but with replaced values instead of missing values.
The selected timeframe for the analysis spans from 2017 to 2023, strategically chosen to
evaluate the impact of the Covid-19 pandemic. This period encompasses three years before,
during, and after the onset of the pandemic, facilitating an examination of any significant
effects on the determinants under study.
We have categorized each company into one of three industries within the industrial sector:
Capital Goods, Commercial and Professional Services, and Transportation. Based on the
results of the Hausman test, which indicated the appropriateness of using fixed effects, we
have included the industry categories in the fixed effects specification. Consequently, we have
not included them as separate dummy variables in our regression models.
32
8 Conclusion
This thesis aims to analyse the significant factors influencing the capital structure of publicly
listed Swedish industrial firms, with a specific contribution to the subject of how the impact
of the Covid-19 pandemic has affected the determinants. Through an empirical analysis
spanning the years 2017 to 2023, several key insights emerged, providing a deeper
understanding of capital structure dynamics in the context of unprecedented economic
disruptions.
Firstly, our findings confirmed that tangibility plays a significant role in determining the
capital structure of firms. Before the pandemic, firms with higher tangible assets were more
likely to have higher leverage due to the lower risk perceived by lenders. However, the
Covid-19 pandemic altered this dynamic, as firms faced increased financial distress and
higher interest rates. This led to a necessity for firms to sell tangible assets to maintain
liquidity and avoid insolvency. Consequently, the coefficient for tangibility decreased,
indicating that while tangible assets remain crucial, their impact on leverage was moderated
by the economic uncertainty introduced by the pandemic.
Profitability consistently showed a negative relationship with leverage across all regression
models, supporting the pecking order theory. This suggests that firms prefer internal financing
through retained earnings over external debt. The persistence of this relationship, even during
the Covid-19 pandemic, highlights the fundamental preference of firms to minimise debt in
times of increased profitability.
Growth, as measured by growth in revenue, did not exhibit significant results, suggesting that
its relationship with leverage is complex and influenced by factors beyond those captured in
this study. This aligns with previous research indicating mixed theoretical predictions about
the impact of growth on leverage.
Non-debt tax shields emerged as significant determinants with a negative relationship to
leverage. This finding supports the theory that non-debt tax shields can substitute the tax
benefits of debt financing, reducing the incentive for firms to take on additional leverage.
The size of the firm, indicated by the logarithm of revenue, became a significant determinant
only when accounting for the pre-Covid period. Larger firms were better positioned to secure
33
debt financing before the pandemic, likely due to their perceived stability and lower risk. This
relationship underscores the importance of firm size in capital structure decisions, particularly
in stable economic environments.
Finally, our analysis emphasises which capital structure determinants for publicly listed
companies within the Swedish industrial sector, especially in response to significant economic
disruptions like the Covid-19 pandemic, affects the capital structure the most. The study
contributes to the literature by highlighting how traditional determinants such as tangibility,
profitability, and non-debt tax shields remain relevant, while also illustrating the irrelevance
of the growth and size. Simultaneously, also consider external economic shocks in capital
structure analysis. These insights are valuable for investors and corporate managers in
understanding and navigating the financial strategies of firms in varying macroeconomic
conditions.
Future research should consider using long and short-term debt to see the differences in the
performance of the two. Since we did not include them in our study and rather concentrated
on the examination of such determinants as affecting the chosen leverage ratio (Debt/Equity),
implementing these variables could have offered a broader perspective into the firm’s capital
structure. Extending the review period to include events like the 2008 financial crisis could be
useful. This wider temporal frame enables the analysis to consider different macroeconomic
conditions, which is helpful in understanding which factors affect leverage ratios in different
situations. Moreover, expanding the model’s scope by incorporating other factors that we did
not factor in our analysis could provide a more comprehensive view of the problem. Some of
the literature sources that we used for inspiration have discussed many factors that can be
considered as potential areas for future research and model enhancement. Documenting these
recommendations and integrating them into subsequent studies will enhance the knowledge of
the factors that affect firms’ leverage ratios in various economic environments.
34
References
Chakraborty, I. (2010). Capital structure in an emerging stock market: The case of India.
ScienceDirect, [online] 24(3), pp.295–314. Available at:
https://www.sciencedirect.com/science/article/pii/S0275531910000036?casa_token=f7UHz5h
H4JkAAAAA:B5r-nnyDBGzyg7Vwky5wsryi2j6MsP_CQQIcjlfi_toU9Lly5Dpb3eEnmmhAd
bVMDs2aGVfaEEY.
Corporate Finance Institute. (2022). Pecking Order Theory. [online] Available at:
https://corporatefinanceinstitute.com/resources/valuation/pecking-order-theory/.
Daoud, J.I. (2017). Multicollinearity and Regression Analysis. Journal of Physics, 949, pp.
1-5. [online] Available at:
https://iopscience.iop.org/article/10.1088/1742-6596/949/1/012009/pdf.
DeAngelo, H. and Masulis, R.W. (1980). Optimal capital structure under corporate and
personal taxation. Journal of Financial Economics, 8(1), pp.3–29.
doi:https://doi.org/10.1016/0304-405x(80)90019-7.
Deesomsak, R., Paudyal, K. and Pescetto, G. (2004). The determinants of capital structure:
evidence from the Asia Pacific region. Journal of Multinational Financial Management,
14(4-5), pp. 387–405. Available at:
https://www.sciencedirect.com/science/article/pii/S1042444X04000039?casa_token=xnmiujv
p-EgAAAAA:6O4CLkmoFccVNmPTzJD1YoAAOgDecezDC1OFv_K3ML3yxHtnRjF8REc-
cK-Sd9rnOhsAZQ56ytk
Edberg, C. and Kjellander, O. (2022). The Impact of Covid-19 on Corporate Capital
Structure. [online] Available at:
https://www.diva-portal.org/smash/get/diva2:1672331/FULLTEXT01.pdf.
Fama, E.F. and French, K.R. (2002). Testing Trade-Off and Pecking Order Predictions about
Dividends and Debt. The Review of Financial Studies, [online] 15(1), pp.1–33. Available at:
https://www.jstor.org/stable/2696797?seq=1.
35
Frank, M.Z. and Goyal, V.K. (2009). Capital Structure Decisions: Which Factors Are Reliably
Important?. The Journal of Financial Management, [online] 38(1), pp.1-37. Available at:
https://www-jstor-org.ezproxy.ub.gu.se/stable/pdf/20486683.pdf?refreqid=fastly-default%3A
36db38e0750e93ddccf1a0c0a1c41f61&ab_segments=&origin=&initiator=&acceptTC=1
Grigore, M.Z. and Stefan-Duicu, V.M. (1976). Agency Theory and Optimal Capital Structure.
Journal of Financial Economies, pp.863–868. Agency theory and optimal capital structure
HARRIS, M. and RAVIV, A. (1990). Capital Structure and the Informational Role of Debt.
The Journal of Finance, 45(2), p.323.
doi:https://doi.org/10.1111/j.1540-6261.1990.tb03693.x.
Jensen, M.C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.
The American Economic Review, [online] 76(2), pp.323–329. Available at:
https://www.jstor.org/stable/1818789.
Long, M. and Malitz, I. (1985). Chapter Title: Investment Patterns and Financial Leverage.
[online] Publisher, p.337. Available at:
https://www.nber.org/system/files/chapters/c11425/c11425.pdf.
Modigliani, F. and Miller, M.H. (1958). The Cost of Capital, Corporation Finance and the
Theory of Investment. The American Economic Review, [online] 48(3), pp.261–297.
Available at: https://www.jstor.org/stable/1809766?seq=1.
Mostafa, H.T. and Boregowda, S. (2014). ResearchGate. [online] ResearchGate. Journal of
Recent Sciences, 3(10), pp. 113-118. Available at:
https://www.researchgate.net/profile/Mostafa-Hashemi-Tilehnouei/publication/274138497_A
_Brief_Review_of_Capital_Structure_Theories/links/5c0a9bbd4585157ac1b03e4c/A-Brief-R
eview-of-Capital-Structure-Theories.pdf.
Myers, S.C. (1984). The Capital Structure Puzzle. The Journal of Finance, 39(3),
pp.574–592. doi:https://doi.org/10.1111/j.1540-6261.1984.tb03646.x
Rajan, R.G. and Zingales, L. (1995). What Do We Know about Capital Structure? Some
Evidence from International Data. The Journal of Finance, [online] 50(5), pp.1421–1460.
Available at: https://onlinelibrary.wiley.com/doi/full/10.1111/j.1540-6261.1995.tb05184.x.
36
Song, H.-S. (2005). Capital Structure Determinants. [online] Available at:
https://www.diva-portal.org/smash/get/diva2:565199/FULLTEXT01.pdf.
Statistiska Centralbyrån. (n.d.). Konsumentprisindex. [online] Available at:
https://www.scb.se/hitta-statistik/statistik-efter-amne/priser-och-konsumtion/konsumentprisin
dex/konsumentprisindex-kpi/pong/tabell-och-diagram/konsumentprisindex-kpi/kpi-12-manad
ersforandring-inflationstakten/.
37