Christian Jonassen and Pelle Cornéer 
 
 
A quantitative study of how capital structure is affected by intangible 
assets - The Case of Sweden 
 
 
 
 
 
 
 
Master of Science in Accounting and Financial Management 
Graduate School 
 
Master Thesis Spring 2021 
Supervisor: Stefan Sjögren 
 
 
 
 
 
 
 
 
 
i 
Abstract 
This thesis examines the effect that intangible assets have on capital structure on the Swedish 
market, and in doing so contributes to the existing body of research on capital structure and its 
determinants. The sample consists of 1 065 different firms, from 9 different industries between 
the period of 2000-2020. We find that intangible assets have a supportive role when it comes 
to securing debt for all of our proxies for leverage. In addition, we find it hard to find any 
consistent differences of the association between intangible assets and leverage in different 
industries. However, we find significant results for the firms active in the industries with the 
highest sophistication of innovation, indicating that the industries with high levels of 
innovation will be more likely to use intangible assets in order to secure leverage. 
 
Keywords: Intangible Assets, Capital Structure, Leverage, Tangible Assets, Knowledge-
Economy, Industry Comparison, Innovation  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ii 
Acknowledgements 
First, we want to express our profound gratitude to our supervisor, Stefan Sjögren, who has 
been a pillar of knowledge and assisted us with precision throughout the emergence of this 
thesis. We have been supported by his expertise in the field and his insights have been greatly 
appreciated. Second, we would like to extend our gratitude towards our fellow students who 
have contributed to this thesis by asking questions, giving us inspiration as well as remarks on 
the thesis. Last but not least, we want to express our humble recognition to our friends and 
families who have stuck by us in these intense times.  
  
iii 
Table of Content 
1. Introduction .......................................................................................................................... 1	
1.1 Background ................................................................................................................................... 1	
1.2 Problem ......................................................................................................................................... 2	
1.3 Purpose .......................................................................................................................................... 3	
1.4 Research question ......................................................................................................................... 3	
2. Theory and hypothesis ......................................................................................................... 4	
2.1 Theoretical framework .................................................................................................................. 4	
2.1.1 Modigliani and Miller theorem	......................................................................................................................................	4	
2.1.2 Trade-off theory	..................................................................................................................................................................	5	
2.1.3 Pecking order theory	..........................................................................................................................................................	6	
2.2 Previous empirical findings and hypothesis .................................................................................. 6	
3. Methodology ....................................................................................................................... 10	
3.1 Data ............................................................................................................................................. 10	
3.2 Variable definition ...................................................................................................................... 11	
3.3 Descriptive data ........................................................................................................................... 13	
3.4 Model specification ..................................................................................................................... 14	
4. Results and Analysis .......................................................................................................... 16	
4.1 Development over time ............................................................................................................... 16	
4.2 Regression results and analysis ................................................................................................... 18	
4.3 Robustness .................................................................................................................................. 27	
5. Conclusions and further research .................................................................................... 31	
5.1 Conclusions ................................................................................................................................. 31	
5.2 Future Research ........................................................................................................................... 32	
References ............................................................................................................................... 33	
Appendix ................................................................................................................................. 36	
 
 
 
 
 
 
 
iv 
1. Introduction 
1.1 Background 
The technologic incubation, driven by the increased knowledge of humans, has led to the 
emergence of valuable digital assets for firms. This has led us towards the knowledge economy 
(also called the knowledge-based economy), which is recognized as products and services that 
occur by knowledge-based activities (Powell and Snellman, 2004). The emergence of the 
knowledge economy becomes evident when looking into the proportion of tangible versus 
intangible assets on the S&P 500. In 1975, the market value of intangible assets only accounted 
for 17% of the total market value, whereas in 2020 the same figure had risen to 90% (Tomo, 
2020) and according to the World Bank (2007), firms are less dependent on physical assets 
than before. Today, large portions of firms' intangible assets are not always recognized in the 
balance sheets but can still constitute a great part of their value. Probably one of the most well 
known companies, Coca Cola’s trademark is estimated to be worth SEK 700 billion in 2007, 
which was about 60 % of their market capitalization. Other firms, like Nike, BMW and Apple, 
are estimated to have market values of intangible assets of approximately 70 % of their market 
capitalization (Melin and Hamrefors, 2007). Reasonably, this comes from changing industries 
and new technologies, which are harder to define and value in monetary terms and therefore 
hard to recognize in the firm's balance sheets. In turn, this creates consequences for firms when 
it comes to securing debt, but also for credit institutes when it comes to granting loans. 
Therefore, it is important to understand the shift towards the knowledge economy, and will be 
incremental for businesses when it comes to securing funding. The trade-off theory, tangible 
assets have been viewed as a more secure and backed collateral due to its redeployability and 
low transaction cost (Harris and Raviv 1991; Frank & Goyal 2008; Parsons & Titman 2009; 
Myers, 1984). Providing sufficient collateral will in the view of borrowers not only decrease 
the likelihood of credit rationing but can also work as a mechanism of signaling the quality of 
a firm (Stigliz and Weiss, 1981). The challenge of backing funding with intangible assets lies 
in the reliability of the value in the assets as well as determining how collateral can be utilized 
by the financier in the case of default. Due to the accounting-complexity in determining the 
real value of intangible assets, as there in many cases is no clear market where these assets are 
traded at arm’s length, backing funding with a high proportion of intangible assets are 
associated with a higher cost of capital and will affect the amount of debt demanded 
1 
(Tokmakciougle et al, 2007). In connection to this, the redeployability of an intangible asset is 
implied to have an effect on the cost of debt (Benmelech & Bergman, 2009).  
1.2 Problem 
Even though capital structure is a well studied field, there is still ambiguity between how the 
nature of assets influence firms leverage (Frank and Goyal, 2009). This stems from the fact 
that the trade-off theory and the pecking order theory are predicting adverse directions of the 
association between leverage and tangibility. The trade-off theory argues that tangible assets 
lowers the costs of financial distress and, in addition, lowers the problems associated with 
agency costs. While the pecking order theory predicts that tangible assets are associated with 
lower information asymmetry, the issuance of equity becomes less expensive leading to a 
decrease in leverage in the presence of tangible assets (Frank and Goyal, 2009). However, this 
is not completely clear, since tangible assets can also lead to increased adverse selection and 
therefore debt becomes relatively cheaper. Due to this, there is still of interest to investigate 
what effect intangible assets may have on the debt levels for firms.  
 
Moreover, we are going to investigate the Swedish market for public firms. The Swedish 
market is of interest, since it has not been researched comprehensively in the field of capital 
structure. However, Skog and Swärd (2015) conducted a study of the Swedish market and 
found that in Scandinavia, the 30 largest Swedish firms have on average a smaller proportion 
of tangible assets and on average a higher debt level. Finally, the knowledge economy is 
emerging, which in turn leads to assets becoming more intangible. While the nature of the 
assets are adjusting, the capital market also needs to be adjusted, and as suggested by Loumioti 
(2012), pledging intangible assets are today economically sustainable and, therefore, the 
importance of intangible assets may have increased during the years for lenders and borrowers. 
Hence, the technological incubation makes it further interesting to study if intangible assets 
may have a supporting effect on leverage.  
 
As aforementioned, the values of intangible assets have become clearly predominant in 
comparison to tangible assets (Tomo, 2020). This raises the question of how intangible assets 
are acknowledged as collateral from a lender's perspective. The empirical literature on capital 
structure is inconclusive regarding intangible assets impact on leverage, which is likewise 
argued by Lim et al (2020).  Traditionally, lenders have favored tangible assets to secure debt, 
2 
since they seem to be associated with less risk than can be argued to be partially true. However, 
there are important characteristics of intangible assets to facilitate more debt. It therefore 
becomes important to examine how the asset-structure of different firms will affect the way a 
firm is leveraged based on which type of assets are used as collateral.  
1.3 Purpose 
The purpose of this thesis is to investigate if Swedish firms utilize intangible assets to secure 
debt. We want to understand if the characteristics of intangible assets are sufficient when 
securing debt by firms.  
1.4 Research question 
The research question is the following:  
 
• Are Swedish firms utilizing intangible assets to secure funding and does this vary across 
industries? 
 
 
 
 
 
  
3 
2. Theory and hypothesis 
2.1 Theoretical framework 
2.1.1 Modigliani and Miller theorem 
The discussion of the determinants of debt was originally awoken by Modigliani and Miller 
(1958), when they argued that the capital structure was irrelevant when it comes to the value 
of the firm. To defend their theory, they developed several assumptions to make their theory 
hold, which is usually referred to as a perfect capital market (Berk and DeMarzo, 2014). The 
assumptions are the following:  
● Investors are able to trade the same securities at competitive prices and their value is 
decided on the present value of the future cash flows. 
● Taxes, transaction costs and costs for issuing securities are absent. 
● Financial decisions do not affect cash flows or do not reveal additional information 
about the firm which was not known before. 
 
From these assumptions Modigliani and Miller developed their two propositions. The first 
proposition states that, under the conditions of a perfect capital market, the value of the firm is 
unaffected by the choice of capital. Thus, increased leverage for a firm will only result in a 
reallocation of where the future cash flows are being extracted (Berk and DeMarzo, 2014). This 
means that the increased cash flow today, due to increased leverage, will be offset by decreased 
cash flows in the future, caused by the increased interest payments in the future. Furthermore, 
they argued that the total amount of cash flow paid out to equity holders and debtholders is 
equal to the total cash flow generated by the firm's assets and, therefore, due to the law of one 
price, the value of the firm should not be affected whether the firm holds debt or equity. 
Moreover, if the investors are not satisfied with the capital structure of the firm, they can use 
homemade leverage to obtain their desired capital structure (Berk and DeMarzo, 2014). This 
means that, if the investor can borrow at the same rate as the firms, the investor can replicate 
any capital structure of a firm by purchasing shares at margin.  
 
The second proposition is related to the expected return of the firm and suggests that a higher 
debt-to-equity ratio has a positive linear relationship with expected return. However, the 
increased expected return is associated with increased risk and therefore leveraged equity has 
4 
a higher cost of capital which means that the value is being discounted to a higher degree. This 
decreases the value of the firm to the same extent as the increased expected return increases 
the value of the firm (Berk and DeMarzo, 2014). Thus, the capital structure of the firm is 
irrelevant in a perfect capital market.  
 
However, this perfect world Modigliani and Miller explained was disputed due to the frictions 
usually observed in practice. Modigliani and Miller (1963) corrected their original proposal 
with an additional paper and stated that debt has tax advantages, since the interest payments 
reduce the profit that is being taxed. They do, however, highlight that to maximize the value of 
the firm, the firm should not be fully financed by debt. Other sources of finance, such as 
retained earnings, may in some circumstances be cheaper even though taxes are present. 
2.1.2 Trade-off theory 
Kraus and Litzenberger (1973) developed the trade-off theory from Modigliani and Miller’s 
capital structure propositions. In addition to taxes, they also introduced bankruptcy costs as 
another imperfection. Taxes were still the driving force for increasing leverage for the firms, 
but on the contrary to taxes, the risk of bankruptcy is increased by the leverage and, therefore, 
could implicate additional costs for the firm. Thus, the optimal level of leverage is where the 
additional value created by tax savings on the margin is offset by the potential costs of 
bankruptcy.  Myers (1984) further developed these thoughts, by introducing that a firm sets a 
target debt-to-value ratio, which it gradually adjusts until there is a balance between the tax 
shield and the bankruptcy costs. The reasons why firms gradually adjust towards an optimal 
debt-to-value ratio is due to the adjustment costs that exist (Myers, 1984). If the trade-off theory 
holds and there are no costs with adjusting towards the optimal level of debt, the optimal level 
of debt would be the current level of debt. According to Frank and Goyal (2009), the optimal 
level of debt is not observable and, therefore, hard to determine. Nonetheless, according to 
Myers (1984), conclusions can be drawn about the magnitude of the bankruptcy costs. First of 
all, risky firms seem to borrow less than safe firms, where the “riskiness” of the firm is defined 
by the variance of the expected market value of the firm's assets. The reason behind this lies in 
the fact that the cost of bankruptcy is caused by the threat of default. Secondly, firms that hold 
larger proportions of assets-in-place, with an active second hand market, are also having a 
higher optimal debt level. Thus, firms that possess specialized assets, intangible assets or 
growth opportunities will hold less debt. The reason behind this is due to the fact that the loss 
5 
of financial distress is not only determined by the probability of default but, also the expected 
costs of when default occurs. (Myers, 1984) 
2.1.3 Pecking order theory 
The pecking order, by Myers and Majluf (1984), states that a firm will decide to acquire capital 
in a predetermined order. Firstly, a firm will use their retained earnings to fund new 
investments, secondly they will use debt, and lastly equity. The reasoning behind this order is 
the costs that the different sources of capital will entail. Retained earnings comes directly from 
the company and, hence, there is no information asymmetry. This as oppose to the both external 
sources of funding, debt and equity, where the firm has to pay a premium in order to 
compensate the investors for the perceived risk of not having the same level of information 
about the firm as inside managers. When comparing the two external sources of funding; debt 
and equity, the difference in cost for the firm stems from two different sources: Level of 
information asymmetry as well as claim to assets in case of default. The relationship between 
debtholders and firms are closer than between a firm and its shareholders, which means a lower 
level of information asymmetry, in connection with the debtholders having a higher claim of 
assets if the firm would default, this leads to debtholders demanding lower prices than 
shareholders does. Thus, explaining the reasoning behind the pecking order theory and the costs 
that different sources of capital will incur on the firm. (Myers & Majlyf, 1984) 
2.2 Previous empirical findings and hypothesis 
Earlier studies in the field have examined how the nature of assets affects capital structure to a 
rather great extent, however, these studies have not been able to depict a clear and consistent 
relationship, why the effect remains rather unclear (Siblikov, 2009). Titman and Wessels 
(1988) investigated the explanatory power of the theories among capital structure. More 
precisely, they examined the determinants which the theories predict are affecting the firm's 
leverage, where the nature of assets were one of the determinants. They used intangible assets 
to total assets and gross property plant and equipment plus inventories to total assets, as proxies 
for measuring the collateralizability. The first proxy had an expected association to be negative 
and the latter had an expected association to be positive. However, due to lack of significant 
results, these associations were not possible to confirm. Gaud et al (2005), however, were able 
to confirm that tangible assets are positively related to leverage. Their result was in line with 
6 
the trade-off theory, however, more support is still needed to develop a complete capital 
structure theory. 
 
One common characteristic of tangible assets is that they are more liquid than intangible assets 
in the case of bankruptcy. According to Sibilkov (2009), the degree of liquidity in a firm's 
assets has an ambiguous effect on the leverage. Less liquid assets could be argued to sell with 
a higher transaction cost and this increases the cost of liquidation, bankruptcy and debt. 
Therefore, this speaks for a positive effect between asset liquidity and leverage, which has been 
argued by Williamson (1988) and Shleifer and Vishny (1992). However, Myers and Rajan 
(1998) and Morellec (2001) argue that this relationship is only positive if managers cannot use 
their discretion on assets if liquidation is required. When managers cannot do this, Morellec 
(2001) found that asset liquidity has a negative effect on debt, while Myers and Rajan (1998) 
argue that the association is rather positive to a certain level and then becomes negative (i.e., 
the relation is curvilinear). Thus, only when assets are pledged as collateral this effect is 
predicted to be positive. In addition, this further strengthens the importance of assets being 
used as collateral when securing debt.  
 
Traditionally, and as already mentioned, tangible assets have been more commonly used as 
collateral, when securing debt, for several reasons already mentioned (Rajan and Zingales, 
1995). However, using intellectual property as collateral has been around for quite some time, 
but is gaining more interest due to the continuously increasing value of intangible assets 
(Baldwin, 1995; Bahrick, 1986). The same thoughts are brought up by Loumioti (2012), who 
found that using intangible assets as collateral has helped alleviate financing frictions as lenders 
to a higher degree have started to accept intangible assets as collateral. In addition, using 
intangible assets as collateral was associated with increased cost of debt and increased supply 
of credits to firms. Nonetheless, their result showed that using intangible assets as collateral is 
fully economically sustainable on the modern capital market. Similarly, Lim et al (2020) 
investigated the impact of identified intangible assets on debt through a change in a specific 
accounting rule. This allowed them to observe reported market values of intangible assets and 
find that intangible assets were positively related to leverage. Additionally, their result 
suggested that intangible assets support debt as much as tangible assets do. In the same field, 
Matemilola and Ahmad (2015) emphasized the importance of both tangible and intangible 
assets when securing debt for firms. Their result indicates that goodwill as well as fixed assets 
have a positive effect on total debt and long-term debt, however, the coefficient for fixed assets 
7 
is higher than the coefficient for goodwill. In connection to these findings and in terms of using 
different types of intangible assets as collateral, Mann (2018) finds that specifically patents, 
due to their inherent liquidity, withhold the attributes of securing funding. Mann (2018) further 
concludes that intangible assets serving as collateral may in fact have an important economic 
value. These findings are also in line with Loumioti (2012), who suggested that goodwill 
reduces information asymmetry between borrowers and lenders and, therefore, alleviates the 
accessibility of capital for firms. Adding to the importance of the relationship between 
borrowers and lenders regarding the issue of information asymmetry is, both Boot and Thakor 
(1994) and Berger and Udell (1995), who emphasize that the longevity of this relationship will 
alleviate additional funding.  
 
The importance of intangible assets is further stressed by Larkin (2013), who found that 
powerful brands can generate a positive attitude among consumers can increase the marginal 
price-setting and, hence, lower the volatility of cash-flows, incurring an increase in the credit 
rating of firms. In turn, this had an effect on the financial policies by increasing net debt while 
increasing leverage and decreasing cash holdings. These findings suggest that intangible assets 
not only need to have an increasing effect on leverage due to the collateralizability, but also 
due to making cash-flows more predictable, facilitating lenders' ability to provide funding.  
 
What could be concluded is that the empirical literature agrees that if assets are collateralized 
they foster the issuance of debt, nevertheless, intangible assets that cannot be pledged as 
collateral may still have a positive effect on cash-flows. Moreover, the proxy for the 
collateralizability is usually the relation between tangible assets and total assets. The few 
studies that have investigated how intangible assets influence leverage presents ambiguous 
results. With this in consideration, we aim to further investigate the relationship of intangible 
assets on leverage and, thus, our first hypothesis will be;  
 
𝐻!: The proportion of intangible assets has a positive relationship with leverage 
 
Further, it is evident that different types of firms in different industries will have incumbent 
preferences when it comes to financing. Industries with a high ratio of tangible assets, i.e., real 
estate will in general be more leveraged than for example firms in the technology industry that 
in general are more dependent on intangible and knowledge-based assets. Differences in 
leverage between industries have been studied to a great extent and are confirmed by many 
8 
studies (see for example, Scott and Martin (1975) and Li and Islam (2019)). How the nature of 
assets influence leverage in different industries seems, however, rather unexplored. Different 
types of intangible assets will have diverse potential effects on leverage whether or not they 
will be admissible as collateral or not. For example, Mann (2018) emphasize, that patents are 
shown to facilitate the issuance of debt. The different characteristics in different industries, 
such as highly innovative industries, will likely be more dense in intellectual capital, such as 
for example patents. Therefore, our second hypothesis will aim to investigate the differences 
that intangible assets may have in terms of securing debt in different industries; 
 
𝐻": The impact of the proportion of intangible assets on leverage has different effects in 
separate industries 
 
  
9 
3. Methodology 
3.1 Data 
The panel data was collected from Bloomberg for the years the years 2000-2020. There are 
several benefits of using panel data. The panel data takes into account individual heterogeneity, 
which is not possible when studying relations cross-sectionally. In addition, it is more 
informative, adds more variability to the data, saves degrees of freedom and there is usually 
less collinearity between variables (Baltagi, 2008).  The panel data is suitable since it allows 
for a more dynamic analysis. This is not easily obtained from normal time-series data, since it 
usually requires a rather large time period (Brooks, 2015). Further, Gaud et al (2005) prefer 
panel data. They argue that the decisions about capital structure are naturally dynamic making 
panel data is appropriate to use in these studies. In order to avoid survivorship bias, the sample 
consists of all public Swedish firms that were active for each given year. The income statement 
and balance sheet items were collected on an annual basis for the first trading day each year 
for the firms, with the requirement that the financial information is the latest information filed 
from Bloomberg. The firms are classified by the Industry Classification Benchmark (ICB) 
introduced by Dow Jones and Financial Times Stock Exchange. Information which refers to 
market prices are gathered from the first trading day each year and all the data is gathered in 
local currency. Financial firms and utility firms were excluded from the sample based on the 
logic presented by Fama and French (1991)1 and due to standard practice (Frank and Goyal, 
2003). In addition, derivatives of indices and firms were manually excluded. The data sample 
consisted of some missing values that were omitted, which may have affected the results. There 
may be several reasons behind the missing values from the sample. One source of the problem 
may be traced to issues with the detection of certain items by Bloomberg, after observing some 
companies with missing values while comparing with the filed annual reports. One additional 
bias to consider while interpreting the results is the fact that the sample consists of only public 
firms. There could be differences in capital structure determinants among private firms that this 
omits.  
 
1Financial  and utility firms may distort the result since the underlying rationale behind leverage is not consistent with other 
industries. Their financial leverage may be affected by regulators and therefore should be excluded (Fama and French, 2002)  
10 
3.2 Variable definition 
Leverage has been defined in many different ways in the literature on capital structure. It is 
important to understand exactly what is meant by leverage, since it may have an effect on the 
inference that will be drawn from the results. Rajan and Zingales (1995) argue that the 
appropriate measure depends on the objective of the analysis. The earlier studies have been 
using a wide range of debt ratios, but there is also a distinction whether book values or market 
values should be used when defining the debt ratio. However, concluding which type of 
measure for leverage is superior is beyond the scope of this paper. Although, in line with Frank 
and Goyal (2009), we intend to explore both book values and market values as well as long-
term and total debt. This means that we will study four different dependent variables for 
leverage; 1) Total debt to book value of assets (TDA), 2) Total debt to market value of assets 
(TDMV), 3) Long-term debt to book value of assets (LTDA) and 4) Long-term debt to market 
value of assets (LTDMV). We do, however, make some adjustments to the market value of 
assets, since we do not have the same source of data. If the identical measure would be used, it 
will result in a large proportion of missing values that may be inappropriate. However, we are 
nevertheless interested in a broader measure of the market value of assets, since the Swedish 
market is rather unexplored when it comes to capital structure studies. Debt is used, both total 
debt and long-term debt, instead of liabilities, since it is assumed to be more appropriate 
according to Rajan and Zingales (1995). When leverage is constructed with liabilities, the ratio 
will be influenced by account payables and pension liabilities that are not desired in our 
objective of understanding capital structure. With respect to this, we will define the market 
value of assets as the market value of equity and add the book value of total debt. Therefore, 
the corresponding measure for leverage could be seen as a proxy for the amount left for 
shareholders if the firm is liquidated. It would be optimal to use the market value of debt, but 
this was not possible due to data limitations. However, according to Titman and Wessels (1988) 
this misspecification caused by book value of debt is probably small as market debt and book 
value of debt are strongly correlated. Further, the reader should bear in mind that this ratio may 
overstate the amount of leverage and, therefore, may not be appropriate when taking probability 
of default into account (Rajan and Zingales, 1995).  
 
Moreover, the independent variable of interest for intangibility is Intangible assets ratio. 
Intangible assets ratio is simply the item intangible assets exported from Bloomberg divided 
11 
by the total book value of assets2, when we estimate the model with TDA and LTDA as 
dependent variables, market value of assets, when estimating TDEV and LTEV. In addition, to 
validate the causality of the variable of interest appropriate control variables have been used 
based on the earlier literature. These are variables that are predicted to influence the capital 
structure decisions of the firm. Market-to-book ratio will be used as control, since it is a 
commonly used proxy for growth opportunities. Growth opportunities in firms are increasing 
the cost of financial distress and, therefore, Market-to-book should be associated with lower 
levels of debt. Profitability is predicted to influence leverage in an ambiguite way. From a tax 
and agency cost perspective, more profitable firms will take on more leverage. From a pecking 
order perspective, internal generated funds infer less costs, why more profitable firms will be 
less leveraged. Further, Size will be used for control and it is expected to be positively 
associated with leverage, since larger firms are more diversified lowering the probability of 
default. However, larger firms are expected to be older and, therefore, predicted to have more 
retained earnings and, hence, the association is inverse from a pecking order perspective. 
Finally, our last control variable is Risk, which is expected to have a negative association with 
leverage. More volatile cash flows are considered to increase the cost of financial distress, 
making debt more expensive for more risky firms according to the trade-off theory. However, 
the pecking order theory predicts the opposite association, since more risky firms are having 
more problems with adverse selection that increases the cost of equity. (Frank and Goyal, 
2009). These variables are selected, since they are argued to be the most significant factors 
from previous literature (Rajan and Zingales, 2005). The variables of interest are first regressed 
against each other, and afterwards the controls are added one by one and regressed again. If the 
significance changes, the direction of the association is changed or the magnitude of the 
coefficient is rubbed significantly, it will be considered. The control variables are decomposed 
in Appendix I.  
  
 
2 We use a similar definition of intangible assets ratio as Titman and Wessels (1988), who used intangible assets to total assets ratio as well.  
12 
3.3 Descriptive data 
Table I shows the descriptive statistics for all the variables of interest. The total sample consists 
of 1 336 firms over the chosen time period. Outliers of the sample were managed by 
winsorizing by the 5th percentile and the 95th percentile. 5 % in each tail of the data was 
reasoned to be appropriate after observing the number of outliers.  
 
 Table I: Descriptive statistics The statistics consist of observations between 2000-2020 and have been managed for missing values and outliers. Missing values were 
 removed and therefore the amount of observations are the ones being left. The sample has been winsorized by the 5th percentile and the 95th percentile in order to reduce the problem of extreme values but still be able to keep a sufficient amount of observations.  
 Variable Obs Mean Std. Dev. Min Max
TDA 7 630 0.1771964 0.1828967 0 0.5801173
TDMV 7 496 0.1655404 0.2002908 0 0.6421298
LTDA 7 620 0.126935 0.1564682 0 0.5144421
LTDMV 7 486 0.1176164 0.1647901 0 0.5549092
Intangibles BV 7 527 0.2888342 0.2331139 0.0005929 0.7562771
Intangibles MV 6 453 0.2336977 0.2629301 0.0002687 0.9361383
Tangibles BV 6 496 0.2669759 0.2266009 0.0014621 0.7358804
Tangibles MV 5 738 0.2450713 0.2813949 0.0002683 0.9662634  
Source: Bloomberg 
 
The independent variables do in general have a higher magnitude than the dependent variables, 
but also a larger standard deviation. Moreover, from the minimum and maximum column, it 
could also be observed that there are no observations with 0 tangible or intangible assets. Table 
II illustrates statistics of the sample divided into the different industries for all the variables of 
interest. The industries containing the greatest number of firms are the Industrials and Health 
care industries, while Consumer Staples, Energy and Telecommunications are the industries 
with relatively small amounts of firms in our sample. For an indefinite reason, Bloomberg was 
not able to specify a large portion of the firms into any industry. These firms are categorized 
as N/A. This may be a limitation with the sample, since it will be more difficult to analyze 
differences between certain industries and their capital structure. Further, the most leveraged 
industry, on average, is Real Estate when all measures for leverage are considered. The least 
leveraged industries, on average, are Technology and Health care, but the order depends on 
which measure for leverage is used. Moreover, the industries with the highest proportion of 
intangible assets are; Health Care, Technology and Telecommunications when intangible assets 
are measured by book values. However, if measured by market values, the industries with the 
highest proportion of intangible assets are Consumer Discretionary, Technology and 
Telecommunications. In contrast, the industries with a the highest proportion of tangible assets 
13 
when book values are considered are; Basic Materials, Energy and Real Estate which is also 
valid when market values are used. 
 
 Table II: Industry statistics 
The table displays the mean value for all the variables of interest divided into the different industries after the data have been managed for 
 missing values and outliers. Missing values were removed and therefore the amount of observations are the ones being left. The sample has 
been winsorized by the 5th percentile and the 95th percentile in order to reduce the problem of extreme values but still be able to keep a 
 sufficient amount of observations. 
 
 
Source: Bloomberg 
3.4 Model specification 
The main model for estimating the effect intangible assets have on leverage is presented in 
Equation 1. The main model is estimated four times in order to capture the different dependent 
variables described. In Equation 1, i denotes the cross-sectional dimension, which in our case 
are firms, and t denoting the dimension of time, in terms of years. The occurrence of the 
different proxies of leverage stems from earlier empirical studies which have found different 
effects depending on which debt is taken into account and if market values or book values have 
been used. Common for all the models is that the variables are expressed as a ratio. When the 
denominator in the dependent variable consists of market values, the independent variables will 
likewise have market values as the denominator to be consistent.    
 
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒#$ 	= 𝛽0 + 𝛽1𝐼𝑛𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑒𝑠#$ + ∑𝛽2→5𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠#$ 	+ 𝑖. 𝛽6→14𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦	 + Ɛ#$ (1)
  
In order to decide which estimation method to be used, the Hausman-test and Breusch and 
Pagan Multiplier test has been conducted. The Breusch and Pagan multiplier test is first applied 
in order to determine if pooled estimation or random effects estimation is deemed more suitable 
for the model (Breusch and Pagan, 1980). The test is significant, which implies that there is 
variation between groups in the sample and that a random effects estimation is more suited for 
the model. Further, the Hausman-test is conducted to decide if random effects or fixed effects 
14 
estimation is more appropriate (Hausman, 1978). By rejecting the null hypothesis, it is found 
that the fixed effects estimation predicts the most unbiased variables and, hence, the fixed 
effects-model will be used. In addition, it is usually also argued that the fixed effects model is 
more suitable when the sample is not randomly selected and instead constitutes the entire 
population (Brook, 2015). In our case, we are studying all the public firms in Sweden and, 
therefore, the fixed effects model could be argued to be appropriate.  
 
Further, in order to answer if there is any difference in the association between the proportion 
of intangible assets and leverage between industries, additional regressions have been 
estimated. The secondary regression model is presented in Equation 2. The equation is similar 
to Equation 1, but the industry dummies are excluded since we are conducting the regression 
for each industry separately. 
 
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒#$ 	= 𝛽0 + 𝛽1𝐼𝑛𝑡𝑎𝑛𝑔𝑖𝑏𝑙𝑒𝑠#$ + ∑𝛽2→5𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠#$ + Ɛ#$                   (2) 
 
Leverage is the dependent variable and the proportion of intangible assets is the independent 
variable for firm i.  
  
  
15 
4. Results and Analysis 
4.1 Development over time 
Visualizing the dependent and independent variables over time will allow for a more 
comprehensive analysis of the results of this thesis. Starting with the dependent variables, 
Figure I and II are presenting the two different proxies for leverage measured by book values 
of assets, i.e TDA and LTDA. The graphs display the average debt levels of the sample over the 
time period. Both are illustrating a similar decreasing pattern over the period with a peak at the 
beginning of the period and lowest at the end of the period. Thus, leverage has decreased for 
the average firm until the end of the period. The lowest level of debt was observed in 2018 for 
TDA and in 2015 for LTDA. However, for both TDA and LTDA, from 2018 leverage has 
distinctly started to increase for TDA in particular. Due to the relatively long time horizon, we 
are able to observe the effects of economic instability. The burst of the IT-bubble can be traced 
at the beginning of the period. Both total debt and long-term debt were negatively impacted by 
the event and recovered quickly the following years. The financial crisis of 2007-2008 appears 
to have had a major impact on both TDA and LTDA for the average firm. LTDA did, however, 
recover slower than for TDA and can be observed in Figure I and II. 
 
 Figure I: Total-debt-to-asset ratio Figure II: Long-term-debt-to-asset ratio 
 The graph is illustrating the average Total-debt-to-asset ratio The graph is illustrating the average Long-term-debt-to-asset between 2000-2020 within the sample. The average is ratio between 2000-2020 within the sample. The average is 
 calculated post winsorizing the sample by the 5th percentile calculated post winsorizing the sample by the 5th percentile and the 95th percentile. and the 95th percentile. 
  
 
 
 
 
 
16 
 
 Figure III: Total-debt-to-market-value ratio Figure IV: Long-term-debt-to-market-value ratio 
The graph is illustrating the average Total-debt-to-market- The graph is illustrating the average Long-term-debt-to-
 value ratio between 2000-2020 within the sample. The average market-value ratio between 2000-2020 within the sample. The 
is calculated post winsorizing the sample by the 5th percentile average is calculated post winsorizing the sample by the 5th 
 and the 95th percentile. percentile and the 95th percentile. 
  
 
Figure III and IV illustrate the leverage proxies measured by market values of assets, i.e., 
TDMV and LTDMV. In general, a similar pattern could be observed where the trend has 
decreased over the period. Both TDMV and LTDMV reached their lowest in 2018 and slightly 
increased towards the end of the period. However, the opposite pattern could be traced 
regarding the effects of economic instability. The debt levels are heightened during these events 
and subsequent years after the event before recovering. The potential cause for this is the low 
market values for firms due to the stock market crashes during these events. Even though the 
time period we are studying is relatively long, it is nonetheless important to take into account 
that these events may have an effect on the inference. Also, important to consider is the 
differences between industries when it regards debt levels. These differences in the different 
proxies of leverage can be seen in Appendix II. Furthermore, in Figure V and VI the 
independent variables of interest are presented as averages among all the firms over time. Both 
proxies of intangible assets ratio have increased during the period. This illustrates the relative 
increased proportion of intangible assets used by firms and confirms Baldwin’s (1995) view 
about the increased interest of intangible assets and that the trend has not vanished. The increase 
is more obvious when observing intangible assets ratio measured with book values (Figure V). 
As can be seen from Figure VI, the effect of economic instability is major when intangible 
assets ratio is measured with market values. Reasonably, when the economic instability is 
causing the market values for firms to decline, it inflates the ratio of intangible assets that can 
be seen from both the burst of the IT-bubble and the financial crisis of 2007-2008. In addition, 
it is worth noting that the intangible assets ratio measured by book values have been quite stable 
after the financial crisis, while intangible assets measured by market values have been declining 
17 
until the end of the period. Indicating that the financial crisis of 2007-2008 impacted the market 
values of equity to a large extent relative to book values of assets. 
 
 Figure V: Intangible assets ratio book value  Figure IV: Intangible assets ratio market value  
The graph is illustrating the average Intangible assets ratio The graph is illustrating the average Intangible assets ratio 
 measured by book values for the total assets between 2000- measured by market values for the total assets between 2000-
2001. The average is calculated post winsorizing the sample 2001. The average is calculated post winsorizing the sample 
 by the 5th percentile and the 95th percentile.                                by the 5th percentile and the 95th percentile.                                
  
 
As with the different proxies for leverage, the proxies for intangible assets do as well vary 
depending on which industry the firm operates in. The average intangible assets ratio, measured 
by book values and market values, can be seen in Appendix III for the different industry 
classifications. The graphical illustrations of leverage and intangible assets make it hard to 
predict any strict relations between intangible assets and leverage. Although all proxies for 
leverage have decreased over time, there are fluctuations over the period that are difficult to 
match with patterns from the average intangible assets over time, since book values differ much 
from when market values are used. What could potentially be observed are peaks in leverage 
and intangible assets for assets measured by market values under high economic instability. 
This indicates that in these periods of uncertainty, when asymmetric information is high, 
intangible assets have a positive effect on leverage. However, the cause of this could also be 
traced to large decreases in market values for firms, at least when measured by market values.     
4.2 Regression results and analysis 
The results of the main regression model (Equation 1) is displayed in Table III below. The 
results are statistically significant at 1 %. The control variables included in the results table are 
all at least statistically significant at 5 %. From the results, intangibles indicates a positive 
relationship towards leverage for all the different proxies of leverage as the dependent variable. 
If intangibles are increased by one unit TDA, TDMV, LTDA and LTDMV are increased by 0.07, 
18 
0.12, 0.03 and 0.08, respectively. In relation to this, the effect seems to be stronger when 
leverage is measured by TDMV and LTDMV, thus, when market values are considered, 
intangibles increases leverage even more. Another interesting observation is that intangibles 
seems to have a stronger effect on TDA and TDMV, which is an indication of different 
components of short-term debt being supported by intangibles. In the literature, the nature of 
assets is usually proxied by the proportion of tangible assets3 to total assets and is expected to 
have a positive relationship to leverage. It therefore makes it reasonable to assume that 
intangible assets in that case should have a negative impact on leverage. However, our results 
are suggesting the opposite and, therefore, contradict the traditional view of capital structure 
and, the trade-off theory, suggested by Myers (1983) and the empirical findings found by Frank 
and Goyal (2009), Gaud et al (2005) and Rajan and Zingles (1995). However, it is important 
to consider that different proxies for the nature of assets have been used, compared to the 
literature above, which may have impacted the results.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3 Tangible assets are in turn usually measured by gross property plant and equipment plus investories (Frank and Goyal, 2009). 
19 
 
T able III: Main regression model of intangibles as predictor of capital structure  
This table presents the result of Equation 2 from the section of methodology. The results are estimated using the OLS regression with the 
f ixed effect model and the time period is between 2000-2020. The sample contains of 1 065 firms when TDA and TDMV are used as dependent 
variables and 1 064 when LTDA and LTDMV are used as dependent variables. Intangibles are intangible assets divided by book value of 
t otal assets when TDA and LTDA are used as dependent variable and while TDMV and LTDMV are used as dependent variable intangibles 
a re defined as intangible assets divided by market value of total assets (see section 3.2 for more information regarding definitions). The 
variables; ROA, Size and MB are control variables and definitions for these can be seen in section 3.2. Before the regressions are run, all the 
v ariables are winsorized by the 5th percentile and the 95th percentile to avoid misleading outliers. * indicates significance at 10 % level, ** 
i ndicates sig  nificance at 5 % level and *** indicates significance at 1 % level. The t-statistics are presented in the parenthesis.  
 
VARIABLES TDA TDMV LTDA LTDMV
Intangibles 0.0694*** 0.1208*** 0.0314*** 0.0753***
(5.391) (12.326) (2.889) (8.778)
ROA -0.0862*** -0.1307*** -0.0466* -0.0644**
(-2.829) (-4.255) (-1.807) (-2.393)
Size 0.0234*** 0.0331*** 0.0198*** 0.0217***
(8.091) (11.465) (8.074) (8.560)
MB -0.0023** -0.0077*** -0.0027*** -0.0059***
(-2.387) (-7.384) (-3.350) (-6.401)
Risk 0.0006*** 0.0009*** 0.0003*** 0.0005***
(9.433) (14.628) (5.621) (9.160)
Constant -0.3754*** -0.5941*** -0.3067*** -0.3737***
(-4.837) (-7.611) (-4.674) (-5.469)
Observations 5,642 5,642 5,638 5,638
R-squared 0.045 0.162 0.030 0.092
Number of firms 1,065 1,065 1,064 1,064
Industry fixed effects YES/NO YES YES YES YES  
Source: Bloomberg 
 
Our result is more in line with Loumioti (2012), Larkin (2013) and Lim et al (2020), who are 
all more in favor of the more supportive role intangible assets could have on leverage. The 
rationale behind our results could be several. One reason, suggested by Lerkin (2013), is that 
the intangible assets (such as brands) may have positive cash-flow effects for the firm and will, 
therefore, enhance the credit ratings for the firms and, in turn, decrease the cost of debt. Another 
possible explanation could be traced to increased liquidity of intangible assets, which could be 
argued to be reasonable due to increased interest in intangible assets in the recent years. As 
Myers and Rajan (1998), Shleifer and Vishny (1992) and Williamson (1988) found, high 
liquidity of assets is associated with more debt, at least if managers cannot use their discretion. 
Therefore, our results could be argued to indicate that intangible assets can and are in fact used 
as collateral for securing debt. Furthermore, the emergence of new technology has made it 
20 
possible for new markets to open, in order to facilitate trade with intangible assets that can also 
be a source of increased liquidity. In addition, Myers (1984) emphasized the importance of 
assets traded frequently at a second-hand market and higher optimal debt levels for firms. In 
that sense, it is not unreasonable to believe that the characteristics of intangible assets may have 
changed, which can be attributable to a more mature market for intangible assets, and may not 
therefore be seen as equally risky as before. Another argument for this contradicting result is 
supported by Berger and Udell (1995), who found that firms that have a closer relationship 
with the banks do not need to provide collateral to the same extent as firms that do not have a 
close relationship with the banks. This close relationship will facilitate monitoring and provide 
the bank with valuable information. Therefore, collaterals are not needed to the same extent as 
information asymmetries will be lower if the relationship is close between the bank and the 
firm. With respect to this, Sweden could be argued to be a bank-oriented nation, with a long 
history of banking and with the world's oldest Central bank (Riksbanken, 2021). In addition, 
Boot and Thakor (1994) illustrated with their model that the longer the relationship between 
the bank and the firm, the requirement of collateral will be lower. Thus, assets that are not 
possible to collateralize, which is usually more common for intangible assets, could therefore 
nevertheless be a driver for issuance of debt and could possibly be an explanation for our results 
for the Swedish market.  
 
According to Frank & Goyal (2009) and the pecking order theory, tangibility of assets 
decreases information asymmetry and, therefore, equity becomes less costly to issue. Thus, the 
pecking order theory predicts that firms with more tangible assets should have less debt in 
relation to equity, firms with more intangible assets should have more debt, since they should 
be more associated with higher asymmetric information. With this in mind, our results indicate 
that intangible assets can play a definite role when it comes to securing funding. However, 
Frank and Goyal (2009) also discuss the ambiguity of the pecking order theory and that the 
direction of the association is not completely clear. This depends on what intangible assets 
actually represent. When tangibility represents assets in place, more tangible assets increase 
adverse selection leading to more debt and, hence, intangible assets should be negatively 
related to leverage (Frank and Goyal, 2009). The results indicate that intangible assets do, even 
if un-compared to fixed assets, foster the issuance of debt in all of our four models. By 
acknowledging the value of intangible assets by creditors accepting it as collateral, it also 
provides the intangible assets with a claimed redeployability, hence, indicating that a more 
holistic market for intangible assets should be feasible and is economically sustainable, which 
21 
is in accordance with Loumioti (2012). The main regression results from Table III have 
provided us with enough evidence to be able to reject our first null hypothesis that intangible 
assets have a negative relationship with leverage and, therefore, intangible assets seem to 
support leverage. 
 
Further, Tables IV-VII present the results from the regressions estimated by Equation 2. The 
results display the association between intangibles and the different proxies of leverage for 
each industry. Thus, the results are separating the industries to see if there is any difference 
between intangible assets as a predictor of capital structure due to different characteristics in 
different industries. TDA and LTDA as dependent variables are illustrated in Tables IV and V, 
respectively, while TDMV and LTDMV as dependent variables are presented in Tables VI and 
VII, respectively. Starting with Table IV, which uses TDA as the dependent variable, intangible 
assets are significant at our critical level for Energy, Health Care, Industrials, Technology and 
Telecommunications. The coefficients are -0.154, 0.096, 0.135, 0.094 and 0.111. The Energy 
industry is the only industry in which firms seem to have a negative association between 
intangible assets and leverage. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
22 
 
Table IV: Secondary regressions of intangible assets as predictor of by industry  
T his table presents the result of Equation 2 from the section of methodology when TDA is used as dependent variable. The independent variable 
of interest is intangible assets and is defined as intangible assets to total book value of assets. The results are estimated using the OLS regression 
w ith the fixed effect model and the time period is between 2000-2020. Before the regressions are run, all the variables are winsorized by the 
5th percentile and the 95th percentile to avoid misleading outliers. * indicates significance at 10 % level, ** indicates significance at 5 % level 
a nd *** indicates significance at 1 % level. The t-statistics are presented in the parenthesis.  
 
 Total debt to total assets
Industry Intangibles ROA Size MB Risk Constant
Basic Materials -0.0364 -0.0996 0.0295*** -0.0081** 0.0008*** -0.4621***
(-0.905) (-0.848) (4.616) (-2.167) (3.599) (-3.312)
Consumer Discretionary -0.0165 -0.2297*** 0.0524*** -0.0009 0.0010*** -0.9083***
(-0.563) (-2.682) (10.061) (-0.393) (5.566) (-8.277)
Consumer Staples -0.1280* 0.1336 0.0300** -0.0041 0.0010** -0.4211
(-1.891) (0.566) (2.358) (-0.685) (2.378) (-1.588)
Energy -0.1540** 0.0661 0.0137 0.0204*** 0.0006* -0.1928
(-2.231) (0.336) (1.058) (4.680) (1.922) (-0.738)
Health Care 0.0959*** -0.0271 0.0372*** 0.0020 0.0004*** -0.6711***
(4.571) (-0.405) (7.914) (1.323) (2.730) (-7.149)
Industrials 0.1351*** -0.1094 0.0298*** -0.0110*** 0.0006*** -0.4648***
(5.063) (-1.531) (7.451) (-4.307) (4.341) (-5.490)
Uncategorized Industry 0.0232 -0.0131 0.0270*** -0.0048** 0.0004*** -0.3877***
(1.122) (-0.292) (7.192) (-2.559) (4.260) (-5.066)
Real Estate -0.5310 0.5808* 0.0443*** -0.0049 0.0006* -0.5942
(-1.603) (1.693) (2.805) (-0.362) (1.665) (-1.635)
Technology 0.0942*** 0.0060 0.0022 -0.0028 0.0003* 0.0139
(3.311) (0.078) (0.408) (-1.496) (1.717) (0.132)
Telecommunications 0.1111** -0.1131 0.0126 0.0026 0.0003 -0.1643
(2.019) (-0.755) (1.340) (0.477) (0.978) (-0.821)  
Source: Bloomberg 
Table V illustrates how intangibles are associated with long-term debt measured by book 
values, thus, when LTDA is used as the dependent variable. Statistically significant results were 
obtained by Health Care and Technology, and the coefficients are indicating a positive 
association with LTDA. Furthermore, comparing these results with Table IV, it could be 
observed that the results for total debt (TDA) as the dependent variable generated more 
statistically significant results for more industries compared to long-term debt (LTDA). While 
evaluating the industries that were both statistically significant for total debt and long-term 
debt, that is Health Care and Technology, the coefficients were higher when total debt was 
considered. Health Care and Technology estimated coefficients of 0.096 and 0.094, 
respectively, when TDA was used. In contrast, the coefficients for LTDA were estimated to 
0.064 and 0.07, respectively. This indicates that intangible assets have a relatively larger effect 
on total debt at least when book values are considered. In turn, this further raises the question 
23 
mentioned in the results from Equation 1, if intangible assets are influencing short-term debt 
as well.  
 
Table V: Secondary regressions of intangible assets as predictor of leverage by industry  
T his table presents the result of Equation 2 from the section of methodology when LTDA is used as dependent variable. The independent 
variable of interest is intangible assets and is defined as intangible assets to total book value of assets. The results are estimated using the OLS 
r egression with the fixed effect model and the time period is between 2000-2020. Before the regressions are run, all the variables are winsorized 
by the 5th percentile and the 95th percentile to avoid misleading outliers. * indicates significance at 10 % level, ** indicates significance at 5 
%  level and *** indicates significance at 1 % level. The t-statistics are presented in the parenthesis.  
 
 Long term debt to total assets
Industry Intangibles ROA Size MB Risk Constant
Basic Materials -0.0318 -0.0587 0.0212*** -0.0097*** 0.0007*** -0.3335***
(-0.884) (-0.550) (3.792) (-2.850) (3.791) (-2.711)
Consumer Discretionary -0.0162 -0.1886** 0.0467*** -0.0006 0.0007*** -0.8322***
(-0.618) (-2.424) (10.206) (-0.270) (4.532) (-8.595)
Consumer Staples -0.0575 0.6926*** 0.0148 -0.0047 -0.0002 -0.1224
(-0.881) (2.982) (1.322) (-0.774) (-0.426) (-0.520)
Energy -0.0362 -0.1524 0.0261*** 0.0102*** 0.0003 -0.4505**
(-0.676) (-1.048) (2.637) (3.129) (1.478) (-2.269)
Health Care 0.0643*** -0.0240 0.0277*** 0.0012 0.0002 -0.4961***
(3.667) (-0.427) (7.167) (0.912) (1.541) (-6.398)
Industrials 0.0411* -0.0107 0.0243*** -0.0066*** 0.0004*** -0.3816***
(1.807) (-0.176) (7.167) (-3.043) (3.173) (-5.305)
Uncategorized Industry 0.0193 -0.0181 0.0207*** -0.0027* 0.0002** -0.2922***
(1.138) (-0.495) (6.598) (-1.756) (2.128) (-4.582)
Real Estate -0.3163 0.1590 0.0245 -0.0255 -0.0006 -0.1657
(-0.767) (0.301) (1.183) (-1.378) (-0.992) (-0.348)
Technology 0.0704*** 0.0277 -0.0006 -0.0021 0.0001 0.0495
(3.330) (0.483) (-0.151) (-1.498) (0.926) (0.632)
Telecommunications 0.0657 -0.1178 0.0134* 0.0030 -0.0000 -0.1958
(1.533) (-0.981) (1.958) (0.710) (-0.124) (-1.325)  
Source: Bloomberg 
Moreover, Table VI illustrates the results when total debt is measured by market values, thus, 
when TDMV has been used as the dependent variable. The industries which are statistically 
significant are Consumer Discretionary, Health Care, Industrials and Technology. The 
coefficients were estimated to 0.117, 0.114, 0.183 and 0.122, respectively, indicating intangible 
assets effect on total debt when measured by market values to be positive. In addition, 
significant results were also obtained for Uncategorized Industry. Compared to the regression 
with TDA as the dependent variable, Health Care, Industrials and Technology are statistically 
significant in both regressions. Another finding is illustrated by the relatively higher 
coefficients for TDMV than for TDA. This indicates that when market values are used for 
measuring leverage for total debt, intangible assets influence leverage to a larger extent. 
However, no significance can be attributed to the Energy industry, whilst Consumer 
24 
discretionary indicates a positive significant relationship to leverage compared to when TDA 
was used as dependent variable.  
 
Table VI: Secondary regressions of intangible assets as predictor of by industry  
T his table presents the result of Equation 2 from the section of methodology when TDMV is used as dependent variable. The independent 
variable of interest is intangible assets and is defined as intangible assets to total book value of assets. The results are estimated using the OLS 
 regression with the fixed effect model and the time period is between 2000-2020. Before the regressions are run, all the variables are 
winsorized by the 5th percentile and the 95th percentile to avoid misleading outliers. * indicates significance at 10 % level, ** indicates 
 significance at 5 % level and *** indicates significance at 1 % level. The t-statistics are presented in the parenthesis.  
 
 Total debt to market value of assets
 Industry Intangibles ROA Size MB Risk Constant
 Basic Materials 0.0327 -0.2054 0.0422*** -0.0204*** 0.0016*** -0.7660***
(0.806) (-1.440) (5.616) (-4.185) (6.121) (-4.679)
Consumer Discretionary 0.1165*** -0.0743 0.0482*** -0.0108*** 0.0014*** -0.8849***
(4.771) (-0.883) (9.495) (-4.255) (8.002) (-8.240)
Consumer Staples 0.0844* 0.2189 0.0192** -0.0274*** 0.0020*** -0.3191*
(1.733) (1.202) (2.265) (-5.394) (5.633) (-1.761)
Energy -0.0249 0.0620 0.0243*** -0.0076** 0.0003 -0.3714**
(-0.666) (0.467) (2.707) (-2.295) (1.362) (-2.028)
Health Care 0.1142*** -0.0013 0.0291*** -0.0011 0.0003*** -0.5229***
(7.255) (-0.026) (8.588) (-0.976) (2.637) (-7.727)
Industrials 0.1826*** -0.2279*** 0.0298*** -0.0216*** 0.0006*** -0.4731***
(8.815) (-3.098) (7.385) (-7.547) (4.586) (-5.500)
Uncategorized Industry 0.0530*** -0.0615 0.0473*** -0.0130*** 0.0010*** -0.8095***
(2.947) (-1.304) (12.231) (-6.240) (9.511) (-10.338)
Real Estate -0.2176 -0.5965 0.0326* -0.0499*** 0.0005 -0.2913
(-0.166) (-1.177) (1.646) (-2.908) (0.788) (-0.634)
Technology 0.1224*** -0.0819 0.0110** -0.0066*** 0.0004** -0.1722*
(5.693) (-1.142) (2.289) (-3.263) (2.554) (-1.802)
Telecommunications 0.0580 0.0175 0.0160** -0.0115** 0.0008*** -0.2530
(1.585) (0.133) (2.127) (-2.159) (3.089) (-1.567)  
Source: Bloomberg 
Lastly, Table VII presents the results when long term debt measured by market values are used 
as the dependent variable. Statistically significant coefficients were estimated for Health Care, 
Industrials and Technology, and the coefficients are 0.091, 0.082 and 0.096, respectively. 
Similar to LTDA, the regression for LTDMV indicates that long-term debt measured by market 
values are influenced by intangible assets positively in the Health Care, Industrials and the 
Technology industry. In addition, the similar pattern could be traced to smaller coefficients for 
when long-term debt measured by market values are considered compared to total debt, and, at 
the same time, larger coefficients for market values of debt compared to book values of debt 
for the significant industries (Health Care and Technology). 
 
25 
 
Table VII: Secondary regressions of intangible assets as predictor of by industry  
T his table presents the result of Equation 2 from the section of methodology when LTDMV is used as dependent variable. The independent 
variable of interest is intangible assets and is defined as intangible assets to total book value of assets. The results are estimated using the 
O LS regression with the fixed effect model and the time period is between 2000-2020. Before the regressions are run, all the variables are 
winsorized by the 5th percentile and the 95th percentile to avoid misleading outliers. * indicates significance at 10 % level, ** indicates 
s ignificance at 5 % level and *** indicates significance at 1 % level. The t-statistics are presented in the parenthesis.  
 
 Long term debt to market value of assets
 Industry Intangibles ROA Size MB Risk Constant
 Basic Materials 0.0364 -0.0935 0.0294*** -0.0164*** 0.0013*** -0.5312***
(1.014) (-0.731) (4.560) (-3.781) (5.388) (-3.759)
Consumer Discretionary 0.0284 -0.1034 0.0431*** -0.0082*** 0.0011*** -0.7941***
(1.293) (-1.370) (10.184) (-3.741) (6.777) (-8.792)
Consumer Staples 0.0456 0.2188 0.0120* -0.0209*** 0.0006* -0.1406
(0.995) (1.307) (1.720) (-4.308) (1.684) (-0.929)
Energy 0.0113 -0.0089 0.0090 -0.0030 0.0000 -0.1236
(0.421) (-0.093) (1.394) (-1.239) (0.168) (-0.937)
Health Care 0.0911*** -0.0151 0.0223*** -0.0002 0.0002* -0.4062***
(6.544) (-0.363) (8.015) (-0.186) (1.722) (-7.248)
Industrials 0.0821*** -0.0958 0.0216*** -0.0158*** 0.0003*** -0.3315***
(4.372) (-1.441) (6.325) (-6.108) (2.828) (-4.496)
Uncategorized Industry 0.0303** -0.0263 0.0327*** -0.0090*** 0.0005*** -0.5447***
(2.055) (-0.682) (10.280) (-5.268) (6.418) (-8.468)
Real Estate -0.3248 -0.5200 0.0127 -0.0541*** -0.0002 0.1074
(-0.239) (-0.967) (0.647) (-3.094) (-0.261) (0.235)
Technology 0.0963*** -0.0213 0.0077** -0.0031** 0.0002 -0.1285*
(5.900) (-0.397) (2.195) (-2.040) (1.598) (-1.831)
Telecommunications 0.0265 -0.0587 0.0125** -0.0090** 0.0002 -0.1827
(0.861) (-0.527) (2.325) (-2.023) (1.110) (-1.526)  
Source: Bloomberg 
Encapsulating the results, a clear majority of the statistically significant industries are 
indicating a positive relationship to leverage. As discussed earlier, the trade-off theory is 
predicting a negative association between intangible assets and leverage while our results are 
showing the opposite. However, the only results obtained that are in line with the trade-off 
theory is the one generated from Table IV for the Energy industry, when TDA is used as the 
dependent variable. Furthermore, higher coefficients were found when leverage was measured 
by market values. This question the appropriateness of the proxies. Frank and Goyal (2009) 
found less reliable results when book values were used for measuring leverage and argued that 
a possible explanation for this could be traced to that book values are backward looking. Debt 
is granted based on the firm's ability to repay the debt in the future and, therefore, leverage 
measured by market values of assets could be seen as a more superior proxy. Moreover, the 
most robust results are extracted from the Health Care and the Technology industry since it is 
statistically significant, and the direction of the association is equal for all proxies of leverage. 
26 
We emphasize the results of these industries to the similar environmental characteristics the 
industries are operating in. These industries are operating in an innovative environment, where 
new technologies are emerging that requires firms to use patents to protect their 
competitiveness. The importance of collateral has been highlighted in many empirical studies 
before (Frank and Goyal, 2009; Rajan and Zingales, 1995; Titman and Wessels, 1988; Gaud et 
al, 2005). However, the ability to pledge assets to secure debt is by these studies associated 
with tangible assets and not intangible assets, suggested by our results. However, as argued by 
Mann (2018), patents are useful for firms when securing financing which in that sense support 
our results in these industries. 
Based on our results from Tables IV-VII, we are able to, even though some of the coefficients 
are insignificant, observe that the importance of intangible assets can have a diverse effect on 
leverage, conditional to one specific proxy for leverage. The lion share of the significant 
coefficients is suggesting a positive association between intangible assets and leverage in the 
different industries, however, in the Energy industry the opposite could be observed from the 
results. This only applies for when TDA is used as the dependent variable, therefore, we would 
only be able to reject our second null hypothesis based on Table IV, however, when considering 
the total results this disparity between the associations is not completely evident. Despite the 
different magnitudes of the coefficients, it is doubtful that the results are able to represent any 
differences between industries with precision. 
4.3 Robustness  
Prior to attributing the estimated relationships to any concrete conclusions, it is also important 
to value the validity of the results. In order to strengthen the causality between leverage and 
intangible assets, we have introduced appropriate control variables to see if the influence of 
intangibles on leverage persist. The variables used for control were; ROA, Size and Market-to-
book-ratio and Risk which are included in the regression results (see Tables III-VII). These 
variables were introduced one by one to see if the significance of the results remained or if 
there were any major changes in the coefficients. The allowance of the control variables did 
neither affect the significance of the results nor caused any major changes to the coefficients. 
Therefore, the relationship does not seem to be spurious based on the used control variables. 
This reduces some of the endogeneity concern that important to consider. However, the reader 
should bear in mind that additional forces may have an impact on leverage, and we have not 
evaluated whether leverage could have any predicting power to intangibles (thus, whether 
27 
simultaneous equation bias exists) and is, therefore, important to consider. Furthermore, if the 
assumptions behind the OLS regressions are not fulfilled then the inference may not be 
completely correct. This can partially be evaluated by checking the residuals for normality as 
well as heteroskedasticity. The former to determine if the error-term is normally distributed or 
not and the latter if the error-term has a constant variance or not. The residuals plotted in 
histograms can be seen from Figures VII-X for our variables of interest. From this graphical 
illustration, our sample indicates to have problems with residuals being normally distributed, 
and all the independent variables are skewed to the right. To further clarify this statement, the 
Jarque-Bera test for normality was performed and confirmed that the residuals are not normally 
distributed, and the test can be found in Appendix IV. The violation of the normality 
assumption should be considered when interpreting the results and could have affected the 
estimated coefficients. However, this could be argued to not be a severe problem, since our 
sample is rather large. (Brook, 2015) 
 
 
Figure VII: Plotted residuals of TDA Figure VIII: Plotted residuals of LTDA 
 The residuals of the main regression with Total-debt-to-assets The residuals of the main regression with Long-term-debt-to-
ratio measured by book values of assets together with a drawn assets ratio measured by book values of assets together with a 
 line of a normal distribution. The regressions are performed drawn line of a normal distribution. The regressions are 
with winsorized variables by the 5th percentile and the 95th performed with winsorized variables by the 5th percentile and 
 percentile.  the 95th percentile.  
 
 
 
 
 
 
 
 
 
28 
 
Figure IX: Plotted residuals of TDMV Figure X: Plotted residuals of LTDMV 
 The residuals of the main regression with Total-debt-to-assets The residuals of the main regression with Long-term-debt-to-
 ratio measured by market values of assets together with a assets ratio measured by market values of assets together with drawn line of a normal distribution. The regressions are a drawn line of a normal distribution. The regressions are 
 performed with winsorized variables by the 5th percentile and performed with winsorized variables by the 5th percentile and the 95th percentile.  the 95th percentile.  
 
The graphical visualization of the residuals of the dependent variables together with predicted 
values for the dependent variables are illustrated in Figures XI-XIV. It could be observed from 
the figures that our data violates the assumption of homoscedasticity. The implications of 
heteroscedasticity become evident in the standard errors which could be misleading and is also 
important to consider when interpreting the results. However, the coefficients estimated will 
still be unbiased and consistent, but they will not be the best linear unbiased estimate. (Brooks, 
2015) 
 
 Figure XI: Scatter plot of residuals and Figure XII: Scatter plot of residuals and 
linear predictions of TDA linear predictions of LTDA 
 This scatter plot is visualizing the residuals together with the This scatter plot is visualizing the residuals together with the 
linear predictions of TDA for the main regression model linear predictions of LTDA for the main regression model 
 presented in Equation 1. Residuals are on the Y-axis presented in Equation 1. Residuals are on the Y-axis 
 an d prediction s are on th e   X  - a x i s . and predictions are on the X-axis. 
  
 
 
 
29 
 
Figure XII: Scatter plot of residuals and Figure XIII: Scatter plot of residuals and 
linear predictions of TDMV  linear predictions of LTDMV 
This scatter plot is visualizing the residuals together with the This scatter plot is visualizing the residuals together with the 
linear predictions of TDMV for the main regression mode l linear predictions of LTDMV for the main regression model 
presented in Equation 1. Residuals are on the Y-axis presented in Equation 1. Residuals are on the Y-axis 
and predictions are on the X-axis.  and predictions are on the X-axis. 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30 
5. Conclusions and further research 
5.1 Conclusions 
The field of research on capital structure is rather dense, however, how the importance 
intangible assets will affect this structure have yet to be fully explored, especially the 
importance of intangible assets. The Swedish market is fairly unexplored in this aspect, why 
this area becomes important to investigate. In an effort to contribute to this gap, this thesis 
examines the importance of intangible assets when considering the capital structure among 
Swedish firms. It is our belief that intangible assets will only continue to grow in importance 
and utilizing them when determining firm’s capital structure will be of key importance. Our 
first hypothesis regarding if intangible assets do in fact not support the issuance of debt on the 
Swedish credit market was possible to reject and we do conclude that intangible assets have a 
positive relationship with leverage. These results are rather contradicting to the traditional 
capital structure theories, especially the trade-off theory. However, we argue that due to the 
complex nature of capital markets of today and the emergence of the knowledge-based 
economy, traditional theories might not be completely suited to explain assets that in many 
ways did not exist when the theories were composed. The reasoning between why intangible 
assets increase the amount of leverage available for firms can be argued to be twofold; Firstly, 
in accordance with both Loumioti (2012) and Lim et. al (2020), the innovation of pledging 
intangible assets as collateral is something we argue will function as a driver of debt. Secondly, 
we as well as Larkin (2013) argue that intangible assets will have an incumbent value in the 
view of creditors as these assets in many cases are of great value for the operations of the firm 
and hence creates enhanced conditions by, for example, decreasing volatility of cash flows. 
Moreover, since we have found a stronger relationship between total debt compared to long-
term debt, we suspect that intangible assets could influence short-term debt as well. In addition, 
another interesting finding is the stronger relationship for intangible assets and leverage when 
market values are considered. 
 
In our second hypothesis, where differences in industries are the focal point, we find our results 
to be rather inconclusive and not providing us with a clear distinction to whether or not 
intangible assets foster issuance of debt differently across industries. Only one of our four 
different models, TDA, generated statistically significant differences between industries and, 
therefore, we are unable to reject the second hypothesis, even if we can observe tendencies in 
31 
some industries in which intangible assets are believed to affect the capital structure decisions 
more greatly than in others. Finally, the most robust results were generated for the Technology 
and Health Care industries, two of the industries with the largest proportion of intangible assets 
are found to be statistically significant and have a positive relation with leverage.  
5.2 Future Research  
There is still room for exploring the importance of intangible assets as a determinant of debt. 
First, it could be interesting to further examine different components of intangible assets and 
collect a more comprehensive set of data of specific intangible assets. This would be interesting 
to investigate, since different intangible assets may support different components of debt in 
various ways and has been evident from our results. Given the inherent nature of different 
intangible assets, some of them may be more or less suited to be pledged as collateral given 
their level of redeployability. Second, the Swedish market is rather specific and differs much 
from the US market, which is fairly explored. It could, in that sense, be valuable to examine 
other Scandinavian countries to see if differences could be observed. Third, it would be 
interesting to also investigate whether a difference in cost of capital can be observed when debt 
is backed by firms with mainly tangible versus intangible assets.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32 
References  
Baldwin, S. K. (1995). To Promote the Progress of Science and Useful Arts: A Role for 
Federal Regulation of Intellectual Property as Collateral. University of Pennsylvania Law 
Review, 143(5), 1701-1738. 
 
Baltagi, B. (2008). Econometric analysis of panel data. John Wiley & Sons. 
 
Bahrick, T. L. (1987). Security Interests in Intellectual Property. AIPLA QJ, 15, 30. 
 
Benmelech, E., Bergman, N., 2009. Collateral Pricing. Journal of Financial Economics 
91(3), p. 339-360. 
 
Berger, A. N., & Udell, G. F. (1995). Relationship lending and lines of credit in small firm 
finance. Journal of business, 351-381. 
 
Berk, J. B., & DeMarzo, P. M. (2014). Corporate finance. Harlow, England: Pearson. 
 
Boot, A. W., & Thakor, A. V. (1994). Moral hazard and secured lending in an infinitely 
repeated credit market game. International economic review, 899-920. 
 
Brooks, Chris, 2015, Introductory Econometrics for Finance, Third Edition, Cambridge 
University Press, UK.ISBN-13: 978-1107661455 
 
Frank, M. Z., & Goyal, V. K. (2008). Trade-off and pecking order theories of debt. Handbook 
of empirical corporate finance, 135-202. 
 
Frank, M. Z., & Goyal, V. K. (2009). Capital structure decisions: which factors are reliably 
important?. Financial management, 38(1), 1-37. 
 
Gaud, P., Jani, E., Hoesli, M., & Bender, A. (2005). The capital structure of Swiss 
companies: an empirical analysis using dynamic panel data. European Financial 
Management, 11(1), 51-69. 
 
Harris, M., & Raviv, A. (1991). The theory of capital structure. The Journal of Finance, 
46(1), 297-355. 
 
Hausman, J. A. (1978). Specification tests in econometrics. Econometrica: Journal of the 
econometric society, 1251-1271. 
 
Kraus, A., & Litzenberger, R. H. (1973). A state-preference model of optimal financial 
leverage. The journal of finance, 28(4), 911-922. 
 
Larkin, Y. (2013). Brand perception, cash flow stability, and financial policy. Journal of 
Financial Economics, 110(1), 232-253. 
 
Li, L., & Islam, S. Z. (2019). Firm and industry specific determinants of capital structure: 
Evidence from the Australian market. International Review of Economics & Finance, 59, 
425-437. 
 
33 
Lim, S. C., Macias, A. J., & Moeller, T. (2020). Intangible assets and capital structure. 
Journal of Banking & Finance, 105873. 
 
Loumioti, M. (2012). The use of intangible assets as loan collateral. Available at SSRN 
1748675. 
 
Mann, W. (2018). Creditor rights and innovation: Evidence from patent collateral. Journal of 
Financial Economics, 130(1), 25-47. 
 
Matemilola, B. T., & Ahmad, R. (2015). Debt financing and importance of fixed assets and 
goodwill assets as collateral: dynamic panel evidence. Journal of Business economics and 
Management, 16(2), 407-421. 
 
Melin, F., & Hamrefors, S. (2007). Den värdeskapande varumärkesstrategin, Sveriges 
informationsförening, Available online: 
https://sverigeskommunikatorer.se/globalassets/dokument/forskningsrapporter/reflektion_1.0 
7.pdf, Gathered [2021-01-08]. 
  
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the 
theory of investment. The American economic review, 48(3), 261-297. 
 
Modigliani, F., & Miller, M. H. (1963). Corporate income taxes and the cost of capital: a 
correction. The American economic review, 53(3), 433-443. 
 
Morellec, E. (2001). Asset liquidity, capital structure, and secured debt. Journal of financial 
economics, 61(2), 173-206. 
 
Myers, S. C. (1984). Capital structure puzzle (No. w1393). National Bureau of Economic 
Research. 
 
Myers, S. C., & Majluf, N. S. (1984). Corporate financing and investment decisions when 
firms have information that investors do not have (No. w1396). National Bureau of Economic 
Research. 
 
Myers, S. C., & Rajan, R. G. (1998). The paradox of liquidity. The Quarterly Journal of 
Economics, 113(3), 733-771. 
 
Ocean Tomo. 2020. Intangible Asset Market Value Study. Retrieved 2021-03-10 from: 
https://www.oceantomo.com/intangible-asset-market-value-study/ 
 
Parsons, C., & Titman, S. (2009). Empirical capital structure: A review. Now Publishers Inc. 
 
Powell, W. W., & Snellman, K. (2004). The knowledge economy. Annu. Rev. Sociol., 30, 
199-220. 
 
Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some 
evidence from international data. The journal of Finance, 50(5), 1421-1460 
 
Riksbanken. (2021). Historia. Retrieved 2021-04-30 from: https://www.riksbank.se/sv/om-
riksbanken/historia/ 
34 
 
Shleifer, A., & Vishny, R. W. (1992). Liquidation values and debt capacity: A market 
equilibrium approach. The Journal of Finance, 47(4), 1343-1366. 
 
Sibilkov, V. (2009). Asset Liquidity and Capital Structure. Journal of Financial and 
Quantitative Analysis, 44(05), 1173-1196.  
 
Skoogh, Jennifer, & Swärd, Philip. (2015). The Impact of Tangible Assets on Capital 
Structure - An Analysis of Swedish Listed Companies (Bachelor’s essay). University of 
Gothenburg. 
 
Stiglitz, J. and A. Weiss (1981). Credit rationing in markets with imperfect information. The 
American Economic Review 71(3), 393–410.  
 
Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. The Journal 
of finance, 43(1), 1-19. 
 
Tokmakcioglu, K., Tas, O., & Gozlu, S. (2007, August). Intangible Assets and Cost of 
Capital: An Application for a Technology Utilizing Firm. In PICMET'07-2007 Portland 
International Conference on Management of Engineering & Technology (pp. 45-51). IEEE. 
  
Williamson, O. E. (1988). Corporate finance and corporate governance. The journal of 
finance, 43(3), 567-591. 
 
World Bank. 2007. Building Knowledge Economies : Advanced Strategies for Development. 
WBI Development Studies. Washington, DC: World Bank. © World 
Bank.  https://openknowledge.worldbank.org/handle/10986/6853 License: CC BY 3.0 IGO  
 
 
 
 
 
 
 
 
 
 
35 
Appendix 
Appendix I 
Control variable Definition 
Return on Assets (ROA) EBITDA/Total Assets 
Size The natural logarithm of Total Assets 
Market to Book Value (MB) Market Value of Equity/Total Assets 
Risk Volatility the last 250 trading days 
 
Appendix II 
 The graph is illustrating the average leverage for each proxy The graph is illustrating the average leverage for each proxy 
between 2000-2020 in the Basic Materials industry. The between 2000-2020 in the Consumer Discretionary industry. 
 average is calculated post winsorizing the sample by the 5th The average is calculated post winsorizing the sample by the percentile and the 95th percentile. 5th percentile and the 95th percentile. 
  
 
 
 The graph is illustrating the average leverage for each proxy The graph is illustrating the average leverage for each proxy 
between 2000-2020 in the Consumer Staples industry. The between 2000-2020 in the Energy industry. The average is 
 average is calculated post winsorizing the sample by the 5th calculated post winsorizing the sample by the 5th percentile 
 percentile and the 95th percentile. and the 95th percentile. 
  
 
 
36 
 
 The graph is illustrating the average leverage for each proxy The graph is illustrating the average leverage for each proxy 
between 2000-2020 in the Health Care industry. The average between 2000-2020 in the Industrials industry. The average is 
 is calculated post winsorizing the sample by the 5th percentile calculated post winsorizing the sample by the 5th percentile 
 and the 95th percentile. and the 95th percentile. 
  
 
 
 The graph is illustrating the average leverage for each proxy The graph is illustrating the average leverage for each proxy 
between 2000-2020 in the Uncategorized industry. The between 2000-2020 in the Real Estate industry. The average is 
 average is calculated post winsorizing the sample by the 5th calculated post winsorizing the sample by the 5th percentile 
 percentile and the 95th percentile. and the 95th percentile. 
  
 
 
 The graph is illustrating the average leverage for each proxy The graph is illustrating the average leverage for each proxy 
between 2000-2020 in the Technology industry. The average between 2000-2020 in the Telecommunication industry. The 
 is calculated post winsorizing the sample by the 5th percentile average is calculated post winsorizing the sample by the 5th 
 and the 95th percentile. percentile and the 95th percentile. 
  
 
 
 
37 
Appendix III 
 The graph is illustrating the average intangible assets ratio The graph is illustrating the average intangible assets ratio 
 measured both by book values and market values between measured both by book values and market values between 
2000-2020 in the Basic Materials industry. The average is 2000-2020 in the Consumer Discretionary industry. The 
 calculated post winsorizing the sample by the 5th percentile average is calculated post winsorizing the sample by the 5th 
 and the 95th percentile. percentile and the 95th percentile. 
  
 
 The graph is illustrating the average intangible assets ratio The graph is illustrating the average intangible assets ratio 
 measured both by book values and market values between measured both by book values and market values between 
 2000-2020 in the Consumer Staples industry. The average is 
2000-2020 in the Energy industry. The average is calculated 
calculated post winsorizing the sample by the 5th percentile post winsorizing the sample by the 5th percentile and the 95th 
 and the 95th percentile. percentile. 
  
 
 The graph is illustrating the average intangible assets ratio The graph is illustrating the average intangible assets ratio 
 measured both by book values and market values between measured both by book values and market values between 
2000-2020 in the Health Care industry. The average is 2000-2020 in the Industrials industry. The average is 
 calculated post winsorizing the sample by the 5th percentile calculated post winsorizing the sample by the 5th percentile 
 and the 95th percentile. and the 95th percentile. 
  
 
 
38 
 The graph is illustrating the average intangible assets ratio The graph is illustrating the average intangible assets ratio 
 measured both by book values and market values between measured both by book values and market values between 
2000-2020 in the Uncategorized industry. The average is 2000-2020 in the Real Estate industry. The average is 
 calculated post winsorizing the sample by the 5th percentile calculated post winsorizing the sample by the 5th percentile 
 and the 95th percentile. and the 95th percentile. 
  
 
 
 The graph is illustrating the average intangible assets ratio The graph is illustrating the average intangible assets ratio 
 measured both by book values and market values between measured both by book values and market values between 
2000-2020 in the Technology industry. The average is 2000-2020 in the Telecommunication industry. The average is 
 calculated post winsorizing the sample by the 5th percentile calculated post winsorizing the sample by the 5th percentile 
 and the 95th percentile. and the 95th percentile. 
  
 
 
 
Appendix IV 
 
 
 
 
  
 
39 
 
40