Firm debt structure in times of unconventional monetary policy and low interest rates By Mike J. Beld

(1)

Firm debt structure in times of unconventional monetary policy and low interest rates By Mike J. Beld

ABSTRACT

The recent years have seen a new way of monetary policy in the European Union with extreme low interest rates and massive asset purchasing programmes. Using a firm level dataset I analyse how firms have changed their debt structure in response to the expansionary monetary policy of the ECB in the period 2011-2016. I also check whether bank-dependent firms respond differently to this event. I find that bank-dependent companies have lowered their bank debt ratios in the sample period and partly substituted this with nonbank debt. The reactions are strongest in the peripheral countries Italy, Spain, and Greece. I also find that smaller sized companies substitute less between various sources of debt. (JEL E52, G32)

Supervisor: prof. dr. K. F. Roszbach Co-assessor: dr. C. G. F. van der Kwaak

Institution: University of Groningen, Faculty of Economics and Business Submission: 7 June 2018

Course code: EBM000A20

(2)

(this page is intentionally left blank)

(3)

I. Introduction

Ample research has been done on how crises affect the credit supply and composition of corporate debt. Literature shows that in times of crisis a lack of external finance slows economic growth, especially among firms that rely heavily on debt financing (Kroszner et al., 2007).

Firms have two alternative possibilities to mitigate the harmful consequences of a credit crunch:

raise more equity capital or switch to other sources of debt. Fernández et al. (2018) explore the latter option. They show that during the financial crisis of 2008, bank-dependent firms that experienced a credit crunch saw an opportunity to partly evade this deterioration of banks’

willingness to lend by switching towards nonbank debt sources. More studies on this topic (Leary 2009, Becker and Ivashina 2014) find the same substitution away from bank debt towards other sources of debt, most common corporate bonds. Lyer et al. (2014) on the other hand find that especially smaller firms could not substitute towards other sources of debt after the 2008 financial crisis.

My thesis adds to the existing literature on debt substitution by analysing the substitution of bank and nonbank debt during the period of unconventional monetary policy following the 2008 financial crisis. I make a distinction based on the bank dependency of the individual firm to capture the particular reactions of firms with various debt structures. Firms’ debt is either bank or nonbank debt, with nonbank debt consisting of corporate bonds and nonbank private debt. The particular period is characterized by a recovering economy, unconventional monetary policy and low interest rates. Apart from the general knowledge that monetary policy affects the loan supply, it would be informative to know how then this affects the debt structure of an individual company. Recent research (Fernández et al., 2018) has shown that shifts in debt structure occur during a financial crisis. However, to my best knowledge, no research has yet been done on debt substitution during the period after a financial crisis.

I want to answer the following questions: First, did the loosening of the monetary stance in the period 2011-2016 lead to firms changing their corporate debt structure? Second, did the loosening of the monetary stance in the period 2011-2016 lead to a different change in debt structure for more bank-dependent firms?

A hurdle for this analysis is to make a distinction between supply and demand side

fluctuations. By only analysing changes in the composition of corporate leverage, and not

(4)

leverage itself I focus on supply side effects. The key assumptions that I make to isolate the supply side effects is that firms’ relative demand for bank or nonbank debt should not change.

This assumption has been previously used by Kashyap et al. (1993), Becker and Ivashina (2014) and Fernández et al. (2018); Furthermore, I focus my research on firms with various levels of bank dependency to find out whether bank-dependent firms react differently to changes in the monetary stance, using firms with lower reliance on bank debt as a control group. Last, I split up the sample in two groups around the median of firm size to see whether large and small firms show different behaviour in their debt decisions. Kashyap et al. (1993) find that there exists disparities in how these two groups determine their optimal debt structure.

Using a time series analysis I want to find out whether the loosening monetary stance has had an effect on how firms choose their debt structure. My dataset consists of 741 Firms from 8 countries in the European Union. Similar to Fernández et al. (2018), I calculate a firms’ bank dependency as the ratio of bank debt to total assets in 2010, one year prior to the start of my sample period in order to avoid endogeneity with the sample period itself. The 3-month interbank market rate is used as a proxy for the monetary stance. Additionally, I also include four firm control variables that are able to explain debt structure. I regress various types of debt ratios, which I fully explain below, on the proxy for the monetary stance and its interaction term with firms’ bank dependency. Further regressions also include several robustness tests and use various subsamples based on firm size, profitability, and growth opportunities.

My results show that expansionary monetary policy reduces the relative presence of bank debt among bank-dependent firms. The ratio of bank debt to total assets declines for the more bank-dependent firms, and a substitution towards other sources of debt takes place; Furthermore, I find that this reduction in bank debt ratios is strongest in the peripheral countries, Greece, Italy, and Spain. The results are robust to diverse ways of measurement. I also show that the results are not driven by supply side fluctuations in the loan and bond market. Last, I find that smaller companies experience more difficulties in substituting towards other sources of debt.

My findings are similar to those of Fernández et al. (2018), who find that after the 2008

financial crisis bank-dependent firms decreased their bank debt ratios and partly substituted this

with nonbank debt sources. In my paper I find a similar reduction in bank debt in the recovery

period after the crisis. My results are in line with the literature showing that a decline in interest

rates leads to profitable firms decreasing their amount of bank debt.

(5)

The thesis is structured as follows: section 2 discusses the relevant literature; section 3 explains the model; section 4 describes the data; section 5 presents the results; section 6 concludes.

II. Literature review

In this section I briefly summarize the existing literature, divided into three subjects:

Factors influencing the debt choice of a firm in general; the cyclicality of bank loans and corporate bonds; and the effect of monetary policy on firms’ debt decisions.

Factors influencing the debt choice of a firm

There is plenty literature on how firms make debt choices. From classical finance theory we know that firms more or less have three options for external finance: bank loans, corporate bonds and private nonbank debt; moral hazard and adverse selection are central in the availability of debt instruments; and firms always prefer the cheapest source of debt. Denis and Mihov (2003) conclude that the credit quality of the borrower is the primary determinant of the debt sources it uses. Firms with high credit ratings prefer bond financing, firms with mediocre credit ratings borrow from banks, and those with low credit ratings borrow from nonbank private lenders. A firms’ choice between bank and nonbank debt depends on the need for monitoring.

Banks are more efficient in monitoring companies than individual investors on the bond market.

Companies that need little monitoring, those with the highest credit ratings, therefore choose

public debt sources because of its cost advantage. A conclusion supported by Diamond (1991),

who further says that firms with mediocre credit ratings rely on bank loans; and firms with high

credit ratings tend to switch to bank loans when interest rates are high and future profitability is

low due to a higher demand for monitoring. This same conclusion is supported by a paper from

Cantilo and wright (2000), who observe an increase in the use of bank debt when interest rates

are high. Something that is contradicting the most important prediction from the bank lending

channel, which says that lower interest rates make bank financing a more attractive alternative.

(6)

On a supplementing note, Houston and James (1996) find that debt decisions depend on firms’ size, importance of growth opportunities, intangible assets, overall leverage, the number of bank relationships and whether it has access to public debt markets. Bank loans are more prevalent among the smaller firms with lower leverage. Suggesting that banks specialize in giving credit to smaller, less risky firms. Houston and James (1996) also point out that banks are able to “create durable transaction-specific information as part of an ongoing relationship.”

This information can have important benefits for both parties, although additional costs can be imposed on the borrower when one lending bank has the sole possession of this specific information.

The cyclicality of bank loans and corporate bonds

Bank loans and public debt follow a distinct cyclical pattern. Becker and Ivashina (2014) find that bank loans tend to be cyclical whereas the market for public debt is fairly stable over time. The relative amount of credit obtained from banks rises and falls in line with bank health and economic conditions. Firms substituting from bank debt to corporate bonds is therefore interpreted as evidence for deterioration of credit conditions. An improvement of credit conditions would likely raise the relative presence of bank loans as part of total firm debt.

The findings of cyclicality are supported by Kroszner et al. (2007) who conclude that sectors that highly depend on debt financing find themselves in greater troubles during a banking crisis in countries with relatively deep financial systems. Especially young firms and firms with many intangible assets experience these amplified troubles in raising new external funds.

Evidence of a substitution away from bank loans when credit conditions tighten is further underpinned by research done by Fernández et al. (2018). After the onset of the 2008 financial crisis, they find that firms partially substitute away from bank credit into multiple sources of nonbank debt: public debt and private debt. This substitution is dominantly prevalent in bank- dependent firms and firms with a public debt rating. A similar conclusion is found by Leary (2009), who notes that especially larger firms tend to switch away from bank loans towards public debt in times of tight credit.

Leary (2009) additionally finds that the leverage of bank-dependent firms follows the

same cyclical pattern as the bank loan market. These firms increase (decrease) their leverage by

(7)

taking out more (less) bank loans when the availability of bank loans increases (decreases). On the contrary, firms with public debt access do not show these large variations in their leverage in the same time periods. These findings suggest that the access to different debt markets is important for a firms’ capital structure and again supports the argument that firm size is an important indicator for the firms’ debt decision because it gives information about the bank dependency and bond market accessibility.

The effect of monetary policy on firms’ debt decisions

Monetary policy is key to the availability of bank loans to the corporate sector. This is the interpretation of Jiménez et al. (2012) from their research on the identification of the bank balance-sheet channel. The paper uses an interesting variable to proxy for the stance of monetary policy, the interbank market rate. A rise in the interbank market rate is depicted as a tightening of the monetary stance. The amount of loans granted to the corporate sector decreases (increases) after an increase (decrease) in the short term interbank market rate.

Kashyap et al. (1993) draw similar conclusions. Tighter monetary policy leads to firms issuing more commercial paper while the amount of bank loans declines. Meaning that contracting monetary policy can indeed reduce the amount of bank loans supplied. This paper finds additional evidence for the interest rate to be a good proxy for the monetary stance.

Duca et al. (2016) study the issuance of global corporate debt. Their conclusions are striking. They find that the large asset purchases from the US quantitative easing programme crowds out investors from the government bond market. The results support the “gap-filling”

theory from greenwood et al. (2010), where government bonds that would normally be in an investor’s portfolio are replaced by corporate bonds. This leads to a surge in demand for corporate bonds and hence a decline in yields.

Concluding remarks

Summarizing the above literature, I expect the above effects to have distinct impacts on

changes in bank debt ratios. First, low interest rates and high future profitability lead to firms

decreasing the amount of bank debt on their balance sheet. Second, looser credit conditions

(8)

induce firms to rise their bank debt holding. Last, looser monetary policy leads to firms issuing more bank loans, but also boosts the demand for corporate bonds.

The literature also gives some suggestions on the use of certain variables. First, the use of the interbank market rate as a proxy for the monetary stance is put forward. A decrease (increase) of the interbank market rate should be interpreted as a loosening (tightening) of the monetary stance. Next, the literature stresses the importance of including firm controls variables that can explain variance in firms’ debt structure. I include firm size, profitability, the amount of gtowth opportunities, and the relative presence of intangible assets in my regression.

III. Methodology

I am going to perform a time series analysis in which I regress the monetary stance on the structure of corporate debt during the period 2011-2016. I use the 3-month European interbank market rate as a proxy for the monetary stance. This period is characterized by a decline in the interbank market rate and thus an easing of the monetary stance. During the sample period, the interbank market rate declined by about one percentage point. The assumption I make to make this methodology viable is that there should not be a change in a firms’ relative demand for bank and nonbank debt. This same assumption is also used by Kashyap et al. (1993), Becker and Ivashina (2014) and Fernández et al. (2018).

A pitfall in this model is to ignore the homogeneity assumption implicit in this model as described by Kashyap et al. (1993). This assumption says that there is heterogeneity among firms in their decision to change their financing. Larger firms often rely more on nonbank funding and may not ever need to switch funding sources. Some smaller firms on the other that rely mostly on bank funding may not be able to switch to nonbank funding because it is too costly to issue corporate bonds. This means that there is no meaningful trade-off in choosing their preferred source of debt. I therefore also include a subsample regression based on the median of firm size to see whether these groups respond different to changes in the monetary stance.

I follow the approach of Fernández et al. (2018). The model controls for firm-level

variables that explain capital structure. Firm-fixed effects are used to capture unobservable

variation between companies regarding their decision on debt structure. The equation is as

(9)

follows:

Debt

i,t

= β

0

+ β

1

*IB

t

+ β

2

*IB

t

*Bankdep

i

+ β

3

*firmcontrols

it-1

+λ

j,t

+ θ

k,t

+ γ

i

+ μ

i,t

(1) Where the subscripts i, j, k, and t indicate firm, industry, country and year respectively. I use three definitions for the dependent variable Debt. It is the ratio of either bank or nonbank debt to total assets, or the ratio of bank debt to total debt. IB measures the year on year change in the three month interbank market rate and is a proxy for the monetary stance. It is calculated as the average value of the rate this year minus the average value of the rate last year. A higher interbank rate means a tighter monetary stance. A positive value for IB therefore means that the interbank rate has risen and that the monetary stance has tightened. Bankdep is the main proxy for bank dependency, it is calculated as the ratio of bank debt to total assets at the end of 2010.

Firmcontrols include a set of widely used firm-specific variables such as firm profitability, firm size, growth opportunities and asset tangibility, which I discuss in section IV. These firm controls are lagged by one year “to avoid simultaneity with corporate leverage variables”

(Fernández et al., 2018). Instead of using a wide variety of dummy variables, I include two factor variables with the same function as a ray of dummies in my model. I include industry-year (λ

jt

), country-year (θ

kt

), and firm-specific (γ

i

) fixed effects. These are both time variant and invariant and similar to the model of Fernández et al. (2018). These factor variables catch any industry or country specific shocks in a specific year. The factor variables should control for most omitted variables.

In my regression, β

1

captures the effect of changes in the interbank market rate on the various debt ratios of firms. β

2

then captures the different effect of companies that are more bank-dependent. The sign of these coefficients depends upon several factors that influence the firms’ decision for a particular type of debt. First, according to the literature, a tightening (loosening) of credit conditions leads to firms decreasing (increasing) their bank debt ratios.

Credit conditions in Europe have not seen a particular trend towards either looser or tighter credit in the period 2011-2016, according to the ECB bank lending survey. Changes in credit conditions will therefore only marginally affect the relative demand for bank and nonbank debt.

Second, looser monetary policy will lead to a boost in the demand for both bank loans and

nonbank debt. With the increase in the latter due to the “gap-filling” theory. Last, literature finds

that a decrease (increase) in the interest rate and high (low) future profitability will lead to firms

(10)

preferring corporate bonds over the use of bank debt. This means that I expect the ratio of bank debt to total assets to decline, and the ratio of nonbank debt to total assets to increase. Since firms with little bank debt cannot reduce their bank debt by much, I expect that only bank- dependent firms are affected by this.

Hence, I expect β

1

to be positive when bank debt and nonbank debt to assets are the dependent variables, due to an overall surge in demand for debt. I expect the sign of β

1

to be ambiguous when bank debt to total debt is the dependent variable, because it is not clear which type of debt will see the largest growth in demand. Next, I expect the coefficient β

2

to be positive when bank debt to total assets and bank debt to total debt are the dependent variables. I expect β

2

to be negative when nonbank debt to total assets is the dependent variable. Formulating this into a hypothesis, I am going to test the following:

Hypothesis 1: H

0

: Unconventional monetary policy in the period 2011-2016 has not led to firms changing their debt structure.

H

1

: Unconventional monetary policy in the period 2011-2016 has led to firms changing their debt structure.

Hypothesis 2: H

0

: Unconventional monetary policy in the period 2011-2016 has not led to a change in debt structure for bank-dependent firms.

H

1

: Unconventional monetary policy in the period 2011-2016 has led to a change in debt structure for bank-dependent firms.

IV. Data and variables

In this section I describe the dataset I use in my research as well as the most important variables that I include in my model.

Data

I combine data from various sources. All balance sheet and profit and loss items for

publicly traded firms are available on Orbis. Orbis reports bank debt as a separate item as part of

the companies’ long term debt. The data does not distinguish between any sorts of nonbank debt.

(11)

The focus of my study therefore lies on changes in long term bank and nonbank debt ratios of individual companies. Data on the average Euribor rate is extracted from global-rates.com.

My research focuses on the ECB quantitative easing programme, and subsequently the decline in the Euribor rate that is has produced. At first, I select all 19 countries from the Euro area. I delete 11 countries that have less than 35 firms in the sample. This results in 8 countries being included in my research.

The analysis focusses on the period 2011-2016. Since I use lagged firm specific data I only use firms that have complete year-on-year data for the period 2010-2016. This is the period just after the financial crisis of 2008. In order to avoid any effects from the crisis itself, I start collecting data from 2010. Since I need one year lagged data for the control variables, my sample period starts in 2011. I take 2016 as the end date of the period because this is the most recent data available. Firms with negative book and market equity, negative assets and negative debt as of December 2011 are excluded. Additionally, following the methods from Fernández et al.

(2018), financial industry firms (SIC codes 6000-6999), regulated enterprises (SIC codes 4000- 4999) and governments enterprises (SIC codes above 8000) are exempted due to their special characteristics. Eventually, I select firms that belong to the 20 industrial sectors with the most observations, based on the first two digits of its SIC code. I exempt industries with insufficient observations because the inclusion of industry specific shocks in the model requires all sectors to have sufficient observations.

In my research I compare the development of firms’ bank and nonbank debt to total assets in comparison with changes in the European interbank market rate. According to previous research by Jiménez et al. (2012), Kisan and Opiela (2000), and Jayaratne and Morgan (2000) the interbank market rate can act as a proxy for the monetary stance. Positive (negative) changes in the interbank market rate mean a contraction (expansion) of the monetary stance. Using a time series analysis I try to capture an effect on the relative presence of bank and nonbank debt levels resulting from changes in the interbank market rate.

Variables

I now fully explain the proxies for the main variables that are suggested by Fernández et

al. (2018): bank dependence, debt structure, and firm control variables.

(12)

Firm bank dependency

The ratio of bank debt to total assets at the start of my sample period proxies for the bank dependency of firms. Following the same methods as Duchin et al. (2010), Kahle & Stulz (2013), and Fernández et al. (2018), I classify firms as bank-dependent immediately before the start of my sample period so that the bank dependency cannot be endogenous to events that happened in the sample period itself. I assume that a larger value of relative bank debt to total assets means that the company is relatively more bank-dependent. Column 8 of panel A in Table 1 shows the mean level of bank debt to assets at the end of 2010 by country. In 2010, the mean level of bank debt to assets was 7.5% for the whole sample. There is heterogeneity in the mean values of bank dependency among countries. Spain and Italy are countries with high levels of bank debt at the end of 2010, whereas Dutch and French companies have relatively little bank debt on their balance sheet.

Debt structure

I use three variables to measure the debt structure of individual companies: bank debt to total assets, nonbank debt to total assets and bank debt to total debt. Bank debt to total assets and nonbank debt to total assets are the main variables used to provide information about changes in the firms’ debt structure. The ratio of bank debt to total debt measures the changes in relative size of both categories. Panel A of Table 1 shows the per country mean values for these variables.

The average percentages of bank debt to total assets, nonbank debt to total assets and bank debt

to total debt in panel B are 6.6%, 7.4% and 49.0% respectively for the whole sample. Columns 3,

5 and 7 in panel A show the changes in ratios between the years 2011 and 2016 per country. I

find that in six out of eight countries the relative presence of bank debt has shrunk in the sample

period, although this effect is only significant for Italy. Overall I find a significant cutback of

bank debt ratios of 0.8% for the whole sample. Nonbank debt ratios have risen significantly in

(13)

six countries. For the whole sample, nonbank debt ratios have shown a significant gain of 2.1%.

In four countries nonbank debt ratios increased at the cost of bank debt ratios, which is shown by a significant negative change in bank debt to total debt in column 7 of panel B.

Table 1

Descriptive statistics of debt structure and changes during the sample period 2011-2016.

Panel A shows the by country mean values of the main variables. Bank debt to assets measures the ratio of bank debt to total assets. Nonbank debt to total assets measures the ratio of nonbank debt to total assets. Bank debt to total debt measures the ratio of bank debt to total debt.

Bankdebt10 is the ratio of bank debt to total assets at the end of 2010. Column 3, 5 and 7 show the differences between 2016 and 2011 for the above explained variables. Panel B shows the mean values for both the whole sample as well as subsamples based on whether the firm is above or below the by country median of bank dependence.

Panel A: Debt structure by country

firms

Bankdebt to assets

bank debt to total assets D 2016-2011

Nonbank debt to assets

nonbank debt to assets D 2016-2011

Bank debt to total debt

Bank debt to total debt D

2016-2011 Bankdep

1 2 3 4 5 6 7 8

Belgium 40 0.0486 -0.0141 0.0962 0.0302* 0.4046 -0.1651** 0.0704

Germany 194 0.0583 -0.0100 0.0666 0.0121* 0.5053 -0.0665* 0.0636

Spain 48 0.1483 -0.0055 0.0509 0.0263* 0.7643 -0.0754 0.1401

Finland 54 0.0985 0.0022 0.0662 0.0102 0.5837 0.0263 0.0954

France 220 0.0526 0.0034 0.0757 0.019** 0.4304 -0.0471 0.0556

Greece 67 0.0518 -0.0106 0.0932 0.0131 0.3967 -0.0263 0.0618

Italy 82 0.0842 -0.0301** 0.0653 0.0507*** 0.6128 -0.2312*** 0.1298

Netherlands 36 0.0405 -0.0226 0.1125 0.0334* 0.2538 -0.1427* 0.0412

Panel B: Debt structure and financial dependence of firms Bankdebt to

assets

bank debt to total assets D 2016-2011

nonbank debt to assets D 2016-2011

2016-2011 Bankdep

1 2 3 4 5 6 7

Mean - whole sample 0.0662 -0.0079** 0.0743 0.0211*** 0.4898 -0.0788*** 0.0750

Mean - firms with high

bank dependence 0.1027 -0.0376*** 0.0680 0.0409*** 0.6310 -0.2334*** 0.1396

mean - Firms with low

bank depence 0.0328 0.0200*** 0.0771 0.0022 0.3636 0.0800* 0.0148

*** significance level of 1%

** significance level of 5%

* significance level of 10%

Divergent results occur when I split up the sample in two groups depending on the bank

dependency of the firm. Firms that were above the by country median of bank dependency,

measured in 2010, lowered their bank debt ratios significantly. Bank-independent companies on

the other hand, actually boosted their bank debt ratios, shown in column 2 of panel B. Column 6

of panel B shows that bank-dependent firms actually substituted towards nonbank debt, since

they decreased the amount of bank debt to total debt by 23.3 percentage points. Firms with little

bank dependency did the exact opposite and augmented the relative presence of bank debt on

their balance sheet.

(14)

Figures 1 and 2 visualize the bank and nonbank debt ratios over time by country. Figure 1

shows the changes of bank debt and nonbank debt ratios over time for firms with high bank

dependency. Figure 2 shows the same information for firms with little bank dependency. In

every country a clear downward trend is visible in the bank debt ratios of high bank-dependent

countries, whereas nonbank debt ratios have increased in every country. The figure suggests a

substitution

(15)

Figure 1. The ratio of bank debt to assets and nonbank debt to assets for firms with high bank dependence in the period 2011-2016.

0.05.1.15

Belgium Finland France Germany Greece Italy Netherlands Spain

11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 Bank debt to assets

0.05.1.15.2

11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16

(16)

Figure 2. The ratio of bank debt to assets and nonbank debt to assets for firms with low bank dependence in the period 2011-2016.

Bank debt to assets

0.05.1.15

11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16

0.05.1.15

11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16 11 12 13 14 15 16

(17)

between different sources of debt being present at all countries, especially in Belgium and Italy.

For firms with little bank dependency, an opposite trend is visible in every country, with bank debt ratios rising in all countries. Nonbank ratios have shrunk in only three countries, meaning that substitution played a smaller role among firms that are less bank-dependent. Although only descriptive statistics, the figures give a good overview of the existing trends among countries.

Table 2 presents the summary statistics for two subsamples based on the by country median of firm size. Some interesting numbers can be observed from this table. There is a clear distinction in how small and large sized firms changed the composition of their debt during the sample period.

At the start of the sample period, bank debt levels of smaller and larger sized firms were rather similar. During the sample period, smaller firms did not change their bank debt levels, whereas larger firms lowered their bank debt levels by 2.14 percentage points on average. Splitting up the two subsamples around the median of bank dependency reveals that the pattern previously found is still present. Bank-dependent firms decreased bank debt levels, whereas firms with little dependency on bank debt boosted their bank debt levels. Nonbank debt has higher presence among larger sized countries. Changes in nonbank debt levels have been similar for both small and large firms; Furthermore, column G shows that substitution away from bank debt and towards nonbank debt is stronger in larger sized firms (-11.16 percentage points) than in smaller sized firms (-4.35 percentage points).

Firm control variables

I use four firm control variables similar to Fernández et al (2018). First, profitability proxies for higher tax benefits and more opportunities to fund investments with internal sources.

It is calculated as the earnings before interest and taxes plus depreciation divided by the companies’ total assets. Second, the market to book asset ratio (Tobin’s Q), a proxy for the growth opportunities of a company. It is calculated as the market value of assets divided by the book value of assets. Third, firm size, proxying for lower bankruptcy costs and less information asymmetry. It is calculated as the logarithm of total assets. Lastly, I include asset tangibility.

This variable is calculated as the firms’ net property, plant and equipment divided by its total

(18)

assets. This variable can be used as a proxy for financial risks, as more collateral is present in the case of a bankruptcy.

Table 2

Descriptive statistics of debt structure and changes during the sample period 2011-2016: subsample based on the by country median of firm size.

Bank debt to assets measures the ratio of bank debt to total assets. Nonbank debt to total assets measures the ratio of nonbank debt to total assets.

Bank debt to total debt measures the ratio of bank debt to total debt. Bankdebt10 is the ratio of bank debt to total assets at the end of 2010.

Column 3, 5 and 7 show the differences between 2016 and 2011 for the above explained variables.

Panel A: firm size below the by country median Bankdebt to

assets

bank debt to assets D 2016-2011

nonbank debt to assets D 16- 11

2016-2011 Bankdep

1 2 3 4 5 6 7

Mean - whole subsample 0.0666 0.0060 0.0562 0.0190*** 0.5428 -0.0435* 0.0654

Mean - firms with high

bank dependence 0.1056 -0.0177** 0.0520 0.0388*** 0.7060 -0.1962*** 0.1286

bank depence 0.0355 0.0246*** 0.0590 0.0025 0.4114 0.1005** 0.0151

Panel B: firm size above the by country median Bankdebt to

assets

bank debt to assets D 2016-2011

nonbank debt to assets D 16- 11

2016-2011 Bankdep

1 2 3 4 5 6 7

Mean - whole subsample 0.0654 -0.0214*** 0.0921 0.0237*** 0.4397 -0.1116*** 0.0836 Mean - firms with high

bank dependence 0.0997 -0.0540*** 0.0812 0.0429*** 0.5697 -0.2618*** 0.1480

bank depence 0.0298 0.0145*** 0.098 0.0019 0.3121 0.0538 0.0145

V. Results

In the section I describe the most important results. First, I dive into the substitution between bank and nonbank debt. Second, I discuss whether the results found can be due to changes from the demand side instead of the supply side. Last, I discuss the disparities between larger and smaller sized firms.

Substitution between bank and nonbank debt.

(19)

In Table 3 I report the results of equation (1) presented in section III. I estimate the model using three different dependent variables: the bank debt ratio, nonbank debt ratio, and the bank debt to total debt ratio. I include the four firm control variables profitability, Tobin’s Q, firm size and tangibility of assets in all regressions. First, I only include the Δ interbank rate as the main variable in the regression. These results are shown in column 1, 3 and 5. The coefficients in these regressions show what effect changes in the interbank market rate, which proxy for the monetary stance, have on the presence of the particular type of debt for the sample overall. Next I also include the interaction term Δ interbank rate bankdep, shown in column 2, 4 and 6. Now, the* coefficient Δ interbank rate captures the effect of the interbank market rate on changes in the various ratios for firms without any bank debt. The coefficient Δ interbank rate bankdep then* captures the differential effect of the interbank market rate on changes in the above three ratios for firms that are bank dependent. Since the interbank market rate declines in my sample period, and the bank dependency of the firm is strictly positive, a positive coefficient for these variables thus implies that the presence of this type of debt has fallen

None of the coefficients of the Δ interbank rate variable in columns 1, 3 and 5 are significant. This means that in the sample period the interbank market rate is unable to explain any of the changes in firms’ corporate debt structure for the sample overall after controlling for the four specific firm characteristics. Based on this result, I cannot reject the H

0

of hypothesis A.

Based on the overall sample, loosening of the monetary stance has not led to a change in firms’

debt structure in general. A reason for this is the 2010-2012 sovereign debt crisis: although firms wanted to raise their debt levels, they were not able to do so because banks and investors were unwilling to provide credit in these years. Supplementing on this theory is the opposite behaviour of firms in changing their bank and nonbank debt ratios, depending on their bank dependency, which is visible in column 2 and 5 of Table 1. Firms with little bank debt raised their bank debt ratios and lowered their nonbank debt to total assets ratio in the sample period. Firms that are more bank dependent decrease their bank debt ratios and expanded their nonbank ratios in the sample period. Since these effects are offsetting each other, I do not find a significant result when I do not control for the variation in bank dependency between firms.

After controlling for the bank dependency of the firm, all the coefficients of Δ interbank

rate in column 2, 4 and 6 remain insignificant. This means that the monetary stance does not

affect the debt structure of firms that are not bank dependent.

(20)

In column 2 and 6 I find a significant positive coefficient for the variable Δ interbank rate

* bankdep. This indicates that during the sample period, firms with a high bank dependency ratio

were actually affected by changes in the monetary stance. These firms shrank their ratios of bank

debt to assets, observed in column 2. A 100 basis points decline in the interbank market rate

hence

(21)

Table 3

Changes in bank and nonbank debt during the sample period 2011-2016.

This table shows the regression results for the effects of monetary policy on firms’ debt structure. There are thee dependent variables: Bank debt to assets, nonbank debt to assets and bank debt to total debt. Δ interbank rate is a measure of the change in the European 3-month interbank market rate, which proxies for the monetary stance. Bankdep is the ratio of bank debt to total assets measured at the end of 2010. Profitability is the ratio of earnings before taxes plus depreciation divided by total assets. Q is the measure for growth opportunities, it is calculated as the ratio of market value of assets to book value of assets. Size is calculated as the natural logarithm of total assets. Tangibility is calculated as the ratio of net property, land and equipment over total assets and measures the tangibility of the assets. The results also include industry-year and country- year fixed effects. The standard errors are clustered at the country level. The t statistics are between brackets.

Bank debt to assets nonbank debt to assets bank debt to total debt

1 2 3 4 5 6

Δ interbank rate -0.003 -0.0068 -0.0034 -0.0006 -0.135 -0.0357

(-0.22) (-0.46) (-0.20) (-0.03) (-0.15) (-0.4)

Δ interbank rate*bankdep 0.1034* -0.076 0.3042*

(2.25) (-1.55) (2.3)

profitability -0.0075 -0.0075 -0.0678*** -0.0678*** 0.0945 0.0943

(-0.52) (-0.55) (-3.61) (-3.69) (0.98) (0.98)

Q -0.0032 -0.0031 0.0096** 0.0095** -0.0154 -0.0149

(-0.87) (-0.83) (2.37) (2.36) (-0.72) (-0.69)

Size 0.0132*** 0.0127** 0.003 0.0034 0.0374 0.0361

(3.03) (2.96) (0.65) (0.72) (1.90) (1.79)

Tangibility 0.0208 0.0205 0.027 0.0272 -0.1072 -0.1085

(0.41) (0.40) (0.65) (0.65) (-0.57) (-0.58)

Industry-year Yes Yes Yes Yes Yes Yes

country-year Yes Yes Yes Yes Yes Yes

cluster country Yes Yes Yes Yes Yes Yes

R^2 0.061 0.0669 0.0647 0.0676 0.0728 0.0749

#obs 4446 4446 4446 4446 4298 4446

#firms 741 741 741 741 733 741

results in a 1.24 percentage points reduction in the bank debt to assets ratio for a firm in the

highest quarter of bank dependency, compared to no reduction for a firm in the lowest quarter of

bank dependency, which is a firm without bank debt. The results imply that more bank-

dependent firms do indeed show a different change in debt structure after changes in the

monetary stance. The positive coefficient found in column 6 for the variable Δ interbank rate *

bankdep means that bank-dependent firms have substituted their reduction in bank debt levels by

increasing their levels of nonbank debt. The above firm with a bank dependency level in the

highest quarter of the sample reacts to a 100 basis points cut in the interbank market rate by

decreasing its bank debt to total debt ratio by 3.6 percentage points more than the firm in the

lowest quarter of bank dependency. Firms that are more bank-dependent lower their bank debt

ratios more than firms that are less bank dependent if the monetary stance loosens. Therefore, I

reject the H

0

of hypothesis B and can conclude that there is indeed a different reaction to the

loosening of the monetary stance depending on the bank dependency of the individual firm. One

explanation for this finding is found in the literature, which says that firms switch away from

(22)

bank debt financing when interest rates are low and future profits high (Diamond, 1991). In this scenario bond financing becomes more attractive. Since firms that depend highly on banks profit more by changing their debt structure, these firms face the strongest incentives to do so. Another explanation is given by Denis and Mihov (2003), who note that “the primary determinant of the debt source is the credit quality of the issuer.” Since lower interest rates can improve the credit quality of the borrower because it lowers the interest burden, this can lead to the firm preferring corporate bonds over bank debt.

Most of the firm control coefficients in Table 4 are in line with theory predictions, although not all are significant in my regression. Profitability is negative and significant when nonbank debt is the dependent variable. An out of table regression also shows that total leverage reduces with profitability. This suggests that more profitable firms depend less on nonbank financing and on debt in general. This is in line with the pecking order theory of corporate finance which says that internal funds are used as the primary source of finance. The coefficient of Tobin’s Q, which proxies for the growth opportunities of a company, is only significant when the nonbank debt ratio is the dependent variable. The value is positive and in line with the pecking order theory because “higher growth opportunities increase financing needs and the relevance of information asymmetries in the firm” – Fernández et al. (2018). The coefficient for size is positive in all regressions but only significant when the bank debt ratio is the dependent variable. The positive relation is explained by Rajan and Zingales (1995), who argue that larger firms have a lower probability to default, implying that they can take on more leverage. The coefficient of tangibility is positive as expected, but not significant. A higher ratio of tangible assets lowers the risk of the company’s debt due to higher collateral value (Rajan and Zingales, 1995), meaning that a firm can have more leverage.

In Table 4 I present the results of the robustness tests. I only show the results of the

regression that includes both the change in the interbank market rate and its interaction term with

bank dependency. The robustness tests in column 1 and 4 show the results when excluding the

country-year fixed effects from the regression. What stands out is the significant negative

coefficient of Δ interbank rate of -0.0152 in column 1. This implies that a firm that is not bank-

dependent raises its debt ratio by 1.52 percentage points after a cut in the interbank market rate

by 100 basis points. It shows the opposite reaction of these two types of firms on changes of the

(23)

monetary stance. The result suggests that the country-year fixed effect partly absorbed the effect

of changes in the interbank market rate on the bank debt ratios of companies, probably due to the

(24)

Table 4

Changes in bank and nonbank debt during the sample period 2011-2016: robustness tests

Bank debt to

assets nonbank debt to assets

Without country- year fixed effects

without clustering

SE Periphery

dummy

Without country-year fixed effects

without clustering

SE Periphery

dummy

1 2 3 4 5 6

(Δ)interbank rate -0.0152* 0.0329 0.0192 0.0101 -0.1819 0.0099

(-1.96) (0.31) (0.72) (0.61) (-0.84) (0.59)

(Δ) interbank rate*bankdep 0.1125* 0.1034*** 0.0437* -0.1006 -0.0760*** -0.0179

(2.01) (3.81) (2.17) (-1.67) (-3.49) (-0.70)

Δinterbank rate*periphery -0.0453*** 0.0184

(-3.5) (1.5)

Δ interbank

rate*bankdep*periphery

0.1940* -0.1892*

(2.06) (-2.07)

profitability 0.0009 -0.0075 -0.0071 -0.0702*** -0.0678*** -0.0682

(0.07) (-0.28) (-0.051) (-4.02) (-2.6) (-3.58)

Q -0.0040 -0.0031 -0.0030 0.0094* 0.0095* 0.0094*

(-1.09) (-0.70) (-0.81) (2.24) (1.87) (2.34)

Size 0.0120** 0.0128* 0.0127** 0.0029 0.0034 0.0034

(2.97) (1.74) (2.88) (0.56) (0.43) (0.72)

Tangibility 0.0191 0.0205 0.0222 0.0248 0.0272 0.0255

(0.38) (0.51) (0.44) (0.61) (0.68) (0.61)

country-year No Yes Yes No Yes Yes

cluster country Yes No Yes Yes No Yes

R^2 0.0434 0.0669 0.0714 0.0507 0.0676 0.0714

#obs 4446 4446 4446 4446 4446 4446

#firms 741 741 741 741 741 741

simultaneous change of the interbank rate in all countries. Next, due to the construction of the standard error clustering, the coefficients itself in column 2 and 5 do not change when standard errors are not clustered, only the standard error itself changes. The coefficient Δ interbank rate * bankdep becomes more significant in both column 2 and 5 of Table 4. Nonbank debt levels now show a significant increase after a decline of the interbank market rate.

I now test the model after including a periphery dummy in the model. The results are in

Table 4. In an economic letter, Nechio (2011) writes that large divergences exist between the so

called “core” and “peripheral” countries in the European Union. With peripheral countries

having both larger unemployment gaps and higher inflation rates. Inclusion of a periphery

dummy should capture a potential peculiar effect of the monetary stance on the firms’ debt

(25)

decisions in these peripheral countries. I designate three out of eight countries as peripheral: Italy, Spain and Greece. I find that these peripheral countries react strongest to changes in the interbank market rate. This is shown in column 3 and 6 of Table 4. The coefficients of the variables that interact with the periphery dummy are significant and higher than those in the original equation. This means that the previous findings that bank-dependent companies reduced their bank debt ratios in times of decreasing interest rates is particularly present in the periphery countries. Firms’ in these countries have subsequently seen a stronger enlargement in nonbank debt ratios. It highlights the contrasting effects monetary policy has among the various countries in the European Union. These three peripheral countries suffered more from the financial crisis and were in deeper need for looser monetary policy. My results capture this different reaction.

Changes in the demand from the firm side and substitution between bank and nonbank debt

Following Fernández et al. (2018) I analyse the data to see whether the shift in bank and nonbank debt ratios may be due to changes in firms’ demand for a particular type of credit instead of from changes in the monetary stance. According to the literature, this should not be the case because changes in a firms’ demand for credit do not change the firms’ relative demand for a particular type of credit, both bank and nonbank debt. I split up the sample in two groups around the median level of profitability and investment opportunities (Q). I take the average profitability and Q of the firm over the whole sample period to decide whether the firm is above or below the by country median level.

I report the results in Table 5. According to the insights of Fernández et al. (2018), firms with profitability and investment opportunities above the median “would be less affected by reductions in demand for credit”, because these firms have higher demand for finance in general.

This implies that the substitution between bank and nonbank debt has to be lower in firms with

high Q and profitability if it is driven by a reduction from the demand side. In column 6 of Table

5, the variables Δ interbankrate and Δ interbankrate bankdep are both higher and more*

significant for firms with profitability above the median (panel B) than for companies with

profitability below the median (panel A), implying lower substitution in the latter. With the

insight of Fernández et al. (2018) it thus means that the results are not driven by changes in

demand for a particular type of credit. The result for Tobins’ Q gives inconclusive evidence. The

(26)

coefficient of Δ interbankrate bankdep in column 3 is higher for firms’ with low Tobins’ Q,* although it is not

Table 5

Changes in bank and nonbank debt during the sample period 2011-2016: Subsamples based on the median of tobin’s Q and profitability.

Panel A: below the median of Tobin's Q and profitability

Q low Profit low

1 2 3 4 5 6

Bank debt to

assets nonbank debt to

assets bank debt to

total debt Bank debt to

assets bank debt to

total debt

Δ interbank rate 0.0201 -0.0435 -0.0208 -0.0016 -0.0004 -0.0835

-0.4200 (-1.40) (-0.7) (-0.17) (-0.04) (-1.01)

Δ interbank

rate*bankdep 0.1006 -0.0735 0.3933 0.0919 -0.0813 0.3162

(1.54) (-1.15) (1.73) (1.01) (-1.02) (1.13)

profitability 0.0658** -0.0941* 0.3830* 0.0046 -0.0954 0.2310

(2.36) (-2.20) (2.29) (0.24) (-5.18) (1.50)

Q 0.0012 0.0184 -0.0911 -0.0023 0.0118 -0.0510

(0.16) (1.18) (-1.12) (-0.33) (1.42) (-0.92)

Size 0.0179* 0.0111* 0.0409 0.0141 -0.0033 0.0388

(1.99) (2.17) (1.10) (1.45) (-0.51) (0.86)

Tangibility 0.0495 0.0446 -0.0292 -0.0227 0.0178 -0.1333

(1.05) (0.62) (-0.11) (-0.33) (0.27) (-0.48)

R^2 0.1086 0.0952 0.1116 0.0847 0.0813 0.0895

#obs 2100 2100 2056 2340 2340 2247

#firms 350 350 348 390 390 386

Panel B: above the median of Tobin's Q and profitability

Q high Profit high

1 2 3 4 5 6

Bank debt to

assets bank debt to

total debt Bank debt to

assets bank debt to

total debt

Δ interbank rate -0.0145 0.0154 -0.0142 -0.0487** 0.0135 -0.1952**

(-0.37) (0.54) (-0.17) (-2.81) (0.86) (-2.40)

Δ interbank

rate*bankdep 0.0958* -0.0741 0.2308** 0.1175*** -0.0855** 0.3507**

(2.18) (-1.78) (2.31) (4.50) (-2.67) (3.00)

profitability -0.0560 -0.0577*** -0.0714 -0.0305 0.0255 -0.2067

(-1.80) (-5.01) (-0.62) (-0.73) (0.63) (-0.88)

Q -0.0023 0.0022 0.0107 -0.0032 0.0048 0.0203

(-0.33) (0.61) (0.45) (-0.90) (0.84) (0.58)

Size 0.0091 -0.0065 0.0469 0.0210* 0.0158* 0.0445*

(1.01) (-0.86) (1.80) (2.15) (2.03) (2.06)

Tangibility 0.0108 -0.0100 -0.0991 0.0950 0.0297 -0.1010

(0.14) (-0.25) (-0.69) (1.82) (1.10) (-0.59)

R^2 0.0932 0.1141 0.1164 0.123 0.1293 0.1326

#obs 2346 2346 2242 2106 2016 2051

#firms 391 391 385 351 351 347

(27)

significant. Hence, I conclude that my results above are not driven by fluctuations in the demand side.

Differences between larger and smaller sized firms

In order to see whether the homogeneity assumption, as described by Kashyap et al.

(1993), is violated, I analyse the dataset to see whether there is a difference between larger and smaller sized firms. The homogeneity assumption can be violated because smaller firms probably have no access to public debt markets, whereas the larger firms might already be fully financing themselves with public debt. This causes the chances of substitution to be low because smaller firms have no option to switch debt sources, whereas larger firms have little incentive to switch to bank debt when already fully financed with cheaper public debt. There is heterogeneity in the variance of both groups, making it harder to find any results for the substitution of debt sources. I therefore expect that smaller sized firms show lower rises in nonbank debt than larger firms. Also, I expect that larger firms show lesser decreases in bank debt than smaller firms after an expansion of the monetary stance. Table 6 shows the results. I present the results for smaller and larger firms in panel A and B, respectively.

The descriptive statistics in Table 2 show that both larger and smaller sized firms have similar levels of bank debt to total assets. Larger firms lowered their bank debt ratio, whereas smaller firms did not change their bank debt ratio on average. Larger firms have more nonbank debt in general and thus more leverage on average. This means that bank debt is relatively more important for smaller firms than it is for larger firms. This is in line with theory saying that smaller firms have less access to public debt, which is the larger constituent of the nonbank debt variable.

First, I discuss the firm control variables. The coefficients of size in column 2 of Table 6

indicate that the effect of firm size on the presence of bank debt is only important for smaller

firms. As the firm grows large enough, the importance of firm size in the amount of bank debt

present fades away. Profitability is also more important in determining nonbank debt ratios for

smaller firms, visible in column 4. An explanation of this can be that a small unprofitable firm is

(28)

seen as a high risk investment, leaving investors with little motivation to provide credit to these firms.

Next, the regression shows that larger and smaller firms react diversely to changes in the monetary stance. The coefficients of Δ interbank rate bankdep in column 2 of Table 6 show* that

Table 6

Changes in bank and nonbank debt during the sample period 2011-2016: Subsamples based on the by country median of firm size

Panel A: size small

1 2 3 4 5 6

Bank debt to assets nonbank debt to assets Bank debt to total debt

Δ interbank rate 0.0098 -0.0020 -0.0379 -0.0321 0.0729 0.0564

(0.23) (-0.05) (-0.74) (-0.62) (0.61) (0.43)

Δ interbank

rate*bankdep 0.1227** -0.0612 0.2343

(3.18) (-1.06) (1.68)

profitability -0.0076 -0.0084 -0.0755*** -0.0750*** 0.1553 0.1531

(-0.49) (-0.54) (-4.84) (-4.87) (1.80) (1.81)

Q -0.0040 -0.0037 0.0093 0.0092 -0.0251 -0.0245

(-1.03) (-0.98) (1.75) (1.73) (-0.90) (-0.88)

Size 0.0159** 0.0152** 0.0021 0.0025 0.0557 0.0540

(2.57) (2.54) (0.26) (0.31) (2.13) (2.02)

Tangibility -0.0120 -0.0109 0.0277 0.0272 -0.2391 -0.2372

(-0.15) (-0.13) (0.46) (0.45) (-1.06) (-1.06)

R^2 0.0982 0.1045 0.0813 0.0827 0.1132 0.1141

#obs 2220 2220 2220 2220 2080 2080

#firms 370 370 370 370 359 359

Panel B: Size large

1 2 3 4 5 6

Bank debt to assets nonbank debt to assets Bank debt to total debt

Δ interbank rate 0.0185 0.0049 0.0163 0.0297 -0.0638 -0.1369

(0.45) (0.12) (0.48) (0.79) (-0.48) (-1.10)

Δ interbank

rate*bankdep 0.1034 -0.1018* 0.4145*

(1.74) (-2.17) (2.01)

profitability -0.0241 -0.0223 -0.0553 -0.0570 -0.0451 -0.0371

(-0.60) (-0.56) (-1.19) (-1.23) (-0.28) (-0.23)

Q -0.0001 -0.0003 0.0070 0.0072 0.0042 0.0032

(-0.01) (-0.04) (0.97) (1.02) (0.12) (0.09)

Size 0.0068 0.0067 0.0031 0.0033 0.0266 0.0263

(0.93) (0.90) (0.28) (0.29) (0.93) (0.9)

Tangibility 0.0895 0.0872 0.0195 0.0217 0.1626 0.1531

(1.51) (1.49) (0.57) (0.64) (0.74) (0.69)

R^2 0.1194 0.126 0.1087 0.1145 0.1256 0.1304

#obs 2220 2220 2220 2220 2206 2206

#firms 370 370 370 370 372 372

(29)

the overall decline in bank debt ratios during the sample period was especially prevalent among the smaller sized bank-dependent firms. Because the interbank market rate has declined in my sample period, the positive coefficient means that the bank debt ratio has also declined in my sample period. On the contrary, surges in the levels of nonbank debt among bank-dependent firms are stronger in larger sized firms (column 4, panel A and B), although the coefficient for smaller firms is not significant. This confirms the importance of public debt market access, which is easier accessible for larger sized firms due to their scale advantages. This result is supported by the coefficients of Δ interbank rate bankdep in column 6. Even though smaller* firms show a larger reduction in bank debt ratios in response to the decreasing interbank rate, substitution away from bank debt is higher among larger firms because it is easier for this group to increase their nonbank debt ratios.

The above results shows the lack of homogeneity between smaller and larger sized firms.

Smaller sized firms shrink their bank debt ratios more than larger firms while they show less substitution towards other sources of debt, indicating that it is harder for these firms to switch to public debt due to their smaller size. This heterogeneity among firms makes it harder to find the significant results above.

VI. Conclusion

In my thesis I describe how the corporate debt structure has evolved in the Euro area over the period 2011 - 2016. This period was characterized by exceptionally low, and decreasing interest rates. The period has furthermore seen unconventional monetary policy with the ECB buying billions of euro denominated government bonds. I use a firm-level dataset in which I analyse two types of corporate debt used by firms from 8 euro area countries. I analyse both bank and nonbank debt. Then, I analyse the results on a per country based level, with the interbank market rate being the most important variable, acting as a proxy for the monetary stance in the euro area.

My findings are in line with the literature with firms showing a decrease in the use of

bank debt in times of low interest rates, but only so for firms that are bank-dependent. A reason

(30)

for this is the improved credit quality of firms due to depressed interest rates, which leads to firms preferring corporate bonds over bank debt (Denis and Mihov, 2003); Furthermore, an easing of the monetary stance is not able to explain deviations in debt structure for the sample as a whole. This is because my sample period contains the sovereign debt crisis, a period in which bank and bond lending was constrained. Additionally, firms with various levels of bank dependency react opposite to changes in the monetary stance. These opposite reactions then offset each other so that I do not find a significant result. Anyway, The more bank-dependent a firm is, the more it reduces its relative presence of bank debt in this period of expansionary monetary policy. These firms have substituted their bank debt for other sources of debt because their ratios of nonbank debt have increased in the sample period. The results suggest that bank- dependent firms are not literally bank-dependent, because a lot of substitution towards other sources of debt has taken place. Further, I find that smaller companies have shown a larger decrease in their bank debt ratios without an accompanying rise in nonbank debt ratios. This highlights the difficulty for smaller firms to raise nonbank debt. Last, I find that firms in the peripheral countries, Greece, Italy and Spain, react much stronger to changes in monetary policy.

With a stronger substitution from bank debt towards nonbank debt being present in these countries.

As for further research, I suggest that another study could be done on the effects of debt

substitution during time periods that are characterized by a tightening of monetary policy and

periods in which the monetary stance remains unchanged. It would be interesting to investigate

these periods because only then one will be able to find out whether monetary policy is the real

driver of these changes in debt structure or that the changes I find are due to other characteristics

omitted in my regression.

(31)

Reference list

Becker, B., and Ivashina, V., 2014, "Cyclicality of creditsupply: Firm level evidence," Journal of Monetary Economics 62, 76-93.

Cantillo, M., and Wright, J., 2000, "How Do Firms Choose Their Lenders? An Empirical Investigation," The Review of Financial Studies 13, 155-189.

Denis, D. J., and Mihov, V. T., 2003, "The choice among bank debt, non-bank private debt, and public debt: evidence from new corporate borrowings," Journal of Financial Economics 70, 3-28.

Diamond, D. W., 1991, "Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt," Journal of Policital Economy 99, 689-721.

Duca, M. L., Nicoletti, G., and Martínez, A. V., 2016, "Global corporate bond issuance: What role for US quantitative easing?," Journal of International Money and Finance 60, 114-150.

Duchin, R., Ozbas, O., and Sensoy, B. A., 2010, "Costly external finance, corporate investment, and the subprime mortgage credit crisis," Journal of Financial Economics 97, 418-435.

Fernández, A.L., González, F., and Suárez, N., "Bank supply shocks and the substitution between bank and nonbank debt," Journal of Corporate Finance 48, 122-147.

Greenwood, R., Hanson, S., Stein, J. C., 2010, "A Gap‐Filling Theory of Corporate Debt Maturity Choice," The Journal of Finance 65, 993-1028.

Houston, J., and James, C., 1996, "Bank Information Monopolies and the Mix of Private and Public Debt Claims," The Journal of Finance 5, 1863-1889.

Jayaratne, J., and Morgan, D. P., 2000, “Capital Market Frictions and Deposit Constraints at Banks.” Journal of Money, Credit, and Banking 32, 74–92.

Jiménez, G., Ongena, S., Peydró, J.-L., and Saurina, J., 2012, "Credit Supply and Monetary Policy: Identifying the Bank Balance-Sheet Channel with Loan Applications," American

Economic Review 102, 2301-2326.

Kahle, K. M., Stuls, R. M., 2013, "Access to capital, investment, and the financial crisis,"

Journal of Financial Economics 110, 280-299.

Kashyap, A. K., Stein, J. C., and Wilcox D. W., 1993, "Monetary Policy and Credit Conditions:

Evidence from the Composition of External Finance," The American Economic Review 83, 78-98.

Kishan, R. P., and Opiela, T. P., 2000, "Bank Size, Bank Capital, and the Bank Lending Channel,” Journal of Money, Credit, and Banking 32, 121–141.

Kroszner, R. S., Laeven, L., and Klingebiel, D., 2007, "Banking crises, financial dependence, and growth," Journal of Financial Economics 84, 187-228.

Leary, M. T., 2009, "Bank Loan Supply, Lender Choice, and Corporate Capital Structure," the Journal of Finance 114, 1143-1185.

Lyer, R., Lopes, S., Peydró, J.-L., Schoar, A., 2014, "Interbank liquidity crunch and the firm credit crunch: evidence from the 2007–2009 crisis," Review of Financial Studies 27, 347–372.