• Nenhum resultado encontrado

FINANCIAL CRISIS SPILLOVERS TO THE CORPORATE SECTOR

N/A
N/A
Protected

Academic year: 2019

Share "FINANCIAL CRISIS SPILLOVERS TO THE CORPORATE SECTOR"

Copied!
28
0
0

Texto

(1)

A Work Project, presented as part of the requirements for the Award of the Masters Degree in

Finance from Nova – School of Business and Economics

FINANCIAL CRISIS SPILLOVERS TO THE

CORPORATE SECTOR

BANK-DEPENDENT BORROWERS VERSUS BOND MARKET

PARTICIPANTS

ANA CATARINA HENRIQUES SIMÕES MARQUES #426

A Project carried out on the Financial Markets Area with the supervision of:

Prof. Miguel Ferreira and Prof. João Santos

(2)

1

FINANCIAL CRISIS SPILLOVERS TO THE CORPORATE SECTOR

BANK-DEPENDENT BORROWERS VERSUS BOND MARKET PARTICIPANTS

Acknowledgements

I would like to express my deepest gratitude to my supervisor, Professor Miguel Ferreira, who

provided insightful comments and suggestions since the formative stages of this project, to this

final draft. His availability, guidance and advice were invaluable. I would also like to thank my

family and friends for the unconditional love and support they offered me, not only throughout

the course of this thesis, but during my entire life.

Abstract

This paper provides empirical evidence on the impact that shocks to capital providers have on

their borrowers’ performance. We use the recent financial crisis, which was originated in the

U.S. mortgage market, as reasonably exogenous shock to the performance of European

Union-based companies, therefore allowing us to disentangle credit supply and demand-side frictions.

Our results show that bank-dependent firms were more adversely affected by the financial crisis

in terms of stock market valuation, investment decisions and profitability, in comparison to

firms with access to public debt markets. Overall, we highlight the role that financial

intermediaries play in the propagation of financial shocks to the real economy, even across

distinct regions.

(3)

2

“Credit availability has the ability to build a modern economy. Lack of credit has the ability to

destroy it, swiftly and absolutely.”

Ben Bernanke, Chairman of the Federal Reserve Board

September 18, 2008

1. INTRODUCTION

In 2007, the world witnessed the beginning of what many claim to be the worst financial crisis

since the Great Depression. It resulted in downturns in most equity markets, in the collapse or

bailout of financial institutions known as too big to fail, and in the freezing of credit and money

markets. After all, because banks’ balance sheets were tremendously affected, this financial crisis turned out to be a banking crisis characterized by a large lending tightening to the

corporate sector, followed by a consequent slowdown of economies worldwide.

Bearing this in mind, the recent financial crisis brought renewed interest in studying the role that

financial intermediaries play in the propagation of financial shocks to the real economy, which

will be our main focus in this paper. We provide new evidence on the impact of the recent

financial crisis on firms’ performance, in terms of: (1) stock market valuation; and (2) investment decisions and profitability. Particularly, we investigate if bank-dependent firms were

more adversely affected by this event when compared to firms with access to the public debt

markets1.

The remainder of the paper is organized as follows. Section 2 explains the theory behind our

study. Section 3 clarifies the relevance of the financial crisis. Section 4 proceeds with the

methodology. Section 5 describes the data sources and characterizes the sample. Section 6

presents our results. Section 7 concludes.

1

(4)

3

2. THEORETICAL FRAMEWORK

The seminal investigation of Modigliani and Miller (1958) assumes that, in frictionless markets,

the supply of capital is perfectly elastic, so that firms can borrow as much as they want at the

same cost of capital. This implies that lending is purely a function of credit demand factors, such

as size, growth opportunities and profitability. Nevertheless, in the real world, market frictions

such as information asymmetry and moral hazard do exist, affecting both the availability and the

pricing of credit (see for example, Stiglitz and Weiss 1981; Faulkender and Petersen 2006;

Lemmon and Roberts 2010).

In fact, many researchers maintain that market imperfections in financial markets prevent firms

from easily and costlessly access alternative financing channels in case their providers of capital

unexpectedly curtail access to credit. For firms that are either financially constrained or heavily

reliant on external financing, this particularly results in suboptimal corporate decisions – such as not expanding inventories or foregoing positive investment opportunities –, with clear negative consequences for those firms.

For instance, in the context of the recent financial crisis, Campello, Graham, and Harvey (2010)

survey 1,050 CFOs worldwide and find that financially constrained firms planned deeper cuts in

employment, technology and capital expenditures; Almeida et al. (2012) show that firms whose

most long-term debt matured right after the onset of the crisis experienced greater reductions in

their investment than other firms; and Buca and Vermeulen (2012) find that, for small and

medium firms from Southern European countries, which are in general more bank-dependent,

there was an unprecedented drop in aggregate fixed capital formation in 2009.

Bearing all this in mind, in a rational market it is expected that any unanticipated shock to the

banking sector should result in a greater stock price decline for firms that are dependent on bank

(5)

4

debt markets. For instance, Slovin, Sushka and Polonchek (1993) document negative effects of

Continental Illinois Bank’s failure on stock performance of the bank’s borrowers; Chava and

Purnanandam (2011) show that, during the Russian Crisis of 1998, bank-dependent firms

experienced larger stock market valuation losses than their bank-independent counterparts; and

Kang and Stulz (2000) find similar results for the Japanese banking crisis in 1990-1993. To the

best of our knowledge, the impact of the recent financial crisis on firm’s stock market valuation is not thoroughly investigated until date.

This paper adds to the existing literature by trying to identify to which extent variations in the

supply of capital after the onset of the recent financial crisis had a greater impact on the

performance of bank-dependent firms in comparison to firms with access to bond markets, in

terms of stock market valuation, investment behaviour and profitability. However, investigating

the impact of financial distresses on firm’s performance poses a major challenge of properly disentangling the effect of supply-side shocks felt by banks from demand-side shocks faced by

firms, mainly because they usually occur simultaneously (e.g. common economic shocks).

Our identification strategy aims at overcoming this problem in different ways. First and

foremost, we use the recent financial crisis, which originated from the subprime mortgage

market, as a reasonably exogenous event to the performance of European Union firms, hence

making it possible to identify the effect of credit supply on corporate performance2. To further rule out any demand-driven effect, we control for firm-specific factors in all our market

valuation regressions and, in an alternative approach, a matched sample analysis is also

performed; when studying firms’ real outcomes, a firm fixed effect model is estimated, which accounts for observable and unobservable heterogeneity in firms’ characteristics. In the next section, we elaborate on the relevance of the recent financial crisis for this study.

(6)

5

3. THE 2007-2008 FINANCIAL CRISIS

The most recent financial crisis had its origins in 2007 with the meltdown in the U.S subprime

mortgage market, which primarily resulted from the widespread of the so called

mortgage-backed securities. The risk mispricing of those securities became clear with the bursting of the

housing bubble, which generated several defaults and forced banks to large write-offs. By the

end of 2007, the largest U.S. banks had already announced write-downs in excess of $100bn.

During the following months, the U.S. government was forced to take over Fannie Mae,

Freddie Mac and AIG, and many banks either got bankrupt or were acquired by others. But it

was certainly the collapse of the Lehman Brothers on the 15th of September 2008 that triggered the whole panic in financial markets. Almost immediately, equity markets experienced a sudden

plummet, market volatility reached historical peaks, and credit markets seized up, as suggested

by Figures 1 to 3, in the Appendices.

This environment gradually caused lending in the interbank market to dry up which, associated

with high deleveraging pressures, ultimately resulted in a credit squeeze characterized by a

significant reduction in the supply of loans and in tightened lending standards to the corporate

sector. For instance, Ivashina and Scharfstein (2008) show that U.S. bank lending fell by 68%

during the peak of the crisis (2008Q4) relative to the peak of the credit boom (2007Q2); and

Campello et al. (2009) report that the financial crisis resulted in an increase on U.S.

commitment fees by 14 basis points, on mark-ups over the LIBOR rate by 69 basis points, and

in a decline of credit maturity by 2.6 months on average.

Summing up, this meltdown in financial markets and the associated losses to the banking sector

resulted in a sudden and severe capital rationing to the corporate domain, hence highlighting the

importance of studying the impact of credit supply shocks on corporate performance. In the

(7)

6

4. EMPIRICAL METHODOLOGY

This section outlines the basic methodology adopted in order to test our key hypothesis.

Hypothesis: the recent financial crisis had a larger impact on the performance of firms that

relied on banks as their main source of capital, as opposed to firms with access to public debt

markets, after controlling for firm-specific factors.

In order to test this hypothesis, our methodology is divided in two main parts. Part I investigates

the impact of the recent financial crisis on firms’ stock market valuation during a short-term window, which is the core contribution of this paper. Part II investigates the associated

medium/long-term effects on firms’ real outcomes – investment decisions and profitability. In both parts, firm performance is modeled as a function of demand and supply-side frictions, in an

attempt to clearly disentangle the effect of the credit contraction of banks from correlated shocks

to the corporate realm.

4.1. Effects on Stock Market Valuation

Following Chava and Purnanandam (2011), we concentrate on stock returns during the crisis

period as the key outcome variable, since it allows to identify the unanticipated impact of the

financial crisis on firms’ market valuation, and it is unaffected by later interventions of policy-makers. Given such, the following cross-sectional regression model is estimated:

i K

k i k i

i BOND F

r   

 

1 1

0 (1)

Where is the market model adjusted stock return of firm i, during the crisis period; is

a measure of bank-independence and our main goal is to estimate 𝛽 , the coefficient on this variable; is a set of firm-specific control variables. In the following sub-sections, additional

(8)

7

4.1.1. Market Model Adjusted Returns

The construction of the dependent variable of our cross-sectional regression – i.e. abnormal returns – relies on a standard event-study methodology3, which imposes the selection of an event and an estimation window.

A 20-trading day event window is defined starting on the day of the Lehman Brothers’ bankruptcy which represented an immediate and massive shock to the financial system (…)

larger than anything seen over nearly two decades (Sterling 2009) – and ending on October 10th – a day that marks the end of a black trading week4, and precedes the announcement of the

Troubled Asset Relief Program (TARP) by the U.S. Government.

Then, for every firm in the sample, the market-model beta is estimated using 250 trading days

for the year of 2007. By choosing this period, we aim at preventing the estimation window from

being contaminated by the turmoil already being felt in the markets between late 2007 and

2008.

Afterwards, using the beta estimates, we compute the abnormal returns i.e. the difference between the observed and the expected returns – over the event window period. The cumulative abnormal returns from 15-Sep-2008 to 10-Oct-2008 represent the dependent variable of the

cross-sectional regression model to be developed.

3

Event studies attempt to measure the abnormal performance associated with an event, departing from the primary assumption that markets process information about the event in an efficient and unbiased manner. For further information, see Kothari and Warner (2005).

4 During this week, the Dow Jones Industrial Average dropped by 18.1%, and the S&P500 fell by roughly 20%.

01.01.07

Event Window Estimation Window

15.09.08 10.10.08 13.10.08

End of a black week

TARP announced Lehman Brothers’

(9)

8

4.1.2. Measure of Bank-independence

A critical part of our methodology is the measure of firm’s bank-independence, which should

reflect whether a firm has access to public bond markets. Following Santos (2006), we

implement this through the inclusion of the dummy variable BOND, which takes the value of

one if the firm has ever5 issued a public bond, and zero otherwise. According to our hypothesis, this variable should present a positive coefficient, meaning that bank-independent firms

experienced greater stock market returns than bank-dependent firms during the crisis period.

In further analyses, we also break down the group of firms with access to public debt markets

into two distinct groups: (a) firms whose last bond issued was rated (RATED); and (b) firms that

most recently issued a non-rated bond (NRATED). Indeed, even though firms that issued

unrated bonds are classified as bank-independent in our first specifications, their lack of credit

rating may act to bond holders as a signal of poor reliability, especially when public debt

markets are under a considerable pressure – as it is evident by the uncommonly high levels of paper-bill spread during this period (Appendices, Figure 4) – and investors are shifting their

capital to safer assets –flight-to-quality. Consequently, we expect the coefficient on the dummy

variable RATED to be more positive than the respective one on NRATED.

4.1.3. Firm-specific control variables

The decision to obtain financing in public debt markets is likely to be related with various firm

characteristics and, therefore, in order to completely isolate the effect of bank-independence on firm’s performance we need to control for these factors in our regression, as they may help to

explain any returns’ differential between these two groups of firms.

5

(10)

9

Since bank-dependent firms are, in general, considerably smaller than bank-independent firms,

we control for firm size using log of firm’s sales in million euros (LNSALES) as proxy. Indeed,

small firms are less likely to be known by analysts and investors, and are more likely to face

asymmetric information problems, which may limit their access to alternative sources of funds.

Additionally, because older firms may benefit from an established reputation, we include

LNAGE(the log of the firm’s age, computed as the difference between the year of the crisis,

2008, and the year the firm first appeared in Datastream). Similarly, we add EBITSALES (EBIT

divided by sales) as more profitable firms have a more comfortable cushion to service their debt.

Because of the aforementioned reasons, we expect these three factors to have a positive impact

on firms’ valuation during crisis.

Also, since firms with high levels of leverage are generally more sensitive to economic

downturns than their low-leveraged counterparts, we expect those firms’ valuation to be more adversely affected by the crisis6. A similar argument applies in terms of default risk. During crisis periods, high default-risk firms are expected to suffer larger valuation losses, not only due

to their increased probability of default but also because they are more exposed to the

risk-aversion and flight-to-quality effects widely noticed during crises. To control for leverage, we include firm’s debt-to-asset ratio in our regressions (LEVER); we control for default risk by

including the firm’s credit-worthiness percentile rank based on Altman Z-Score7 (ZSCORE).

Additionally, firm’s market-to-book ratio (MKTBOOK) is also incorporated, which proxies for

future value creation through growth opportunities. The effect of growth opportunities on firms’ valuation is twofold. First, it is expectable that firms’ growth opportunity set shrinks during crisis periods; second, even if growth opportunities remain, the shortage of credit may force

6

For further information, see Opler and Titman (1994).

(11)

10

firms to forego some intended projects. In either case, market’s response to firm value is expected to be lower for high-growth firms.

We include shares turnover (SHARESTURN) to account for differences in stock liquidity across

firms, i.e. the degree to which an asset can be bought or sold in the market. During crisis

periods, the effect of stock liquidity on returns can be ambiguous. On the one hand, because

liquidity becomes particularly valuable at times of great turmoil in the markets, less liquid

stocks may experience a significant stock price decline. On the other hand, in order to obtain

immediate cash inflows, investors will generally get rid of their most liquid risky assets, leading

to larger price drops for these stocks.

We complement this set of firm controls with two variables related with stock market valuation,

STOCKVOL and STOCKRET, which are respectively the standard deviation and the average of

the firm’s stock return over the past twelve months. Since stocks that are either more volatile or have higher past returns should experience superior price declines during crisis, we expect both

variables to have a negative impact on firms’ valuation.

Finally, we include country and Fama-French industry fixed effects in all our regressions, as to

ensure that our results are not driven by any unobservable country and industry characteristics.

To control for dependence in the error terms for firms within the same country, we use robust

standard errors clustered at the country-level in all the analyses performed.

4.2. Effects on Real Outcomes – Investment and Profitability

In order to complement our market valuation analysis, we study the impact of the recent

(12)

11 iq iq iq iq q

i q

i q

q i i

iq BOND C C BOND C BOND C LEV SG CF

Y   1 1 2 2 1 * 1 2 * 2     (2)

Where measures firm’s performance in real terms, for firm i in quarter q. Two distinct

measures of performance are used: (1) capital expenditures scaled by lagged assets, as proxy for

investment; and (2) operating income scaled by lagged assets, as proxy for profitability.

𝛼 measures firm fixed-effects. is a measure of access to bond markets, constructed as

in the previous specifications8; it is time invariant in this regression model, and thus subsumed by the firm fixed-effects. equals one for quarters after 2008Q3, and zero otherwise (crisis

period). Additionally, we include , which equals one for quarters between 2007Q3 and

2008Q3, and zero otherwise, as to control for any variations occurring during this mixed period

(early stage of the crisis).

Since firm fixed effect models only capture firm-specific factors that are either time-invariant

(e.g. industry) or slow changing (e.g. size)9, we incorporate and , which are firm varying characteristics that control respectively for leverage, sales growth and operating cash flow, of firm i in quarter q. These variables are scaled by the amount of assets at the

beginning of the period, as to maintain a common scale factor.

This model is estimated using data from 2007Q2 to 2009Q4. We use standard errors that are

heteroskedasticity-consistent and clustered at the firm level. Our main goal is to estimate the

coefficient on , which measures changes in real outcomes during the crisis period,

for firms with access to bond markets in comparison to bank-dependent firms; therefore, we

expect it to be positive in both models (investment and profitability).

8 As in the previous part of our methodology, in some specifications we also segregate access to bond markets by

using the dummy variables RATED and NRATED.

(13)

12

5. DATA, SAMPLE CONSTRUCTION AND DESCRIPTIVE STATISTICS

This section provides information on the data and sample construction. For further detail on the

definitions of the variables used, please refer to Table 1, present in the Appendices.

5.1. Data and Sample Construction

The data for this project has Datastream database as its main source. Our initial sample covers

all the European Union listed companies comprised in this database, having data on stock

returns for the crisis period, as well as for the year before. Afterwards, some adjustments were

performed to the sample selected.

Firstly, both financial firms (SIC codes between 6000 and 6999) and utilities (SIC codes

between 4910 and 4940) were eliminated from the sample, since their nature does not match

our interests in this study. To remove the effect of the bid-ask bounce from the analysis – which is more relevant for less liquid securities -, we also remove firms trading at less than 1€ stock price at the end of 2007. All-equity financed firms were also excluded from the sample since,

according to our classification, they would always be considered as bank-dependent and it is not

possible to distinguish whether these firms could have accessed the market for public debt if

they wished to. Furthermore, in order to insure that the results are not be driven by any

demand-side condemand-siderations, all firms reporting primary operations in the United States were also

eliminated from the sample. Finally, we remove all junk-rated firms as, due to their poor

credit-quality, they are not representative. This leaves us with a sample of 2,648 firms for our base

case analysis.

For the first part of this study, we obtained accounting and financial data as of December 2007;

for part II, quarterly data was gathered from 2006Q2 to 2009Q4. Information regarding bond

issuance up to September 2008 was also collected, as well as the respective bond rating. When

(14)

13

used as proxy. To prevent outliers from affecting our results, the data is winsorized at 1% and

99% in all the analyses performed.

5.2. Descriptive Statistics

Table 2 illustrates our sample by comparing bank-dependent firms with firms that accessed

public debt markets, along some key characteristics.

First of all, it is worth mentioning that the sample is composed by approximately 85%

bank-dependent firms and only 15% firms with access to public bond markets, which clearly

illustrates the bank-based system prevailing in most European Union countries.

The results reported in Table 2 also confirm that, as previously discussed, there are clear

differences regarding key characteristics across the two groups of firms. In terms of size, the

average bank-independent firm has annual sales of 10,400 million euros, which is roughly 13.5

times more than the average bank-dependent firm. Also, firms with access to bond markets are

on average 8.14 years older then bank-dependent firms, and are likewise more profitable.

Concerning leverage and default risk, firms that issued bonds have on average greater

debt-to-asset ratios and present an increased probability of default. There are also other remarkable

differences in terms of shares turnover, past stock market returns and volatility.

Overall, we find that these two groups of firms present considerable disparities across various

key dimensions, which can per se explain any return differential between them during the crisis

period. To proceed with our analysis, we control for these differences in the next section, using

(15)

14

6. RESULTS

In this section, we analyse stock market performance for the full sample of firms, followed by a

matched sample procedure. Afterwards, we provide our regression results concerning changes

on firms’ real outcomes – investment decisions and profitability.

6.1. Effects on Stock Market Valuation – Full Sample Analysis

We begin our analyisis by examining the impact of bank-independence on stock market returns,

ignoring all other factors. From the analysis of Table 2, we conclude that, during the crisis

period, the mean (median) firm with access to public debt markets experienced market-model

abnormal returns of -6.85% (-6.73%), in comparison to -10.76% (-9.15%) grossed by

bank-dependent firms. The differences in both the mean (3.91%) and median (2.42%) returns across

the two groups of firms are statistically significant at the 1% level of significance.

Table 3 reports the OLS regression results from regressing market-model adjusted returns

(ABNRET) on the measure of bank-independence and on the control variables previously

discussed. In Model 1, we determine bank-independence by whether a firm has ever issued a

public bond (BOND). In Model 2, we segreggate bond issuers according to whether the last

bond issued by the firm was rated (RATED) or not (NRATED), due to the previously mentioned

reasons. Both models include country and Fama-French industry fixed-effects. The t-statistics

reported are computed using robust standard errors clustered at the country-level.

Model 1 shows that, during the crisis period, firms with access to public debt markets registered

on average 2.71% higher returns than bank-dependent firms, after controling for firm-specific

factors. Even though the magnitude of the return differential between the two groups of firms

(16)

15

Methodology section. They confirm that stocks of bigger and older firms performed better

during the crisis period; whilst liquid stocks, stocks with high past returns, and stocks of

highly-leveraged firms experienced a larger drop.

In Model 2, the public debt market access is decomposed through the introduction of the

dummy variables RATED and NRATED, which represent the marginal impact of contemporary

public issuances of rated and unrated bonds, respectively. The results are in accordance with the

expectations. In this specification, the estimate on RATED equals 5.68% and is statistically

significant at the 1% significance level. This indicates that firms that most recently issued rated

bonds earned on average 5.68% higher returns than bank-dependent firms, after controlling for

firm-specific risks. Still, there is no evidence that stocks of firms that had issued unrated bonds

(NRATED) performed better than those of bank-dependent firms during the crisis period.

All in all, the results suggest that, after a shock that affects the health of the banking system,

bank-dependent firms suffer greater market valuation losses than firms that have access to

alternative capital sources. Furthermore, after distinguishing between rated and unrated bonds

our results become stronger, which highlights the greater trust that markets deposit upon bonds

that possess a credit rating, especially at times when financial markets are under great pressure.

6.2. Effects on Stock Market Valuation – Matched Sample Analysis

Although our results in the full sample analysis do support our hypothesis, some objections can

be raised, namely regarding the assumption on the exogenous access to bond markets. In fact,

given the large dissimilarities across bank-dependent and bank-independent firms, it is possible

that our results are driven by unobservable factors, which may lead to biased estimators. We

address this potential problem using a matched sample analysis based on a matching of

(17)

16

The first step of the matching procedure is to estimate a probit model of the dependent binary

variable indicating whether a firm has accessed or not the market for public debt. We consider

as determinants of this choice10: firm’s size, age, growth opportunities, profitability, leverage, default risk, stock market liquidity, past stock return and volatility. Additionally, we include the

Fama-French industry classification and the country dummies, as to control respectively for

industry and country-specific factors.

The first two columns of Table 4 Panel A, denoted Pre-Match, present the estimation results of

the probit model. Through the analysis of the aforementioned table, it is evident that various

factors that we used to explain abnormal stock market returns in the previous regressions are

determinant to the access of public debt markets. Larger and older firms are better established,

which boosts their access to public debt markets. Moreover, firms that are more profitable and

have greater growth opportunities are also more attractive to investors in bond markets. The

probability of accessing bond markets is also positively correlated with firm leverage, since

those firms have eased-access to credit, and with volatility of returns, which may be only a

result of the unique conditions of the markets at the time. Firms with high risk of bankruptcy are

less attractive to bond holders, hence less likely to access credit markets. We obtain a pseudo

R-square of 36.14%, suggesting a reasonable fit of the model.

Afterwards, we obtain the propensity score i.e. probability of accessing bond markets for each

firm in the sample, which allows us to find, for every independent firm (treated), the

bank-dependent firm (control) that is the most similar in terms of the observable characteristics (i.e.

the closest propensity score). We match firms without replacement which, despite sacrificing

the number of firms in the sample, maximizes the statistical power of the tests performed. The

(18)

17

matching procedure yields a sample of 812 firms (406 bank-independent firms matched with

406 bank-dependent firms).

The second two columns of Panel A, Table 4 (Post-Match), present the results of estimating the

probit model on the matched sample, which is only carried out for validity purposes. From the

analysis of this informstion, we can confirm the effectiveness of our matching technique, since

none of the variables is significant in this specification this indicates that, after the match, there are no pair wise statistically significant differences across the two groups of firms.

Table 4, Panel B, provides the difference-in-difference estimates of the abnormal returns, before

and after matching bank-independent and bank-dependent firms along key observable

dimensions. From the matched sample analysis we conclude that, during the crisis period, the

mean bank-dependent firm (control) experienced a cumulative abnormal return of -8.92% in

comparison to -6.89% for firms with access to public debt markets (treated). The difference

(2.03%) is significant at the 1% significance level, though smaller than for the Pre-Match

analysis (3.53%). Additionally, when performing the same analysis but using firms that issued

rated bonds as the treatment group, our results become stronger, with a difference-in-difference

estimate of 6.54%, statistically significant at the 1% significance level11.

All in all, we find evidence that, even for a sample of homogenous dependent and

bank-independent firms across observable characteristics, the stock market differential for these two

groups of firms during the crisis remains, with bank-dependent firms being clearly more

adversely affected than their bank-independent counterparts.

6.3. Effects on Investment and Profitability

The estimation results concerning the impact of the financial crisis on firms’ real outcomes are provided in Table 5, Models 1 to 4.

11 For space saving purposes, the results of the

(19)

18

Through the analysis of this table, it is evident that the financial crisis had a strong negative

impact on firm’s real outcomes, since C2 and C1 are negative and highly significant in both

models. Regarding the control variables, we confirm that sales growth and operating cash-flow

have a positive effect on investment and profitability, whilst leverage has the opposite impact on

both measures.

More importantly, Model 1 reports that firms with access to public debt markets were able to

navigate the crisis with better levels of capital spending than their bank-dependent counterparts.

In fact, the mean bank-dependent firm experienced a drop of roughly 1.70% in their capital

expenditures during the crisis, as compared to 1.30% faced by firms that issued public bonds

(1.09% for issuers of rated bonds –Model 2)

Additionally, Model 3 also documents a significantly positive coefficient on BOND*C2,

indicating that the profitability of bank-independent firms was less affected by the financial

crisis than that of their bank-dependent peers. Indeed, whilst the profitability of bank-dependent

firms decreased on average 5.35% between the fourth quarter of 2008 and the homologous

period of 2009, that reduction was of only 4.55% for bank-independent firms (4.18% if we

consider firms that issued rated bonds Model 4).

Notwithstanding, it is worth mentioning that these results have some clear limitations. Firstly,

we assume that the two groups of firms remained unchanged over the period analysed, which is

not completely accurate, as some firms that previously financed themselves uniquely with bank

financing, may have started to issue public bonds during this period. Secondly, because this

analysis comprises a vast period, we are not able to assure that the contraction experienced by

firms in terms of capital expenditures and profitability was due to variations in external

(20)

19

7. FINAL REMARKS

The financial crisis that hit the U.S. in 2007-2008 was a significant shock to the global financial

system, resulting in several defaults and huge losses of equity capital to the banking system.

This ultimately resulted in a credit squeeze characterized by a significant reduction in the supply

of loans and in tightened lending standards to the corporate sector. Because this natural event

had its origins in the U.S. mortgage market, it can be considered as a reasonably exogenous

shock to the performance of European Union-based firms, therefore allowing us to study the

impact of banks health on borrowers’ performance. Particularly, we compare the performance of firms dependent on bank financing with that of firms with access to public debt markets.

We find evidence that, during the 20 trading days after the Lehman Brothers’ collapse, firms with access to public debt markets registered on average 2.71% higher adjusted returns than

bank-dependent firms, after controling for firm-specific factors (2.03% using a matched sample

analysis). Furthermore, bank-independent firms were able to navigate the crisis with better

levels of capital expenditures (0.4%) and profitability (0.8%) than their bank-dependent

counterparts. Our results become stronger when we compare bank-dependent firms with firms

that issued rated bonds, which stresses the importance of a credit rating in providing firms with

a greater visibility and reliability in capital markets, especially during crisis periods.

Due to the particular effects of contagion present in this crisis, we also bring new evidence on

the role that the global integration of financial markets plays in the propagation of shocks across

regions, through the banking channel. Most importantly, our findings have strong implications

for the current sovereign-debt crisis in Europe and for policy-makers and monetary authorities

in general, highlighting the importance of efficient and well-developed corporate bond markets

in helping companies and the economies to better resist to financial and banking crises, and to

(21)

20

8. REFERENCES

[1] Albertazzi, U., and D. Marchetti. 2010. “Credit Supply, Flight to Quality and Evergreening: An analysis of Bank-Firm Relationships after Lehman”. Bank of Italy, working paper.

[2] Almeida, H., M. Campello, B. Laranjeira, and S. Weisbenner. 2012. “Corporate Debt

Maturity and the Real Effects of the 2007 Credit Crisis”. Critical Finance Review. 1:3-58.

[3] Bernanke, B., M. Gertler, and S. Gilchrist. 1996. “The Financial Accelerator and the Flight to Quality”. The Review of Economics and Statistics 78:1-15.

[4] Buca, A., and P. Vermeulen. 2012. “Corporate investment and bank-dependent borrowers during the recent financial crisis”. European Central Bank, working paper.

[5] Campello, M., J. Graham, and C. Harvey. 2010. “The Real Effects of Financial Constraints: Evidence from a Financial Crisis”. Journal of Financial Economics 97:470-87.

[6] Chava, S., and A. Purnanandam. 2011. “The Effect of Banking Crisis on Bank-Dependent Borrowers”. Journal of Financial Economics 99:16-35.

[7] Faulkender, M., and A. Petersen. 2006. “Does the Source of Capital Affect Capital Structure?”. The Review of Financial Studies 19(1):45-79.

[8] Gertler, M., and S. Gilchrist. 1993. “The Role of Credit Market Imperfections in the Monetary Transmission Mechanism: Arguments and Evidence”. The Scandinavian Journal of Economics 95:43-64.

[9] Holmstrom, B., and J. Tirole. 1997. “Financial Intermediation, Loanable Funds, and the Real Sector”. The Quarterly Journal of Economics 112:663-91.

[10] Ivashina, V., and D. Scharfstein. 2010. “Bank Lending during the Financial Crisis of 2008”. Journal of Financial Economics 97:319-38.

[11] Judge, A., and A. Korzhenitskaya. 2011. “Credit Market Conditions and Impact of Access to the Public Debt Market on Corporate Leverage”. Working paper

(22)

21

[13] Lemmon, M., and M. Roberts. 2010. “The response of corporate financing and investments to changes in supply of credit”. Journal of Financial and Quantitative Analysis

45:555-87.

[14] Love, I., L. Preve, and V. Sarria-Allende. 2007. “Trade Credit and Bank Credit:

Evidence from Recent Financial Crises”. Journal of Financial Economics 83(2):453-69.

[15] Modigliani, F., and M. Miller. 1958. “The Cost of Capital, Corporation Finance and the Theory of Investment”. The American Economic Review 48:261-97.

[16] Ongena, S., D. Smith, and D. Michalsen. 2003. “Firms and their Distressed Banks: Lessons from the Norwegian Banking Crisis”. Journal of Financial Econometrics 67:81-112.

[17] Peek, J., and E. Rosengren. 2000. “Collateral Damage: Effects of the Japanese Bank Crisis on Real Activity in the United States”. The American Economic Review 90:30-45.

[18] Petersen, M., and R. Rajan. 1994. “The Benefits of Firm-Creditor Relationships: Evidence from Small Business Data”. Journal of Finance 49:3-37.

[19] Puri, M., J. Rocholl, and S. Steffen. 2011. “Global Retail Lending in the Aftermath of the U.S. Financial Crisis: Distinguishing between Supply and Demand Effects”. Journal of Financial Economics 100:556-78.

[20] Santos, J. 2011. “Bank Loan Pricing Following the Subprime Crisis”.Review of Financial Studies 24:1916-43.

[21] Santos, J., and A. Winton. 2006. “Bank Loans, Bonds and Information Monopolies Across the Business Cycle”. Journal of Finance 63:1315-59.

[22] Stiglitz, J., and A. Weiss. 1981. “Credit rationing in markets with imperfect information”.

American Economic Review 71:393–410.

[23] Slovin, M., M. Sushka, and J. Polonchek. 1993. “The Value of Bank Durability:

Borrowers as Bank Stakeholders”. The Journal of Finance 48(1):247-66.

[24] Sterling, W. 2009. “Looking Back at Lehman: An Empirical Analysis of the Financial

(23)

22

9. APPENDICES

Figures 1 and 2 –Equity Prices12 and Implied Volatilities13, during 2006-2008. Source: Federal Reserve Bank of St. Louis.

Figures 3 and 4 –LIBOR-OIS and TED Spreads14, andPaper-Bill Spread15, during 2006-2008.

Sources: Bloomberg; Federal Reserve Bank of St. Louis.

12

Equity indices are expressed in local currency; 1 January 2006 = 100.

13 Volatility implied by the price of at-the-money call option contracts of the indices; expressed in percentage points. 14

The LIBOR-OIS spread reflects the difference between the rate at which banks will lend to each other (LIBOR), compared with the overnight indexed swap (OIS) rate; the TED spread equals the difference between the U.S. Treasury Bill and the Eurodollar rate. Both are expressed in percentage points.

15 Paper-Bill spread represents the difference between commercial paper and Treasury-bill rates, in percentage points.

0 25 50 75 100 125 150 175 200 225 Ja n-0 6 Ap r-0 6 Ju l-06 Oc t-0 6 Ja n-0 7 Ap r-0 7 Ju l-07 Oc t-0 7 Ja n-0 8 Ap r-0 8 Ju l-08 Oc t-0 8 Equity Prices

DJ Eurostoxx Nikkei 225

S&P500 0 10 20 30 40 50 60 70 80 90 100 Ja n-0 6 Ap r-0 6 Ju l-06 Oc t-0 6 Ja n-0 7 Ap r-0 7 Ju l-07 Oc t-0 7 Ja n-0 8 Ap r-0 8 Ju l-08 Oc t-0 8 Implied Volatilities

VXJ VSTOXX VIX

-6 -5 -4 -3 -2 -1 0 1 2 3 4 Ja n-06 A pr -06 Ju l-06 O ct -06 Ja n-07 A pr -07 Ju l-07 O ct -07 Ja n-08 A pr -08 Ju l-08 O ct -08

Libor-OIS and TED Spreads

Libor-OIS Spread TED Spread

0,0 0,3 0,5 0,8 1,0 1,3 1,5 Ja n-06 A pr -06 Ju l-06 O ct -06 Ja n-07 A pr -07 Ju l-07 O ct -07 Ja n-08 A pr -08 Ju l-08 O ct -08 Paper-Bill Spread

(24)

23

Table 1 – Variable Definitions

Panel A: Effects of the Financial Crisis on Stock Market Valuation

ABNRET Stock Market Adjusted Returns between 15-Sep-2008 and 10-Oct-2008.

BOND Dummy variable that takes the value of one for firms that have ever publicly issued

a bond.

RATED Dummy variable that takes the value of one if the firm most recently issued a rated

bond.

NRATED Dummy variable that takes the value of one if the firm most recently issued an unrated bond.

LNSALES Natural logarithm of the firm’s sales (sales measured in million euros).

LNAGE Natural logarithm of firm’s age; age being computed as the difference between the

crisis year (2008) and the year the firm first appeared in the Datastream database.

MKTBOOK Market value of equity divided by book value of equity.

EBITSALES EBIT divided by sales.

LEVER Ratio of total debt to total assets.

ZSCORE ZSCORE = 1.2Z1 + 1.4Z2 + 3.3Z3 + 0.6Z4 + 0.999Z5.

Z1 = Working Capital / Total Assets; Z2 = Retained Earnings / Total Assets; Z3 =

Earnings Before Interest and Taxes / Total Assets. Z4 = Market Value of Equity /

Book Value of Total Liabilities. Z5 = Sales/ Total Assets.

STOCKRET Annualized mean of daily stock returns.

STOCKVOL Annualized standard deviation of daily stock returns.

SHARESTURN Total number of shares traded over the past 12 months, divided by the average

number of shares outstanding for the period.

Panel B: Effects of the Financial Crisis on Investment and Profitability

BOND, RATED,

NRATED

Defined as aforementioned.

CAPEXASSETS Capital expenditures divided by lagged assets.

EBITASSETS EBIT divided by lagged assets.

SG Sales growth divided by lagged assets.

CF Operating cash-flow divided by lagged assets.

(25)

24

Table 2 – Descriptive Statistics

Mean 25th pctl 50th pctl 75th pctl Std. Dev.

Panel A: Bank-Dependent firms (N=2238)

SALES 773.193 32.688 121.588 436.162 2,557.455

AGE 16.346 10 14 22 9.3857

MKTBOOK 2.7246 1.2341 2.0115 3.2699 2.8518

EBITSALES 0.1308 0.0622 0.1116 0.184 0.1675

LEVER 0.2158 0.0787 0.1975 0.3189 0.1604

ZSCORE 0.5071 0.256 0.513 0.761 0.2903

SHARESTURN 1.4541 0.0758 0.2581 0.697 6.2066

STOCKRET 0.0115 -0.2323 -0.0162 0.2232 0.4361

STOCKVOL 0.407 0.2808 0.3572 0.4587 0.2051

ABNRET -0.1076 -0.217 -0.0915 0.0155 0.1929

Panel B: Firms with access to public debt markets (N=410)

SALES 10,400 727.619 3,220.278 11,200 16,200

AGE 24.4829 15 23 39 12.14

MKTBOOK 2.9121 1.471 2.2759 3.4737 2.7758

EBITSALES 0.1791 0.0981 0.1714 0.1828 0.1601

LEVER 0.2756 0.1941 0.2766 0.3409 0.1321

ZSCORE 0.4216 0.203 0.4035 0.604 0.2525

SHARESTURN 1.7047 0.1864 0.8605 1.7583 5.1062

STOCKRET -0.2014 -0.1836 -0.001 0.157 0.2805

STOCKVOL 0.3161 0.2361 0.2875 0.3542 0.1256

ABNRET -0.0685 -0.1705 -0.0673 0.0281 0.1681

This table reports summary statistics of key variables used in our analysis, for bank-dependent firms and firms with access to public debt markets (Panel A and B, respectively). The sample consists of 2,648 firms, for which we were able to obtain financial and market information in Datastream. This sample does not include: (1) Financial borrowers and utilities; (2) zero-debt firms; (3) firms trading at less than 1€ at the end of

(26)

25

Table 3 – Effects of the Financial Crisis on Stock Market Valuation: Full Sample

Model 1 Model 2

Estimate t-val Estimate t-val

BOND 0.0271 2.43**

RATED 0.0568 3.84***

NRATED 0.1039 0.79

LNSALES 0.0057 2.55** 0.0046 1.99**

LNAGE 0.0248 2.99*** 0.0252 3.03***

MKTBOOK 0.0029 2.09** 0.0028 1.95*

EBITSALES -0.0006 -0.47 -0.0005 -0.38

LEVER - 0.0756 -2.45** -0.0724 -2.35**

ZSCORE 0.0130 0.70 0.0154 0.83

SHARESTURN - 0.0018 -2.25** -0.0018 -2.30*

STOCKRET - 0.0958 -8.23*** -0.0966 -8.34***

STOCKVOL 0.0441 1.55 0.0448 1.57

R2 0.1485 0.1506

N 2,648 2,648

Industry Fixed Effects IN IN

Country Fixed Effects IN IN

This table presents regression results from regressing firm’s market adjusted stock returns (ABNRET) on a

measure of bank-independence and on a set of firm-specific control variables. Model 1 measures

bank-independence by the dummy variable BOND; Model 2 segregates bond issuances though the dummy

(27)

26

Table 4 – Effects of the Financial Crisis on Stock Market Valuation: Matched Sample

Panel A: Matching Estimation Results

Pre-Match Post-Match

Estimate t-val Estimate t-val

LNSALES 0.4224 12.45*** 14.774 -1.77

LNAGE 0.1951 2.68*** 3.0443 0.4

MKTBOOK 0.0213 1.97** 2.8342 0.82

EBITSALES 1.1667 3.67*** 0.1780 0.66

LEVER 0.8638 2.49** 0.2550 1.34

ZSCORE -0.6628 -2.95*** 0.4207 0.4

SHARESTURN 0.0099 1.07 1.7080 -0.46

STOCKRET 0.0044 0.05 -0.0211 -1.3

STOCKVOL 1.0670 3.89*** 0.3169 0.59

Pseudo R2 0.3614 0.013

N 2,273 812

Industry Fixed Effects IN IN

Country Fixed Effects IN IN

Panel B: Abnormal Returns for Treatment (Treat=1) and Control (Treat=0) groups

This table presents the results of the matched sample analysis concerning stock market adjusted returns of bank-dependent and bank-inbank-dependent firms around the crisis period (15-Sep-2008 to 10-Oct-2008). Panel A

presents the results of the probit regression, with access to public debt markets (BOND) as the binary dependent variable. In the Pre-Match model all the firms with non-missing observations are used, whilst in the

Post-Match model only firms that could be matched based on the propensity score from the Pre-Match model are

considered. Panel B presents the results from comparing bank-independent (treatment) and bank-dependent

firms (control), before and after the match. * Significant at 10%; ** significant at 5%; *** significant at 1%.

Treatment = BOND Treatment = RATED

Pre-Match Post-Match Pre-Match Post-Match

ABNRETTREAT=0 -0.1042 -0.0892 -0.1007 -0.0996

ABNRETTREAT=1 -0.0689 -0.0689 -0.0342 -0.0342

ABNRETTREAT=1-

ABNRETTREAT=0

0.0353 0.0203 0.0665 0.0654

t-test ABNRET 3.45*** 2.26** 4.53*** 3.50***

(28)

27

Table 5 – Effects of the Financial Crisis on Real Outcomes: Investment and Profitability

Models 1 and 2: Investment Models 3 and 4: Profitability

Estimate t-val Estimate t-val Estimate t-val Estimate t-val

C1 -0.0026 -3.86*** -0.0026 -3.85*** -0.0215 -14.60*** -0.0215 -14.60***

C2 -0.0170 -25.19*** -0.0170 -25.19*** -0.0535 -35.85*** -0.0535 -35.85***

BOND*C1 0.0006 0.40 0.0045 1.20

BOND*C2 0.0040 2.36** 0.0080 2.16**

RATED*C1 0.0019 0.78 0.0076 1.40

NRATED*C1 -0.0003 -0.13 0.0021 0.45

RATED*C2 0.0061 2.51** 0.0117 2.16**

NRATED*C2 0.0023 1.08 0.0053 1.11

SALESG 0.0010 16.05*** 0.0010 16.05*** 0.0016 11.42*** 0.0016 11.42***

OPERCF 0.0170 5.59*** 0.0170 5.60*** 0.4992 74.33*** 0.4992 74.34***

LEVER -0.0116 -3.96*** -0.0116 -3.97*** -0.1238 -19.10*** -0.1239 -19.11***

R2 0.0319 0.0319 0.1978 0.0319

N 39,246 39,246 39,246 39,246

Firm Fixed Effects IN IN IN IN

This table presents the firm fixed effect regression results for the effect of the crisis on firm’s investment and profitability. This regression model is estimated using quarterly data for 2,648 firms, from 2006Q2 to 2009Q4. The dependent variables are: CAPEXASSETS, quarterly capital expenditures scaled by lagged assets (in Models

Referências

Documentos relacionados

As Table 3 suggests, increases in the ratios of debt (both long-term and short-term), interest expenses and impairment losses in accounts receivable to turnover are

The probability of attending school four our group of interest in this region increased by 6.5 percentage points after the expansion of the Bolsa Família program in 2007 and

Staying with the results reported in Table 4 for Panel 1, we can look at two other financial ratios that clearly go in line with the GDP growth rate, namely the

Table 1 shows how the difference in prices of an asset decreases with risk aversion (the Tree price difference decreases from 0.83 to 0.45 and 0.10 as one moves from logarithmic

researchers enable to model and to examine the possible impact of the 2008 global financial crisis; this examination showed that the general effect of the crisis on the

As the Latin American financial sector did not face any severe market turmoil comparable to the uS or the Eu in the financial crisis since 2007, the existence of SIFIs has

Os objetivos específicos são: caracterizar e analisar a produção brasileira de carne bovina (dando destaque para os maiores produtores); analisar o comportamento das exportações

On the other hand, a complex unnoticed contractual relationship may be defined using our contract model, by exploiting the whole structure including