Liquidity management and employment : an empirical analysis for years 2005 and 2008.

(1)

Liquidity Management and Employment

An Empirical Analysis for years 2005 and 2008

MSc Thesis

Master Thesis in Business Economics (20 ECTS)

By: Ivo Spil (5603803) Supervisors: Dr. Erasmo Giambona

Dr. Jeroen van de Ven University: Universiteit van Amsterdam

Faculty: Faculty of Economics and Business Tracks: Finance

(2)

Liquidity Management and Employment

An Empirical Analysis for years 2005 and 2008

By Ivo Spil

Abstract

This paper attempts to reveal new evidence on liquidity management and therefore uses manually acquired data about credit line usage of firms in 2005 and 2008. Firms that are large, profitable, investment graded and have easy access to credit lines tend to have higher levels of credit lines. Cash is a substitute source of liquidity to credit lines,

however this changes during the crisis. The paper is inconclusive about how drawdown activity is influenced by the amount of credit lines a firm has. Firms that hoard liquidity tend to cut on planned spending on employment. This theory is supported by actual employment levels in 2005 and in 2008.

Acknowledgements

Written by one person, a MSc thesis is never a solo project. My thanks go to Dr. Erasmo Giambona for being my thesis supervisor, for letting me participate in this research on liquidity, for his patience and advice. I would also like to thank my second supervisor Dr. Jeroen van de Ven for his advice and wise council. Further, I especially would like to thank Frans and Myrna Spil, Timo Klein, Pierre van den Oord and Maurits Dijkstra for their advice and support. You were great inspiration and sparring partners. Finally, I would like to thank friends and family for supporting me during this writing process.

(3)

Table of content

1. Introduction ...5

2. Review on the Relevant Literature ...7

2.1 What are Credit Lines? ...7

2.2 Liquidity Management during the Financial Crisis ...8

2.3 Liquidity Crisis ... 10

2.4 When Credit Lines are used ... 11

2.5 The relation between Cash Flow and Credit Lines. ... 12

2.6 Liquidity and Employment ... 14

3. The Empirical Research: Data, Methodology and Hypothesis ... 15

3.1 Introduction ... 15

3.2 Data ... 15

3.3 Methodology and Hypothesis ... 16

4. Analysis and Results ... 20

4.1 Introduction... 20

4.2 Descriptive Statistics Credit Lines ... 20

4.3 Subsample Analysis Credit Lines ... 21

4.4 Univariate Analysis Credit Lines ... 23

4.5 Regression Analysis on Credit Lines ... 24

4.5.1. Full Sample Regression Analysis on Credit Lines ... 24

4.5.2 Regression Analysis of Sample Firms with Lines of Credit ... 26

4.5.3 Regression Analysis Credit lines and Drawdowns ... 28

4.6 Descriptive Statistics Employment ... 29

4.7 Univariate Analysis Employment ... 30

4.8 Regression Analysis on Employment ... 31

4.8.1. Full Sample Regression Analysis on Employment ... 31

4.8.2 Regression Analysis on Employment for firms with Credit Lines... 32

4.8.3 Regression Analysis Employment t+1 levels on Liquidity ... 33

5. Conclusion, Discussion and Limitations ... 34

5.1 Conclusion ... 34

5.2 Discussion and Limitations ... 36

6. Literature List ... 39

Appendix ... 41

Table 1: Distribution of cash holdings and credit lines as a percentage of assets in 2005 ... 41

Table 2: Variable definitions ... 42

(4)

Table 4: Usage of credit lines: analysis on firm characteristics ... 44

Table 5: Correlation matrix of lines of credit, cash, cashflow and drawdowns for 2005 and 2008 ... 46

Table 6: Full sample analysis, OLS regression on Lines of Credit ... 47

Table 7: Probit regression on Lines of Credit dummy [0,1] ... 48

Table 8: Sample of Firms with a line of credit, OLS regression on Lines of credit ... 49

Table 9: Sample with division among the Credit Line distribution, OLS regression on Lines of credit ... 50

Table 10: Sample of firms with Lines of Credit, OLS regression on Drawdowns ... 52

Table 11: Descriptive Statistics Employment ... 53

Table 12: Correlation matrix of employment, cash, lines of credit, sales and employees for 2005 and 2008 ... 54

Table 13: Full Sample, OLS Regression on Employees ... 54

Table 14: Sample firms with credit line, OLS regression on Employees ... 55

(5)

1. Introduction

The financial crisis last decade raised questions about how channels of liquidity dried up. Brunnermeier (2009) researched how this crisis emerged and explains how this crisis could be described as a liquidity crisis. This work will focus on liquidity management, how the levels of cash holdings and credit lines changed through time for the companies, how they used credit lines and if managers had difficulties with renewing or initiating credit lines. This paper examines the role of credit lines and attempts to find out how extensive their role in corporate liquidity management is, in times of uncertainty and when sources of liquidity are scarce e.g. in times of a financial crisis.

This work is inspired by the research of Campello et al. (2010). They studied firms in the years 2008 and 2009 on how they used different sources of liquidity during these times. The way liquidity was managed by firms had influence on hiring employees and other investment decisions. The research of Campello et al. (2011) and so this work, focus on different sources of liquidity instead of default ratios like classical credit line research

Jemenez, Lopez and Saurina (2009) performed. In this way this work tries to find evidence on interaction between different sources of liquidity in times of financial crisis. When observing corporate decisions it seems that credit lines are associated with cutting on planned spending. Firms with limited access to credit lines, in contrast, appear to choose between saving and investing during the crisis and this could have impact on employment levels.

Liquidity management has influence on employment levels. Campello et al.(2010) confirm a negative relationship between levels of cash, credit lines and planned corporate spending. Firms tend to hoard liquidity when liquidity is scarce and cut on corporate spending as employment. While using data about actual employment levels we will attempt to confirm this theory and test if cuts on planned employment spending reduces employment levels. Bennet (2011) states that firms use firing mechanisms as liquidity adjustment tools. Less employment means less operational spending and so firms increase liquidity while hoarding it to survive economical down fall. The influence of credit lines and firm specific characteristics as size and industry will be measured by OLS regression analysis. With this analysis we might find indicators that can predict cuts on spending on employment. With the last analysis we can link liquidity management to actual employment levels.

(6)

While companies rely heavily on credit lines (Shockly and Thakor, 1997) the importance of credit lines in liquidity management grew during the financial crisis (Ivashina and

Scharfstein, 2010). Those results will be compared with the results of the moment when the financial crisis was not initiated yet (2005).

This work will elaborate about the characteristics of credit facilities, how companies substitute between credit lines and internal liquidity, cash holdings and cash flow, when they will have liquidity shortage and how this influences corporate investments like employment expenditures. Firms tend to save on corporate expenditures when they hoard liquidity. This research will attempt to proof this and will also try to give proof for the statement that liquidity influences employment levels

To support daily operations a firm needs a certain amount of liquidity. With the growth of importance of credit lines, having a credit line to continue these operations could mean the survival or bankruptcy of a firm. With this research we can see the influence of a crisis on liquidity management and how levels of cash holdings and cash flows influence the amount of credit lines a firm establishes. On the second part we attempt to unravel how employment levels, which play an important role in making daily operations possible, are managed and impacted by the crisis. The paper attempts to find out how the acquisition of credit lines influences the driving force of daily operations, the amount of employees.

This paper will use a unique dataset with data acquired out K-10 files proved by the Securities and Exchange Commission of the United States. No database about credit lines exists until this day, so by using these unique data this paper attempts to find new, remarkable outcomes.

This paper is organized as follows. Section 2 reviews the relevant literature on about credit lines, the financial crisis, internal and external sources of liquidity and influences on employment. Section 3 describes how the used data is collected and describes the models tested... It will describe the methodology used, the models that will be tested and the hypotheses tested in this paper. In Section 4 the empirical research is conducted. This will consist out of two parts. The first part will analyze influences on credit lines, cash and cash flow. After that, correlations of the key variables are analyzed. Last, a regression analysis will be conducted using the models described in Section 3. The second part of the empirical research will analyze influences on employment, cash, credit lines and sales. First correlations are examined and a regression analysis is done using the models presented in Section 3. Finally, Section 5 will conclude the research, based on the results described in Section 4.

(7)

2. Review on the Relevant Literature

This section starts with a review on the relevant literature. First a description of credit lines is given and also the determinants of credit lines would be provided. Thereafter liquidity

management during the last financial crisis will be described. Third, amplification

mechanisms that help explain the causes of the financial turmoil will be outlined. Fourth, an explanation about what drives liquidity and when firms begin to use credit lines will be given. Fifth, the relation of a firm having credit lines and the firm’s rate of cash flows will be

discussed. The negative relation between liquidity and planned spending will be discussed and the specific effect that firing costs can have on cutting on employment

2.1 What are Credit Lines?

Credit lines are provided by a banks or financial institutions as a form of credit. A credit line can also be referred as a credit agreement, other terms are presented in Section 3 of this work. The credit lines are drawn from a fund whose size is agreed upon when the credit facility is acquired. The total credit facility is an off-balance sheet item, but the drawn credit lines on this facility are stated as a liability on the balance sheet of the firm who uses it. Credit lines could be used to finance daily operational needs or investments, and the acquisition of other parties. Credit lines have a maturity date and so there is a limited time in which they have to be paid back. A firm with such a credit facility can use the credit lines at its own discretion without the obligation to ask for permission from the lender. Because of the convenience on how these credit lines can be used, firms often make use of this form of credit (Laurent, 2011).

When a credit line is acquired it is common for the lender and borrower to agree upon six common issues considering the credit line: 1. the borrowing purpose, 2. the borrowing base, with 3. financial covenants, 4. the interest rate over the borrowed amount and 5. the level of commitment fees paid over the unborrowed amount and 6. the maturity date. When acquiring a credit line, the lender and borrower start negotiation about these six points which are based on financial information about the borrowing party in most of the times. For the lender this financial information is important to construct a reliable estimate of the chances the facility is paid back and to estimate the chance of default on the borrowed amount.

(8)

obtainability of a credit line (Sufi, 2005). How better these ratios are, the higher the chance for a firm to acquire a credit line. Argawal et al (2004) also link the financial performance (profitability) of a firm to enable acquisition of larger credit lines.

To control the risks of default and to increase the chance of being paid back the lender constructs several, most of the times financial, on which the borrower has to comply. The violation of a covenant could bring as result that the credit line is terminated by the lender. The borrower has to pay back its borrowings immediately and loses its access to the credit facility (Laurent, 2011). Covenants on revolving lines of credit imply that the use and

availability of the credit lines for the firm depend significantly on the operational performance of the company.

Having access to credit lines incur costs for the borrowing firm which are negotiated and agreed upon. Costs of credit lines consist out of interest rates paid over the borrowed amount of the facility. Those rates are usually linked to the London Interbank Offered Rate or Prime rate plus a margin dependent which is based on the financial data of the borrower. These costs are closely related to market interest rates. While over the used part of the credit facility interest rates are paid, a commitment fee has to be paid based on the unused amount of the facility. Most of the times this commitment fee is a percentage of the unused part of the facility (Laurent, 2011).

The financial performance and ratios determine the chance of access to a credit line for a firm. The lenders are banks or financial institutions and construct covenants on the credit lines which the borrowing firms have to meet. The costs of these credit lines are linked to LIBOR or the Prime Rate in most of the times, while on the unused part of the facility a commitment fee has to be paid.

2.2 Liquidity Management during the Financial Crisis

The focus of the last two paragraphs laid on the explanation of credit lines. In this paragraph this work focuses on liquidity management of which credit lines are part of. Campello et al. (2010) did research on liquidity management in the financial crisis. They attempt to explain how internal and external sources of liquidity interact and gathered data about credit lines, corporate spending and drawdowns on credit lines. Data of firms from the U.S. was used and with this they came to new insights how liquidity management influenced investment

(9)

decisions. An important outcome out of their research is that companies that have liquidity shortage tend to cut on corporate investments.

Companies switch between several sources of liquidity when liquidity is scarce. The data that Campello et al. (2010) used shows how firms substitute between internal and external funds in the crisis period. Demiroglu and James (2011) claim that cash is an

imperfect substitute for credit lines. But firms tend to choose not to use credit lines when they have enough internal funds, implying a cost wedge between these two sources of liquidity. The price of loans increased during the crises as is shown in Figure 1 (Beenders, 2011). This meant that better performing firms with ample funds had easier access to loans compared to less performing firms.

Figure 1: Bank spreads on small and large loans

Source: European Central Bank

With this information they were able to differentiate between different types of firms and focused on characteristics like amount of sales of the company and their profitability. Because access to credit lines is generally based on financial ratios, small, less performing firms with constrained access to loans and credit lines had to manage their funds of liquidity in a different way than large, better performing firms. The cost of acquiring a loan or credit line was more expensive for less performing firms because of higher determined commitment fees and higher margins (Campello et al, 2010).

An important outcome of the research of Campello et al (2010) is that their research seems to point that levels of liquidity influenced corporate spending in a notable way. Firms

(10)

with high level of funds tend to have lower costs on credit lines and so have easier access credit lines in comparison to firms that have a liquidity shortage. The firms that acquire credit lines tend to spend those on corporate investments while those firms with restricted access to credit tend to cut on corporate expenditures and investments while they choose to save and hoard their scarce funds.

2.3 Liquidity Crisis

Paragraph 2.2 focused on quantitative facts about liquidity management. This paragraph connects these liquidity issues with the financial crisis that could be outlined as a liquidity crisis, Brunnermeijer (2009). He describes a sequence of events that led to a shortage of liquidity. This shortage inclined while expert investors were not able to acquire enough funds and so risk was not optimally shared. This amplified the effect of shocks, leading to the crisis as we now know it. Brunnermeijer (2009) divides the concept of liquidity into two categories: funding liquidity and market liquidity.

Funding liquidity could be defined as a measure that states the restrained access of

funds to expert investors. How easier it is for investors to acquire funds, how higher the funding liquidity is. Market liquidity could be defined as the level of difficulty that is encountered when one tries to find somebody to complete an attempted trade. When market liquidity is low, the price of assets which is sold actually drops because the asset becomes less scarce combined with a lack of demand on the asset. The mechanisms of funding liquidity and market liquidity and the way these two interact could explain how small shocks can cause a liquidity shortage in the market with a liquidity crisis as effect (Brunnermeier 2009).

The creation of securitized and structured financial products like Credit Debt

Obligations (CDO’s) which made it easier to allocate risk to the ones who were most able to

bear it, led to an huge increase in credit expansion. This credit expansion led to an increase of housing prices. A mortgage crisis was initiated by this. The assumption was that housing prices could only rise and so when a borrowing party would default, the outstanding mortgage could be refinanced with the extra value the house had gained trough time. Financial

institutions eased the conditions on granting a mortgage because of this assumption. These events are described by both Adrian and Hyun Song (2010) and Brunnermeier (2009). After lowering lending standards, the amounts of defaults on mortgages increased. The assets of the lending parties decreased in value, forcing the financial institutions to increase their lending

(11)

standards and increase their margins. These two events ignited fire-sales of houses, funded with mortgage money. Prices of houses dropped significantly, which again forced financial institutions to increase lending standards more to the point where they stopped lending

money. This started bank runs which were fatal for some institutions, with a liquidity crisis as outcome (Brunnermeier 2009).

With the mechanisms of last liquidity crisis in mind it becomes clear that in a down turning economy credit rationing seems to be important. Thakor (2005) shows that when real interest rates are low and the market environment is friendly, this situation can lead to

excessive credit supply. Financial institutions have to consider their reputation when lending loans and could create an effective equilibrium with restricting access to these funds. When banks ration credit, it could be more attractive for a firm to use internal funds instead.

2.4 When Credit Lines are used

A liquidity crisis can influence the availability of credit lines. When banks do not have funds, they ration credit and cannot provide credit lines to lenders. This paragraph describes when credit lines are used. Lins et al. (2010) show several factors that influence the use of credit lines around the world and the use of cash for their corporate liquidity. First, they state that credit lines are a common used source for liquidity over the world, with a mean of 15% to total assets for an average firm.

Table 1 about here1

Second, they identify the reasons for firms to acquire credit lines by survey chief financial officers from 29 countries. These CFO’s were interviewed about how they managed liquidity and for what purposes they used credit lines. In summary Lins et al. (2010) identify four main reasons to acquire and use credit lines.

1

Source: Lins et al. (2010). Panel A lists the various categories of total cash to assets provided to survey respondents (column (i)) and the percentage of firms that fall into each category (column (ii)). Column (iii) contains the median percentage non-operational cash to total cash for each category of total cash to assets. Panel B lists the various categories of total credit lines to assets provided to survey respondents (column (i)) and the percentage of firms that fall into each category (column (ii)). Column (iii) contains the median percentage non-operational cash to total cash for each category. This percentage is computed by assuming that a firm’s level of cash to assets and non-operational cash to total cash is at the midpoint of its indicated range.

(12)

The first thing CFO’s are stating that they have credit lines when they expect high future external funding needs. An example is the use of a credit line as an instrument to fund future investment opportunities.

Second, when CFO’s believe their equity is undervalued they tend to have higher lines of credit. The relationship between undervaluation and credit lines appears to be just an equity phenomenon because there is no relationship between credit lines and a perception that credit spreads are too wide or debt ratings are too low2. One potential explanation for this lack of a positive relation is that if a firm believes it pays too much for debt it may not want a larger credit line because, being a debt contract, it too could seem very costly.

Third, the certainty of funding for acquisition opportunities is an important factor for CFO’s line of credit decisions. Credit lines tend to hedge against the possibility that frictions in obtaining external finance may prevent a firm from funding valuable future investment opportunities in potential good times ahead. In contrast, non-operational (excess) cash holdings are not positively related to either a firm’s need for future external funds or a belief that its equity is undervalued

And fourth, the most prominent reason cited by CFO’s when deciding on excess cash holdings is that it acts as a buffer against future cash flow shortfalls which Faulkender and Wang (2006) confirm in their paper. Non-operational cash tends to hedge against the

possibility that capital market frictions will prevent a firm from funding its current operations in potential bad times ahead. Broadly speaking, lines of credit appear to be held to fund future growth options and so this non-operational cash appears to be held as general purpose

insurance (Lins et al., 2010).

2.5 The relation between Cash Flow and Credit Lines.

In paragraph 2.4 numbers about holding credit lines are given and four major reasons for what situations they are used are given. While one of the reasons bank lines of credit are used to secure liquidity for potential future downfalls in cash flows, there also seems to be a relation between having high cash flows and credit lines. Sufi (2009) finds that firms with high cash flow tend to substitute credit lines for cash, while firms with low cash flow tend to use cash as main source of liquidity. As Berger and Udell (2002) and Cole (1998) describe, smaller firms tend to rely more on build relationships with the lender than on financial metrics as cash flow. Sufi (2009) was among the first to research the interaction between cash and credit lines. He

(13)

sampled 300 firms and collected information from their annual reports about credit line holdings, how much was drawn down on these lines and information about covenant violations (Beenders, 2011). Two main conclusions of the research of Sufi (2009) are important for this research.

First, he finds strong evidence for cash flows being the main predictor in whether a firm has a credit line or not. When a firm performs relatively well and so has high cash flows, it tends to use more credit lines. Further, when the firm expects a period of distress2, credit lines are only used as source of liquidity when they own ample cash.

Second, Sufi states that this correlation is based on the fact that to acquire credit lines, a firm has to meet certain covenants based on financial ratios as cash flow. When a covenant is violated banks tend to stop providing these credit lines (Beenders, 2011). The effect of covenant violation is visualized in Figure 2.

Figure 2: The effect of covenant violation on availability of credit3

2

Distress is measured on the basis of whether they are below or above the sample median z-score as in Altman’s (1968) z-score excluding leverage, given that leverage is a direct function of the proportion of used and unused lines of credit. More specifically, z-score is calculated as ZSCORE = 3.3∗ EBIT totalassets + 1.0 ∗ sales

totalassets + 1.4∗ etained earnings totalassets + 1.2 ∗ working capital total assets where retained earnings is

item 36 and working capital is item 179 from Compustat. ZSCORE has a mean of 1.21 and a standard deviation

of 2.80 in the random sample. Altman’s (1968) z-score excluding leverage, given that leverage is a direct function of the proportion of used and unused lines of credit. More specifically, z-score is calculated as ZSCORE = 3.3∗ EBIT total assets + 1.0 ∗ sales total assets + 1.4 ∗ retained earnings totalassets + 1.2 ∗ working capital

Total assets where retained earnings is item 36 and working capital is item 179 from Compustat. ZSCORE has a

mean of 1.21 and a standard deviation of 2.80 in the random sample.

3

Figure 2 maps the availability of lines of credit around a covenant violation. Time t = 0 is the year in which the covenant violation takes place, and the y-axis represents the availability of the line in years before and after the

(14)

Figure 2 shows that when a covenant is violated, the availability of the credit line decreases with a mount of circa 20 percent. Remarkably, the figure shows that before a breach of covenants firms tend to use a higher level of their credit lines. Cash flow levels could explain why firms with high cash flow volatility have more difficulties to acquire credit lines. These firms cannot always meet the financial covenants linked to the credit lines and so have restricted access to these lines. Sufi (2009) shows that firms with low cash flows acquire lower amounts of credit lines. Low cash flows increase the chance for a firm that they violate specific covenants on the line. Lenders then restrict access to these credit lines and so these firms have a lower amount of credit lines (Beenders, 2011).

2.6 Liquidity and Employment

Campello et al. (2010) also researched the relation between having liquidity out of cash holdings and credit lines to planned spending. One of these spending items is employment. They show that firms with higher levels of cash seem to spend more on investments with the use of credit lines, which are easy to acquire because of the high levels of cash and covenants on credit lines that require these cash levels. In contrast to firms with less access to credit that tend to hoard liquidity in the form of cash and credit lines and cut on planned spending on employment.

An imported reason for this negative effect could be firing is used as a liquidity adjustment tool. Bennet (2011) explains that firing can be used to adjust production to its efficiency level or is used as a bargaining tool. However, firing can also be understood as a liquidity adjustment tool to create net liquidity. Firing employees reduces costs on

employment and so the firms can spend the saved money on short-term net working capital for example. This could mean that when a firm needs liquidity, it might value liquidity higher than the present value of output of a certain employee. Then firing this employee allows the firm to increase its net liquidity and satisfies its liquidity needs. Bennet (2011) defines this feature as labor's liquidity service.

This section outlined the relevant literature about the relationship between liquidity

management and expenditures on employment. First this section explained what credit lines are and what their determinants are. Credit lines play a significant role in liquidity

management and are important during a liquidity crisis. When liquidity is scarce, banks try to ration credit and so firms can find themselves in financial distress. They could try to keep

(15)

liquidity on a certain level by hoarding liquidity and cut on planned spending on employment, by lowering employment levels.

3. The Empirical Research: Data, Methodology and

Hypothesis

3.1 Introduction

The next section describes the empirical research. First, the paper gives insight in what data it uses and how this data is collected. After that the paper describes the methodology and the objectives of this paper e.g. which hypotheses are tested. The empirical research is divided in two parts. The first part will follow Sufi’s (2009) methodology and the methodology of Campello et al. (2010). In the second part of the research the relation of liquidity on employment will be researched.

3.2 Data

The data I will use about credit lines is not collected in a precise database like Wharton Compustat or LPC-Dealscan. Those databases only reveal data about originations. The data about credit lines and drawdowns on credit lines is manually collected via www.sec.gov. Via this site I collected the data out of annual reports, 10-K filings, which state information about liquidity, for example credit lines and how much is borrowed down on these credit lines. While analyzing the 10-K filings I follow the methodology Sufi (2009) recommends and search for the definitions stated in his work:

‘credit lines; credit facility; revolving credit agreement; bank credit line, working capital facility; lines of credit and line of credit’.

Next to these terms this work will also collect information about total amount of credit lines the firms has and how much is borrowed on this line. Accounting data of the companies in the sample is collected from Wharton Compustat. The definitions of the variables we focus on in this work are defined in Table 2.

(16)

Table 2 about here

The sample we use contains data about 2100 firms out of 2005 and 1782 firms out of 2008 for the first part of the research and of 938 firms out of 2005 and 945 firms out of 2008 for the second part of the research. The data I will use are data of non-financial firms, so banks or other financial institutions will be left out since the financial crisis will have an exaggerated effect on these firms. We picked 2008 to sample the crisis and picked 2005 to sample for a normal period. Instead of Sufi (2009) who uses log assets to control for size, this paper will use sales to non-cash total assets in the second part of the empirical research to control for size.

3.3 Methodology and Hypothesis

The first part of the research will follow Sufi’s (2009) methodology. The paper will focus on the main determinants of credit lines. These determinants are the size of facilities, the use of available credit lines, the level of cash holdings and the size of cash flows of the firm. After this the paper will investigate how firms manage liquidity during the crisis. I start with mean difference tests and correlation analyses. To research how internal funds like cash and cash flow interact with external funds like credit lines I will perform OLS regression analysis.

In this work we will use the same definitions as Campello et al. (2011) use. To describe firms they use the definitions: 1. big, if their sales are less than $1 billion, 2. investment graded, if their bonds are rated equal or higher than investment grade BBB- , 3.: easy access to credit, if they have a credit line and higher cash flow levels than the mean of the sample, and 4. profitable, if they report profit. After this I will give basic evidence on interactions between credit lines, drawdowns, and cash holdings.

Hypothesis 1: Firms with higher cash flows make greater use of credit lines.

We analyze interactions between different sources of liquidity and will check if these interactions are dependent of the four defined firm characteristics. This is done via a

regression analyses based on OLS. I regress the amount of lines of credit to cash, cash flow and an interaction term of cash level and cash flow. The variables are ratios to non-cash equivalent assets. In this way the variable cash is not a denominator for all the variables in the model. The interaction term is included to check how cash flow interacts with credit lines for

(17)

different levels of cash. The matrix included will correct for several firm-specific measures. The specification can be written as follows:

( 1 ) LC/(Total Assetsi = c + β1* Cash Holdingsi + β2* Cash Flowi+

β3* (Cash Holdings * Cash Flow)i + γXi + εi

c represents a constant. Xi is a matrix which contains variables to control for size of the firm, profitability, easy access to credit and having a credit rating. εi is an error term. Expected is that firms with higher cash flows tend having higher levels of credit lines. This because profitable firms have easier access to credit lines because they violate covenant less often.

Hypothesis 2: Firms with higher cash holdings and cash flow tend to have a higher chance to have obtained a credit line.

This is measured via a probit regression. This measurement uses a maximum

likelihood procedure. The dependent variable can only take two values when we use this kind of estimation. The dependent variable equal 1 when the firms owns a credit line, and 0 if they do not. The model has the following form:

( 2 ) Credit line [0,1] = c + β1* Cash Holdingsi + β2* Cash Flowi+

c represents a constant. Xi is a matrix which contains variables to control for size of the firm, profitability, and having a credit rating. εi is an error term. The control dummy for easy credit is left out. Because one of the conditions for this dummy to be one is that the credit line dummy, the dependent variable is one. Therefore, the covariant pattern has only one outcome and so is a false variable for credit line to predict the chance to be one.

Expected is that how higher the cash levels of a firm are, how more lines of credit it has because specific covenants on the credit lines based on financial ratios are met. To draw conclusions we will measure the marginal effects of the variables in model (2). This enables us to see what effect the firm characteristics have on the possession of credit lines.

Hypothesis 3: Firms with high internal funds of liquidity draw relatively less on their credit lines than firms relatively less internal funds of liquidity.

(18)

We will replicate this test used by Campello et al. (2010) to test the relationship between level of internal funds of liquidity, e.g. cash holdings and cash flow to drawdowns on credit lines. The OLS regression model we test for this is defined as:

( 3 ) Drawdowns/Credit Linesi = c β1* Cash Holdingsi + β2* Cash Flowi+

c represents a constant. Xi is a matrix which contains variables to control for size of the firm, profitability, easy access to credit and having a credit rating. εi is an error term. Expected is that the direct negative relationship proposed by Campello et al. (2010) could be confirmed.

In the second part of the empirical research this paper focuses on the relation of liquidity on number of employees. These determinants will be the size of facilities, how large the cash holdings are and to normalize for size we take sales to total assets as parameter. The effect of firm characteristics is first analysed with mean difference test. After this I will follow-up with correlation analyses and OLS regression analysis. To control for the effects of type of

industry the firm belongs to, we will include industry dummies. The reason for this is that type of industry can influence the number of employees the company has. Out of the collected data four types of industry dummies are defined: a dummy for companies in the consumer goods industry, a dummy for manufacturing companies, a dummy for IT companies and a dummy for pharmaceutical and biochemical companies. Companies get an industry type if they meet conditions stated in Table 1. These dummies are chosen based on the distribution of SIC codes in the sample. Not every firm out of the sample is given an industry dummy to prevent the dummy trap. Those are the proxy firms with other industry codes.

The paper starts with mean difference tests and correlation analyses. Then I will use regression analysis to search for explanations for how firms’ liquidity influences their number of employees.

Hypothesis 4: Firms with lower liquidity have a lower number of employees.

Companies who hoard liquidity tend to cut planned costs on operational spending. The first thing companies cut on is employee expenditures. Bennet (2009) states that companies use firing as a liquidity adjustment tool. This is tested by the following model:

(19)

(4) Number of Employeesi = c + β1*Cash Holdingsi + β2 *LCi+ β3* Sales i + γXi + εi

Where LC is the amount of credit lines to assets, c is a constant, Xi is a matrix containing control variables for type of industry. εi is an error term. Expected is t how higher the liquidity level of the firm is (by hoarding), the less it will spend on employment expenditures and so have a lower number of employees.

The paper also checks for delayed effects in the number of employees. Companies could have a delay time in reacting to liquidity shortages and so cannot react in the same year. This could mean that not the number of employees of the same year is effected by liquidity of that year, but that the number of employees the year after is effected. With this analysis we could check if planned cuts on employment indeed take place. This will be tested by the following model:

(5) Number of Employeesi+1 = c + β1* Cash Holdingsi + β2* LCi+ β3* Sales i + γXi + εi

Where LC is the amount of credit lines to assets, c is a constant, Xi is a matrix containing control variables for type of industry. εi is an error term. Expected is that how less the liquidity level of the firm is, the more it will save on planned employment expenditures and actually do and so have a lower number of employees a year later. This model will only be tested for the crisis year 2008. This work assumes that firms probably try to hoard liquidity in the crisis and because of this we expect the negative relation between the amount of credit lines a firm holds and their employment levels, to be strong.

Section 3 described the data and how this was collected. It also explained the methodology, the models that are tested and which hypothesis is tested. In the next section the types of analysis and their results are described.

(20)

4. Analysis and Results

4.1 Introduction

In the upcoming section the paper analyses the data and will describe the results. Since the empirical research consists of two parts, the paper will first describe the statistics about the credit lines firms have and how much is withdrawn of the lines of credit. After that, the paper does a subsample analysis on credit lines. Third, the paper will do a univariate analysis on credit lines. Fourth, a regression analysis will be done on liquidity management. Part two of the research that analyses employment and liquidity will analyse the data in three ways. First the general statistics are described. Second, a univariate analysis will be done on employment and third a regression analysis is done on employment.

4.2 Descriptive Statistics Credit Lines

In 2005, 67.62% of the sample of firms had a credit line. The size of their credit lines was 15.77 % to their total assets on average. In 2008, 65.54% of the firms in the sample have a credit line at place. The mean size of credit lines was 15.95% to total assets. The firms with credit lines draw down up to 12.32% and 14.46% of their credit line on average for 2005 and 2008, respectively. The average size of loan commitments and credit line borrowings seems to have increased slightly during the crisis period.

Cash levels, the amount of cash equivalent items divided by non-cash total assets seem to be significantly higher for companies without a credit line in comparison to companies with a credit line in 2005. In the crisis year 2008 this trend seems to be disappeared.

When observing cash flow levels, observed is that in 2005 companies without a credit line have negative cash flow compared to their total assets. Firms with a banking line of credit have a mean cash flow level of 5.78% compared to -49.70% for firms without such a facility. This seems to confirm that better performing firms tend to have better access credit lines.

When observing the summary statistics, the year 2005 visualizes indications for a negative relationship between internal and external liquidity sources: companies that have banking

(21)

lines of credit tend to have lower levels of internal funds of liquidity. The average level of cash holdings for these firms is lower (30,69%) than for the firms out of the sample without credit lines (98,81%).

Like Campello et al. (2011) this work attempts to divide the firms in two groups, defined as regular borrowers/firms and irregular borrowers/firms. In section 2 this work explained how firms with credit lines have to comply with certain covenants defined by the lender. Financial ratios are an important part of these covenants. Better performing firms have a higher chance to acquire a credit line. These unconstrained firms are defined in this work, like in the work of Campello et al (2011) as 1. being big: having sales equal or higher than 1 billion dollars (22.43% and 38.05% of firms in 2005 and 2008). 2. having easy access to credit: having a credit line and higher cash flow than the mean out of the sample (52.38% and 50.06% of the firms in 2005 and 2008). 3. having a credit rating of BBB- or higher (22.57% and 33.22% of the firms in 2005 and 2008). And 4. being profitable which are 59.86% and 59.99% of the firms out of the sample in 2005 and 2008, respectively. The firms that do not meet these conditions are defined as irregular firms and will also be appointed as constrained firms. We will use both terms in this work.

4.3 Subsample Analysis Credit Lines

In this paragraph the results of the subsample analysis will be given. This analysis focuses on the usage of credit lines. The results are presented in Table 4. Columns 1, 3, and 5 are for the year 2005 and columns 2, 4, and 6 are for the year 2008. The firms that have a credit line are split by the four company characteristics mentioned in paragraph 4.2 e.g. size, easy credit, investment grade and profitability.

In column 1 and 2 of Panel A in Table 4 it is shown that ‘regular borrowers’ defined in paragraph 4.2 more often have a banking line of credit than ‘non-regular firms’. This is consistent with the research Campello et al. (2010) conducted. Mentioned must be that in the year 2008 small firms dominated the group of firms that owned a credit line. Those

differences between the two periods are significantly different. Column 3 and 4 show that in 2005 regular borrowers have higher level of average loan commitment than non-regular firms.

(22)

Except for the difference between easy credit and non-easy credit firms in 2005 and 2008, all the differences between the years 2005 and 2008 are significantly different.

Column 5 and 6 show that firms with easy credit had higher levels of drawdowns on their loan commitment in both 2005 and 2008. For both years this difference is larger than 25% and significant on the 1% level. What is revealing is that for both years the dominance tends to switch groups. It is not exactly clear why this happens and it must be mentioned that the differences between years 2005 and 2008 are all significantly different.

The level of drawdown activity is measured in columns1 and 2 of Panel B in Table 4. The amount of drawdowns increased for almost all groups over the years 2005 and 2008, except for large firms, with more than 3%. Although this increase is not rather high, the differences are all significant. For unprofitable firms the difference was the largest with 7.78% change with a significance level to less than 1%. Drawdowns increased while the crisis emerged and firms found themselves in liquidity shortage. This is consistent with the theory that states that external sources are used to obtain liquidity levels when inside forces cannot provide this liquidity in economic downturns. Loan commitments increased and drawdown levels on credit lines increased too. It can be concluded that during the crisis, companies in this sample relied greatly on liquidity by credit lines in order to make up for liquidity shortages.

Cash holding levels are shown by columns 3 and 4 of Panel B. Restricted or non-regular firms tend to have significantly higher cash holdings compared to non-restricted firms or regular firms. Also when 2005 and 2008 are compared with each other it is clear that cash holdings decreased during the crisis regardless the company characteristics. Two outliers are observed for small and non-investment grade firms. All differences are statistical significant. The gap of the differences for regular firms versus non regular firms became slightly smaller in 2008 however this does not count for every firm characteristic. When liquidity shortage became an issue for non-regular firms, they could use their higher levels of cash holdings to provide for their liquidity needs.

Finally, column 5 and 6 for Panel B of Table 4 compare for cash flow levels with respect to firm characteristics. Generally for regular firms cash flow is higher. The gap between the cash flow levels of regular and non-regular firms is significantly large. This gap is almost 50% measuring difference of firm characteristics. The differences are all significant. However, end of year results do not change much. While the crisis of 2008 was quite severe, the 2008 end-of-year results do not show that average performance of companies suffered in the economic crisis. Everyday operations were not influenced by any resource shortage.

(23)

4.4 Univariate Analysis Credit Lines

In this paragraph correlations between lines of credit, cash holdings and cash flows of the sample firms in years 2005 and 2008 are measured. In Panel A of Table 5 the correlations of the full sample are shown between amount of credit agreements, cash holdings and cash flow. In Panel B the correlations between amount of credit agreements, cash holdings, cash flow and amount of drawdowns is shown, thus only companies with a credit line are used in Panel B.

The results in Panel A can be interpreted as cash being a substitute for lines of credit. The negative coefficients of -0.0437 for the sample out of 2005 and -0.0610 for the sample out of 2008 seem to point to that. A higher cash level goes with a lower level of lines of credit. The estimated correlations between cash flow and credit lines have positive values, 0.1065 for the sample out of 2005 and 0.0620 for the sample of 2008. Higher cash flow levels tend to go with higher credit lines. This could be linked to the fact that firms have to meet certain

covenants to acquire a credit line. Better performing firms, with higher cash flows have easier access to credit lines. These results seem to confirm that. We have to remark that the values of these estimated correlations are rather low, however, they are in line with the theories of Campello et al. (2011) and Sufi (2009). This is in compliance with the findings of Malinova (2012) on the same research with comparable data.

In Panel B we measure the correlations between credit lines, cash, cash flow and drawdowns in the years 2005 and 2008 for firms with a credit line. Remarkable is the positive correlation coefficient of 0.2093 between cash and credit lines in 2005. This seems to indicate that in normal times, firms with credit lines did not substitute these for cash. Another remarkable outcome is the negative coefficient of -0.3146 between cash flow and credit lines in 2005. This result seems to indicate that unrestricted firms tended to have fewer lines of credit when they had higher cash flows. This can be interpreted as better performing firms using less credit lines in normal times. In 2008 both correlation coefficients switch sign and are then in line with the theories described by Campello et al (2011) and Sufi (2009). The correlation coefficients between drawdowns and cash seem to reject Hypothesis 3. For the sample of 2005 this correlation coefficient has a value of 0.0141 and in 2008 it has a value of 0.0.314.

(24)

This seems to indicate that firms with higher cash levels tend to draw down more on their lines of credit. This is not in line with the theories that describe that the use of external funds of liquidity and internal funds of liquidity are substitutes and negatively related to each other.

A part of the results of the univariate analyses seems to reject the theories about drawdowns, however, the correlation coefficients of credit lines and cash are in line with the theories we test. We try to gain more clarity on the theory testing further in the next section.

4.5 Regression Analysis on Credit Lines

In this paragraph we will analyse the regression on credit lines and drawdowns on credit lines. First we will analyse the full sample. Thereafter we will analyse the firms with a credit line and will also separate these group in firms that have an above median level of cash flow and a below median level of cash flow. Finally, the drawdown behaviour of firms with a line of credit will be regressed.

4.5.1. Full Sample Regression Analysis on Credit Lines

Table 6 presents the estimated coefficients of model (1) described in Section 1. The regression is done on data of the full sample.

The relationship between credit lines and cash based on the OLS analysis describe a slight negative relationship in Section 1. In normal times a substitution effect between these sources of liquidity as described by Sufi (2009) and Campello et al (2011) is confirmed. It must be said that the coefficient is close to zero. Firms that hold higher cash levels do not have an urge to borrow. The cost of drawdowns in comparison to using own cash could be a reason for this. The relationship however is not significant. In column 6 this negative relationship is switched to a positive one, which does not support the theory. An explanation could be that firms foresaw a coming liquidity shortage and stacked up cash and also drew on their credit lines. The negative relationship between lines of credit and cash flow could be explained by companies that pay a commitment fee over the unused amount of credit lines next to paying interest on the amount of credit lines used. Most of the companies have easy access to credit. They have a larger loan commitment and so pay higher commitment fees.

(25)

Observing the relation between credit lines and cash flow we see a negative relation between these two. This result is not significant at the 1% level and supports the theory that more profitable firms acquire credit lines easier. Covenants influence availability to credit lines and those are based on financial ratios and targets, for example debt-to-EBITDA. This could be an explanation of this relationship. In the crisis the coefficients of most dummies have a negative sign which does not support the theory. Those coefficients are significant to the 5% level.

The theory of Campello et al. (2010) that states that credit lines increase with cash flow for profitable firms does not seem necessarily true. This is tested with the inclusion of the interaction term between cash holdings and cash flow. Both the coefficients in years 2005 and 2008, column 3 and column 6 of Table 6 respectively are close to zero and statistically significant.

Both the outcomes of column 3 and column 6 of Table 6 show that in this sample the theory breaks down in the crisis. More research is needed about crisis years to test this theory. Since data about credit lines is not available in a database yet, more data has to be collected to research crisis years as 2007 and 2009.

Table 7 shows the results of the probit regression, model (2) described in Section 3 of this paper is tested. The dependent variable is a credit line dummy that has value 1 if the firm has a credit line and value 0 otherwise. The outcomes show the marginal effects of the variables cash and cash flow and could be interpreted has how the chance a firm has a credit line is increased by the level of the parameter. Column 3 of Table 7 shows a negative relationship between cash and having a line of credit (-0.260) and is significant at the 1% level. This supports the theory that states that cash and credit lines are substitute sources of liquidity. However, in 2008 the coefficient for cash switches to one with a positive sign. This relationship is not significant.

In 2005 there is a positive relation between cash flow and having a line of credit. The coefficient had a value of 0.197 and is significant at the 1% level. This supports the theory described in Section 2. In 2008 this coefficient switches to a negative sign with value of -0.027 but is not significant.

(26)

having a credit line and a firm with a investment grade has 20.30% higher chance of having a banking line of credit. These results support the outcomes of the subsample analysis that showed that regular firms are acquiring credit lines easier than non-regular firms. In 2008 the influence of the dummy variables brake down and are all close to zero. They are not

significant.

Contrary to the theory, also observing the results of the regression that tests model (1), it is inconclusive if the theory still holds in the crisis. While the described theories are

supported by the data of normal times it is not clear if cash and cash flow levels influence the availability of credit lines when a crisis hits the economy.

4.5.2 Regression Analysis of Sample Firms with Lines of Credit

This subparagraph focuses on firms with lines of credit. After concluding that the univariate analysis supports the theory, and observing that the theory breaks down in crisis years for the full sample, we will exclude firms without a credit line to test if the theory only holds for firms with a credit line. The relation of credit lines, cash and cash flow are tested for the sample firms, using regression model (1) described in Section 3 of this paper. Table 8 columns 1 to 3 show the results of year 2005, while columns 4 to 6 describe results of 2008.

The results of the regression on credit lines for firms with credit lines do not support the theory. The coefficient’s signs are positive for cash in both column 3 and column 6. In both years the coefficients are statistically significant to the 1% level. Omitting the interaction term in column 2 and column 5 does not change the signs of the coefficients and even omitting the coefficient for cash flow from model (1) does not change the sign for cash.

For cash flows we see the opposite results. The signs in both years 2005 and 2008 are negative with a statistical significance to the 1% level. Omitting the interaction term in column 2 and column 5 does not change that. This implies a substitution effect between lines of credit and cash flows which could be explained by commitment fees which lower the level of EBITDA.

(27)

The interaction term in column 3 and column 6 is close to zero. This result does not differ with the results of the full sample analysis and suggest that there is no diminishing positive effect of profitability on lines of credit when firms hold more cash.

The values of the dummy variable coefficients differ trough the several columns of Table 8 and never exceed an absolute value higher than 6%. The results of the subsample analysis in paragraph 4.3 are not supported by this regression analysis.

The effect of cash flows on the amounts of credit lines is inconclusive. In the next part we will focus on the relationship between the variables of model (1) for different levels of profitability to see if firms with low levels of cash flow have different coefficient values for model (1) than firms with high levels of cash flow.

The next analysis will focus on firms with credit lines among the cash flow distribution. The firms with credit lines are separated in two groups: a group with cash flows below the median cash flow level of the sample. These results are shown in column 1 to column 6. Columns 1 to 3 show the results of the regression for firms with below median level of cash flow for the year 2005 and column 4 to 6 show the results of the regression for firms with below median level of cash flow for the year 2008. Column 7 to 9 show the results of the regression for above median cash flow level firms for year 2005 and column 10 to 12 show the results of the regression for above median cash flow level firms for year 2008. Column 1 and column 4 show a negative relationship between the amount of lines of credit and cash flow for below median cash flow level firms in both years 2005 and 2008. This suggests that higher cash flows for less profitable firms result in a lower amount of lines of credit. In the proxy year 2005 this negative relationship is slightly stronger. While prices for credit lines went up for credit lines during the crisis, higher cash flows implied a weaker negative effect to the amount of credit lines. This result is supported by the subsample analysis. Firms acquired more credit lines in the crisis year. This acquisition of credit lines could be the reason for the weaker relationship between credit lines and cash flow in the year 2008 compared to 2005.

In column 7 to 12 the results for above median cash flow level firms are presented. The coefficient for the relationship of cash flow to the amount of credit lines is switched to a positive sign in column 9, (0.166), and is statistically significant tot the 5% level. In the year 2008 (column 12) the coefficient is close to zero but not statistically significant. In the year

(28)

2005 more profitable firms had more credit lines which supports the theory. For the year 2008 this theory is inconclusive for above median cash flow level firms.

Cash has a negative sign coefficient only in the year 2008. In 2005 this relationship is positive and statistically significant to the 5% level. In 2008 the sign is negative; showing a substitution effect between cash and line of credit but this coefficient is not statistically significant.

This analysis for firms divided in two groups along the cash flow distribution supports the evidence that amount for credit lines, cash and a cash flow depends among other factors on cash flow levels. However only for above median cash flow level firms the theory described in Section 2 is supported, and for those firms, in 2008 the coefficient of cash flow is not statistically significant. For less profitable firms the relationship of cash flows is negative on credit lines. The substitution effect of cash on credit lines described by the theory is not supported by this analysis.

4.5.3 Regression Analysis Credit lines and Drawdowns

This subparagraph will analyse the relation between drawdowns, cash and cash flow. Model (3) described in Section 3 of this paper is tested. The univariate analysis showed remarkable outcomes. To test if the theory on drawdowns breaks down, and to see if Hypotheses 3 has to be rejected, we will perform this regression. The results are shown in Table 10. Columns 1 to 3 represent the year 2005 and columns 4 to 6 represent the year 2008.

For the years 2005 and 2008, column 3 and 6 respectively, the coefficients of both cash and cash flow are near zero. For year 2005 the coefficient is 0.06 and for the year 2008 it is -0.08 for cash which are both statistical significant. Because of the near zero coefficients and a positive sign for cash in year 2005, it is hard to state that the more cash a firm has the less it will draw down on its credit line, as theory describes. Cash flow coefficients are -0.06 and -0.09 for years 2005 and 2008 respectively. Both coefficients are statistical significant. Profitable firms and firms with a credit rating seem to draw down less on their credit lines. Those results are highly statistical significant. The negative coefficient of the profitability

(29)

dummy in this analysis with drawdowns has even a higher absolute value for the sample in the crisis period. This could indicate that more profitable firms chose to use liquidity out of revenues more than drawing on their, in the crisis, more expensive credit lines.

The subsample made clear that for both regular and irregular firms the average drawdowns in 2008 increased. The interpretation of this could be that in times of crisis when liquidity is scarce, firms use their credit lines more often. Because of the small difference of the estimated coefficient for cash for years 2005 and 2008 it is inconclusive if we can reject Hypotheses 3.

In the upcoming paragraph we will analyse the number of employees on sources of liquidity e.g. cash holding to non-cash total assets and line of credits to non-cash total assets. We test the theory that states that firms with hoard liquidity tend to cut on operational costs like employment. The first firms cut on their employees by firing them. We will test this theory by doing OLS regression analysis of employee numbers to cash, total assets and we will correct for size and type of firm. To control for size we include as variable the amount of sales to non-cash total assets in the regression model. For type of firm we use four dummies. One for firms producing consumer goods, one for manufacturing firms, a third for IT firms and firms in the computer industry and the last one for pharmaceutical, and biochemical firms.

4.6 Descriptive Statistics Employment

Lines of credit are defined as the total loan commitment that a company can use, divided by non-cash total assets. In 2005, 60.76% of the sample of firms had a credit line and an average loan commitment of 28.62 % of total assets. For 2008 72.17% of the firms in the sample have a credit line at place. The mean size of credit lines is 18.67% of total assets.

When we observe the cash levels, the amount of cash equivalent items divided by non-cash total assets seem to be higher (21.84%) for companies without a credit line (70.65%) in comparison to companies with a credit line in 2005. In the crisis year 2008 these percentages do not have changed much. The sample shows 18.67% average cash holdings to non-cash total assets for firms with credit lines and almost 87.58% average cash holdings to non-cash

(30)

total assets for firms without credit line. This is consistent with the theory that firms without credit lines have to hoard cash to sustain their liquidity needs.

For the purpose to separate between different types of borrowers and their use of credit lines, the sample of companies is divided with respect to company characteristics: consumer goods, manufacturing, IT and pharma. Firms in the consumer goods business correspond to 13.95% and 14.50% of the total sample for 2005 and 2008. Companies pooled in the manufacturing business are represented by 26.73 % and 23.60% for the years 2005 and 2008 respectively. Companies are considered as ‘ITcomputer’ when they have a SIC number above 3500 and 3600. Those consist of 7.67% and 11.85% of the firms in 2005 and 2008 respectively. Finally, a firm is said to be in the pharmaceutical and biochemical industry when their SIC number is between 2770 and 3500. Those represent 31.84% and 33.97% of the companies in the sample of years 2005 and 2008 respectively

4.7 Univariate Analysis Employment

In this paragraph the correlations between number of employees, lines of credit, and cash holdings and cash flow of the sample out 2005 and 2008 are measured. In Table 12 the correlations of the full sample are shown between amount of employees, cash holdings and lines of credit for the years 2005 and 2008.

Table 12 shows near zero estimated coefficients for cash holdings of 0.0464) for 2005 and (-0.0072) for 2008. Also a negative correlation between lines of credit and employees is

showed in Table 12. The relationship is significant at the 1% level for normal times with a coefficient of -0.13 for compared to a coefficient that is near zero in 2008 and is not

significant. Both sources of liquidity show a negative effect to employees. This does support the theory that firms that hoard liquidity cut on employees by lowering their amount. This is revealing because this means a less liquid firm has more employees. Maybe the crisis was detected too late by firms and they were hiring more and more employees because they thought there was still a growing trend. Because this results do not give us enough insight in what is going on, this paper will do a regression analysis in the next paragraphs.

(31)

4.8 Regression Analysis on Employment

In this paragraph we will analyse the regression on employees, cash, lines on credit lines and sales. We will do this by using the regression equation described in Section 3 of this paper by model (4). First we will analyse the full sample. Thereafter we will analyse the firms with a credit line. Finally, the number of employees in the year after 2008 to cash, lines of credit and sales will be analysed.

4.8.1. Full Sample Regression Analysis on Employment

Table 13 presents the outcomes of the full sample analysis. Employees, Credit lines, cash holdings and sales are regressed based on model (4).

The relationship between credit lines and employees on the OLS analysis is negative which column 3 shows. The coefficient is -4.686 and is significant at the 1% level. Cash holdings also show a negative relationship with a coefficient of -0.379 and is significant to the 1% level. Remarkable, sales to total assets show a high negative relationship but also this

coefficient is not significant. The signs of the results in column 6 are not different from those in column 3. Cash holdings show a negative relationship to amount of employees which is significant to the 1% level and has a coefficient of -1.318. While total lines of credit and sales to total assets show again significant negative values. The results suggest that in the normal year and in the crisis the number of employees has a negative effect on liquidity and sale levels.

When sales to assets ratio is omitted in model (4) the magnitude of the relationships do not change much. Cash still shows a negative sign in column 2 and 5, for the years 2005 and 2008 respectively, and line of credits to employees are still negative related and highly significant. The dummy variables in column 3 are all statistical significant and show a firm positive relation to employees. This can be interpreted as follows: for the type of firms we correct, those firms tend to have higher average level of employees.

The results support the theory that liquidity is negatively related to employment numbers. This issue has to be analysed further which will be done in the next subparagraph

(32)

4.8.2 Regression Analysis on Employment for firms with Credit Lines

To test the results for firms with credit lines we leave out firms without lines of credit and regress employment on cash, lines of credit and sales. This is done by using model (4) described in Section 3 of this paper. Table 14 shows the results of the regression analysis. Column 1, 2 and 3 represent the outcomes for firms with credit lines in 2005. Column 4, 5 and 6 represent the outcomes for firms with credit lines in 2008.

In column 3 we see the results of the sample for firms with credit lines of 2005. Cash has a negative coefficient of -1.551 and is statistically significant to 1%. Lines of credits influence on employees are near zero but also this coefficient is not significant. Sales to assets has a positive coefficient of 403.65 but not significant. When we omit the sales to assets variable out of model (4) results for cash and lines of credit do not show a substantial change in

magnitude. By omitting lines of credit (column 1), the coefficient of cash is -2.262 and highly significant. Cash shows a consistent negative relationship to employee numbers.

The regression of firms out of this sample of the year 2008 shows almost the same results as the regression for the year 2005. Cash has a negative coefficient of -5.417 and lines of credit has a coefficient of -3.344. Both values are significant at the 1% level. Sales to assets shows a strong negative relation to employment levels and is highly significant. When sales to assets ratio is omitted of model (4) the coefficient of both cash and lines of credit stay

negative (-6.184 and -3.476 respectively) and are both significant to 1%.

The industry dummies are all positive related to employment. Most of those dummies are significant to the 1% level. This means that the selected industries have higher

employment levels on average than the proxy firms, firms in other industries, out of this sample.

The results of this analysis show a negative relationship between liquidity and employment levels. To test the theory if cuts on planned employment indeed turn out in lower future employment, we will test the employment level after the crisis year 2008. Labor contracts or other agreements could be the reason for this one year delay. To test if planned cuts on employment turn out in actual lower employment levels, we will test the relation of employment in 2009 to 2008 liquidity levels.

(33)

4.8.3 Regression Analysis Employment t+1 levels on Liquidity

The last subparagraph showed a negative relation of employment to liquidity. To test if

planned cuts turn out in actual cuts, we will test the employment level a year later to the tested crisis year 2008, in this paper. For this regression analysis we use model (5) described in Section 3 of this paper. Table 15 shows the results of this regression analysis.

In column 3 we see the results of the regression in which all the variables of model (5) are used. Cash has a negative coefficient of -1.205 and is statistically significant to the 1% level. The relation of lines of credits on employees a year later is negative with a value of -1.514. This coefficient is significant to 1%. Sales to assets have a high negative coefficient of -780.896 and is significant to the 10% level.

When we omit the sales to assets variable of model (5) the results in column 2 of Table 15 show a value of -1.197, significant to the 1% level. The coefficient of lines of credit does not change much compared to column 3 with a value of -1.686 and is highly significant. Column 1 shows a coefficient of -1.172 for cash which is significant to the 1% level. We can conclude that cash holdings have a negative relationship to employment a year after 2008. De industry dummies that correct for industry specific factor all generally show the same results for column 1, 2 and 3 of Table 15. All the industry dummies coefficient values range from 1 to 7 and have a positive sign. The dummy that corrects for pharmaceutical and biochemical firms and the dummy that corrects for IT firms and firms active in the computer industry are not significant through columns 1, 2 and 3 of Table 15.

In Section 4 of the paper we analyzed data on liquidity levels. First we conducted a subsample analysis on types of firms. This showed that regular firms, this means, large, profitable firms with an investment grade equal or above BBB-, and easy access to credit generally had more credit lines than non-regular firms. The univariate model and the regression models had different outcomes. In the crisis the results of the univariate analysis support the theory that credit lines are a substitute for cash for the full sample of firms in 2005. However, the

regression analysis does not necessarily support this theory. The probit regression did neither for the crisis year. When we focus on firms with credit lines the regressions do not suggest a